Datasets:

CambridgeMolecularEngineering

/

StressEngineeringLiterature

like 0

Follow

Molecular Engineering 11

⩽ 16.7. When N is close to or smaller than 1, material behaves like single crystals. When N is larger than 10–15, the material properties do not fall into the size effect occurrence range as shown in . Therefore, it needs to emphasize that the defined material parameters in this paper are valid when the value of N falls into the size effect occurrence range. Based on Eqs. , the fracture stress (σc) can be expressed as:, the grain size is expressed as a function of fracture stress (σc) and strain (εc) in the following:The relationship among the grain size, fracture stress and the strain of macro-sized specimen with a large number of grains in thickness direction can be modeled by Eq. . When the material thickness is decreased to micro-scale and there are only a few grains in the thickness, the material model becomes invalid. Assuming that there is a function, f(N)

=de, correlating the equivalent grain size (de) of macro-scaled material to the specific ratio of specimen size to grain size (N) in the modeling of size effect on fracture behavior. In other words, the function actually describes how the experimental results deviate from the results predicted by the conventional model due to size effect. In the function, de represents the grain size of macro-sized material, which has the same fracture stress and strain as that of the micro-sized material with a specific value of N. The function can be deduced based on Eq. and the experimental results of fracture stress and strain with the different values of N. The material constants in Eq. , C1/Ls and C2, can be obtained from literature and the identified fracture stress and strain by experiment, the relationship between N and de is determined and shown in where g is a material constant. h describes the change rate of de with N. Based on Eqs. , the fracture stress as a function of fracture strain and N is formulated as:In the extreme case that there is only one grain in the thickness direction, the term, C2b(g + h/N) in Eq. related to the grain boundary strengthening effect, becomes insignificant. Based on Eqs. , the fracture strain as a function of N is obtained in the following:, the relationship among the fracture stress, fracture strain and the ratio of specimen size to grain size can be modeled. shows the comparison between the estimated results with the proposed models and the experimental results from Klein et al. The conventional understanding and the established knowledge of material deformation behavior in macro-scale are no longer valid or accurate in micro-forming due to size effect. It is thus necessary to study material size effect and have an in-depth understanding of the micro deformation behavior in the micro-forming process. In this study, the tensile test of pure copper foils with different thicknesses and grain sizes has been conducted. It is found that the flow stress, ductility and the number of micro-voids on fracture surface decrease with the decreasing ratio of specimen size to grain size. The decrease of flow stress is due to the decreasing of grain boundary hardening effect with the increasing grain size. In addition, the influence of deformation mechanism of grain boundary sliding and grain rotation at the surface grains becomes significant to the overall material deformation behavior with the decreasing specimen size.For the samples with a small ratio of specimen size to grain size, only a few micro-voids and typical knife edge rupture are observed. The phenomena could be addressed based on the facts that the grain boundary acts as a barrier to dislocation motion and stress concentrates at the grain boundary which lead to the formation of micro-void. The volume fraction of grain boundary decreases with the decreasing ratio of specimen size to grain size. It results in the decreasing number of micro-voids. The models to describe the interactive effect of specimen and grain sizes on fracture stress and strain have been proposed. The accuracy of the proposed models has been validated. The research advances the understanding of size effect on deformation and fracture behaviors in micro-forming process.Cyclic fatigue behavior of carbon fiber reinforced vinylester resin composites manufactured by RTM and VARIThe mechanical behavior of [0°]8 unidirectional, [+45°/0°/−45°/90°]S quasi-isotropic and [+45°/−45°]3S angle-ply laminates produced by pressure or vacuum assisted resin infusion from toughened vinylester resin systems with unidirectional carbon fiber fabric reinforcement was investigated. The study included the determination of ultimate in-plane tension, compression and shear properties as well as the characterization of the cyclic fatigue behavior under stepwise increasing and constant amplitude loading. The experimental results showed a better fatigue performance of an epoxy-terminated butadiene-nitrile rubber modified VE resin system. Furthermore, the cyclic strength of the composites produced by VARI was lower compared to RTM composites. From the S–N curves of the quasi-isotropic material for R

= +0.1, −1 and +10 the fatigue life diagram was determined.Infusion processes, like resin film infusion, resin transfer molding (RTM) and vacuum assisted resin infusion (VARI), are gaining increasing interest for the cost-efficient production of fiber reinforced polymer matrix composites (e.g., Airbus A380 rear bulkhead). Meanwhile, purpose-built thermoset resin systems with adequately low processing viscosities are available Cyclic fatigue failure in composites results from the accumulation of damage within the material, the fatigue behavior is influenced by its resistance to microcracking and the manufacturing quality. Although showing a better fatigue resistance composites from epoxy resins tend to be expensive while polyester and vinylester resin composites are more brittle and, hence, less fatigue resistant but attract due to their lower material costs. Following previous research on the toughness improvement of vinylester based resin systems As matrix materials vinylester (VE) and vinylester-urethane hybrid (VEUH) resin systems were used. The properties of the VE system were optimized by generating interpenetrating networks (IPN) after modification with epoxy (EP) resins, which were either aliphatic or cyclo-aliphatic in nature. The commercially available bisphenol-A type VE resin diluted in styrene (30 wt%) was a product of DSM composite resins (Daron® XP-45-A-2). Polyprox R3 aliphatic (Al-EP) and Polyprox R11 cyclo-aliphatic (Cal-EP) EP were supplied by UPPC, Germany. The fracture toughness increase of both 1/1 ratio VE:Al-EP and VE:Cal-EP IPN systems was rather promising An innovative 2-component vinylester-urethane hybrid resin was presented by DSM-BASF structural resins. The combination of the radical polymerization of styrene onto the double bonds of the vinylester resin with the polyaddition between the isocyanate groups of a polymeric diisocyanate (Daron® XP-40-B-1) and the hydroxyl groups of the VE results in a narrow network with a heat resistance increased from 140 to 220 °C approximately. The addition of a liquid, epoxy-terminated butadiene-nitrile rubber (ETBN) led to a considerable increase of the resin toughness at room temperature compared to the baseline VEUH system As reinforcements dry unidirectional (UD) carbon fiber warp rovings (HTA 5131, 400 tex) which were stabilized by E-glass rovings (EC 9 1383, 34 tex) in the weft direction were used (). The use of this material is a necessary compromise regarding the RTM and VARI manufacturing routes to the composite and the intended introduction of the experimental data into a critical element based fatigue life analysis with the single UD ply as the building element to multidirectional laminates To limit the amount of experimental work fiber dominated [0°]8 as well as matrix dominated [+45°/−45°]3S laminates were screened with regard to their static properties. Furthermore, the cyclic fatigue properties of these laminates were determined under stepwise increasing loading with R

= −1 alternating stress amplitudes as well as under R

= +0.1 constant amplitude tension–tension loading. Herein, R denotes the ratio of the minimum (Smin) and the maximum (Smax) nominal axial stress components in one load cycle.Based on these results the properties of all composites were assessed from which the resin system with the most interesting property combination (denoted as reference VE resin system) was selected and subsequently tested under R

= +0.1, R

= −1 and R

= +10 (compression–compression) cyclic fatigue in [0°]8 and [+45°/0°/−45°/90°]S laminates as well as under R

= +0.1 in [+45°/−45°]3S laminates. The quasi-isotropic laminate was selected as a laminate with structural relevance. As the 0° ply is supposed to constitute the critical element of such laminates in the context of the critical element concept, S–N curves and the remaining strength of the UD ply ([0°]8 laminate) needs to be determined. For all cyclic tests, specimen fracture was defined as the fatigue failure. As the quasi-isotropic specimens tested under R

= +0.1 and R

= −1 delaminated during cyclic testing the fracture of the load bearing 0° plies was used to define cyclic failure in this case.For mechanical testing unaged and unnotched specimens with a length and width of 250 mm and 25 mm were used () which were cut out of laminated 540 mm square plates using a water cooled saw with diamond coating. The plates were produced by resin injection via RTM (injection pressure 5 bar) or via VARI (5 mbar) and were cured according to the cure cycle shown in . Glass fiber/epoxy tabs were bonded to both end surfaces with a 15° chamfer adjacent to the gauge section of the specimen to minimize stress concentrations at the load introduction. For strength testing of the UD laminate 10 mm wide specimens were used to reduce the required force of the test machine.Mechanical testing was carried out on 40 or 100 kN Schenck servohydraulic cylinders under ambient laboratory conditions. The tests were controlled by the MTS MultiPurpose Testware software. Quasi-static tensile and compressive tests on the UD laminate were carried out at a crosshead speed of 1 mm min−1. The quasi-static shear properties were determined by carrying out 1 mm min−1 uniaxial tension and compression tests on the angle-ply specimens parallel to the x direction of the laminate. Stepwise increasing and constant amplitude cycling was carried out under load control with a sine wave load profile where the maximum frequency was limited to 5 Hz to minimize the generation of hysteretic heat in the specimen volume. For tests with excursions to the compression range anti-buckling guides as shown in For the mathematical description of the cyclic strength data the functionwas used where S and N denote the cyclic strength and cycle number to failure. The curve fitting parameters A and B were determined from the test data for 5%, 50% and 95% probability of survival including run-outs using the least squared error method. The fatigue life diagram was approximated based on the S–N curves for R

= +0.1, −1 and +10 using a parametric model for carbon fiber reinforced plastics as published by Harris et al. wherein Fa=σa/Xt denote the nominal stress amplitude σa normalized by the quasi-static ultimate tensile strength Xt, Fm=σm/Xt the normalized nominal stress mean σm and xc=Xc/Xt the normalized ultimate compressive strength Xc. As above, the free parameters f, U and V in were determined by applying the least squared error method to the S–N data.To model damage accumulation in variable amplitude fatigue loading the remaining strength model of Hahn and Kim In this equation R(n) denotes the remaining quasi-static strength after constant amplitude fatigue to n cycles at a maximum stress amplitude Smax without failure, R(n

= 0) is the quasi-static strength of the virgin material, and K is an exponential coefficient unequal to 1 determined by the least squared error approximation to the experimental data. results of the quasi-static tension, compression and shear tests on the RTM manufactured VE systems with carbon fiber reinforcement are given. From the quasi-static data there is no evidence that one of the VE resin systems shows a certain advantage on the mechanical behavior in comparison to the others (CF/VE:Cal-EP best in tension stiffness and strength, CF/VE:Al-EP best in compression strength associated with a large scatter, CF/VEUH:ETBN best in shear strength). The fact that the shear test data under uniaxial tension and compression (measured without and with anti-buckling guides, respectively) compare well is a good indication that the use of the anti-buckling guides does not affect the test results evidently.A clear distinction of the mechanical performance of the investigated VE matrix systems was found under cyclic step loading of the RTM manufactured [+45°/−45°]3S laminates with R

= +0.1 and, even more clearly, with R

= −1 as shown in . The figure shows a plot of the cyclic load steps applied to the specimens where each data point represents a failed specimen within the specific load step. While the CF/VE:EP IPN specimens failed early (30,000–40,000 cycles to failure) at rather low stress amplitudes (30–35 MPa) the stress amplitudes and cycle numbers to failure of the ETBN modified CF/VEUH systems were considerably higher (Sx, max

= 65–80 MPa and N

= 110,000–150,000). Furthermore, the cyclic performance of the carbon fiber reinforced VE resin systems was investigated using the fiber dominated [0°]8 laminate configuration by R

= −1 cyclic step loading tests (). Again, the CF/VE-EP IPN specimens failed early at nearly equivalent stress amplitudes whereas, also in this case, the CF/VEUH:ETBN specimens reached higher stress levels. In R

= +0.1 constant amplitude fatigue test results are plotted for the [0]8 unidirectional CF/VE laminates. From these results it can be seen that the CF/VEUH:ETBN achieves the best properties as well as, at lower cyclic stress levels, the VE:Cal-EP resin system outperformes the VE:Al-EP system. From these results the VEUH:ETBN system was selected as the reference VE matrix system to be used for all further investigations (see shows the experimental data for constant amplitude fatigue testing of the RTM and VARI manufactured quasi-isotropic CF/VEUH:ETBN and the corresponding S–N curves (50% probability of failure) according to wherein Sx, max versus N was plotted for R

= +0.1 and R

= −1 while Sx, min was plotted for R

= +10. While the slope of the S–N curves for constant R of the two manufacturing routes is more or less equivalent the cyclic strength of the VARI laminate is lower compared to the RTM specimens for all tested R ratios. The S–N curve parameters A and B (5%, 50% and 90% probability of survival) according to for the quasi-isotropic CF/VEUH:ETBN laminate are given in . Because of the delamination failures observed under R

= +0.1 and −1 loading the corresponding S–N results need to be interpreted as a first approximation, but certainly not as the cyclic strength values in a strict material science sense.The fatigue life diagram for 50% probability of survival of the quasi-isotropic CF/VEUH:ETBN laminate is shown in , the parameters f, U and V of the constant fatigue life curves according to shows the experimental data after R

= +0.1, −1 and +10 cycling of the unidirectional CF/VEUH:ETBN laminate from which the S–N curves for 50% probability of survival were determined according to . As found for the quasi-isotropic laminate the slopes of the S–N curve of the RTM and VARI manufactured unidirectional CF/VEUH:ETBN materials are comparable. Again, the cyclic strength of the RTM laminate is higher than that of the VARI material while this effect cannot be observed in their quasi-static strength data. shows the remaining strength data measured on the unidirectional CF/VEUH:ETBN laminate fatigue cycling of the specimens with R

= +0.1 and S1, max

= 1100 MPa without failure to the cycle numbers indicated in the diagram. The scatter of the data for a constant n is quite large together without an evident effect of the fatigue cycling on the remaining strength except from the very high cycle numbers. Nevertheless, curve fitting of the experimental data to The results of R

= +0.1 constant amplitude loading of [+45°/−45°]3S angle-ply CF/VEUH:ETBN laminates manufactured by RTM and VARI together with the 50% S–N curves are shown in . As can be seen the constant amplitude fatigue behavior of the angle-ply laminate is similar to the behavior of the quasi-isotropic and unidirectional laminates, namely the higher cyclic strength of RTM compared to VARI laminate as well as comparable slopes of the corresponding S–N curves.The quasi-static and constant amplitude cyclic fatigue behavior of carbon fiber reinforced vinylester resin systems which were toughened by either the generation of interpenetrating networks with aliphatic and cyclo-aliphatic epoxy resins or the addition of a liquid, epoxy-terminated butadiene-nitrile rubber was investigated. The composite laminates were produced by resin infusion via RTM and VARI. Quasi-static in-plane tension, compression and shear testing of [0°]8 and [+45°/−45°]3S laminates resulted in an unclear picture with regard to the mechanical performance of the investigated resin systems. On the other hand, R

= −1 cyclic step loading provided a definite indication of a considerably higher cyclic fatigue performance of the CF/VEUH:ETBN composite which, consequently, was selected for detailed mechanical testing.Constant amplitude R

= +0.1 tension–tension, purely reverse and R

= +10 compression–compression cyclic fatigue testing was carried out on [+45°/0°/−45°/90°]S quasi-isotropic and on [0°]8 unidirectional laminates. Furthermore, [+45°/−45°]3S angle-ply laminates were tested under R

= +0.1 loading. The fatigue life diagram of the quasi-isotropic CF/VEUH:ETBN laminate for 50% probability of survival was derived. Together with the remaining strength degradation of the unidirectional laminate as damage metric necessary input data for critical element based fatigue life analyses were determined.Compared to the RTM laminates the fatigue strength of the VARI laminates was considerably lower under all cyclic loading profiles which was attributed to two effects. Obviously, the higher pressure difference associated with RTM resulted in a better consolidation of the dry fiber reinforcement and a higher fiber volume content (about 59% in contrast to 54%). In addition, as shown in , voids parallel to the weft rovings up to a length of 150 μm could be observed in the VARI specimens which were not present in the RTM laminates and which affect the cyclic strength of the these materials.Diffusivity and micro-hardness of blended cement materials exposed to external sulfate attack► The expansion in control mortar was about 10 times higher than the control paste. ► Fly ash blended mixtures were less sensitive to the aggregate addition. ► PIXE was found to be a useful NDT technique for measuring the diffusivity. ► Diffusion of Na and S and leaching of Ca and Si occurred during sulfate attack. ► Hardness increased in paste and decreased in mortar in the exterior layer.Many degradation processes in cement based materials include the diffusion of one or more chemical species into concrete and consequent chemical reactions which alter the chemical and physical nature of the microstructure. External sulfate attack is mostly described by a coupled diffusion–reaction mechanism which leads to the decomposition of hardened cement constituents and cracking of the paste. This paper discusses the significance of diffusion properties and chemical changes in external sulfate attack in blended cement based composites. A method based on Particle Induced X-ray Emission (PIXE) was developed to measure the diffusion properties in a non-destructive test method. Quantitative Energy Dispersive Spectrometry (EDS) and micro-hardness technique were also used to study the chemical and mechanical changes from sulfate attack. Diffusion coefficients and rates of reaction were determined for paste and mortar mixtures, showing higher diffusion rates and lower hardness values in mortar compared to paste for control mixtures. Partial replacement of cement with fly ash improved the transport properties and reduced the level of damage in exposure to sulfate attack.The durability of cement based materials is directly influenced by the resistance to transport of chemical species throughout the porous multi-component system. External chemical attack involves the transport of aggressive media into concrete via the interaction among various competing mechanisms of ionic diffusion, gas diffusion, liquid sorption, gas permeability, and liquid permeability When concrete is subjected to external sulfate attack, ionic diffusion properties become major indicators of transport characteristics, serviceability, and design life. Sulfate attack proceeds by the diffusion of sulfate ions and inward movement of a reaction front, accompanied by decomposition of major cement paste constituents and formation of expansive products that lead to microcracking of the material Ionic diffusion is the predominant transport mechanism in most chemical attack cases for cement based materials with Fick’s law used as a conventional tool to characterize the process shows the general 3-D form of Fick’s second law for non-steady state (transient) diffusion in Cartesian coordinates, indicating that the rate of change of concentration as a function of time is related to concentration gradient. In the general form of this equation, t is time, (x,y,z) are space coordinates, C

=

C(x,y,z,t) is the concentration, and D

=

D(x,y,z,t,C) is the coefficient of diffusion.For most concrete elements with known geometrical shapes and exposure conditions, the simpler 1-D or 2-D problems of diffusion are commonly considered. Since the general mechanism of diffusion of sulfate ions into concrete is similar to the diffusion of chloride ions for 2-D geometry with additional reaction terms. The solution to these equations depends on the initial and boundary conditions which are discussed by Crank ∂C(x,y,t)∂t=D0∂2C(x,y,t)∂x2+∂2C(x,y,t)∂y2+R(C)The reaction can be of different orders, i.e. zero-order, 1st, and 2nd order as described in . The first two types can be solved using closed-form solutions based on error function or series. The solution to the 2nd order problem needs numerical formulation as expressed by Tixier and Mobasher for a simplified case of Ci

= 0. In this equation, D is the apparent coefficient of diffusion and k is the chemical reaction coefficient. Typical concentration profiles (assuming D

= 10−12

m2/s and k

= 10−8

s−1) are shown in with and without consideration of the reaction term.CC0=12exp-xkD·erfcx2D·t-k·t+expxkD·erfcx2D·t+k·tThe reported values for coefficients of diffusion for sulfate ions in concrete vary depending on the mixture deigns (aggregate size and volume fraction, pozzolan type, fineness and replacement level, cement fineness, w/cm, additives, etc.) and curing conditions (time, moisture, temperature, etc.). The effective diffusivity can be related to the porosity of cement paste as described by Garboczi and Bentz . In this equation, ‘D’ is effective ionic diffusivity, ‘D0’ is diffusivity of ion in unconfined water, ‘φ’ is total capillary porosity and H(x) is Heaviside function defined as: H(x

> 0) = 1 and H(⩽0) = 0. Note that D0 for sulfate ions can be considered in the order of 10−9

m2/s The apparent coefficients of diffusion for ions have been traditionally measured by wet chemical methods (e.g. NaCl for chlorides and Na2SO4 for sulfates) from a singly exposed side to simulate 1-D diffusion Having calculated the coefficient of diffusion and the reaction coefficient for a particular cementitious material, one can use diffusion–reaction based models to predict the level of damage from external sulfate attack. For instance, models proposed by Krajcinovic et al. The current work studies the chemical and mechanical changes due to external sulfate attack for some cementitious materials. Paste and mortar mixtures were prepared both for control (Portland cement) and blended cement (class F fly ash). Measurement of the ionic diffusion was performed for major elements and the concentration profiles in 1-D and concentration contours in 2-D were generated. Particle Induced X-ray Emission (PIXE) and Energy Dispersive Spectroscopy (EDS) were used for the measurement of diffusion and chemical changes. One-dimensional diffusion data were compared with the predicted values obtained from diffusion equations. Changes in the macroscopic expansion and micro-hardness were also determined in order to observe the mechanical damage from sulfate attack.Four categories of mix designs including paste and mortar with and without fly ash replacement were used for this study. Type I/II Portland cement and low calcium class F fly ash were used with water to cement (W/C) ratio of 0.5 and 28 days of wet-curing prior to testing. Sand to cement (S/C) ratio of 2 was used for mortar mixtures while fly ash to cement (FA/C) ratio of 0.3 was used for blended mixtures. Experiments were conducted to measure the linear macroscopic expansion, 1-D and 2-D diffusion, and micro-hardness properties of specimens exposed to sodium sulfate (Na2SO4) solution. Different specimen size and exposure conditions were used for different tests as presented in . It is noted that the current work is a part of a more comprehensive study in a form of dissertation Sulfate resistance of cementitious materials has been traditionally tested following ASTM C 1012 Particle Induced X-ray Emission (PIXE) is a quantitative and non-destructive technique that relies on the spectrometry of characteristic X-rays emitted when high-energy beams of proton ions (H+) with 0.3–10 MeV energy ionize atoms of a specimen to permit only a 1-D diffusion profile from the top surface. The specimens were then placed in 10% Na2SO4 solution at 75 °C for an accelerated diffusion process. Cylinders were removed from the solution after 3 months of exposure and were cut in two halves using a saw. To obtain the concentration profiles of elements using PIXE, for each half-cylinder, seven spots were scanned with an aperture size of 2 mm along the longitudinal axis. shows a typical PIXE test result for a variety of major elements (Na, Al, Si, S, K, Ca) obtained for the X spots. The concentration profiles of sodium and other elements can then be plotted vs. the depth of penetration for measuring the diffusion coefficients. For verifying the accuracy of PIXE scanning process, specifically in the mortar mixtures where aggregates might affect the results, three palettes were prepared and tested. For this case, samples were drilled at three depths of penetrations (8, 24, and 40 mm); the collected powders were used for making palettes. These palettes with diameter of 10 mm and thickness of 2 mm could be representatives of the overall compositions of the mixture and were made by pressing the ground powder using a hydraulic press.After 12 months of exposure, the 25 × 25 × 280 mm specimens used for the expansion test were also used for chemical analysis using quantitative Energy Dispersive Spectroscopy (EDS). Thin 25 × 25 × 10 mm samples were cut from original specimens exposed to sodium sulfate solution as shown in . These samples were polished using 300 and 600 grit polishing papers and the 25 × 25 mm cross sections were marked using a sharp tipped pencil to generate a grid mesh as schematically shown in . The exterior layer is hereby called EXT and the interior portion is called INT for comparison purposes. A novel SEM-EDS setup was used and the quantitative compositional analyses of exposed samples were obtained using window scanning option with an area of 1 mm2 for each grid point.The 25 × 25 × 10 mm thin specimens used for EDS were also used for micro-hardness testing to understand the mechanical changes on the exposed samples. Micro-indentation is a common method of evaluating the quality of materials for engineering purposes, in particular ductile materials (i.e. metals) but also brittle materials such as concrete in which P (Kgf) is the applied force, α is the indenter diagonals angle equal to 136° and D (mm) is the average of diagonals of the indentation In this study, a 0.2 Kgf load was applied on the samples for 15 s, followed by a measurement of indentation size using an optical microscope. For each grid point, three replicate indentations were made as shown in a and the average values were used for calculating the hardness values. b shows an SEM image of the indentation performed on a cement paste sample.The linear expansions for control and blended mixtures are shown in a and b. The overall expansion values obtained for mortars are one order of magnitude larger than those obtained for paste for the control mixtures, implying the role of aggregates in generating higher porosity in ITZ in the absence of fly ash. Where an increase of aggregates fraction increases the expansion for control mixtures, the fly ash blended mixtures seem to be less sensitive to this parameter. This can be explained by the beneficial effect of fly ash in improving the ITZ characteristics. The averaged expansion values vary between 0.1% and 1.8% for control but only 0.1% and 0.15% for blended cement systems.The elemental analyses obtained from PIXE were used and weight percent of the major elements (Ca, Si, Al, Na, and S) were plotted for different depths of penetration. a and b shows the concentration profiles for calcium, aluminum, and silicon for paste and mortar mixtures, after 3 months of exposure. Fly ash blended systems were relatively higher in the levels of Si and Al and lower in Ca contents due to the modification of the overall chemical compositions. The variations of these elements are not significant at various depths of penetration; however paste mixtures show a smoother trend compared to the mortar mixtures. This could be due to relatively short exposure time (3 months) in PIXE test setup. a and b on the other hand present the variations of sodium and sulfur contents in paste and mortar mixtures. The level of Na in control and blended systems are not very different, however the concentrations of S are changing dramatically in the two systems. a and b shows the variations of sodium and sulfur in mortar and paste mixtures obtained from the analysis of powder palettes. The results are in partial agreement with the concentrations obtained from regular (non-destructive) PIXE test setup. It should be noted that this agreement is more reliable for higher depths of penetration (i.e. 40 mm), however for shallower depths (i.e. 10 mm), there is a 4% difference in the concentration values obtained from the two methods.The concentrations of sulfur were plotted and error function was used for fitting and measuring D and k values. D which is the coefficient of diffusion plays a more significant role in the diffusion reactions compared to k which is the chemical reaction rate. Original experimental data were used for mortar; however, a normalization factor was used for modifying the data for paste mixtures. This is due to the alternation of chemical and porous characteristics in mortar compared to paste from the addition of aggregates. The coefficients of diffusion and rates of reaction are shown in for S/C ratio of 2. The results show that the coefficients of diffusion for mortar mixtures are as much as 3.5–4.5 times higher in mortar compared to paste.The major elements obtained from EDS (S, Na, Ca and Si) were plotted using interpolated contours on the cross section of the specimens. shows the typical results obtained from quantitative EDS for paste blended mixture after 12 months of exposure. The average value of sulfur (S) is 6.7% in the exterior layer but only 2.2% in the interior of the sample. These values are 19.6% and 22.3% for calcium (Ca), implying that this element has leached out of the specimen into the solution. The results are presented graphically for major elements. a and b shows the contour plots for sulfur and sodium, demonstrating the diffusion of these elements while a and b shows the leaching of calcium and silicon outside of the samples. The non-symmetrical distribution of these elements can be attributed to the cracking of the sample on the bottom surface and higher diffusion of S and Na, as well as higher leaching of Ca and Si through this surface. The non-symmetry which is explained by one-sided cracks can be observed in most of the exposed specimens and shown typically in The values for micro-hardness were calculated and for one typical case (control paste after 12 months of exposure), presented in with each cell representing the average and standard deviation of three indentations (6 diagonals). The average and standard deviations for the exterior (EXT) layer and the interior (INT) layer are presented in for all the mixtures along with the percentage of change in hardness value (ΔH). The values corresponding EXT and INT are the basis for comparison in the following statements. The hardness values for paste mixtures increased (EXT vs. INT) as sulfates diffused into the sample and ettringite crystals were formed before cracking took place. This increase was higher in blended systems (10.6% after 3 months and 29.3% after 12 months) compared to the control (2.6% and 9.1%). On the other hand and for mortar mixtures, the hardness has decreased (EXT vs. INT) after 12 months of diffusion in the cracked samples. This reduction is much higher in the control (11.8%) compared to blended (1.6%) after 12 months of exposure. The hardness values are interpolated and represented as surfaces in a and b. This change of response between paste and mortar mixtures is in agreement with the higher diffusion rates and expansion levels of mortar in comparison with paste mixtures.The results from expansion tests, diffusion tests and micro-hardness tests were found to be compatible, all indicating lower diffusion rate and less damage in paste mixtures compared to mortar mixtures for control specimens. This can be explained by the effect of interfacial transition zone (ITZ) which exists in mortar due to the presence of aggregates inclusions. Higher porosity and faster diffusion followed by cracking and stiffness reduction was observed in these mixtures. Partial replacement of Portland cement with class F fly ash improved the ITZ, as well as transport properties and reduced the level of damage from sulfate attack.The physical, chemical and mechanical alternations of cement based materials were studied in exposure to external sulfate attack. Paste and mortar mixtures were made with and without fly ash replacement in various specimen forms. The standard ASTM C 1012 test method was followed for measuring the macroscopic linear expansion of the materials which showed that the level of expansion in control mortar mixtures were one order of magnitude higher than the control paste mixtures, possibly due to the ITZ effect in increasing the porosity of the system. Fly ash blended mixtures with improved microstructure were less sensitive to the aggregate addition.Particle Induced X-ray Emission (PIXE) setup was used for the measurement of concentration profiles of the major elements in a 1-D diffusion problem. Being a non-destructive and fast method, PIXE was found to be a very useful technique. The values of diffusion coefficients were 3.5–4.5 times more in control mortar mixtures compared to the paste mixtures. The 2-D contours of concentration of the major elements obtained from quantitative EDS showed the diffusion of sodium and sulfur and leaching of calcium and silicon during the 12 months of sulfate attack. The micro-hardness values of the exposed samples were also determined which showed an increase of hardness (2.6–29.3%) in paste mixtures and a decrease of hardness values (1.6–11.8%) in mortar mixtures in the exterior layers. These studies imply an improvement of the microstructure, i.e. less porous ITZ for fly ash blended mixtures.Numerical and experimental investigation on seismically damaged reinforced concrete wall panels retrofitted with FRP compositesThis paper presents the results of an experimental program on precast reinforced concrete wall panels (PRCWP). These panels were damaged under cyclic lateral loads and thereafter retrofitted or rehabilitated and retested. The experimental program was conceived to analyse the possibilities of using FRP materials for strengthening the PRCWP affected by seismic action. The fibre reinforced polymer (FRP) composites are frequently used in strengthening structural elements because of their superior characteristics and simple technology. The existing literature lacks information concerning reinforced concrete walls (RC) retrofitted by FRP composites compared to other structural members. This paper presents various effective strengthening solutions that can be applied to damaged elements. The retrofitting solutions consist in use of EBR-CFRP strips, combined EBR-CFRP strips with NSM-CFRP plates, textile reinforced mortar (TRM) using glass fibre grid, and TRM using carbon fibre grid. The solutions were proposed with the aim of restoring the wall shear resistance and to provide the confinement effect at the ends. The experimental results indicate that the performance of the elements, repaired and strengthened, were almost equal to or higher than the reference elements in terms of load bearing capacity, stiffness and energy dissipation capacity. A more ductile behaviour compared to the reference elements was recorded for the rehabilitated and retrofitted elements.narrow door opening enlarged to a wide door openingnarrow window opening enlarged to a wide window openingmodulus of elasticity of steel reinforcementsecant stiffness corresponding to the δi displacement amplitude (Ri drift ratio) on the monotonic load-displacement envelopeConsidering the history of the precast reinforced concrete large panel buildings and their behaviour during and after major earthquakes, this type of structural system was well-designed and executed for seismic actions. Fintel, author of “shear walls – an answer for seismic resistance” Although strengthening with externally bonded FRP strips is an easily applicable and practical technique, little information is available in the literature on reinforced concrete walls. Recent research on reinforced concrete walls strengthened with EBR-FRPs were conducted by Mohammed et al. Accounting for the building stock in Romania and Europe, built using precast panels, a theoretical and an experimental program was conducted at the Politehnica University Timisoara to investigate the seismic performance of the precast reinforced concrete wall panels (PRCWP) and strengthening solutions using FRP materials. Starting with the details used during construction, the program investigated initially the behaviour of PRCWP and thereafter a few possible solutions of strengthening the seismically damaged elements. The strategies applied in the current experimental program included the use of externally bonded carbon fibre reinforced polymer reinforcement (EBR-CFRP), near surface mounted carbon fibre reinforced polymer reinforcement (NSM-CFRP), textile reinforced mortar (TRM) with a carbon fibre (CF) grid and textile reinforced mortar using a glass fibre (GF) grid.Six specimens with openings, called precast reinforced concrete wall panels, PRCWP (7–12), were proposed and laboratory tested. This part of the experimental research program continued the previous phase, during which five specimens, known as PRCWP (7–8 and 10–12), were investigated and presented in Todut et al. The paper aims to comprehend the seismic performances of the strengthened specimens compared to the reference specimens. Additionally, the appropriate solutions for the good use of FRPs, including the fibres type, quantity, position, orientation, and anchorage, were analysed and presented. A numerical analysis was performed for the retrofitted elements, and the results were compared with the experimental values. The general behaviour of the finite element models represented by the load–displacement curves were in agreement with the tested walls.The experimental program presented here uses six 1:1.2 scaled elements of PRCWP (7–12) that were reinforced and casted according to Romanian Project Type 770-81 A brief mention of the specimens denotes PRCWP (7-E1-T/R) and (12-E1-T/R) has a narrow door opening (E1), PRCWP (8-E3-T/R) has a large door opening (E3), PRCWP (9-E1/E3-R/T) has an initial narrow door opening that was enlarged to a wide door opening (E1/E3), PRCWP (10-L1/L3-T/R) has an initial narrow window opening that was enlarged to a wide window opening (L1/L3) and PRCWP (11-L1-T/R) has a small window opening (L1). The dimensions of the experimental wall specimens were the following: 2150 mm height, 2750 mm width and 100 mm thickness. The narrow door opening dimensions were 1800 mm in height and 750 mm in length; the wide door opening dimensions were 1800 mm in height and 1750 mm in length; the small window opening dimensions were 1000 mm in height and 750 mm in length. Whereas after enlargement, the dimensions of the wide window opening were 1000 mm in height and 1750 mm in length.The materials used in the specimens were concrete, steel rebar and FRPs. Material tests were performed only on the concrete and steel. The resulting concrete class for the unreinforced specimens was C30/37 for PRCWP (7), C12/15 for PRCWP (8), C16/20 for PRCWP (10, 11) and C25/30 for PRCWP (9, 12). The measured properties of the steel reinforcement are given in . The geometrical and mechanical properties of the Carbon Fibre (CF) fabric, CF plates and epoxy resin matrices used as EBR-CFRP and NSM-CFRP for retrofitting the elements are summarized in presents the properties of the glass fibre and carbon fibre grid used in the TRM. The presented characteristics are based on the manufacturer’s data. The repair mortar used to replace the heavily damaged concrete was a Sika MonoTop 614 for PRCWP (7, 8, 10) and Mapegrout Easy Flow GF for PRCWP (11–12). The mortars had a compressive strength of 60 N/mm2 at 28 days according to the product data sheet.The experimental specimens were positioned between a loading steel concrete composite beam and a foundation steel concrete composite beam. The test set-up scheme is presented in . The retrofitted specimens were tested under quasi-static reversed cyclic lateral loads – displacement controlled, similar with the initial tests performed on the reference elements. Two cycles per drift were performed in a constant increment of 0.1% drift ratio, namely 2.15 mm. In addition to lateral loads, vertical loads were applied to simulate gravity loading and to restrain the rotation of the specimens. The behaviour of the elements was monitored using pressure transducers (P), displacement transducers (D) and strain gauges placed on the rebar and FRP (G). Strain gauges were applied on the steel reinforcement for the unreinforced specimens and on the FRP material for the retrofitted specimens. shows the strain gauge position on the steel reinforcement and FRP material.The experimental specimens revealed shear behaviour in accordance with the design process. A significant number of cracks appeared in all regions of the panel during the experimental test. Thick inclined cracks in the piers were observed. Additionally, local concrete crushing and reinforcement yielding occurred. Failure details of the tested unreinforced specimens are given in . Among these details, concrete crushing at the extremities of the diagonal compressed strut of the panel and cast in place mortar at the bottom right corner of the door opening for PRCWP (7-E1-T) was noted; the dowel effect was observed at the vertical rebars of the welded wire mesh, and a tear occurred through the horizontal rebars of the welded wire mesh. Specimen PRCWP (8-E3-T) developed sliding shear in the left pier, whereas in the right pier, a vertical crack between the panel and the wing element appeared. The top corners of the door opening and the cast-in-place mortar exhibited concrete crushing. Specimen PRCWP (10-L1/L3-T) developed concrete crushing in the left corners of the window opening. For element PRCWP (11-L1-T), concrete crushing occurred in the bottom left corner of the window opening, parapet and in the right wing. Specimen PRCWP (12-E1-T) exhibited concrete crushing in the top right corner of the opening and the cast-in-place mortar at the bottom corners of the door opening. Similar behaviours for the specimens was obtained by Demeter The damaged specimens were repaired before retrofitting, by removing the crushed concrete and replacing the concrete with non-shrink, high-strength repair mortar. The wall panel surface was then polished with a special grinder to achieve a fully smooth surface, and the concrete edges of the element were rounded at a radius of approximately 20 mm to achieve the effectiveness of the confining solution. Local holes were drilled in the wall panel to provide the anchors for the strengthening system, and the surface of the wall was vacuum-cleaned. The cracks were cleaned and filled superficially with epoxy resin, except for PRCWP (11-L1-T/R) in which the cracks were injected with a fluid epoxy resin using mechanical injection packers and a hand pump.The strengthening strategies presented herein were based on the behaviour and failure mode of the reference elements.To restore the initial load bearing capacity of the specimen, horizontal unidirectional EBR-CFRP strips (SikaWrap 230C) of 100 mm width were applied symmetrically on both faces of the wall. To restore the flexural capacity of the element, vertical carbon fibre strips were placed along the door opening and horizontally along the upper edge of the spandrel. Additional short strips were placed at an incline at the top corners of the door opening, horizontally and vertically in the spandrel region. Short CFRP tows were used to anchor the bottom vertical strips to the foundation beam. Horizontal carbon fibre strips were applied on both piers, anchored at their ends by overlapping strips on the door opening side and by short CFRP tows at the wing-side end to restore the shear capacity of the wall. Confinement was provided by CFRP strips at the corners of the opening, in the right pier and at the ends of the wing walls (a). According to this design philosophy, a more ductile behaviour for retrofitted elements is expected instead of the brittle failure of the reference elements.Following the identical retrofitting strategy that was presented for the previous element, unidirectional carbon fibre strips (SikaWrap 230C) were applied around the wide door opening and along the upper edge of the spandrel. The bottom vertical strips were also anchored to the foundation beam by short CFRP tows. Horizontal carbon fibre strips were applied on both piers, wrapping from one face to the other. Additional horizontal strips were wrapped at the wing-side end through the confined wing to the other face. These end strips were also bonded by CFRP tows. Confinement was provided at the inside corner of the piers and at the ends of the wing walls. The details of the applied solution are presented in The experimental specimen simulating a door opening enlargement was not tested initially unreinforced (being first strengthened and thereafter tested) because this element was assumed to be weakened by the cut-out intervention. According to the reinforcement details of the specimen c), unidirectional carbon fibre strips were applied vertically along the left side of the opening and horizontally along the top and bottom edge of the spandrel. The bottom vertical strips were anchored to the foundation beam by short CFRP tows. Horizontal carbon fibre strips were applied on the left pier, wrapped from one face to the other, and anchored at the wing-side end by CFRP tows. Confinement was provided at the corners of the opening and at the ends of the wing walls. The near surface mounted carbon fibre plates (12 mm × 1.2 mm) were applied horizontally along the entire height of the right pier at 180 mm centres, whereas the last two from the top were extended until the middle of the spandrel. All carbon fibre plates were anchored in the wing element using epoxy resin. Additional horizontal confinement strips were provided at the middle of the spandrel to assure the left anchorage of the top carbon fibre plates. Additionally, vertical strips were wrapped from one face of the spandrel to the other, accounting for the lack of the spatial reinforcement cage in the spandrel region corresponding to the cut-out.After repairing the damaged specimen, a mechanical anchorage system composed of threaded rods (6 cm length) was fixed in the wall using resin to connect the precast panel with the TRM system. The anchorage system used was a mechanical and punctual type, composed of threaded rods, nuts and washers. The TRM strengthening system used a glass fibre grid. To provide a bonding bridge for the concrete, the patching mortar Sika Monotop 910 N was applied on the surface of the wall. The first layer of the TRM component mortar (Sika Monotop 722 Mur) was then applied, followed by the glass fibre grid (SikaWrap 350G) fixed with a nut and washer and the second layer of mortar. Detailed data related to the glass fibre grid application are presented in Todut et al. a). In the application process, the grid pieces marked 4 were first mounted on each side of the parapet, followed the number 5 pieces, which were wrapped around the parapet. Grid pieces number 2 were applied to each side of the piers, followed by the number 3 pieces wrapping each side of the opening. Finally, the grid piece marked 1 was wrapped around the spandrel.After previous experience with the punctual anchorage, a surface type of anchorage was selected by using a high-strength steel fibre cord (MapeWrap S Fiocco). Because the TRM component mortar (Planitop HDM) was a two-component material, a high-strength, cement-based mortar (with fine-grained aggregates, special admixtures and synthetic polymers) blended with a liquid (latex) with high adherence and no bonding bridge was necessary. A steel fibre cord of 25 cm and 35 cm in length was impregnated with resin a day prior to application, and the cord was then fixed through the panel using epoxy resin. The first layer of the TRM component mortar was then applied, followed by the fresh/wet impregnated carbon fibre grid (Mapegrid C170) and the second layer of mortar. Detailed data related to the application is presented in Todut et al. b). First, the grid pieces were applied to each side of the parapet (marked 4 on figure); then the number 9 pieces were wrapped around the parapet. Number 2 and 2” pieces were mounted to each side of the pier, followed by the number 3 grid that was wrapped around the right side of the opening. The number 1 grid piece was wrapped around the spandrel, whereas the number 5, 8, 6, and 7 pieces were applied at an incline in the corners of the opening. Finally, the steel anchorage ends were unfolded, and a washer with a concrete nail was fixed in the resin to realize the surface anchorage.d), near surface mounted plates (Carboplate E170/100/1.4) of 10 mm × 1.4 mm were applied horizontally in the piers and in the spandrel in each side of the wall at 200 mm centres. All plates were anchored in the wing element. CFRP confinement strips (MapeWrap C UNI-Ax) were provided at the corners of the opening and vertical, wrapping from one face to the other of the spandrel, and at the ends of the wing walls.Generally, the behaviour of specimens was in accordance with the testing procedure and the strengthening strategy. Among the failure details observed after the experimental test of the strengthened specimens, the following were noted:PRCWP (7-E1-T/R): cracks appeared in the spandrel and in the right pier; fracturing of the right bottom confinement strip of the wing element together with concrete and cast-in-place mortar crushing was observed; horizontal shear strips debonded; the confinement strip from the top right corner of the door opening and bottom corner of the right pier debonded; and sliding shear occurred at the bottom of the right pier.PRCWP (8-E3-T/R): cracking was observed in the spandrel and in the piers; a vertical crack between the left pier and the wing appeared; horizontal shear strips debonded; shear and flexure strips debonded and fractured at the top right corner of the opening; and the repair mortar crushed.PRCWP (9-E1/E3-R/T): cracks occurred in the spandrel, piers and top corners of the opening; local debonding was observed at the top horizontal flexure strip, top right corner of the opening confinement strip, top left corner of the opening flexure and confinement strips; the top left corner of the opening confinement strip fractured with mortar crushing; bottom right wing concrete crushing and bottom right corner of the opening mortar crushing was observed.PRCWP (10-L1-/L3-T/R): all regions of the panel developed cracks; for drift ratios exceeding 0.3%, mortar exfoliation and TRM system debonding was observed; when the punctual anchorage tried to maintain the system tied to the wall, excessive swelling appeared between the threaded rods; after the removal of the TRM system, thick inclined cracks and concrete crushing was observed on each side of the window opening.PRCWP (11-L1-T/R): several inclined cracks appeared across the piers, spandrel and corners of the opening. The specimen was not tested until failure because of the constraints imposed by the testing facilities. The behaviour prior to failure indicates similarities with a solid wall specimen PRCWP (12-E1-T/R): several inclined cracks appeared across the piers, but the specimen failed to reach failure because of the constraints of the testing equipment from laboratory. Similar to the previous specimen, the load bearing capacity of the rehabilitated element was assumed to be highly superior compared to the unreinforced specimen.The failure condition of the strengthened specimens is depicted in The behaviour of the tested strengthened wall panels is shown in as load-drift ratio hysteresis loops in comparison with the reference specimen. The basic results of the tested specimens are presented in . The load bearing capacity results are presented in , showing a variation between 46.5–111% in the positive loading cycles and 93–154 in the negative loading cycles.The cumulative dissipated energy (CED) per half-cycle versus the drift ratio within each experimental test performed is presented in , whereas the calculated average (among two cycles) cumulative energy dissipated is presented in . Several strengthened specimens, namely PRCWP (7, 10–11), developed higher energy dissipation compared to the reference specimens; whereas other specimens, PRCWP (8, 12), the energy dissipation was similar for the strengthened and reference specimens. The PRCWP (9-E1/E3-R/T) specimen dissipated more energy compared to PRCWP (8-E3-T/R), and less energy compared to PRCWP (7-E1-T/R) and (12-E1-T/R).The ductility of the wall specimens was evaluated using the μ0.85 method (presented in presents the ductility coefficient for the strengthened specimens compared to the reference ones. According to the results, the obtained ductility ranges from 1.31 to 1.87. Specimen PRCWP (9-E1/E3-R/T) developed a higher ductility compared to the strengthened PRCWP (7, 8). Two post-damage strengthened specimens did not reach failure during the tests. However, according to the shape of the load-drift ratio hysteresis loops, the behaviour of the specimens appeared rather more rigid than ductile.During the experimental tests, strain was measured on the vertical, horizontal and inclined steel rebars for the unreinforced specimens and on the vertical, horizontal and inclined FRPs for the rehabilitated specimens. shows the steel strain ε (‰) versus the drift ratio for the unreinforced specimens and for the rehabilitated specimens. Although the measurements did not indicate high strain values, the FRPs were active during the experimental tests. In most cases, the maximum strain occurred close to the corners of the opening, notably for PRCWP (7–9, 11–12). For PRCWP (10) strain gauges were applied only in the piers and on each side of the opening at which failure was predominant. In addition to these strain values, the strengthening was efficient according with the general behaviour and observations, even with the appearance of premature fractures or FRP debonding in several cases.According to the stiffness versus drift-ratio diagram (), the strengthened specimens PRCWP (8, 10, 11, 12) exhibited a similar initial stiffness compared to the reference specimens. For PRCWP (7-E1-T/R), the initial stiffness increased by 25% compared to the reference specimen. PRCWP (9-E1/E3-R/T) developed a higher initial stiffness (2.88 times higher) than PRCWP (8-E3-T/R), a lower initial stiffness (2.28 times smaller) compared to PRCWP (7-E1-T/R) and 1.80 times smaller compared to PRCWP (12-E1-T/R).A numerical analysis was performed in ATENA 2D software for the strengthened precast reinforced concrete wall panels tested to predict the maximum lateral load. The bilinear stress–strain law for elastic-perfectly plastic reinforcement was used in the model, whereas SBETA material was used for concrete (). EBR-CFRP strips and NSM-CFRP plates were introduced as independent reinforcement bars, and the TRM system was modelled as a smeared reinforcement. A perfect connection was assumed for all strengthening systems used. An interface (cast-in-place mortar) for bonding the concrete panel and the reaction beams was assumed in the numerical analysis, for which the parameters were calculated according to Atena Program Documentation – Part 1 (Theory) , the numerical analysis was reliable. Therefore, this analysis can be used for standard meshes following the recommendations available in literature and predict the behaviour of the elements.This paper describes the rehabilitation strategies and the experimental test results obtained for six precast reinforced concrete wall panels in comparison with reference walls. All specimens were tested under simulated seismic conditions, first in an unreinforced condition, then in a repaired and rehabilitated state (except PRCWP 9, which was directly strengthened and tested). Several rehabilitation strategies were adopted, and select conclusions were drawn within the limitation of the current research. Reasonably accurate numerical simulations were developed. The major conclusions derived from this study are as follows:Generally, the reference precast wall panels revealed extensive cracking, reinforcement yielding and concrete crushing at the extremities of the diagonal compressed struts and in the corners of the pier near the opening.In the case of the strengthened specimens, horizontal and vertical FRP strips debonding, confinement strips fracturing, TRM debonding when a punctual type of anchorage was used, diagonal cracks, cast in place mortar crushing and concrete crushing in the corners of the pier near the opening was observed.The load bearing capacity of the specimens was restored through the rehabilitation strategies and increased in several cases. The measured results showed a maximum load variation between 46.5% and 111% in the positive loading cycles and 93% and 154% in the negative loading cycles. Additionally, the displacement at failure was higher for strengthened specimens. As the tested specimens were 1:1.2 real scaled elements, the results nearly reflect the real behaviour.A larger number of specimens would be necessary to test the variation of different parameters such as the cut-out intervention, opening dimensions and rehabilitation strategy.The post-damage strengthened PRCWP (7, 10, 11) specimens dissipated more energy compared to the reference samples, whereas the PRCWP (8, 12) specimens developed a similar energy dissipation as the reference specimens.According to the μ0.85 method, the ductility of PRCWP (7, 10) specimens was higher than the reference specimens. Only for the post-damage strengthened PRCWP 8 was the ductility lower than the reference unreinforced specimen; this lower ductility was because in the prior-to-damage testing condition and strengthening system used, the PRCWP 9 developed a higher ductility compared to the strengthened PRCWP (7, 8).Although the measurements did not indicate high strain values, the FRPs were active during the experimental tests. In most of the cases, the maximum strain occurred close to the corners of the opening.Most of the post-damage strengthened specimens exhibited a similar initial stiffness compared to the reference specimens, notably PRCWP (8, 10–12). Whereas for the strengthened PRCWP 7, the initial stiffness increased (25%) compared to the reference specimen.The numerical analysis performed and presented in this paper shows that no sophisticated models are necessary to predict the results in terms of peak lateral load and displacements. Generally, a good agreement was obtained between the numerical analysis and experimental tests of precast reinforced concrete wall panels strengthened with different systems. Similar values were obtained in terms of load and displacement for the EBR strengthening system. For the specimen PRCWP (10), the numerical model produced higher values compared to the experimental specimen, in which debonding of the system was recorded. A better type of anchorage would have assisted in realizing a better connection between the panel and the strengthening system and, consequently, in obtaining higher load bearing capacities and displacement values.Further studies and tests related to the strengthening of elements using various opening types and FRP strategies are in progress. The studies aim to establish the behaviour of the wall panels with different cut-outs and investigate the most convenient solutions of strengthening to ensure efficiency during seismic actions.Polymers are increasingly utilized in various industry sectors, because they feature high specific strength (strength-to-density ratio), corrosion resistance and cost-effectiveness. Many neat polymers have relatively low mechanical properties for structural applications, so they are often reinforced with fillers Graphite consists of infinite number of graphene sheets stacked to each other via van der Waal bonds. Graphite intercalation compounds are produced by introducing guest molecules or ions between graphene sheets. A GIC was first accidently prepared by Schafhäutl in 1840 using sulfuric acid and oxidizing agent . Upon heating, the intercalates are burnt out leaving a worm like structure with high volume where the stages are weakly connected.Elastomer (ethylene-propylene-diene monomer, EPDM 4045) was provided by Jilin Petrochemical Limited China, which contains 53–59% ethylene with Mooney viscosity ML (1 + 4) at 100 °C equivalent to 38–52. Curing chemicals listed in were purchased from market. A commercial graphite intercalation compound (GIC, Asbury 3494) with an average size of over 75 µm and a carbon content of over 80% was kindly supplied by Asbury Carbons, Asbury, NJ. Curing chemicals and the GIC were used as received without further purification.The graphite intercalation compound (GIC) was melt-compounded with the elastomer ethylene-propylene-diene monomer (EPDM) using a laboratory-sized two-roll mill at room temperature. The GIC was gradually added to EPDM during the compounding, and the gap between the two rolls was fixed to produce a ∼0.5 mm-thick sheet; such a process would avoid materials loss and GIC delamination. Mixing time ranges 10–13 min depending on the GIC fraction. Curing chemicals presented in were then added using the same machine. Finally, a vulcanization process was carried out at 150 °C for 30 min under 3 MPa. A preheated stainless steel mold was used to produce sheets each of 10 × 10 × 2 mm in size.Mechanical performance of neat EPDM and its composites is determined by tensile testing where an Instron 5567 equipped with a 2 kN load cell was used at a cross-head speed of 100 mm/min, with a dedicated extensometer to measure the elongation of the middle region of each dumb-bell sample. Young’s modulus was determined as the slope of a stress-strain curve at the initial stage; namely at strain range 0–1%. Tear tests were conducted at the same cross-head speed using non-nicked samples. Tear strength is determined by following Eq. . All recorded values are the average of three valid measurements for each fraction.where Ts, t and Fmax are respectively tear strength, sample’s thickness and the maximum load achieved during the measurement.The fire retardancy of neat EPDM and EPDM/GIC composites was investigated by cone calorimetry (FTT Limited, East Grinstead, UK) according to ASTM international standard. Samples of 100 × 100 × 2 mm in dimension were preconditioned at 23 °C with 50% humidity; a heat flux of 50 kW/m2 was set for all samples. The measurement provided parameters including time to ignition (TTI), peak heat release rate (PHRR), time to peak heat release rate (TPHRR) and total heat release (THR). A fire performance index (FPI) was calculated according to Eq. Oxygen consumption rate, CO and CO2 yields, smoke production rate and total smoke were used to investigate the toxicity. The data recorded here are the average of four replicates.Thermal degradation of neat EPDM, the graphite intercalation compound and the composites was studied by a thermogravimetric analyzer (Perkin Elmer TGA 7, USA). The measurements were conducted with each sample weighing ∼3.5 mg. Temperature increased to 800 °C @ 20 °C/min in nitrogen atmosphere with a purge rate of 40 ml/min.X-ray diffraction (XRD) and scanning electron microscopy (SEM) were used to investigate the structure and morphology of the GIC, EPDM/GIC composites, and the residues from the calorimetry measurement. XRD was performed by an X'Pert PRO MRD/XL-XRD equipped with Cu Kα radiation at accelerating voltage 40 kV and filament current 40 mA. Scanning was performed at wave length λ = 1.5406 Å and rate 2°/min within 2θ

range of 5–50°. Scanning electron microscopy (SEM, Philips XL30 FEG-SEM) was used at 10 kV for the samples coated with a thin layer of platinum.Since neat elastomers usually need reinforcement to meet certain industrial requirements, this study will examine whether the graphite intercalation compound (GIC) can reinforce an elastomer: ethylene-propylene-diene monomer (EPDM). The GIC – a form of intercalated graphite – is prepared by inserting sulfate-like chemicals between stages; each stage contains 3–4 graphene sheets on average contain the representative stress-strain curves and mechanical properties of neat EPDM and its composites. The obvious mechanical property improvement in a is attributed to the GIC that has somewhat similar properties to graphite: Young’s modulus 8–15 GPa and flexural strength 7–100 MPa b appears to increase linearly with the GIC content; it changes from 1.37 to 4.54 MPa at 12.0 vol%, nearly a 230% improvement. The Young’s modulus decreases with GIC fractions to 2.8 vol% and then increases remarkably. For instance, the composite at 12.0 vol% shows a modulus of 6.73 MPa which is nearly 200% higher than the neat elastomer. The reason is elucidated below.A GIC consists of graphene layers interleaved with chemicals. Graphite is well known for its lubrication effect – the separation between layered graphene under shearing – due to the weak van der Waals bonding. The lubrication effect might be more obvious for a GIC duo to its intercalated structure which can absorb air and water, thus likely promoting the separation. Under loading, the matrix of crosslinked macromolecules transfers stress, which can be tension, compression and shear, to the micro-sized GIC particles; shearing should cause the separation of graphene layers, leading to a lubrication effect that would soften the matrix. This explains why the modulus reduces at low GIC fractions. It is worth noting that the lubrication only occurs under shearing. At high GIC fractions, there are many more GIC particles, and this may limit the space for shearing. As a result, lubrication does not play an obvious role at high fractions. Given the difference in modulus between GICs and elastomers, the GIC at high fractions can obviously stiffen the matrix. These two effects are schematically demonstrated in It is worth to mention that adding a GIC into an elastomer increases the matrix physical crosslinks, which ultimately rises the overall crosslink density (chemical and physical crosslinks). This lifts the load-bearing capacity of the composite through effective stress transfer via those crosslinks, leading to high tensile strength. In addition, those delaminated GIC sheets would enhance rupture resistance (tensile strength) as they are stiffer and stronger than the matrix. The lubricating nature of GICs facilitates the mobility of elastomer’s chains upon loading which in turn enhances the elongation at break. However, a high GIC content would reduce the elongation at break due to the confinement effect of the delaminated GIC sheets on chains uncoiling. For example, in c, the GIC at 5.7 vol% increases elongation at break from 154% to 758%, followed by reduction. At 12.0 vol%, the crosslink density may exceed an optimum value leading to a reduced elongation at break. The reason was well explained in the literature (see Section 3.3 in Ref. Elastomers used in automotive industry are prone to cracking and tear failure. Tear strength is a measure of crack-propagation resistance under shear loading. Since the GIC is plate-like and has high mechanical strength compared to the elastomer, it may work as a crack barrier that lengthens the crack path to failure, as schematically shown in d, the composite tear strength improves up to 26.7 N/mm at 5.7 vol% representing a 276% enhancement, and then drops slightly at 12.0 vol% likely because the GIC aggregated providing spots for stress concentrations that would induce micro-cracks nearby.Thermogravimetric analysis was performed to provide onset temperature and char yield (), to analyze the thermal properties of the graphite intercalation compound (GIC), neat elastomer and its GIC composites. All samples were heated from room temperature to 800 °C at 20 °C/min under nitrogen atmosphere. As advised by Perkin Elmer Neat EPDM shows nearly complete thermal degradation with an onset temperature of 446 °C, whilst the composites degrade at higher temperature and leave char residues. At 12.0 vol% GIC, the onset temperature shifts to 452 °C with 19.7 wt% char residue. It confirms that adding the GIC improves the thermal stability of the elastomer. The GIC upon heating expands and covers the elastomer surface, which intervenes the burning cycle and may prevent the matrix from sequential burning cycles. These are elucidated in the next sections. presents heat release rate, peak heat release rate, mass loss rate, oxygen consumption rate, fire performance index, and time to ignition of neat elastomer and its composites, with numerical values reported in The consumption rate and total quantity of oxygen during burning were obtained for each composite. In c, the neat elastomer has a peak of oxygen consumption rate of 0.6 g/s, and the rate reduces to 0.4 g/s for the 12.0 vol% composite. However, the total quantity of oxygen consumed along burning process increases with the GIC fractions as recorded in . Reasons and fire retarding mechanisms are proposed below.Time to ignition is mainly influenced by the onset temperature of pyrolysis of a material and its thermal properties including heat capacity, thermal inertia and absorption coefficient. Heat capacity is the amount of heat required to raise the temperature of 1 kg material by 1 K. Thermal inertia is the tendency of a material to balance its temperature with the surroundings, and it is determined by the square root of the product of specific heat capacity, material density and thermal conductivity. Earlier time to ignition is attributed to two reasons (i) the onset temperature of decomposition of GIC is almost 2.25 times lower than neat EPDM leading to low pyrolysis temperature for GIC/EPDM composites as recorded in TGA analysis – and (ii) the thermal conductivity of elastomers can be enhanced by carbon fillers including carbon black, carbon nanotubes, graphene and graphite consists of three main elements – heat source, oxygen, and fuel. Flame retarding additives work on one or more of the three elements, but discontinuing the burning cycles is essential to stop a fire. Neat elastomer shows complete burning without residue left as presented in can be explained by a fact where the graphite intercalation compound (GIC, also named expandable graphite) formed a protective-char layer hindering oxygen to enter the fuel sink. To elaborate the detailed mechanisms, we need to understand the GIC constituents. A GIC is typically produced by inserting intercalant materials, such as alkaline earth metals, alkali metals and acids, into graphite layers leading to larger basal space between those layers 24nC+mH2SO4+1/2O2→C24n+(HSO4-)(m-1)(H2SO4)+1/2H2OThe GIC contains 2.9 wt% of Sulfur. At temperature over 200 °C, the intercalants decompose into gases generating immense inter-layer pressure for exfoliation to create GnPs. The expansion occurs due to the following redox reaction endothermic graphite and its intercalants may effectively reduce heat from the source andOn post-expansion, the worm-like structure of the expanded GICs on the composite surface performs three functions as illustrated in to fill and cover the holes or cracks resulted from burning the matrix;to protect the composite beneath from further decomposition.Additionally, the expanded graphite being thermally conductive could dissipate the heat accumulated on the surface more rapidly in comparison with neat elastomer. This explains the reduction in heat release rate, peak heat release rate and mass loss rate in The char of EPDM/GIC composites made by the cone calorimetry test was fragile, and its morphology was carefully examined by X-ray diffraction and scanning electron microscopy. In , it is clear that the GIC inside the elastomer matrix has swollen and expanded upon burning due to the gasses released. shows the XRD patterns of the GIC and its EPDM composites before and after the cone calorimetry test. Both GIC and composites show strong diffraction at 26.5° which is assigned to graphitic structure at (0 0 2) plane corresponding to 0.34 nm layer spacing a, all the composites prior to burning display increasingly stronger diffraction intensities with the GIC fractions than the GIC’s intensity, because the GIC particles were orientated and tightly stacked through the processing by a two-roll mill and the crosslinking under pressure. In b, the composites’ residuals exhibit far weaker intensity at 26.5° compared to the GIC. The burning temperature is so high as to expand the GIC immediately. Upon burning, the GIC expands and the intercalates are released in the form of gasses, likely leaving the GIC stages intact but in a random order in form of worm-like structure. It is known that each GIC’s stage contains a number of graphene sheets depending on the stage level. The disordered fashion of the stages after burning explains the low peak intensity in Smoke is often released once polymers have been ignited. Dense smoke prohibits visibility, causes dizziness, and challenges victims to escape. Smoke is a type of fine soot comprising particles of various size originated from the thermal decomposition of polymers and sometimes from filler. In the recent London fire, smoke was so heavy that victims could not breathe and escape from the lift; a photo published online shows dense smoke billowing into the sky from the Grenfell Tower We herein quantitatively investigated the production rates of CO and CO2, the smoke production rate and the total smoke produced. contain the rates and quantity of gases emitted. In a and b, the addition of the GIC delays the production of CO and CO2; the gas release rates slightly increase at low GIC fractions, but they definitely reduce much at high fractions. This means that only high fractions of the GIC such as 12.0 vol% are effective in reducing the CO and CO2 production rate.However, the yield of both gases does not change much in When a GIC is heated up to over 200 °C, graphite reacts with sulfuric acid producing carbon monoxide. A portion of the carbon monoxide is further reacted with oxygen producing carbon dioxide. Other gases emitted during the GIC expansion include water vapor and sulfur dioxide. At low fractions, the expanded product at the top layer may not be sufficient to form a protective char layer. Therefore, most of the matrix would burn out. This consequently produces much smoke from both the matrix and the GIC. At high fractions, there are sufficient expanded product to build up a char layer; this protects the underneath elastomer and GIC respectively from burning and expanding, which reduces the quantity of emitted smoke and gasses.It is not practical to undertake the limited oxygen index (LOI) and vertical burning tests (UL94). In the UL94 testing, all the samples were burnt. Due to the soft nature of elastomer composites, the samples cannot be held vertically for the LOI measurement.Controls on reservoir heterogeneity of tight sand oil reservoirs in Upper Triassic Yanchang Formation in Longdong Area, southwest Ordos Basin, China: Implications for reservoir quality prediction and oil accumulationCompared to conventional reservoirs, pore structure and diagenetic alterations of unconventional tight sand oil reservoirs are highly heterogeneous. The Upper Triassic Yanchang Formation is a major tight-oil-bearing formation in the Ordos Basin, providing an opportunity to study the factors that control reservoir heterogeneity and the heterogeneity of oil accumulation in tight oil sandstones.The Chang 8 tight oil sandstone in the study area is comprised of fine-to medium-grained, moderately to well-sorted lithic arkose and feldspathic litharenite. The reservoir quality is extremely heterogeneous due to large heterogeneities in the depositional facies, pore structures and diagenetic alterations. Small throat size is believed to be responsible for the ultra-low permeability in tight oil reservoirs. Most reservoirs with good reservoir quality, larger pore-throat size, lower pore-throat radius ratio and well pore connectivity were deposited in high-energy environments, such as distributary channels and mouth bars. For a given depositional facies, reservoir quality varies with the bedding structures. Massive- or parallel-bedded sandstones are more favorable for the development of porosity and permeability sweet zones for oil charging and accumulation than cross-bedded sandstones.Authigenic chlorite rim cementation and dissolution of unstable detrital grains are two major diagenetic processes that preserve porosity and permeability sweet zones in oil-bearing intervals. Nevertheless, chlorite rims cannot effectively preserve porosity-permeability when the chlorite content is greater than a threshold value of 7%, and compaction played a minor role in porosity destruction in the situation. Intensive cementation of pore-lining chlorites significantly reduces reservoir permeability by obstructing the pore-throats and reducing their connectivity. Stratigraphically, sandstones within 1 m from adjacent sandstone-mudstone contacts are usually tightly cemented (carbonate cement > 10%) with low porosity and permeability (lower than 10% and 0.1 mD, respectively). The carbonate cement most likely originates from external sources, probably derived from the surrounding mudstone. Most late carbonate cements filled the previously dissolved intra-feldspar pores and the residual intergranular pores, and finally formed the tight reservoirs.The petrophysical properties significantly control the fluid flow capability and the oil charging/accumulation capability of the Chang 8 tight sandstones. Oil layers usually have oil saturation greater than 40%. A pore-throat radius of less than 0.4 μm is not effective for producible oil to flow, and the cut off of porosity and permeability for the net pay are 7% and 0.1 mD, respectively.Tight oil is crude oil trapped in unconventional reservoirs with extremely low porosity and permeability. With the decline of conventional oil production and advances in horizontal drilling and hydraulic fracturing techniques, tight oil reservoirs are now considered to be an important contributor to the global crude oil supply (), such as the Barnett tight oil play in the Fort Worth Basin (), the Bakken tight oil play in the Williston Basin () and the Eagle Ford tight oil play in South Texas (). There are a considerable amount of potential tight oil resource plays in the petroliferous basins of China, such as those in Songliao Basin, Sichuan Basin, Bohai Bay Basin and Ordos Basin (). Tight oil is now expected to be an important emerging source of oil supply in China (Ordos Basin is the second largest sedimentary basin in China, in which, the Upper Triassic Yanchang Formation is an important oil-bearing formation. Tight oil herein refers to oil that accumulated in oil shale or interbedded tight sandstone reservoirs adjacent to source rocks with subsurface matrix permeability of less than 0.1 mD (). The tight oil resource plays in the Ordos Basin are characterized by a wide areal distribution of source rocks, tight sandstone reservoirs, complex pore-throat structures, and high oil saturation (). The proved reserves of tight oil in the Yanchang Formation of the Ordos Basin were more than two billion tons in 2012 (). With the increasing difficulty of exploiting and developing conventional oil fields in the Ordos Basin, tight oil resources will become increasingly realistic and important supplementary sources in the future.With continuing growth in the exploration and development of tight oil sands, fluid storage and flow capability in such low-permeability systems have become a major concern for petroleum geoscientists and engineers. An understanding of the petrophysical properties of tight oil reservoirs is essential for reservoir evaluation and successful exploitation (). Pore-throat, rather than overall pore volume (i.e., porosity), controls flow capability, producible pore volumes and hydrocarbon flow rates in reservoir rock. Therefore, porosity alone is not an accurate predictor of reservoir quality, especially in tight sands with significant diagenetic alterations (). Identification and quantification of various pore types and their contribution to the overall porosity and flow is an essential step in understanding and predicting the producibility of tight gas/oil reservoirs (). Many detailed papers have been published to investigate the relationship between pores, pore-throats, and reservoir properties in tight-sands (). However, characterizing pore structure could be difficult because a wide pore size distribution is typical for tight oil reservoirs, with pore sizes ranging from several nanometers to several hundred microns (). The routine methods developed for conventional reservoirs are limited when applied to tight oil reservoirs. Limitations exist in conventional analytical methods (e.g., pressure-controlled mercury injection and petrographic thin section point-counting) when quantitatively describing the full 3D granular and porous microstructures (). Recent developments in rate-controlled porosimetry experiments and 3D X-ray micro-CT imaging coupled with conventional petrographic analysis allow direct measurements of the throat radius, pore structure and pore connectivity in three dimensions at the pore scale.Compared to conventional reservoirs, pore structure and diagenetic alterations of tight oil sandstone reservoirs are highly heterogeneous, which is responsible for the heterogeneity of oil accumulations within the same sandstone units. Predictions of sandstone reservoir quality have been well discussed in previous studies (), including the characterization of grain size, sorting, composition, early diagenesis, burial history and tectonic fracturing (). Various proposed porosity controls have been reviewed by , such as dissolution of unstable framework grains and early cements, grain-coated cementation, cement inhibition by early emplaced hydrocarbons and decreased thermal exposure. However, difficulties remain when applying these models and control factors to predicting higher porosity and permeability sweet zones within tight oil sandstone reservoirs. Although the effects of sedimentation and diagenetic alterations have been well studied to understand the post-depositional alterations of the reservoir quality of tight oil sands (), less attention has been paid to the diagenetic heterogeneity within similar depositional lithofacies, the characterization of the pore-throat distribution and its control on reservoir quality or the effects of the reservoir quality on the heterogeneity of oil accumulations.To bridge this gap, this study focused on the following questions:How should the pore structure and pore-throat distribution of tight oil reservoirs be characterized?What control factors significantly affect reservoir heterogeneity in tight oil sandstones, especially the diagenetic heterogeneity and the heterogeneity of oil accumulations in similar depositional facies?What are the implications of the reservoir petrophysical properties on the heterogeneity of oil accumulations in tight oil reservoirs?The results of this study will provide insights into reservoir quality prediction and increase the chance of success when exploring and developing tight oil reservoirs in similar settings.The Ordos Basin is a typical cratonic basin with an area of 25 × 104 km2 located in the western part of the North China block (A). The basin is bordered by the Yin Mountains to the north, the Luliang Mountains to the east, the Qinling Mountains to the south and the Liupan-Helan mountains to the west (). The evolution of the Ordos Basin during the Paleozoic–Mesozoic is divided in three stages: (1) a Cambrian to Early Ordovician cratonic basin with divergent margins; (2) a Middle Ordovician to Middle Triassic cratonic basin with convergent margins and (3) a Late Triassic to Early Cretaceous intraplate remnant cratonic basin (). During the Late Triassic, the Liupan Mountains were thrust underneath the southwestern Ordos area, which led to the formation of the Southwestern Ordos Foreland Depression (). By the end of the Late Triassic, the termination of thrusting and the subsequent erosion of the Liupan Mountains led to an isostatic rebound of the Ordos Basin, which produced a regional unconformity between the Triassic and the Jurassic sediments in the basin (). The Ordos Basin is divided into six structural units: the Yimeng Uplift, the Weibei Uplift, the Western Edge Thrusting Belt, the Jinxi Flexural Fold Belt, the Tianhuan Depression and the Shanbei Slope (B). The study area, the Longdong Area, is located in the southwestern section of the Ordos Basin and covers an area of 5 × 104 km2 (B). The current dip of the bulk of the Longdong Area is to the east with a dip angle of less than 1°. Faults and anticlines have not developed over the majority of the study area (During the Late Triassic, the southwestern shallow-lacustrine deltas and northern deltas prograded into the basin and formed stratigraphic successions more than 3000 m (approximately 10,000 ft) in the Southwestern Foreland Depression, while the strata tapered to the east to less than 1000 m (approximately 3200 ft) in the Ordos Basin (). Recent hydrocarbon exploration and outcrop studies have demonstrated that shallow-lacustrine sand-rich deltas developed extensively along the gentle slopes and central part of the basin, forming the main reservoir rocks of the Triassic oil fields (). The Upper Triassic Yanchang Formation is subdivided into ten members: Chang 10 Member to Chang 1 Member from the bottom to the top of the formation (). The vertical facies succession indicates that the Yanchang Formation covers the entire lacustrine life cycle of the Late Triassic Ordos Basin. Five third-order transgressive-regressive cycles have been recognized: Chang 10 (SQ1), Chang 9-Chang 8L (SQ2), Chang 8U-Chang7 (SQ3), Chang 6-Chang 3 (SQ4) and Chang 2-Chang 1 (SQ5). SQ1, SQ2-SQ3, and SQ4-SQ5 represent three stages of the entire lacustrine life cycle: the initial formation stage, the maximum expansion stage and the extension stage, respectively (A). The Chang 7 stage is the climax of the lake development. During this stage, fine-grained shales and mudstones were deposited in semi-deep and deep lake settings, forming high-quality hydrocarbon source rocks with an average TOC of 13.75% and a vitrine reflectance (Ro) in the range of 0.85%–1.15% (). The crude oil generated in Chang 7 source rocks migrated a short distance to the overlying Chang 6 and underlying Chang 8 sandstones, forming tight oil accumulations (). Most of the Chang 8 tight oil fields have been discovered in the Huanxian, Heshui and Qiangyang areas in the central part of the Longdong Area (B). Oil is mainly accumulated in shallow lacustrine delta distributary channel sandstones with thicknesses of 10–35 m (A total of 156 core samples in the main reservoir intervals of the Chang 8 sandstones were collected from 40 wells. The samples were analyzed and measured using a series of analytical techniques, including thin-section petrography, scanning electron microscopy (SEM), X-ray diffractometry (XRD), pressure-controlled porosimetry (PCP), rate-controlled porosimetry (RCP) and 3D X-ray micro-CT imaging.The samples selected for thin-sectioning were prepared by vacuum-impregnation with blue or red epoxy resin to highlight the pores. The thin sections were partly stained with alizarin red for carbonate mineral determination. Through thin section observations, both framework and intergranular pore-filling minerals were identified. In addition, paragenetic sequences for different rock sets and various types of porosity were also determined. The abundance of detrital and porosity components, grain size, roundness and sorting were obtained by counting 350 points on each thin section.Scanning electron microscopy (SEM) was used to characterize the pore geometry, cement morphology and the textural relationships between the minerals. Twenty samples were gold-coated and examined with a JEOL JSM 840A scanning electron microscope at an accelerating voltage of 20 kV and a current emission of 50–100 pA with the back-scattered electron detector.The mineralogical composition of the clays was determined using X-ray diffraction (XRD) with an X'Pert-MPD X-ray diffractometer with Cu-Kα radiation operated at 100 mA and 40 kV. Clay sub-samples (<2 mm) were examined after being air-dried, ethylene glycol-saturated and heated at 550 °C for 2 h.The porosity and permeability of the 142 samples were measured from 2.5 cm diameter core plugs from the reservoir intervals using a nitrogen permeameter. The dry and clean core samples were placed in the permeameter and injected with nitrogen at confining pressures of 100 and 400 psi.The pore structure was characterized by the parameters obtained from the pressure-controlled porosimetry experiment (PCP) and the rate-controlled porosimetry experiment (RCP). PCP was performed on a PoreMaster PM33-13 mercury porosimeter following the Chinese SY/T 5346-2005 standard (). The lengths of the samples were approximately 2.5 cm. The maximum intrusion pressure was 80 MPa, corresponding to a pore-throat size of 9.2 nm. After reaching the peak pressure, the pressure was gradually decreased to let mercury extrude from the samples. In the PCP experiments, both the intrusion and extrusion curves were obtained.RCP was performed on an ASPE-730 mercury porosimeter following the Q/SY DQ1531-2012 standard of the PetroChina Daqing Oil Field in China. The mercury injection rate was at a quasi-static constant value of 5 × 10−5 mL/min. The maximum intrusion pressure was 6.2 MPa (900 psi) to keep the injection rate quasi-static. This corresponds to a pore-throat size of 0.12 μm. Unlike traditional pressure-controlled porosimetry, this approach is based on the fluctuations of the pressures of the injected mercury as the pore structure changes. By injecting mercury into a small core sample at an extremely low constant rate and accurately measuring the mercury pressure and volume, pores and throats can be distinguished according to pressure fluctuations. This approach has excellent applicability in pore structure characterization of conventional sandstones, but has rarely been applied to tight oil reservoirs (). The most significant advantage of RCP is that the total intrusion curve can be divided into two sub-curves: the throat intrusion curve and the pore intrusion curve, from which the size distributions of the pores and throats and the pore-throat ratio can be obtained. The detailed principles and operating procedures of RCP have been thoroughly explained by 3D X-ray micro-CT is a modern, non-destructive analytical method that allows the investigation of rock samples with X-rays and the imaging of pore spaces in 3D. Micro-CT imaging and image analysis is capable of characterizing statistics like the number and length of the pore throats, the pore connectivity and 3D images of the pore structure with voxel size up to the micrometer scale. Four core samples selected for RCP experiments were also prepared and imaged using 3D X-ray micro-CT. From the original 5-cm core plugs, miniplugs of 2 mm diameter were drilled for the acquisition of the micro-CT tomograms. For each miniplug, 901 tomogram images were obtained using an Xradia ULTRAXRM L-200 Nano CT instrument at an accelerating voltage of 90 kV, and the time of exposure for each tomogram is 30 s. Images are all composed of 20483 voxels with voxel sizes down to 1.3-μm resolution.Segmentation of tomography images is a crucial step for characterization and quantitative analysis of pore structures (). The purpose of segmentation is to remove artifacts of imaging, such as concentric shadows in the CT image and to delineate pores and minerals. A binarization process is commonly implemented to separate images into pore and mineral phases. Simple brightness thresholding is often not adequate for this purpose and, in this case, advanced image processing, such as noise reduction, artifact removal and multiband thresholding need to be utilized (). Noise suppression in gray scale CT volumes was achieved with a simple median filter to provide a uniform basis for consistent comparison of different segmentation algorithms. The segmentation of a grayscale image is not unique and segmentation algorithms therefore require manual interaction and quality control. For a review of image segmentation methods applied to porous media, we refer to The output of micro-CT scanning is a reconstructed 3D volume of local X-ray attenuation coefficients, which depend on both material composition and density (). Specialized rendering software --Avizo was used for visual inspection of this 3D volume based on this local linear attenuation coefficient. To obtain quantitative results from the X-ray CT data, the complete 3D volume was analyzed in a dedicated software package, such as Avizo, VGStudio Max (). Quantitative results from 3D analysis can include data on the overall texture of a material, component volume fractions, pore and grain size parameters and morphology, surface texture, and many more. The 3D data visualization of pore structure in a small subset of the miniplugs was shown in , and the length of the sub-volume is 700 μm.Point counting of thin sections reveals that the detrital composition of the Chang 8 sandstones varies from lithic arkose to feldspathic litharenite according to Folk's classification scheme (), with an average composition of Q35.6F36.1R28.3 (). Minor variations in the relative percentage of detrital components exist in different locations (). The Chang 8 sandstones are lithic rich, with lithic populations including metamorphic fragments (av. 14.5% volumetrically), volcanic fragments (av. 10.52%), sedimentary fragments (av. 2.67%) and biotite (av. 3.47%).The Chang 8 sandstones are primarily fine-to medium-grained and moderately to well-sorted. Their grain size ranges from fine-grained (0.1–0.25 mm) to medium-grained (0.25–0.5 mm). Their sorting ranges from moderately to well sorted, and their grains are subangular to subrounded. The grain-contacts are mostly point-linear contacts in lithic arkose and linear to concavo-convex contacts in feldspathic litharenite (The mineralogy and petrophysical properties of the Chang 8 reservoir sandstones have undergone serious diagenetic modification. The major diagenetic alterations are mechanical compaction, quartz cementation, growth of carbonates and authigenic clay minerals and the dissolution of unstable grains.The Chang 8 sandstones have undergone different degrees of mechanical compaction, which usually leads to IGV (intergranular volume) loss. The mechanical compaction is identified by rotation and slippage of grains (grain rearrangement), deformation of ductile grains (A and B) and linear to concavo-convex grain contacts (A and B). Most of the sandstones in the main reservoir interval were deposited as distributary channel sands with large grain size, low matrix or ductile RF content. The detrital grains occurred as point contacts, which may have enhanced the compression resistance of the rock framework and effectively preserved part of the primary porosity during burial diagenesis.Nearly all the sandstones underwent various degrees of cementation. The major types of cements are authigenic clay minerals (e.g., kaolinite, illite, illite/smectite and chlorite), carbonates (e.g., calcite, ferrocalcite and ferrodolomite) and quartz overgrowths (). Authigenic clay minerals are the most prevalent type of cement in the Chang 8 sandstones, constituting 2%–21% (avg. 8.8%) of the total rock volume. Based on the X-ray diffraction and SEM analysis, four types of authigenic clay were identified: chlorite, kaolinite, illite-smectite mixed-layer (I/S) and illite. Authigenic chlorite is the most abundant diagenetic constituent accounting for 26%–75% (avg. 47.4%) of the total clay content. Chlorite occurs mainly as coatings and rims covering the framework grains (C, D and 4E), with few occurrences as rosettes partially filling the pore space (F). The typical characteristics of chlorite rims are: (1) chlorite exists as pore-lining cements forming 8-10-μm-thick rims around detrital grains; (2) chlorite rims are not visible at the detrital grain contacts; (3) quartz overgrowths are absent or minor where chlorite rims cover the grains; (4) residual primary intergranular porosity can be preserved in sandstones with chlorite rims; and (5) the sandstones are rich in volcanic rock fragments (VRF), and the dissolution of VRF and feldspar grains are common; (6) the sandstones with chlorite rims are mostly deposited in the deltaic distributary channel environments (). Kaolinite accounts for 7%–38% (avg. 13.3%) of the total clay content. The monocrystalline kaolinite takes the form of pseudohexagonal plates while the aggregates exhibit vermicular and booklet morphologies (G). Mixed-layer illite-smectite (I/S) accounts for 1%–16% (avg. 8.2%) of the total clay content in the form of lining cement, and the I/S clay coatings have been transformed into lath-like illite crystals (Carbonates are common in the Chang 8 sandstones, constituting 0.2%–33.75% (avg. 3.75%) of the total rock volume. Calcite and ferrocalcite are the two dominant cements, with average volumes of 1% and 2.6%, respectively. Cementation of dolomite and ferrodolomite is minor or absent in the thin sections. Carbonate cement usually fills the primary intergranular pores or secondary intragranular pores when the grains are not coated with clay or when quartz-overgrowth has not already filled the pores, resulting in the destruction of porosity (A and B). Generally, ferrocalcite cement partially replaces framework grains including feldspar and lithic fragments (Quartz cementation in the Chang 8 sandstones mainly occurs as quartz overgrowths (C), forming either incomplete or complete rims around the quartz grains. The volume of quartz cements ranges from 0.2% to 7.2%, with an average of 1.4%. Boundaries between nuclei (detrital grain) and overgrowth cement are visible due to the presence of “dust” lines (i.e., organic matter, clays or iron-oxides) (C). Quartz overgrowths are commonly absent when quartz grains are coated with thick clay rims (D and E). As revealed by SEM, numerous oriented rhombohedral projections can be identified on the detrital grains. The euhedral quartz crystals fill the intergranular porosity blocking the pore throats and ultimately reducing the porosity and permeability of the reservoir sandstones (Dissolution of unstable detrital grains, such as feldspar and lithic fragments, is common in the Chang 8 sandstones, resulting in the formation of intergranular and intragranular secondary pores. Intragranular pores with a honeycombed configuration are usually formed by the selective dissolution of detrital feldspars along the cleavage (E). Compared to the dissolution of feldspars and lithic fragments, the dissolution of quartz is relatively weak due to its strong chemical stability, and as a result, the rims of the quartz grains are only slightly dissolved (Based on textural relationships, the relative timing of the primary diagenetic features in the Chang8 sandstones was reconstructed, and the paragenetic sequence and porosity evolution are summarized in . However, due to the complex burial and diagenetic history, the timing and duration of all of the observed diagenetic effects cannot be precisely determined.Based on the observations from thin sections and SEM, pores were classified into three types: (a) primary intergranular pores between detrital grains, (b) secondary intragranular pores in partially dissolved grains (e.g., feldspars and lithic fragments) and (c) intercrystalline micropores in biotite and clay minerals.The porosity of relative high-quality reservoirs is composed of relatively large, mostly primary interparticle pores in the Chang 8 sandstones. The intergranular pores can be reduced through mechanical compaction, cementation and pressure dissolution during the progressive diagenetic process. Nevertheless, the intergranular primary porosity in distributary channel sandstones has been largely preserved where later cements have incompletely filled the pore space between grains in samples with less compaction (A and B). Residual intergranular pores are typically triangular and range in sizes from 10 μm to 150 μm. Secondary intragranular pores were mainly formed by partial to pervasive dissolution of detrital feldspars along cleavages (E). The size of the feldspar partial dissolution pores is typically less than 30 μm. Some complete dissolution pores are dozens of microns in size and are often filled by authigenic kaolinite (C). Although the secondary pores formed by dissolution along the granular boundaries are not always distinguishable from primary pores, both types of pores contribute to the effective overall porosity (A). Micropores here mainly refer to intercrystal pores in authigenic clay minerals (e.g., kaolinite, illite/smectite, illite and chlorite) (D, E and 7F), with pore sizes ranging from 0.1 μm to 10 μm.The porosity of the Chang 8 sandstones was obtained via two types of measurements: laboratory measurements on core plugs and thin section measurements by image analysis using the point counting method. The core plug porosity of the Chang 8 sandstones averages 8.2% and ranges from 0.4% to 25.8% (fracture porosity). Core plug permeability ranges from 0.001 mD to 49.6 mD (fracture permeability), with an average of 0.55 mD. Correlation between the plug porosity and permeability is positive as expected, with a moderate correlation coefficient (R2 = 0.6535) (A). Comparison of thin-section porosity with core plug porosity (B) shows consistent positive correlations for most samples; a discrepancy because the micropores are too small to be observed using a petrographic microscope (). For example, the porosity of the sample from well Xi 210-4 (2070 m) measured from image analysis is lower (6.45%) (A) than the porosity from core plug analysis (14.31%). This discrepancy exists because micropores were not included in the image analysis of the thin section but were included in the core plug analysis.In this study, pore-throat size distributions of tight oil reservoirs were investigated using SEM, pressure-controlled perimetry (PCP), rate-controlled prosimetry (RCP) and 3D X-ray micro CT imaging.In the PCP experiment, the mercury intrusion volume was converted to mercury intrusion saturation. Pore-throat size distribution (PTSD) curves were calculated based on the mercury intrusion curves. Based on the characteristic parameters (), the PCP curves of 12 samples from 11 wells were classified into four types. The threshold pressures (Pd) increase with decreasing average maximum pore-throat radius (Rd) from type I to type IV samples (). The mercury intrusion pressure of all samples increases from the beginning to the end with no horizontal stage (A1, B1, C1 and D1). When the pressure decreases to the atmospheric pressure once again, the residual mercury saturation and extrusion efficiency vary greatly between the selected samples. The largest residual mercury saturation is in type I and II with an average of 54.5% and 54.88%, respectively (A1 and 9B1), while the moderate is in type III with an average of 43.88% and the smallest is in type IV with an average of 23.82% (The PTSD of the selected samples is wide, with pore-throat sizes ranging from 9.2 nm to 4 μm. The pore-throat radius of type I samples show a distinctly biomodal distribution. The smaller pore-throat radius lies between 0.01 μm and 0.04 μm, with a mode of 0.025 μm, while the larger pore-throat radius lies between 0.4 μm and 2.5 μm with a mode of 0.1 μm (A2). For types II, III and IV, pore-throats with radius greater than 1 μm are rare. The pore-throat radius of type II shows a unimodal distribution with a pore radius mode of 0.2 μm (B2), however, the pore-throat radius of types III and IV show biomodal distribution characteristics. The two pore-throat radius modes for Type III are 0.02 μm and 0.16 μm (C2), and those for type IV are 0.01 μm and 0.1 μm (By monitoring the mercury capillary pressure in rate-controlled porosimetry experiments, the total mercury intrusion curves can be partitioned into two parts: the pore mercury intrusion curve, which depicts the distribution of the pore bodies, and the throat mercury intrusion curve, which depicts the distribution of throats that interconnect with the pore bodies (RCP mercury intrusion curves were classified into two types in this study and are represented by two typical samples in this paper (Xi210-4 and Li87-4 in ). The detailed characteristic parameters of all measured samples are shown in . For sample Xi210-4, which was cored from massive bedded distributary channel sandstone, the threshold pressure is 0.203 MPa. The trend of the total intrusion curve follows the trend of the pore intrusion curve in areas with low capillary pressures. As the pressure increases, the pore intrusion curve becomes steeper while the trend of the total intrusion curve follows the trend of the throat intrusion curve. The final mercury intrusion saturation of the pore space is 39.32%, which is greater than the final mercury intrusion saturation of the throat space (20.25%) (A). In contrast, for Li87-4, which was cored from wavy cross bedded distributary channel sandstone, the threshold pressure is 1.57 MPa, much higher than that of Xi210-4. For Li87-4, the trend of the total intrusion follows the trend of throat intrusion curve, and the final mercury saturation of the throat space is 36.87%, which is greater than the final mercury saturation of the pore space (20.31%) (Pore distribution, throat distribution and pore-throat radius ratios were calculated from the RCP intrusion curves and are shown in . Both pore distribution and throat distribution show normal distribution characteristics. The pore radius is distributed mostly between 75 μm and 225 μm, with a peak of 120 μm. The variations in the pore radius of the samples with different permeability values are minimal (A). The correlation between porosity/permeability and pore radius is expected to be positive. However, the data points scatter widely in the cross-plots and correlation coefficients are low (R2 = 0.3243 and 0.1065) (The throat distribution is in the range of 0.1 μm–12.57 μm, with significant variations between the samples with different permeability values. The massive- or parallel-bedded distributary channel samples usually have higher permeability values and wider distribution ranges of throat radius, while the mouth bar and wavy cross bedded distributary channel samples have lower permeability values and narrow distribution ranges of throat radius (B and C). For a sample with low permeability, such as Li87-4 (0.04 mD), the throat radius is in the range of 0.12 μm–0.5 μm, with an average of 0.325 μm. In contrast, for a sample with high permeability, such as Xi210-4 (1.96 mD), the throat radius is in the range of 0.12 μm–12.57 μm, with an average of 1.965 μm. Throats, with sizes larger than 1 μm, account for 55.9% of the total throat population. The porosity and permeability values positively correlate with throat radius, with high correlation coefficients (R2 = 0.8426 and 0.9405, respectively) (C and D). This indicates that throat size is the most important parameter affecting permeability. Small throat size are likely responsible for the ultra-low permeability in tight oil reservoirs.For samples with different permeability, significant differences in throat radius result in the wide apparent variation in pore-throat radius ratio (the ratio of averaged pore radius to the mean value of throat radius), which ranges from 35 to 1200 (D). In general, samples with higher permeability tend to have lower pore-throat radius ratios. This arrangement of relatively large-pore spaces and relatively small-throats is typical for tight oil reservoirs, which is different from other conventional reservoirs.A comparison between the PCP and RCP results for sample Li210-4 shows that the total mercury saturation from PCP averages 77.38% (), much higher than that from RCP (59.58%) (). The reason for this discrepancy is that the maximum intrusion pressure of RCP is 6.2 MPa, which can only characterize pore spaces larger than 0.12 μm. Although RCP fails to identify nanopores with radius less than 0.12 μm, it is capable of characterizing pore structures with large pore/throat differences.In general, the storage and flow characteristics of tight oil reservoirs are strongly dependent on the interconnectivity of the pore space. Through the steps of image acquisition, image processing and segmentation, the 3D data visualization of pore structure of the sub-volumes were shown in . The methodology used to perform the image processing and image analysis was stated in Section . The results of the 3D visualization of the pore space were illustrated in video form, with pore space in red, throat in green and rock matrix in white. Pore connectivity could then be illustrated in a color-coded 3D visualization of the pore space. Pores connected with each other are indicated in the same color, while disconnected pores, or isolated pores, are indicated in different colors (The results show great variability and the very different natures of the pore fabrics at micro-to nano-scales. From 3D images, pore structure and connectivity of the open and microporous pore space were probed directly in three dimensions at the pore scale. Two typical samples are displayed here (Xi 210-4 (2070 m) and Li87-4 (2346 m) in ), and their pore network structure and throat network structure are illustrated using the 3D images derived from the tomographic dataset. The network shows a clustering of porosity within larger open pores and a connectivity dominated by the long straight throats associated primarily with grain boundaries. The pore connectivity can be illustrated in a color-coded 3D visualization of the pore network structure. Pores that are connected to each other via ‘slot-like’ throats are coded in the same color.For sample Xi210-4 with high permeability (1.958 mD), thehe open porosity analyzed from this image is 14.1%, which is in agreement with the RCP data (14.1%, ) from the same plug. The majority of the pores are in green, indicating the excellent connectivity of the pore space (D). Only a few micropores are in different colors and were interpreted as being disconnected from each other and from the major pore space due to the absence of effective throats (C). On the contrary, sample Xi 87-4 is significantly tighter with a permeability of 0.035 mD. The porosity of this sample was estimated to be approximately 4.1%, which is also in agreement with the RCP data from the same plug (). Compared to the Xi210-4 sample, the majority of its pores are coded in various colors, indicating the pore network is very poorly connected (In general, depositional lithofacies significantly control: (1) the primary porosity and permeability of sandstones, (2) the sand body geometry, grain size, sorting, sand/mud ratio and architecture and (3) the diagenetic alterations after deposition (). The Chang 8 sandstones were deposited in shallow lacustrine deltaic environments (). Sandstones with high reservoir quality were deposited in high-energy environments, including distributary channel and mouth bar facies. Although the majority of the samples were selected from shallow lacustrine deltaic distributary channel or river-mouth bar facies from the sub-units of the Chang 8 sandstones, variations exist in grain size, matrix content and bedding structures, which could account for the heterogeneity within similar depositional facies, as well as different diagenetic assemblages observed within similar lithofacies.In a given depositional facies, the bedding structures of sandstones exert significant controls on the reservoir porosity and permeability. Sandstones with massive or parallel beddings mostly occur in distributary channel sandstones deposited in high-energy environments and have large grain size, are well sorted and have low matrix content. The highest reservoir quality is developed in massive bedded sandstones. The porosity of the majority of the samples is in the range of 5%–15%, and the permeability is between 0.1 and 10 mD (). Parallel-bedded sandstones are associated with moderate reservoir quality. Cross-bedded sandstones usually occur in mouth bar or sheet sand facies, which are characterized by finer grain sizes, higher matrix content and poor primary porosity compared to channel sandstones. Therefore, the poorest reservoir quality is developed in cross-bedded sandstones with porosity no greater than 10% and permeability no greater than 1 mD (In terms of diagenesis, depositional facies have considerable impacts on the distribution of eogenetic and mesogenetic alterations and therefore on the evolution pathways of the reservoir quality and heterogeneity. For instance, distributary channel sandstones provide good pathways for fluid flow and the mass transfer of elements and ions, therefore contributing to the creation of secondary porosity.In addition to the lithofacies-controlled variations in porosity and permeability, the evolution of the reservoir quality of the Chang 8 sandstones is chiefly governed by the extent and pattern of diagenetic alterations during its progressive burial history.The burial history of well B269 reveals that the Chang 8 sandstones were buried quickly to depths of up to 2800 m (). The rate of porosity reduction with increasing depth depends on various factors. Among them, the initial sandstone composition might be the most influential factor. Petrologic analysis shows that the Chang 8 sandstones contain large amounts of ductile fragments (e.g., volcanic, biotite and mudstone fragments), and the abundance of ductile rock fragments exerts a significant influence on the reservoir quality. According to point-counting data from thin-section studies, ductile rock fragments account for 5%–33.2% (avg. 9.7%) of the total rock volume (). The correlations between the abundance of ductile rock fragments and the porosity/permeability are negative (), suggesting that sandstones with abundant mechanically unstable compositions commonly undergo a more rapid reduction in porosity and permeability during burial via mechanical compaction. As shown in the plot of percent cement vs. intergranular volume (IGV%) () and assuming that the well-sorted Chang8 sandstones had an original porosity of 40% (), the porosity reduction caused by mechanical compaction varies among the various rock types. The feldspathic litharenites underwent a 50% reduction in original porosity by mechanical compaction and intergranular pressure, while the lithic arkoses underwent less than 50% reduction in original porosity by mechanical and chemical compaction (Although the destruction of the primary pores during early burial was mainly due to the compaction of ductile detrital RF and clays, compaction could have been inhibited by cementation or the infill of abundant eogenetic silica or carbonate cement (). The cross plots of porosity and permeability versus compaction rate suggest a negative correlation between the reservoir quality and the compaction rate, when the compaction rate is greater than 0.5 (). Low porosity and permeability also occur in samples with a compaction rate of less than 0.5%. In this case, compaction is not the major control of porosity loss. Cementations in the pore space make sandstones more resistant to compaction. However, cementation also leads to a reduction in the total porosity (Chlorite rim cementation is the dominant diagenetic feature of the Chang 8 tight oil sandstones. Hypotheses on the origin of this chlorite include precipitation through the dissolution of volcanic rock fragments (), recrystallization of smectite in association with the smectite-illite transformation (, and alteration of a kaolinite precursor (Burton et al., 1987). During burial diagenesis, chlorites preferentially form in fluids with high pH values and high Mg and Fe ion contents (The Chang 8 sandstones were deposited in a relatively shallow deltaic environment (proximal and distal delta front) as part of a large fluvial system draining volcanics of southwest Liupan Mountain area (). Weathering of these volcanics brought large amounts of particulate Fe to the river mouth. An increase in salinity near the river mouth (or potentially within the deltaic distributary channels) caused the Fe to be flocculated and deposited in the proximal and distal delta-front depositional environments (). The dissolution of VRF and feldspars is interpreted as providing the main source of Fe, Mg, Si and Al ions for the authigenesis of diagenetic chlorite in the Chang 8 sandstones. Volcanic rock fragments and feldspars, commonly seen being replaced by chlorite in thin section. The sandstone samples that are rich in chlorite rim cement are mostly restricted to the deltaic distributary channel sandstones and were deposited in high-energy environments.The chlorite cement in grain coating samples, forms delicate hollow structures outlining the original grain boundaries of dissolved grains (A), indicating that the beginning stage of chlorite growth predates dissolution of feldspar grains. The preservation of delicate grain molds indicates that dissolution occurred after the sandstone framework had become stabilized with respect to mechanical compaction, otherwise these grain molds would have collapsed. No chlorite rims were observed at the grain contacts between the detrital grains, suggesting that the chlorite rims precipitated after the initial phases of mechanical compaction (A). Usually, the chlorite rims around the detrital grains are likely to be stained by oil (Quartz overgrowths in the Chang 8 sandstones preferentially occur where the chlorite rims covering the detrital quartz grains are discontinuous or absent, as recognized in other studies (e.g., ). Correlation between the diagenetic quartz volume and he abundance of pore-lining chlorite is negative, with a moderate correlation coefficient (R2 = 0.5979) (A). There may be factors other than the abundance of chlorite controlling the effectiveness of the chlorite rims inhibition of quartz cementation, such as the continuity of pore-lining and the modal grain size (). Well-developed chlorite rims could act as effective barriers isolating the surface of detrital quartz grains from pore fluids, which could inhibit the precipitation of quartz onto detrital quartz grains (). Chlorite coatings and rims are thin and discontinuous in sandstone samples, allowing nucleation for quartz overgrowth on some parts of the quartz grain surfaces. SEM reveals several minor euhedral quartz crystals that developed on grain surfaces co-existing with chlorite rims (D). Due to their larger specific surface, fine-grained sandstones require larger volumes of pore-lining chlorite than coarse-grained sandstones for the effective inhibition of quartz cementation (). Therefore, chlorite rims were well developed in most medium-grained sandstones in the Chang 8 Member but less developed in the fine to silt sandstones. Conversely to the chlorite coatings, the chlorite rosettes (F) are isolated and occur late in diagenetic history (). Isolated chlorite rosettes are not volumetrically significant and do not contribute to porosity preservation, as they partially fill intergranular spaces, and have no inhibiting effect on quartz cement nucleation and growth (The cross-plots of the porosity and permeability versus the abundance of chlorite rims show that reservoir quality correlates positively with the abundance of chlorite rims when it is less than a threshold of 7%, indicating that pore-lining chlorites play a significant role in porosity preservation in this situation (B and C). However, after the threshold of 7%, porosity and permeability decrease with increasing chlorite content. There is no correlation between ductile rock fragments and pore-lining chlorite abundance (D). The pore-lining chlorite abundance negatively correlated with the porosity lost by compaction (E), but positively correlated with the porosity lost by cementation (F). So compaction played a minor role in porosity destruction in the samples with pore-lining chlorite abundance greater than 7%. In this situation, the pore-lining chlorites obstruct the pore throats and the connectivity, and therefore, significantly reduce the reservoir permeability. In some cases, the porosity preserved by the chlorite rims appear to have been filled by late mesogenetic carbonate cements (B and C). Therefore, only the porosity preserved by chlorite rims and not filled later by quartz or carbonate cement can be effective in the formation of anomalously high porosity and permeability, or sweet zones, at great burial depth.Carbonate cements usually precipitate in primary and secondary intragranular pores, primarily in the form of pore-filling cement (F) and as replacement minerals of detrital grains (i.e., feldspar grains) ( indicate that when the abundance of carbonate cement is less than a certain threshold (approximately 10%) there is no clear relationship between the abundance of carbonate cement and the reservoir quality. However, samples with high carbonate abundance (greater than 10%) tend to have lower porosity (less than 10%) and permeability (less than 0.3 mD) (). Therefore, carbonate cementation is also an important factor that controls the quality of the Chang 8 sandstone reservoirs.Carbonate cement can occur with variable degrees in the entire sandstone unit, ranging from poorly-cemented to tightly cemented. Carbonate cement is preferentially concentrated near the top and base of the sand body, along the contacts with interbedded mudstones. For example, in the interval of 2333.9–2341 m in the Bai 246 well (), carbonate cement is well developed near the top (volumetrically 13.8% at 2334.37 m) and the base (volumetrically 10.9% at 2340.9 m) of the sandstone body, which are close to the sandstone-mudstone contacts. In contrast, the abundance of carbonate cement is only 4.9%–7.4% (with an average of 5.8%) in the central part of the sandstone body. Therefore, the porosity and permeability near the top and base of the sandstone body (9.7%, 0.15 mD and 6.75%, 0.18 mD, respectively) are much lower than those in the central sand body (average: 12.4% and 1.4 mD) (). Similar carbonate-cemented zones have been observed in other wells throughout the study area.Statistical data indicates that the distance from adjacent sandstone-mudstone contacts has a significant influence on the abundance of carbonate cement, as well as on the porosity and permeability (). For sandstones within 1 m from adjacent sandstone-mudstone contacts, the abundance of carbonate cement is greater than 10% (A), while porosity is lower than 10% and permeability is lower than 0.1 mD (B and C). For sandstones that are greater than 1 m from the adjacent sandstone-mudstone contacts, the abundance of carbonate cement is lower than 10% (A), whereas porosity and permeability are larger and relatively constant (B and C). Because carbonate cementation is concentrated towards the top and base of the sandstone intervals, thicker sandstone bodies are commonly characterized by tight cementation along the base and top but retain relatively high porosity and permeability in the majority interval, which would be an effective reservoir, or sweet zone. On the other hand, thin sandstone bodies are pervasively cemented by carbonate cements, and therefore the reservoir quality is poor. For instance, the Bai 246 sample (2333.17 m) was taken from sandstone that is 1 m thick. The abundance of carbonate cement in this sample is 17.7%, while the porosity and permeability are low, 4.7% and 0.08 mD, respectively (Carbonate cement may originate from various potential sources, such as external, internal or mixed sources (). Observations of thin-sections suggest that there was no occurrence of detrital carbonate grains (both carbonate rock fragments and calcite fossil fragments) in the Chang 8 sandstones. The adjacent mudstones would have provided sufficient ions for the authigenesis of carbonate precipitation. The fundamental components needed to form carbonate cements (e.g., Ca2+, Mg2+, Fe2+ and CO32−) in mudstones may have been derived from the maturation of kerogen (), clay mineral conversions such as smectite to illite or chlorite (The carbonate cements in the sandstones are more likely to have originated from pore water and mass fluid migration from adjacent mudstones, which is supported by the stable carbonate isotope data (). The carbon isotopic composition (δ13C) distribution of calcitic or dolomitic carbonate cements is mainly controlled by the carbon source (). CO2 from organics has the lightest carbon isotope ratios, ranging from −18‰ to −33‰ (), whereas the carbon isotope (δ13C) of CO2 in the atmosphere is approximately −7‰ (). During the precipitation process of cements, due to carbon isotope fractionation, the carbon isotope ratios of cements (δ13C) are heavier relative to the original carbon isotope ratios by 9‰–10‰ (). The poikilitic calcite is generally precipitated at temperature less than 80 °C, and they are mainly precipitated during the shallow burial when rocks have not experienced strong compaction (A). The carbon isotopic composition of the poikilitic calcite cements (δ13C PDB) in Chang 8 sandstones is approximately −0.3‰∼-5‰, with an average value of −3.6‰, suggesting an inorganic origin (). Compared with the poikilitic calcite, the pore-filling (both primary and secondary) ferro-calcite has different material sources and precipitation mechanisms. The ferro-calcite is observed to have filled dissolution pores of feldspars, indicating that the carbonate precipitation and feldspar dissolution are genetically correlated (B). Ferro-carbonate cements were precipitated at a relatively high temperature, approximately 100 °C–170 °C. Ferro-calcite cements in Chang 8 sandstones displayed slightly lighter carbon isotopic compositions, ranging from −6‰ to −10‰, with an average value of −7.8‰ (). Organic CO2 and associated feldspar dissolution are the main material source for this carbonate cement precipitation. Ca2+, Fe3+, and Mg2+ released by the clay mineral conversion in the reservoirs are also important elemental sources. Higher ferro-carbonate content indicates stronger effects of organics on carbonate cement precipitation (During the thermal maturation of the source rock, the fluid released by the overlying Chang 7 source rocks was rich in organic acid and carbon dioxide (), which made the pore water acidic and led to the dissolution of unstable detrital grains, such as feldspar, and created abundant intergranular and intragranular dissolution pores in the Chang 8 sandstones (A). Feldspar can be transformed into kaolinite and quartz under acidic conditions (). Kaolinite usually occurs as a pore filling or pseudomorphous replacement of feldspar grains (H). The kaolinite intercrystalline microporosity could form a storage space for oil accumulation (G). The existence of chlorite rims preserve the primary porosity forming sites of enhanced flow for the late basinal fluids, which would form zones of secondary porosity. The dissolution porosity is in the range of 0.2%–4.6% (avg. 1.8%) and accounts for 10%–30% of the total porosity measured by the point-counting method, indicating that dissolution is effective in improving the reservoir quality in the Chang 8 tight sandstones.The oil-bearing properties vary greatly between different parts of the Chang 8 sandstones. Based on laboratory core analyses, different degrees of oil-bearing types have been identified, covering the spectrum of oil-saturated sandstone, oil-stained sandstone, oil-potted sandstone, oil-trace sandstone and fluorescing sandstone (). The oil saturation increases with increasing average porosity and permeability from fluorescing sandstones to oil-stained sandstones.The oil-saturated and oil-stained sandstones with high oil content occur primarily in deltaic distributary channel sandstones. In a given depositional facies, the bedding structures of sandstones not only control the reservoir porosity and permeability but also control the heterogeneity of oil accumulations. Massive- and parallel-bedded sandstones contain abundant oil accumulations, whereas cross-bedded sandstones contain minor oil shows or no oil shows (). In general, the oil distribution is relatively homogeneous in massive-bedded sandstones (A and C), while the oil distribution in parallel- and cross-bedded sandstones is highly heterogeneous. In parallel- and cross-bedded sandstones, the coarser laminas are oil-stained to oil potted, while the finer laminas are oil-trace or non-oil bearing (D to G). The wavy cross-bedded sandstones are commonly non-oil bearing, due to finer grain size, and the finer muddy laminas would also form fluid flow barriers and baffles (In thin-sections, the oil-bearing samples are characterized by weak compaction, point-linear grain contacts, well-developed chlorite rim cementation, intra-feldspar dissolution and a lack of carbonate and quartz cementation (, sandstones with massive or parallel beddings were mostly deposited in distributary channel depositional facies. For distributary channel sandstones, chlorite rims are usually well developed and halt quartz overgrowth. In addition, carbonate cementation is concentrated near the top and base of the sandstone intervals.Because oil flow is primarily controlled by reservoir quality, so a cut off of porosity and permeability should exist for effective reservoirs with the potential to accumulate and flow oil. Cross-plots of the porosity and permeability of oil-bearing intervals and dry intervals () determined by well testing suggest that oil-bearing intervals have higher porosity and permeability than dry intervals. A porosity of 7% and a permeability of 0.1 mD is the cut off for the net pay in the study area (A). Oil saturation (So) of the oil layers ranges from 40% to 75.68%, with an average of 56.78%, while most of the So values of the water layers are less than 40% (B and C). Sandstone with porosity lower than 7% and/or permeability lower than 0.1 mD do not have producible oil. Traditionally, the major oil flow path in tight oil sandstones, which connects primary and secondary porosity, is thought to be the narrow “slot-like” pore throats along the grain boundaries.According to the relationship between permeability and throat radius obtained by the RCP experiments in Section D), there is a cut off on the throat radius of 0.4 μm, which is also the cut off for effective reservoirs to produce oil. For instance, four deltaic distributary channel sandstone samples with different oil accumulation characteristics were selected from the Ning128 and Xi210 wells. Their PCP mercury intrusion curves are shown in . For the two oil-bearing samples (Xi210-4 2070 m and Ning128-2 1802.3 m), the pore-throat radii are in the range of 0.01 μm–0.25 μm, however, the majority of the permeability are distributed in pore-throat radii greater 0.4 μm. In contrast, nearly all the pore-throat radii for the non-oil-bearing samples (Xi210-3 2079.6 m and Ning128-1 1802.2 m) are less than 0.4 μm, indicating that a pore-throat radius less than 0.4 μm is not effective for oil flow, charging and accumulation. In 3D X-ray micro-CT images (), the oil-bearing sample Xi210-4 2070 m is observed to have excellent pore connectivity, and the majority of the pore space is connected via ‘slot-like’ throats with large throat size.The Chang 8 tight oil sandstones in the Longdong Area are fine-to medium-grained, moderately to well-sorted lithic arkose and feldspathic litharenite. The sandstones have undergone significant diagenetic alterations, such as compaction, quartz overgrowth, carbonate and authigenic clay mineral cementation and dissolution of unstable grains.The pore-throat size distributions of tight oil reservoirs are highly heterogeneous. The throat size is the most important factor that affects permeability. Small throat size is believed to be responsible for the ultra-low permeability in tight oil reservoirs. High quality reservoirs are characterized by larger throat sizes, lower pore-throat radius ratios and well-connected pores.The reservoirs are extremely heterogeneous resulting from the heterogeneities of depositional lithofacies, pore structures and diagenetic alterations. Authigenic chlorite rim cementation and the dissolution of unstable detrital grains are two major diagenetic processes that preserve porosity and permeability sweet zones in oil-bearing intervals. After compaction and cementation, residual intergranular primary pores and secondary pores are the major storage spaces.High-quality reservoirs were deposited in high-energy environments including distributary channel and mouth bar facies. In a given depositional facies, the bedding structures of sandstones not only have a significant control over the reservoir porosity and permeability but also over the heterogeneity of oil accumulations.Well-developed, continuous chlorite rims inhibit quartz cementation effectively preserving the primary intergranular pores, whereas discontinuous and thin coatings or rims are not effective in preserving the primary pores. Chlorite rims do not effectively preserve porosity when the chlorite content exceeds a threshold value of 7% because pore-lining chlorites obstruct the pore throats and connectivity and therefore significantly reduce the reservoir permeability.Carbonate cement is preferentially concentrated near the top and base of the sand bodies, along the contacts with interbedded mudstones. Stratigraphically, sandstones within 1 m of sandstone-mudstone contacts are usually tightly cemented by carbonates (carbonate cement greater than 10%) with low porosity and permeability (lower than 10% and 0.1 mD, respectively). The source of carbonate cement was most likely external, possibly derived from the overlying and underlying mudstones. Pore-filling carbonate cements filled the earlier dissolved intra-feldspar pores and the residual intergranular pores, and finally formed the tight reservoirs.Petrophysical properties significantly control the flow capability and the heterogeneity of oil accumulations in the Chang 8 tight sandstones. The cut off of porosity and permeability for net pay are 7% and 0.1 mD, respectively. A pore-throat radius of less than 0.4 μm is not effective for producible oil to flow.A new first-order mixed beam element for static bending analysis of functionally graded graphene oxide powder-reinforced composite beamsIn this paper, a new first-order mixed beam element is established and applied to investigate the static bending of the functionally graded graphene oxide powder-reinforced composite beams. The proposed beam element is developed via a mixed finite element formulation based on first-order shear deformation theory. The new element consists of two nodes and three degrees of freedom per node. In addition, the present beam element uses linear shape functions. The proposed beam element is free of shear locking without using selective or reduced integration. The comparison study demonstrates that the present beam element can predict exact solutions with very few elements, so, the computation cost is reduced. Then the proposed beam element is used to analyze the static bending of functionally graded graphene oxide powder-reinforced composite beams. A comprehensive parameter study is carried out to demonstrate the effects of some parameters such as the boundary conditions, graphene oxide powder weight fraction, graphene distribution patterns, graphene oxide powder dimensions, and the slender ratio of the beams.Functionally graded materials (FGMs) were first introduced and produced in 1984 by Japanese scientists to prepare thermal barrier materials Because there are many advantages of using FGM structures in engineering and industry (Vinh et al. Recently, the GOP is considered as a novel alternative reinforcement with being compatible with polymers matrix. Ebrahimi et al. It can be concluded that an excellent understanding of the mechanical behaviors of the GOP reinforced composite (GOPRC) beams is very necessary. In this study, a new first-order mixed beam element is developed to investigate the bending behavior of GOPRC beams. The organization of the present work is as follows: gives the construction of four types of GOPRC beams and mixed finite element formulation of the proposed beam element. gives several examples to demonstrate the high convergence rate and accuracy of the proposed beam element and some application on the static bending of GOPRC beams. Some important conclusions and suggestion for future works are given in In this study, a beam made of functionally graded graphene oxide powder-reinforced (FG-GOPR) composite with the length of L, the width of b and the thickness of h is considered as shown in The FG-GOPR is composed of a polymer matrix material and GOP reinforced. Four types of the distribution of GOP through the thickness of the beams is considered for analysis in this work, which are denoted by U-GOPR, V-GOPR, X-GOPR and O-GOPR. The effective Young’s modulus of the FG-GOPR composite beams is calculated via modified Halpin-Tsai model as the following formulawhere αL=0.49 and αT=0.51 indicate the GOPR efficiency in the axial and transverse directions. Young’s modulus EL and ET represent longitudinal and transverse Young’s modulus of the composite beam, and they are computed as followsEL=1+ξLηLVGOP1-ηLVGOPEM;ET=1+ξTηTVGOP1-ηTVGOPEMwhere two coefficients ηL and ηT are calculated viaηL=EGOP/EM-1EGOP/EM+ξL,ηT=EGOP/EM-1EGOP/EM+ξTIn which EGOP,EM are respectively Young’s modulus of the GOPR and polymer matrix, and ξL, ξT are the geometry factors, which are computed aswhere dGOP and hGOP denote the diameter and thickness of GOPR, respectively. The effective Poisson’s ratio is computed via the rule of mixture taken bywhere VGOP,VM denote the volume fraction of the GOPR and polymer matric, respectively. The relation between these volume fractions is given bywith VGOP for four patterns of FG-GOPR beam is calculated as the following formulaVGOP=VGOP∗for U - GOPR1+2zhVGOP∗for V - GOPR4zhVGOP∗for X - GOPR2-4zhVGOP∗for O - GOPRwhere VGOP* is the total volume fraction of GOPs that can be calculated via followswhere ρGOP,ρM are mass density of graphene oxide powder and matrix. The effective shear modulus is computed asBased on the first-order shear deformation beam theory, the displacements of any point of the beam are given byThe non-zeros strain field of the beam is obtained asIt is noticed that the symbol (,) is used to denotes the derivatives with respect to the quantity following it.The constitutive relations between the stresses and strains field of the beam can be obtained as followswhere E(z),G(z) are respectively Young’s modulus and shear modulus.The strain energy of the beam can be obtained as the following formulaThe variation of the strain energy of the beam can be computed asIntegrating across the section of the beam, one getswhere P and T1 are the resultant forces, which are calculated viawhere k is the shear correction factors which can take the value of 5/6 and π2/12 or be a function of the material gradient P=b∫-h/2h/21zE(z)u,x+zβ,xdz=b∫-h/2h/21zE(z)1zu,xβ,xdzH=b∫-h/2h/21zE(z)1zdz=b∫-h/2h/2E(z)1zzz2dzA new first-order Mixed Beam Element (MiBE) is developed in this section. The MiBE consists of two nodes and three degrees of freedom per node as show in The nodal displacement vector of the beam element is given byIn which, the coordinate and displacements within the beam element approximate via nodal variables as the following formula (Frikha et al. x=N1x1+N2x2u=N1u1+N2u2w=N1w1+N2w2β=N1β1+N2β2+Nmβmwhere N1,N2 are the linear shape functions, Nm is the quadratic shape function, and βm is a parameter which will be eliminated later. The shape functions are calculated as the following formula (Frikha et al. The resultant shear force is rewritten as follows (Frikha et al. , the variation of the strain energy is expressed as follows (Frikha et al. The variation of the work done by external force is calculated as the following formulaThe Hamilton’s principle is employed to established the discretized equations of motion of the beams as followsδUTδT0δβmKuuKuTKuβKuTTKTTKTβKuβTKTβKββUT0βm-fufTfβ=0Kuu=∫-11BTHBL2dξ;KuT=∫-11BsTL2dξ;Kuβ=∫-11BTHBmL2dξ;KTT=-∫-111HsL2dξ;Kββ=∫-11BmTHBmL2dξ;KTβ=∫-11NmL2dξfu=∫-11NwTqL2dξ+∫-11BsTFL2dξ;fT=-1Hs∫-11FL2dξ;fβ=∫-11NmFL2dξBy eliminating the parameter βm from Eq. K11=Kuu-Kuβ1KββKuβT;K12=KuT-Kuβ1KββKTβ;K22=KTT-KTβ1KββKTβThe parameter T0 is eliminated from Eq. , the static bending equations is achieved as the following formulaIt is noticed that all integrations in Eq. are calculated using full Gauss integration, but the shear-locking phenomenon does not occur. This is an advantage of the proposed beam element MiBE in comparison with many available beam elements.Initially, the bending behavior of a homogeneous beam is considered to demonstrate the high convergence rate and accuracy of the proposed element MiBE. A homogeneous beam with the width of b=1m, Young’s modulus of E=29000Pa, and Poisson’s ratio of ν=0.3 is investigated. Firstly, a homogeneous beam with one end side clamped and other end side subjected to a concentrated load F=100N, is studied. The comparison between the present results and Heyliger with different number of beam elements Ne. Secondly, a homogeneous beam with simply supported at two end sides and subjected to a uniform load q=10N/m is considered. exhibits the comparison between the present results and those of Heyliger , it can be concluded that the proposed element beam MiBE is high convergence rate and the present results are in good agreement with those of Heyliger Next, the static bending of an FG-GOPR composite beam is considered to validate the accuracy and efficiency of the proposed beams element MiBE. An FG-GOPR composite beam made from a mixture of epoxy matrix and GOP is considered. The beam length is L=100m , the depth is b=1m and the high is h. The dimensions of GOP are dGOP=500nm and hGOP=0.95nm. The material properties of GOP are EGOP=444.8GPa , ρGOP=1.09g/cm3 , νGOP=0.165 ; and those of epoxy matrix are EM=3GPa , ρM=1.2g/cm3 , νM=0.34. The convergent rate of the proposed beam element MiBE are presented in for FG-GOPR composite beams with different number of elements (Ne) as well as two cases of clamped–clamped (CC) and simply supported (SS) beams. In which, the length-to-thickness ratio is L/h=10 , the weight fraction of GOP is WGOP=0.1% , and three schemes of distribution of GOPs (U-GOPR, O-GOPR, and X-GOPR) are considered. The FG-GOPR composite beams are subjected to uniformly distributed loads (UL) with intensive of q0. The non-dimensional center displacements of the FG-GOPR composite beams are calculated by wc∗=wc.100EMh3/(q0L4). According to , it is obvious that the proposed beam element MiBE is high convergence speed, it can predict the exact center displacement of the CC and SS FG-GOPR composite beams with only two elements.Now, to validate the accuracy of the proposed beam element MiBE, the comparisons of the non-dimensional center deflections of the FG-GOPR composite beams under UL with a various value of WGOP and the length-to-thickness ratios L/h are given in . In which, the numerical results of Zhang et al. In this section, an FG-GOPR composite beams with the length of L=100m , the width of b=1m , and the thickness of h is considered. The beams are made from a mixture of epoxy matrix and reinforcement of GOP. The material properties of GOP and epoxy matrix are given as follows (Zhang et al. Epoxy matrix: EM=3GPa , ρM=1.2g/cm3 , νM=0.34 ;GOP: EGOP=444.8GPa , ρM=1.09g/cm3 , νM=0.165.The diameter of GOP is dGOP=500nm , and the thickness is hGOP=0.95nm. The following non-dimensional quantities are used through the rest of the paper (where h0=L/10)Four types of boundary condition are considered in this study, which are clamped–clamped (CC), simply-simply (SS), clamped-simply (CS), clamped-free (CF) supported. The FG-GOPR composite beams are subjected to uniformly distributed load (UL) or sinusoidal load (SL) with intensive of q0.Firstly, the non-dimensional maximum deflections of the FG-GOPR composite beams with different types of boundary conditions, the distributed loads, the length-to-thickness ratios and the weight fraction of GOP are given in . According to these tables, the non-dimensional maximum deflections of the FG-GOPR composite beams decrease rapidly as increasing of weight fraction of GOPs. So, a small weight of GOPs can increase strongly the stiffness of the FG-GOPR composite beams, while the mass is almost unchanged. The maximum deflections of CF beams are highest while those of CC beams are the smallest. Besides, the maximum deflections of X-GOPR composite beams are the smallest while those of O-GOPR composite beams are the greatest. It means the distribution of the GOPs through the thickness of the FG-GOPR composite beam has a strong effect on the bending behavior of the beams. The beams with GOPs-rich surfaces are stiffer than those with GOPs-pure surfaces.Next, we analyze the influence of the length-to-thickness ratio L/h on the deflection of the FG-GOPR composite beams subjected to UL. Two cases of the boundary conditions are considered, which are CC and SS conditions. In which, the weight fraction of GOP is WGOP=1.0% , the length-to-thickness ratio varies from 5 to 50. The variations of the non-dimensional center deflections of CC and SS beams versus the length-to-thickness ratio L/h are plotted in . It can see clearly that when the length-to-thickness ratio increases, the deflection of four types of FG-GOPR composite beams increase. The increased speed of O-GOPR composite beams is the fastest and that of X-GOPR composite beam is the slowest. In general, the maximum deflections of SS beams are approximately five times greater than those of CC ones. demonstrates the influence of the weight fraction of GOP WGOP on the non-dimensional center deflections of the FG-GOPR composite beams with L/h=10. The weight fraction of GOP WGOP varies from 0.1% to 5%. It can be seen that when the weight fraction WGOP is small, the center deflections of U-GOPR and V-GOPR composite beams are similar. The center deflections of the FG-GOPR composite beams decrease rapidly when the weight fraction WGOP increases. With only 5% of the weight fraction of the GOP, the maximum deflections of the beams are approximately six times smaller than those of the FG-GOPR composite beams with 0.1% of the weight fraction of GOP.Continuously, the effects of the size of the GOP on the static bending behavior of the FG-GOPR composite beams are investigated. displays the variation of the non-dimensional center deflections of the CC and SS FG-GOPR composite beams as the function of the ratio of dGOP/hGOP. Where the values of the ratio of dGOP/hGOP varies from 20 to 1000. It can be seen that when the aspect ratio dGOP/hGOP increases, the deflections of the FG-GOPR composite beams decrease. It means, the increasing of the size of the GOPs leads to an increase in the stiffness of the materials. Once again, it can be seen that the deflections of the X-GOPR composite beams are the smallest whilst those of the O-GOPR composite beams are the highest. So, we can control the stiffness of the FG-GOPR composite beams with a similar weight fraction of GOPs by regulating the dissemination of the GOPs through the thickness of the beams.In this study, a new mixed beam element (MiBE) has been established for the static bending analysis of FG-GOPR composite beams. The proposed element is developed based on the first-order shear deformation theory and mixed finite formulation with linear approximated shape function. Several examples have been provided to demonstrate the fast-convergency and accuracy of the proposed beam element for both isotropic and FG-GOPR composite beams. Then, the proposed beam element MiBE was applied to analyze the bending behavior of FG-GOPR composite beams. The parameter study shows that the bending behavior of the FG-GOPR composite beams depends strongly on the distribution of the GOP, amount of GOP and the size of GOP. The results of the current study can serve as reference results for designing, testing, and analysis of FG-GOPR composite plates and beams. Due to the efficiency and accuracy of the proposed element, the application of MiBE for the vibration, buckling and dynamic problems can be considered in future works.This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.First-Principles study of structural, elastic and electronic properties for the KT-GaBP2 semiconductor under pressureThe structural, elastic and electronic properties of KT-GaBP2 under pressure ranging from 0 to 70 GPa were investigated by using density functional theory. The elastic stiffness constants were calculated under different pressures and found that the single crystal KT-GaBP2 is elastically stable in the range from 0 to 70 GPa, and afterwards becomes unstable. It can be inferred that the phase transition may occur in the range of 70 ∼ 80 GPa. The universal anisotropy index of single crystal KT-GaBP2, as well as the elastic moduli and Debye temperature for the polycrystal KT-GaBP2 were calculated under different pressures, and the effect of pressure on the elastic properties was studied further. It is found that the single crystal KT-GaBP2 is elastically anisotropic and that the anisotropy increases monotonically with increasing pressure. The polycrystal KT-GaBP2 inclines to resist better volume compression than shape deformation, and it will possess a best thermal conductivity under a certain pressure value within the range of 25 ∼ 35 GPa. Finally, the electronic properties of KT-GaBP2 induced by pressure were studied and found that the energy gap decreases monotonically with increasing pressure. Furthermore, the direct energy gap changes into an indirect one as the pressure up to about 70 GPa.The semiconducting materials composed of elements from II, IV and V groups as well as III and V groups in the ABX2 stoichiometry (II-IV-V2 and III-III-V2) have attracted extensive attentions in the area of photovoltaic (PV) technique due to their higher degrees of tunability of energy gap by foreign element substitution, alloying and phase engineering, as well as excellent ability to absorb visible light in the past few years In view of this, investigating the natures of the GaBP2 compound under a pressure effect is necessary to develop its practical applications. However, the systematic researches about the physical and chemical properties of GaBP2 compound induced by pressure have not been reported at present. In this paper, the KT-GaBP2 is selected as research object to study its structural, elastic and electronic properties under pressure.All calculations were performed by using the plane-wave pseudopotential approach based on the density functional theory (DFT) The KT-GaBP2 belongs to the orthorhombic crystal system with space group PNA21 and possesses four formula units per unit cell. The structure with atomic descriptions of KT-GaBP2 unit cell at 0 GPa pressure is shown in . From this figure, the unit cell is composed of 16 atoms, which includes four Ga atoms, four B atoms and eight P atoms. The atomic positions of KT-GaBP2 are Ga (0.083, 0.625, 1.000), B (0.087, 0.125, 1.000), P1 (0.081, 0.624, 0.374) and P2 (0.082, 0.126, 0.378). tabulates the calculated lattice parameters, volume and density under different pressures, together with previous data In order to understand the impact mechanism of pressure on the structure, the numerical structural parameters, namely X/X0 (X=a, b, c, V), b/a, c/a and b/c were calculated, where X0 represents the lattice parameters and volume at 0 GPa pressure (a) presents the variation of X/X0 versus pressure ranging from 0 to 70 GPa. From this figure, all the a/a0, b/b0, c/c0 and V/V0 decrease with increasing pressure. In particular, the variance rate of a/a0 is largest among that of a/a0, b/b0 and c/c0, suggesting that it is easiest to compress a single crystal KT-GaBP2 along the a-axis direction. (b) depicts the variations of b/a, c/a and b/c versus pressure. As indicated in (b), both the b/a and c/a increase accompanied by rising pressure. Specifically, the b/a and c/a are 1.165 and 0.960 at 0 GPa pressure respectively, and become 1.182 and 0.972 at 70 GPa respectively, which increase by 1.46% and 1.25% respectively, indicating that the compressibility along the a-axis direction is largest. Nevertheless, the b/c decreases first in the range from 0 to 50 GPa and then increases beyond 50 GPa, showing the compressibility along the b-axis direction is larger than that along the c-axis direction as pressure varies from 0 to 50 GPa, and then smaller than that along the c-axis direction beyond 50 GPa.The elastic stiffness constants Cij of single crystal KT-GaBP2 under different pressures were calculated. presents the calculated Cij, together with previous data , the Cij at 0 GPa pressure are in good agreement with previous data. The variation curves of Cij versus pressure are depicted in . From this figure, the C11, C22, C33, C12, C13 and C23 increase monotonically in the pressure range from 0 to 60 GPa, and then decrease beyond 60 GPa, whereas the C44, C55 and C66 are less sensitive to pressure. Besides, all of the C11, C22 and C33 are much larger than C44, C55, C66, C12, C13 and C23, indicating the deformation resistances along the axial directions are stronger than those along the non-axial directions. Furthermore, the relationship C33>C22>C11 implies that the strength of bonding along the c-axis direction is stronger than that along the b- and a-axes directions. And, the C22 are slightly larger than C11, suggesting that the compressibility along the a-axis direction is more compressible than that along the b-axis direction, which is well consistent with the conclusions reached in Section The elastic stability of single crystal KT-GaBP2 was evaluated via Born criteria C11>0,C11C22−C122>0,C11C22C33+2C12C13C23−C11C232−C22C132−C33C122>0,C44>0,C55>0,C66>0., the calculated Cij satisfy the aforementioned criteria, implying the orthorhombic KT-GaBP2 is elastically stable against the elastic deformations with pressure in the range from 0 to 70 GPa. In addition, the Cij at 80 GPa pressure were also calculated and found that the C66 is less than zero, suggesting the criteria are destroyed and implying the KT-GaBP2 becomes unstable. Therefore it can be inferred that the phase transition may occur in the pressure range of 70 ∼ 80 GPa.The average elastic moduli of polycrystal KT-GaBP2 were calculated according to the Voigt–Reuss–Hill approximations , together with previous results available , the BH, GH, E and ν at 0 GPa pressure are in good agreement with those of previous work. The variations of BH, GH and E versus pressure are depicted in . From this figure, all the BH, GH and E increase first and then decrease, indicating the resistance to volume deformation capacity, resistance to shear deformation capacity and rigidity are firstly stronger and then weaker in the range from 0 to 70 GPa. Furthermore, all of the BH are larger than GH over the entire pressure range, suggesting that the polycrystal KT-GaBP2 is globally more resistive to volume compression than to shear deformation. On the other hand, the larger E is, the better stiffness will be. According to the variation of E, it is easy to infer that the polycrystal KT-GaBP2 will reach a strongest stiffness at a certain value within the pressure range of 40 ∼ 50 GPa.The BH/GH ratio and Poisson’s ratio ν are commonly used to predicate ductile and brittle behaviors of materials  exhibits the variations of BH/GH ratio and ν accompanied by pressure. From this figure, the behavior of BH/GH ratio is extremely similar to that of ν. Both the BH/GH ratio and ν are smaller than their respective critical values as pressure varies from 0 to 10 GPa, signifying that the polycrystal KT-GaBP2 possesses a brittle nature in the range of 0 ∼ 10 GPa. Furthermore, a transition from brittleness to ductility occurs before 20 GPa. The variations of H and HV versus pressure are presented in . It is clear that both the H and HV display also almost same changing trend. In contrast to the variations of BH/GH and ν, the H and HV decrease initially and then increase. It turns out that the polycrystal KT-GaBP2 presents a brittle nature in the range from 0 to 10 GPa, which verifies the discussions about the variations of BH/GH ratio and ν versus pressure.The elastic anisotropy and thermal conductivity of KT-GaBP2 were evaluated by using universal anisotropy index AU. From this figure, all the values of AU are larger than 0.14, demonstrating that the single crystal KT-GaBP2 is elastically anisotropic and that the elastic anisotropy increases monotonously with pressure. The elastic wave velocities vt, vl and vm as functions of pressure are shown in (a). From this figure, both the vm and vt have similar behavior. At the same time, the variation of ΘD versus pressure is presented in (b). From this figure, as pressure up to 30 GPa, the ΘD exhibits a non-linear increment, and it is enhanced by more than 5%, which is attributed to the rapid increment of bulk modulus in the range from 0 to 30 GPa. The ΘD decreases rapidly in the range from 30 to 70 GPa, however, which is possibly caused by the decrease of vm in the range from 10 to 70 GPa. Usually, a high ΘD means a large thermal conductivity The effect of pressure on the electronic properties for the KT-GaBP2 were studied finally. The variation of energy gap Eg versus pressure is illustrated in . From this figure, the Eg behaves with a quasi-linear decrease with increasing pressure. The band structures at 0 and 70 GPa pressures are displayed in (a) and (b), respectively. According to (a), both the valence band maximum (VBM) and conduction band minimum (CBM) are located at Γ point, and the value of Eg is 1.02 eV, which is in fairly good agreement with the previous result calculated following GGA-PBE by Kumar et al. (b)), indicating a transition from direct band gap to indirect one occurs in the range from 60 to 70 GPa.The lattice parameters, volume and density of KT-GaBP2 unit cell under different pressures were calculated, and the effect of pressure on lattice parameters and volume were discussed further. It is found that the a-axis direction is the more compressible than the b- and c-axes directions. The elastic stiffness constants Cij were calculated under different pressures and the structural stability was evaluated further. It is found that the orthorhombic KT-GaBP2 is elastically stable in the pressure range from 0 to 70 GPa, and afterwards becomes unstable, it can be inferred that the phase transition may occur in the pressure range of 70 ∼ 80 GPa. The elastic moduli, universal anisotropy index AU and elastic Debye temperature ΘD were calculated under different pressures and found that the deformation resistances along the axial directions are stronger than those along the non-axial directions, and the strength of bonding along the c-axis is stronger than that along the b- and a-axes for the polycrystal KT-GaBP2. Furthermore, the polycrystal KT-GaBP2 inclines to resist better volume compression than shape deformation, and it will reach a strongest stiffness at a certain pressure within the range of 40 ∼ 50 GPa. From the BH/GH ratio, ν, H and HV, there exists a transition from brittleness to ductility induced by pressure for the polycrystal KT-GaBP2. Moreover, the single crystal KT-GaBP2 is elastically anisotropic, and the anisotropy increases monotonically with rising pressure. It will possess a best thermal conductivity as the pressure up to a certain value within the range of 25 ∼ 35 GPa. The electronic properties of KT-GaBP2 were studied finally and found that the energy gap decreases monotonically with increasing pressure. Furthermore, the direct energy gap changes into an indirect one as the pressure up to about 70 GPa.Yan-Tong Bian: Conceptualization, Investigation, Methodology, Formal analysis, Writing - original draft. Sheng-Hui Qian: Investigation, Methodology, Formal analysis. Xin-Xin Ding: Conceptualization, Investigation, Methodology. Guang-Hua Liu: Investigation, Formal analysis, Writing - review & editing, Supervision.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.© 2013 Elsevier B.V. All rights reserved.In the present work, a medium Mn, Fe–Mn–C–Al–Si alloy was subjected to different heat treatment conditions and subsequent deformation to understand the effect of these processes on austenite. It was found that, after intercritical annealing, the microstructure was ferrite plus austenite duplex phase (FADP) regardless of cooling rate to room temperature. When cold rolled, the retained austenite of the FADP structures exhibited both twinning and strain induced transformation (SIT) to martensite. A detailed characterization of the co-existence of twinning and SITing after cold rolling is presented.Over the last few decades, growing demands for weight saving and safety requirement have motivated new concepts of automotive steels to achieve improved mechanical properties in comparison with the existing Advanced High Strength Steels (AHSS). Among various developments, medium Mn content steels are considered to be potential candidates to achieve the performance targets of the so-called third generation AHSS Indeed, several literature studies proved that the medium Mn steels containing 5–10 wt% Mn have enhanced mechanical properties compared to the first generation AHSS due to the occurrence of deformation induced martensitic transformation This paper describes a novel ferrite plus retained austenite microstructure based on the above alloying concepts. The behavior of retained austenite during room temperature deformation of three heat treated variants was investigated, with a focus on twinning and strain induced transformation characteristics.The steel used throughout this work was supplied by CANMET-MTL (Hamilton, Ontario, Canada). Castings were done in an induction furnace. The composition (wt%) of the examined steel is shown in . This is one composition selected from a series of alloys which were designed on the basis of attaining a certain level of stacking fault energy (SFE) in the metastable austenite to promote twinning as opposed to transformation to martensite The cast ingot was sectioned into plates having a thickness of 30 mm, and was homogenized for 3 h at 1200 °C, immediately hot rolled to 6 mm at temperatures between 1100 and 750 °C and then water quenched. The as-hot rolled plates were intercritically annealed at elevated temperatures and then either air (AC), furnace (FC) cooled or water quenched (WQ) prior to mechanical testing and microstructural characterization. shows the detailed heat treatment process of the steel samples. The annealing temperatures were chosen to observe the transformation behavior of austenite at different temperatures and cooling rates (i.e. temperature conditions at highest FCC and/or C content etc.). The main objectives of the heat treatments were firstly to maximize the amount of retained austenite and then, to control the stability of retained austenite. Finally, cold rolling was also conducted on the annealed samples to observe the deformation behavior of the austenite.The microstructural examinations were initially carried out on a Nikon L150 optical microscope, Bruker D8 X-ray difractometer and Philips XL-30 field emission scanning electron microscope (FE-SEM). All samples were mechanically polished down to 0.05 μm with colloidal silica and then chemically etched using a solution of 2% Nital followed by 10% aqueous sodium metabisulfite (Na2S2O5). The etchants used in this study have been widely used for AHSS steels to differentiate ferrite, austenite and martensite. So it is indeed required to differentiate them before analyses The volume percentage of the phases were calculated with the help of Clemex Captive image analyze software and an XRD intensity correlation method The BAHR DIL 805 quench dilatometer was used to evaluate the transformation behavior and to confirm the martensite start (Ms) temperature of steel composition at fast cooling rates.Finally, mechanical tests were performed using an MTS hydraulic machine by using 100 kN load cell. Tensile test samples were machined according to ASTM E-8 sub-size standard with a strain rate of 0.1 mm/s.According to the FactSage predicted phase diagram (), the equilibrium structure at room temperature of this composition has no austenite. However, this composition and the above heat treatments generated a microstructure of ferrite plus austenite duplex (FADP) microstructure. shows the morphology of the FADP structure after the two different heat treatment conditions. The first generation AHSS, relied on bainite to help retain (metastable) austenite, thus leading to an equiaxed ferrite plus bainite structure, with the retained austenite largely adopting morphologies associated with the bainite. In this structure, no bainite is required to retain the austenite, and the structure is comprised of equiaxed austenite and ferrite grains. Since the predicted equilibrium microstructure at room temperature has no austenite, the austenite in this structure is assumed to be unstable., the grain sizes of the ferrite and austenite are more or less comparable to each other, with the ferrite appearing to be a little coarser than the austenite grains. The microstructures also showed equiaxed austenitic grains with a limited number of annealing twins and very small amount of martensite in the austenitic phases, but apparently with more annealing twins after air cooling. One of the most interesting points of these microstructures is that there is no significant effect of cooling rate; the retained austenite levels are very similar (about 50 vol%) even after furnace cooling, indicating that the kinetics of the transformation to ferrite are relatively slow. Particularly striking is the absence of martensite after water quenching, indicating the martensite start (Ms) temperature is well below room temperature.In order to estimate the Ms of the austenite formed at the intercritical temperature, the retained austenite composition in the WQ condition (i.e., as-quenched) was measured by EPMA-WDS from the microstructure in . The partitioning of C and Mn to austenite can be seen in Unfortunately, there is no empirical equation in the literature to predict Ms from the austenite composition that has been created for medium Mn steels. Moreover, all such empirical equations have been developed by experiments that begin with 100% austenite, whereas the transformation that is of interest here is from the two phase ferrite plus austenite condition. Nevertheless, the above composition was used to determine the Ms temperatures according to the following equation The Ms temperature is predicted to be −199.8 °C, therefore the austenite is expected not to transform martensite during quenching to room temperature. The transformation behavior of austenite to martensite was also confirmed with dilatometer experiments at very high cooling rates () but no sign of any transformation was observed.Having noted the above insignificant effect of cooling rate on the amount of austenite it might be argued that this austenite is stable and not unstable as predicted by FactSage. Indeed, in contrast to the FactSage predictions, the ThermoCalc studies of De-Cooman et al. show that the pseudo-binary Fe–C phase diagrams of 10 Mn% steels have single phase austenitic area at elevated temperatures and stable austenite can be seen at room temperature conditions However, after cold rolling, all the microstructures exhibited significant reduction in retained austenite, as measured by XRD (). Since it can be assumed that this was due to strain induced transformation (SIT) to martensite, this indicates that the austenite is unstable (i.e. retained). In the case of the differences of the austenite prediction, there is no obvious kinetic issue that could explain transformation of austenite during room temperature deformation if it is not unstable.Furthermore, cold rolling not only transformed approximately half of the retained austenite to strain induced martensitic, but also generated twinning, as is clearly observed both optically (). The only other reference to the simultaneous observation of twinning and SITing in retained austenite that the authors have found is by Timokhina et al. It is clearly of interest to understand exactly why the retained austenite in the FADP steel can undergo both twinning and SIT during deformation, whereas the retained austenite in TRIP steels undergoes only strain induced transformation and the retained austenite in TWIP steels undergoes only twinning. However, in all previous studies, the microstructures that undergo twinning are 100% austenite which is energetically stable, e.g., TWIP and Hadfield steels Equally important is to determine when twinning and SIT take place during deformation, i.e., the evolution of the microstructure with strain. In fact, there are almost certainly three deformation mechanisms operating: (i) slip, (ii) SIT, and (iii) twinning. In terms of modeling the events, if these mechanisms are treated purely as ways to relieve the applied stress, then the obvious approach is to use the concept of critical resolved shear stress via the Schmidt factor. This approach is well known for slip and twinning; the concept of critical stress is also used in SIT Another way to consider the problem is by a classic Gibbs energy approach; this is commonly used to explain TRIP shows a schematic diagram of the effect of composition on the free energies of these two phases. If the diffusion of solutes is allowed, the two phase assemblage is delineated by the common tangent between these curves, which leads to the two co-existing phases having different compositions; this does not occur in this case because both SIT and twinning are diffusionless ‘transformations’, and the compositions of twinned and SIT phases are therefore the same. Therefore, in order to satisfy the co-existence of twinned austenite (γtwin) and SIT phase (α1), the Gibbs energies of the two phases should be the same or, in practice, very similar for the existing composition of the retained austenite. This scenario maybe feasible as twinning and tripping are mechanisms at high (about 20%) and low (2%) Mn concentrations, respectively. EPMA analysis Since it is accepted that there is a strong effect of Mn and C on the SFE and transformation kinetics during deformation, one other reason why both SIT and twinning are seen in the retained austenite may be due to an inhomogeneous carbon distribution within the retained austenite. This could lead an incomplete transformation of retained austenite and these regions of austenite tend to transform to martensite or twinning at lower strains Engineering stress–strain curves are displayed in summarizes their mechanical properties. Firstly, the WQ condition leads the highest total elongation of more than 30%, which is significant enhancement for medium Mn steel compositions, and it was balanced with a relatively high UTS value of 868 MPa. Annealing with AC led to slightly higher UTS more than 900 MPa but lower ductility values of 23%. In case of FC the total elongation was compatible with AC but at a much lower strength value because of the softer and coarser grain morphology., increasing amount of transformation increases strength, which is basically due to increasing strain induced martensite. However, increasing transformation does not lead to increasing ductility; this demonstrates that increasing SIT increases ductility only if it postpones strain instability. In this case, it appears that SIT may take place well before strain instability, thus contributing little to ductility. In fact, increasing ductility correlates better with the increasing retained austenite, which might be due to FCC being fundamentally more ductile than ferrite or martensite.It is also interesting to note that the ductility of the as-quenched specimens is slightly higher than the air cooled ones for a given RA vol%. This may be related to the fact that in the WQ specimen, the intercritical austenite was formed at higher temperature, thus leading to a lower carbon content The effect of twinning has not been considered but it is expected enhance the mechanical properties. Unfortunately there is not enough information to quantify the separate contributions of twinning and SIT in the microstructure.Fe10Mn generated a ferrite and retained austenite duplex structures (FADP microstructures) after an intercritical anneal, regardless of whether the specimen was quenched or air cooled.The microstructure comprised of significant amount of retained austenite (about 50 vol%) and an equiaxed structure.Cold rolling significantly reduced the retained austenite by strain induced transformation to martensite, and simultaneously twinned some of the retained austenite.Despite the similarity between the microstructures, the mechanical properties of all these steels are slightly different between each other. The results clearly showed that the ductility of these 10% Mn steels increases with the retained austenite fraction of the microstructure, and the strength increases with the increase of transformed strain induced martensitic volume fraction.Poly(arylene-ethynylene) with tuned rigidity/flexibility as reinforcing component in polystyrene-based ionomer blendsPoly(pyridylene/phenylene-ethynylene) polymers of controlled rigidity/flexibility were synthesized and used as molecularly dispersed reinforcing components in ionomer blends with partially sulfonated polystyrene. Homogeneous ionomer blends were formed due to acid–base interactions between the two blend components. The complete protonation of the poly(pyridylene/phenylene-ethynylene) in the acid/base ionomer blends has been proven by UV–VIS absorption and emission spectroscopy. Furthermore, the complete miscibility of the blend components in the ionomer blends was revealed from DSC analysis and transmission electron microscopy. The mechanical properties of the synthesized ionomer blends were determined by stress–strain measurements. The Young modulus of the blends was found to systematically vary with the rigidity/flexibility of the reinforcing polymer i.e. the molecular conformation of the reinforcing polymers as determined by small angle X-ray scattering.Polymer–polymer molecular composites are an attractive approach to composite materials with superior ultimate properties as compared to macroscopic fiber reinforced composite materials A requirement for miscibility in polymer–polymer blends is a sufficiently negative heat of mixing , compatibility may also be achieved if the interaction between the two blend components is sufficiently strong.By introducing specific ionic interactions, e.g. acid–base interactions between the two components . Similar observations have been made for the compatibility of semiflexible and flexible polymers where Coulombic or H-bonding promoted the miscibility The molecular miscibility generates a perfect molecular dispersion of the reinforcing polymer in the matrix (), which is the ideal case of molecular reinforcement, especially if the reinforcing polymer comes close to a true rod as in our previous studies Semiflexible liquid crystalline rigid rod polyesters or polyamides have also been successfully applied as the reinforcing component, utilizing the excellent but anisotropic mechanical properties of the LCPs for a mechanical reinforcement of isotropic and mechanically weaker engineering plastics As the material properties of a molecularly reinforced composite material obviously depend on the rigidity/flexibility of the reinforcing polymer, it is of significant interest to determine to which extent the mechanical properties of a molecularly reinforced polymer/polymer composite material can be controlled by tuning the overall stiffness of the reinforcing polymer. This not only provides information for the design of new reinforcing polymers but also allows to estimate if and/or to what extent semiflexible worm-like macromolecules which are known and already employed in plastic materials would be suited as blend components after having been provided with suitable groups capable of strongly interacting with the matrix.In this paper, we report the blending of novel poly(pyridylene/phenylene-ethynylene) PPyPE polymers of tuned rigidity/flexibility ( with partially sulfonated polystyrene and their application as reinforcing component in ionomerblends. The tuning of the possible conformation with regard to rigidity/flexibility is given by the ratio of para/meta-substituted pyridylene moieties in the macromolecule (mole fraction n/(n+m) and m/(n+m) in the chemical structure ). This is directly reflected from the systematic variation of the Flory exponent that was determined by small angle X-ray scattering ( representing constitutional features of the PPyPE polymers.The synthesis of poly(pyridylene/phenylene-ethynylenes) (PPyPE) of different rigidity/flexibility was carried out according to literature known procedures . Polymers of different rigidity/flexibility were obtained by systematically varying the ratio of 2,5-/3,5-dibromo pyridine. The ratio of the two pyridine isomeric moieties within the backbone of the copolymers was determined by NMR-spectroscopy. Since the hydrogen in 4-position to the nitrogen of the pyridine ring generates signals, which are in case of a meta- or para-linkage well enough separated from each other, the signal intensities allow the quantitative determination of the ratio of the two pyridine isomeric moieties. This is described in detail elsewhere ) have comparable molecular weights as determined by vapor pressure osmometry and gel permeation chromatography ). Since for the PPyPE polymers, the conventional GPC–polystyrene(PS)-calibration does not give the true molecular weights, the number average molecular weight of the PPyPE polymers was determined by vapor pressure osmometry, and these data together with the model compound 1 have been used for correct calibration.Partially sulfonated polystyrene (PS-co-SSH, 11 mol% arylene sulfonic acid) was used as the flexible matrix polymer. Homogeneous acid–base-ionomer blends were obtained by dropping a chloroform solution of the PPyPE (concentration between 0.2 and 0.7wt%) under stirring into a chloroform solution of PS-co-SSH (concentration between 1.2 and 1.8 wt%) up to equivalent amounts of the acid/base-functionalities and collecting the ionomer blend precipitate. All blends were dried at 100 °C for 2 days under vacuum.Transmission electron microscopy (TEM) studies have been conducted on a LEO 912 Ω electron microscope with an accelerator voltage of 120 kV. Microtomed samples were prepared from compression molded specimen (170 °C, 5 min). The experimental procedure was similarly as described elsewhere Stress–strain measurements were carried out with an Instron 4301 (Automated Materials Testing System) in a temperature controlled environment. Measurements were conducted at 150 °C. Samples were again prepared by compression molding (170 °C, 5 min) as were the samples for TEM investigation.Small angle X-ray scattering was performed on beamline 1–4 of the Stanford Synchrotron Radiation Laboratory (SSRL) at the Stanford Linear Accelerator Center (SLAC), in Stanford, CA. The facility offers a focused, collimated X-ray source with a flux of 1010 photons on a spot size of ∼0.5 mm (vertical)×1 mm (horizontal), monochromated by a 1 1 1 Si crystal to a wavelength of λ=1.488 Å. For the solution scattering experiments, spectroscopic grade tetrahydrofuran (Aldrich) was used without further purification. The sample cells were filled with 1% weight concentration solutions. Further details about the sample preparation and data collection in the SAXS experiments are described elsewhere Since the pattern and type of linkages in a polyarylene molecule determine its overall conformation, macromolecules based on arylene moieties in the constitutional unit are ideally suited models with designed and controlled rigidity/flexibility.By varying the ratio of para to meta linkages of the pyridine moiety incorporated in the poly(arylene-ethynylene) (), molecules of different rigidity/flexibility are obtained. A rigid rod molecule is generated if only para linkages are present; the introduction of meta linkages provides a kink, and with increasing number of kinks the structure will ultimately be reminiscent to a flexible coil molecule. The degree of polymerization of all poly(arylene-ethynylene) employed in this work was relatively low on purpose: by keeping the contour length below the persistence length (which based on poly(arylene) data is estimated to be in the range of 13–20 nm Considering the phenylene constitutional unit with p-ethynylene-linkages, the molecule with all m-linkages at the pyridine moiety may also be described as multiple broken rod with pyridylene–ethynylene–phenylene–ethynylenen–pyridylene adopting a coil conformation In the blend formation of the poly(pyridylene/phenylene-ethynylenes) with sulfonated polystyrene from solution, quantitative proton transfer from the styryl sulfonic group to the pyridylene group takes place resulting in a poly(pyridylenium/phenylene-ethynylene) multiplication and polystyrene sulfonate polyanions (The complete protonation of the poly(pyridylene/phenylene-ethynylene) in the acid/base ionomer blends with sulfonated polystyrene has been proven by UV–VIS absorption and emission spectroscopy. As expected from their conjugated structure, the optical properties of the PPyPE polymers depend on the electronic properties of the conjugated arylene–ethynylene-π-electron system and therefore on the extension of the conjugated system Model studies applying a model compound 1 representing characteristic constitutional features of the polymers () and p-toluene sulfonic acid were performed in order to determine the degree of protonation achievable within the PPyPE polymers. A solution of 1 in deuterated chloroform was treated with p-toluene sulfonic acid and the resulting change in the chemical shifts of the pyridylene-proton signals initiated by the protonation of the pyridylene–nitrogen was monitored by NMR-spectroscopy . The absorption and emission spectra steadily change with the protonation of the pyridylene-unit; the spectrum of the 100% protonated 1 distinctly differs from the spectrum of the non-protonated species.Partially protonated 1 displays spectra composed of the corresponding spectrum of the protonated and the non-protonated species with the ratio being related to the obvious equilibrium of the two isolated chromophores. In this context it has to be mentioned that here, in contrast to the NMR experiment, a higher number of equivalents of acid has to be added to obtain the fully protonated 1; this is to some extent an effect of different sample concentrations and will be described in detail elsewhere Since the PPyPE polymers contain the same constitutional unit as 1, optical spectroscopy can be also applied the determine the degree of protonation of the PPyPE polymers. The change in the poly(pyridylene/phenylene-ethynylene) absorption and emission spectra in chloroform solution upon addition of p-toluene sulfonic acid is exemplarily illustrated for the rigid PPPyPE by the series of spectra depicted in Again, this model study shows that the absorption and emission spectra steadily change with the protonation of the pyridylene moiety, and that the spectrum of the protonated polymer distinctly differs from the spectrum of the non-protonated polymer. As observed for the model compound 1, the spectra of the partially protonated polymer species is composed of the spectra of the corresponding protonated and non-protonated species, indicating the presence of an equilibrium between two isolated species The polymer behaves as if it is composed of isolated chromophores Since the PPPyPEH+/PS-co-SS− ionomer blends are insoluble in common solvents, the comparable optical characterization of the blends had to be performed in solid state. The emission spectrum of the solid film of the PPPyPEH+/PS-co-SS− ionomer blend gives a single fluorescent band only (λmax,E=563 nm with λexc=420 nm) which distinctly differs from the emission spectrum of the bulk PPPyPE (). Based on the results obtained for the model compound 1 and the PPPyPE in solution as described above, this is a clear spectroscopical evidence that complete proton transfer has taken place during the ionomer blend formation (stoichiometry of pyridylene and sulfonic acid groups); the same results have been obtained for the other systems where the poly(pyridylene/phenylene-ethynylenes) of tuned rigidity have been employed with the only difference that the absorption as well as emission band maxima varied with the mole fraction of m-linkages The complete miscibility of the blend components in the ionomer blends with sulfonated polystyrene (PPyPEH+/PS-co-SS−) as compared to a phase-separated blend with pure polystyrene (PPyPE/PS) has been first investigated by DSC analysis. The comparison between the DSC curves of the PPPyPEH+/PS-co-SS− blend (curve 3a/b, with the curves of the employed blend components (curves 1 and 2a/b, ) shows for the ionomer blend a single glass transition at around (curve 3b, second heating) which is in between the glass transition of the PS-co-SSH matrix polymer (curve 1; ) and the PPPyPE reinforcer polymer (curve 2b, second heating; ); due to the matrix-rod ionic interactions, the Tg temperature-range of the ionomer blend is broader than that of the pure matrix polymer (compare curves 1 and 3, For comparison and in order to clarify the necessity of rod-coil intermolecular ionic interactions for molecular blending, the DSC curve of the blend of PMPyPE with PS-co-SSH is contrasted to the DSC-curve of the mixture of PMPyPE with polystyrene (PS). Again, the PMPyPEH+/PS-co-SS− ionomer blend exhibits a single glass transition (curve 4b, second heating, ) only, which is in between the glass transitions of the blend components (matrix polymer, curve 1, ). In contrast to this, the DSC curve of the PMPyPE/PS blend shows multiple transition features (curves 5a/b, ) to be expected for an immiscible blend: The DSC trace is reminiscent to the various phase transitions of the PMPyPE polymer (Tg around −10 °C and endothermic transitions at higher temperature, compare curves 6a/b, ) and exhibits also a the strong enthalpy relaxation peak of the glass–rubber transition of polystyrene (around 106 °C, compare curve 5a and curve 7, ). The multiple transition peaks of the PMPyPE polymer phase in the phase-separated PMPyPE/PS blend are attributed to the melting of different crystallites of partially crystalline PMPyPE microphases which may be related to constitutional and chain length distribution effects; this has been discussed elsewhere This microphase separated system of poly(pyridylene/phenylene-ethynylene) domains dispersed in PS matrix which results from the lack of interacting groups and thus the incompatibility of the blend components has been visualized by TEM As it is already inferred from the DSC traces of the PPPyPEH+/PS-co-SS− and PMPyPEH+/PS-co-SS− ionomer blends, the high resolution TEM bright field images of microtomed specimen showed no evidence of heterogeneity, indicating the complete miscibility through attractive ionic interactions (; the granular texture does not reflect a heterogeneity but is typically obtained for homogeneous amorphous materials observed in phase contrast The molecular dispersion of the poly(pyridylene/phenylene-ethynylene) polycations in the polystyrene sulfonate matrix was further revealed from nitrogen net element specific image (ESI) TEM (). The bright areas on black background represent enrichment of the element nitrogen in a nitrogen-free matrix. Since the element nitrogen is only present in the poly(pyridylene/phenylene-ethynylene) blend component, these bright areas can only be associated with molecularly dispersed poly(pyridylene/phenylene-ethynylene) molecules; this is in agreement with the angström-to-nanoscopic dimensions of the size and shape of the polycations. In this context it is interesting to note that the ESI(N)-TEM of the PPPyPEH+/PS-co-SS− ionomer blend infers anisotropic orientation of the rod-like polycations () which is not seen for the coil-like PMPyPEH+ polycations in the PMPyPEH+/PS-co-SS− ionomer blend (). Such a texture would be in accordance with theory which predicts anisotropic one phase ionomer blends when the volume fraction of the rod-component (of a given length i.e. aspect ratio) exceeds a certain value The reinforcement in the polymer–polymer composites due to the rigid rod PPPyPE or the semiflexible, copolymers PCPyPE is evident from the comparison of the stress–strain curves of the sulfonated polystyrene matrix polymer PS-co-SSH and the blends with poly(pyridylene/phenylene-ethynylene) (. The values obtained for the Young modulus E of each blend are also given. Due to the brittleness of the materials at ambient temperature, stress–strain measurements were conducted at 150 °C, i.e. above the glass transition temperatures Tg of the polymer blends and blend components.With the rigid rod poly(para-pyridylene/phenylene-ethynylene) in the blend, the elastic modulus increased from 0.28 MPa of the sulfonated polystyrene matrix polymer (curve 1, PS-co-SSH) to 7.7 MPa (curve 6, PPPyPEH+/PS-co-SS−), while for the blend with the relatively flexible coil-like poly(meta-pyridylene/phenylene-ethynylene) (curve 2, PMPyPEH+/PS-co-SS−) no increase of the modulus as compared to the modulus of the pure matrix polymer was detected. This enhancement of the mechanical properties can directly be related to the stiffness of the reinforcing polymer, since the formation of an ionomer network which might also contribute to the overall reinforcement is present in both cases.This view further implies that the varying reinforcement effect as determined for the other samples containing the poly(co-pyridylene/phenylene-ethynylene) copolymers (PCPyPE) of varying m-pyridylene linkages (see ), could be related to differences in the stiffness of the poly(pyridylene/phenylene-ethynylene) blend components. Consequently considering the Flory exponent being correlated to the fraction of para-linkages In order to correlate the anisotropy and therefore the rigidity/flexibility of the reinforcing PPyPE polymer with the mechanical properties of the reinforced ionomer blend, the Halpin–Tsai equations The aspect ratios of the PPPyPE and PMPyPE homopolymers as well as of the PCPyPE copolymers as compiled in were obtained by using literature known bond length data . In this context, it is also worthwhile to mention that the glass transition temperature of the poly(pyridylene/phenylene-ethynylenes) scales (decreases) with the content of m-pyridylene-linkages in the same way Further studies investigating the elongation induced orientational ordering of the reinforcing PPyPE polymers of different rigidity/flexibility within the ionomer blends were performed applying SAXS. Ionomer blend samples were strained to about 300% elongation and the degree of orientational ordering induced was monitored. While for the not strained samples no orientational ordering was observed indicating isotropy, the strained ionomer blend samples displayed orientational ordering phenomena. The degree of orientational ordering varied hereby systematically with the rigidity/flexibility of the PPyPE reinforcing polymer: with the ionomer blend containing the rod-like polymer PPPyPE display the highest order of orientation while for the ionomer blend with the coil-like polymer PMPyPE no orientational ordering was observed. Since the degree of orientation of the reinforcing polymers was measured at high elongation values while the Young-Modulus was determined at low elongation values, i.e., in a practically unoriented state of the reinforcing macromolecules, the correlation between molecular architecture and bulk material properties is meaningful The study of the series of rod-coil ionomer blends based on poly(pyridylene/phenylene-ethynylene) of tuned rigidity/flexibility as blend component with sulfonated polystyrene has demonstrated that the mechanical properties of a molecularly reinforced blend material can be controlled by varying the rigidity/flexibility of the reinforcing polymer component. Stress–strain measurements showed, that a significant reinforcement effect of almost a factor of 30 was achieved when the blends were made with rigid rod polymer but the mechanical properties of the flexible matrix polymer remained unchanged upon blending with the flexible Poly(pyridylene/phenylene-ethynylene). The Young modulus of the blends was found to systematically vary with the rigidity/flexibility of the reinforcing polymer.Information about the molecular conformation of the reinforcing polymers in solution obtained by small angle X-ray analysis, allowed to directly relate the change of the mechanical properties of the ionomer blends to the overall chain stiffness, i.e. to the rod/coil character of the reinforcing polymers.These findings open interesting perspectives for the design of novel materials with controlled mechanical properties based on molecular rod/coil ionomer blends; this will be emphasized in future studies.Quantitative relationship between anisotropic strain to failure and grain morphology in additively manufactured Ti-6Al-4VThe aim of this study was to identify processing-microstructure-mechanical property links in additively manufactured Ti-6Al-4V. First, the microstructure and mechanical properties of Ti-6Al-4V produced via two laser powder bed fusion (LPBF) additive manufacturing (AM) methods, one using a pulsed laser (P-LPBF) and the other a continuous-wave laser (CW-LPBF), were investigated and compared. Second, existing data from the literature were integrated with the present data in order to identify a general quantitative relationship between anisotropic ductility and grain morphology in additively manufactured Ti-6Al-4V. This revealed that an exponential relationship exists between the anisotropic grain morphology and anisotropic elongation to failure in Ti-6Al-4V. In particular, this relationship shows that a prior-β grain aspect ratio (grain height to grain width) exceeding 6 results in significant anisotropy in elongation. Namely, the columnar grains dominate the fracture mechanics by furnishing significant damage accumulation paths for tension in the longitudinal direction, resulting in higher ductility in the build direction than that in the longitudinal direction. With respect to processing, it was shown that as-built CW-LPBF samples had nearly equiaxed grains while those made by P-LPBF had elongated columnar grains. This resulted in greater yield strength, ultimate tensile strength, and ductility in the CW-LPBF samples compared to P-LPBF samples.Powder bed fusion (PBF) additive manufacturing (AM) of metals uses a laser or electron beam to selectively melt powder metal feedstock to build 3-dimensional (3D) components in a layer-by-layer fashion During processing, the thermal gradient between the melt pool and previously solidified layers, which depends on processing parameters, geometry of the component, and location within the component, results in a location-dependent thermal history throughout the component Laser based PBF systems use either continuous-wave (CW) laser emission Previous research has been performed to investigate the mechanical properties of additively manufactured Ti-6Al-4V in the build (transverse) and longitudinal directions. These studies have shown that the strength in as-built samples is nearly isotropic, while the elongation to failure is higher in the transverse direction than the longitudinal direction The present work focuses on identifying quantitative links between the anisotropy in ductility and the microstructure in additively manufactured Ti-6Al-4V. The microstructure, namely the prior-β grain morphology, and tensile mechanical properties of samples fabricated using CW-LPBF and P-LPBF were investigated. Additionally, studies from literature were analyzed to quantify grain morphology and its impact on anisotropic ductility. From these data, a quantitative correlation describing anisotropic tensile ductility based on grain morphology is developed. Additionally, this study presents the effect of processing conditions on the microstructure and tensile mechanical properties of Ti-6Al-4V fabricated via CW-LPBF and P-LPBF. The effect of microstructure (CW versus pulsed), direction (anisotropic microstructure in CW versus pulsed), and surface roughness (macroscopic structure in P-LPBF) on properties was investigated.To identify the difference of the impact of CW laser versus a pulsed laser processing in PBF on microstructure and properties, Ti-6Al-4V samples were fabricated using both of these laser heat sources. For CW-LPBF (EOSINT M280), 32 mm × 30 mm × 4 mm walls were fabricated on a 280 mm × 280 mm annealed Ti-6Al-4V substrate in an Ar filled chamber to minimize oxygen contamination. Samples were fabricated using the CW laser parameters in . Prior to sample removal from the substrate, the entire build was subjected to a standard PBF stress relief in Ar at 650 °C for 3 h. Uniaxial tension samples were extracted from the deposited walls in the horizontal (longitudinal) and vertical (transverse) directions using wire electrical discharge machining (EDM). The tensile sample geometry used was in accordance with ASTM E8 Conversely, samples made using P-LPBF (Renishaw AM 250) were fabricated directly in the uniaxial tension sample geometry shown in b, which was also in accordance with ASTM E8, with a gauge length of 21.3 mm and a cross sectional area of 8 mm2. Samples were fabricated in two orientations, transverse and longitudinal, on a 250 mm × 250 mm substrate in an Ar filled chamber with the baseplate heated to 170 °C throughout the build process. The fabrication parameters used for the near-net-shaped P-LPBF samples are specified in . Although there was a smaller gauge region in the CW-LPBF samples, the sample dimensions incorporated the same order of magnitude of grains for mechanical testing to be representative of bulk properties. For P-LPBF samples, ten samples were fabricated such that their tensile axes were in the transverse direction, and ten were fabricated such that their tensile axes were in the longitudinal direction. The grips of the longitudinal samples were fabricated directly on the substrate, while support structure was used in the gauge region to maintain sample geometry during fabrication. Prior to sample removal from the substrate, the entire build was subjected to the same stress relief heat treatment as the CW-LPBF samples. Samples were removed from the substrate using wire-EDM, and subjected to a grit blasting to remove un-melted powder. Half of the samples in each orientation were left in this as-built condition for tensile testing, while the gauge regions of the other half of the samples were milled to determine how the surface finish impacted the measured properties.To quantify the mechanical properties that result from CW-LPBF versus P-LPBF processing, samples from each build were subjected to tensile testing. Uniaxial tensile testing of the P-LPBF samples was performed on an electromechanical testing frame (Instron 4202) with a 10 kN load cell (Instron model 2518-804). Uniaxial tension testing of the smaller CW-LPBF samples was performed on a servo-hydraulic load frame (810 MTS) with a 25 kN load cell (MTS model 661 20E-01). All tensile tests were performed under quasi-static conditions with a strain rate on the order of 10-4

s-1. Digital image correlation (DIC), a non-contact method for measuring surface deformations, was used to compute surface strains using correlation software (Vic2D, Correlated Solutions). Sample gauge regions were painted with a white basecoat followed by a random black speckled pattern on top. A digital camera (Point Grey GRAS-50S5M-C) was used to take images of the deforming gauge region of the sample at 1 Hz during each test. The surface deformations in the gauge region of each sample were computed from the digital images using a cubic B-spline interpolation algorithm with a subset size of 21 pixels and a step size of 5 pixels. The axial strain in the gauge section of each sample was measured using a vertical virtual extensometer measuring 5.5 mm for the CW-LPBF samples and 20 mm for the P-LPBF samples.Microhardness was measured using a Vickers indenter (Leco MHT Series 200) with a load of 300 g and a dwell time of 15 s. At least nine locations at varying distances from the substrate were analyzed for microhardness, with five indentations at each height, in both P-LPBF and CW-LPBF samples.The Archimedes method for determining density was used to quantify porosity. The sample density was computed as:where Ma is the measured mass of the sample in air, Mw is the measured mass of the sample in water, and ρw is the density of water, which was assumed to be 1.0 g/cm3. Calculated densities were compared with the theoretical density of Ti-6Al-4V, 4.43 g/cm3For microstructural analysis, samples were prepared using standard metallurgical procedures with a final polish using 0.05 µm colloidal silica, and etched using Kroll's reagent (2 vol% hydrofluoric acid and 3 vol% nitric acid in distilled water). Images of the microstructures were taken using a digital optical microscope (OM, Keyence VHX-2000). The digital OM was also used for 3D surface reconstruction of the samples for surface roughness analysis.To quantify the grain dimensions in the present study and from reported data, selected micrographs were overlaid with a five by five grid pattern aligning with the longitudinal and transverse directions. Grain dimensions were computed based on how many grain boundaries each line intersected. Due to the elongated morphology of grains, the line lengths were dictated by grain boundary locations, so that a line started on a grain boundary and ended on another. Measurements were made for each individual line three times for a total of fifteen measurements in each orientation. To ensure proper statistics of measurements, this procedure was repeated for each micrograph by five individuals. The twenty-five grain dimensions per orientation were used to determine the average width/height of the grains for each micrograph.The measured mechanical properties for all samples are given in , and representative engineering stress-strain curves of CW-LPBF, as-deposited P-LPBF, and machined P-LPBF samples under uniaxial tension are shown in . The variations in both ultimate tensile strength (UTS) and elongation to failure indicate that differences in processing conditions and surface finish influenced the bulk material properties. In general, the CW-LPBF samples had higher yield strength (0.2% offset), UTS, and elongation to failure in both orientations tested compared to the corresponding machined P-LPBF samples. In the same direction, CW-LPBF samples had higher elongations than the as-built and machined P-LPBF samples. The average yield strength of CW-LPBF samples was 8.3% higher, and the average UTS 6.7% higher, than that of the machined P-LPBF samples.Microhardness as a function of vertical position from the substrate was measured to determine if the mechanical properties, and indirectly, microstructure, had a location dependence that would impact the properties within the gauge regions of the samples. In the CW-LPBF samples, the centers of the gauge region for the longitudinal samples and transverse samples were approximately 5 mm and 13 mm from the baseplate, respectively, and microhardness measurements were taken at heights from 3 mm to 24 mm from the baseplate. In P-LPBF samples, the center of the gauge region for the longitudinal and transverse samples was approximately 5 mm and 32 mm away from the baseplate, respectively, and microhardness measurements were taken at heights from 1 mm to 9 mm away from the baseplate. As shown in , the microhardness as a function of distance from the substrate was found to be constant by the time the gauge region was reached in all samples. Thus, the gauge regions are sufficiently far from the baseplate such that the microstructure has reached a steady state in all of the tested samples.The average hardness in the CW-LPBF samples was 403 ± 8 HV compared to 375 ± 7 HV in the P-LPBF samples. These values agree with existing literature, for which the hardness of Ti-6Al-4V fabricated by PBF has been reported to be greater than traditionally processed Ti-6Al-4V (341–369 HV) X-ray CT scans revealed that neither CW-LPBF nor P-LPBF samples contained lack-of-fusion porosity. The representative as-built longitudinal and transverse P-LPBF samples examined with the Archimedes method showed that the samples were 99.7 ± 1.3% and 98.7 ± 2.2% dense, respectively. The CW-LPBF deposited wall had a density of 98.2 ± 1.9% measured via Archimedes method. These analyses indicate samples were near fully dense with only gaseous, spherical porosity present.The tensile mechanical properties of samples were measured in the longitudinal and transverse orientations with respect to the build layers to determine the degree of anisotropy of components made by PBF. As shown in , samples fabricated via CW-LPBF exhibited isotropic yield strength and UTS, but anisotropic strain to failure, with a higher elongation in the transverse direction. Near isotropic yield strength and UTS behavior was observed in the milled P-LPBF samples, and similar to CW-LPBF samples, the elongation in the milled P-LPBF samples was higher in the transverse direction than in the longitudinal direction.The isotropic strength behavior in both sets of samples can be explained by the low strain hardening behavior in Ti-6Al-4V where σ is the true stress, εp is the plastic strain, k is the strength coefficient, and n is the strain hardening exponent. The strain hardening exponents for CW-LPBF and P-LPBF samples were found to be 0.06 and 0.08, respectively. The very low strain hardening rate in both sets of samples, in which the yield strength was isotropic, resulted in isotropic UTS despite anisotropy in elongation to failure. We note that although the prior-β grain morphologies differ, which impacts elongation, as discussed in , the plasticity behavior in Ti-6Al-4V is dictated by the α lath morphology and preferred orientation To explain the differences in the mechanical properties of Ti-6Al-4V samples manufactured using CW or pulsed lasers, the microstructures were analyzed, and the prior-β grain dimensions were quantified. Prior-β grains that grow epitaxially across several build layers in AM have been widely reported in literature , respectively. There is a clear difference in prior-β grain size and morphology between the two processing methods. The microstructure of the sample made via CW-LPBF () contains small, nearly equiaxed prior-β grains, with average measured widths of 96.3 ± 18.0 µm and lengths (in the build direction) of 125.3 ± 14.4 µm, and the prior-β grains contain acicular α laths. The samples made with P-LPBF () also have prior-β grains containing acicular α-laths; however, the prior-β grains are elongated and extend across multiple build layers, with the measured dimensions of 150.3 ± 22.7 µm wide by 1201.8 ± 190.2 µm long.In welding research, it has been shown that the morphology and size of grains are controlled by the relationships between thermal gradient (G) and the solidification growth rate (R) of as-deposited material In CW-LPBF, with increasing scanning speed, the melt pool becomes elongated behind the laser spot, resulting in a decreasing thermal gradient, increasing solidification growth velocity at the melt pool boundary, and decreasing G/R ratio The laser in P-LPBF build does not simultaneously supply energy to the build and vary position, but rather supplies bursts of power at discrete locations. This results in a higher thermal gradient at the boundary of the melt pool compared to a continuously scanning laser, and is similar to spot welding ), which has also been observed in other studies for builds with large G/R ratios To quantitatively compare the two microstructures, the grain aspect ratio, or grain height divided by grain width, was computed for CW-LPBF and P-LPBF samples, as well as microstructures in prior literature, as tabulated in . In the CW-LPBF samples, this aspect ratio was 1.3, indicating nearly equiaxed grains, while grains in the P-LPBF samples had a much higher aspect ratio of 8.0. This significant difference is attributed to the disparate thermal histories of these samples, which were dictated by different energy inputs, different scanning approaches, and different thermal conditions during fabrication, as described in Carroll et al. found that prior-β grain boundaries, which contain semi-continuous α phase, act as damage accumulation sites under load We propose using the grain morphology to describe, and potentially predict, anisotropic elongation behavior in Ti-6Al-4V. To elucidate any quantitative links between the anisotropic microstructure and the anisotropic ductility in Ti-6Al-4V made by AM, the elongation to failure ratio, defined as the strain to failure in the transverse direction divided by that in the longitudinal direction, versus the grain aspect ratio, is plotted in , for data reported in the literature and the present study. Each data point included comes from a study that reported elongation in the transverse and longitudinal directions as well as a micrograph showing complete prior-β grains. Studies in which the grain lengths were not fully encompassed in the micrographs were not included, as the grain dimensions could not be fully quantified. Additionally, in order to isolate the effect of grain morphology only, data for which the elongation in the transverse direction was less than that in the longitudinal direction, resulting in an elongation ratio less than 1, were excluded due to presumed or identified internal defects. In a fully-dense Ti-6Al-4V component, where grains grow vertically across build layers, the elongation in the transverse direction should be greater than that in the longitudinal direction due to the preferential damage accumulation along prior-β grain boundaries The data of anisotropic elongation versus grain aspect ratio in were fitted with an exponential curve passing through the point (1.0, 1.0), corresponding to an isotropic sample. This resulted in the following empirical description of the elongation ratio (y) as a function of grain aspect ratio (x):This expression can be used to describe and predict the strain to failure ratio between two directions in additively manufactured Ti-6Al-4V based only on the prior-β grain aspect ratio. These data show that grain aspect ratios below about 6 do not lead to significant anisotropy in elongation, as there is enough of a distribution in grain sizes in those samples to dilute the impact of anisotropic grains. However, once the grain aspect ratio exceeds 6, the anisotropy in elongation becomes significant as the elongated grains dominate the mechanics by furnishing significant damage accumulation paths when tension is applied in the longitudinal direction.With respect to the present samples, when the P-LPBF samples are loaded in tension along the longitudinal direction, the grains are perpendicular to the loading axis, and are separated by grain boundaries, composed of grain-boundary α. Therefore, the grains are pulled apart in Mode I fracture opening fashion As noted in prior literature, surface roughness of as-built samples, due to both the layered fabrication process as well as partially melted particles adhered to sample surfaces, can result in an overall decrease in the elongation to failure of the material Here, the effect of surface finish was examined for the P-LPBF samples. The layer stepping effect was noticeable on the surface of these as-built samples, and the impact of surface finish on properties is seen when comparing the elongation of the as-built and machined P-LPBF samples in . The transverse as-built samples had an elongation of 6.5 ± 1.0%, which increased to 7.8 ± 1.1% upon milling. Surface roughness was also apparent in the longitudinal P-LPBF samples due to necessity of incorporating support structures for their fabrication, and as such, the elongation to failure increased in these samples from 1.4 ± 0.2% to 2.8 ± 0.8% upon surface machining.A quantitative assessment of the surface roughness was made by analyzing surface line profiles generated from OM 3D reconstruction images taken on the thin edge of the gauge region of the P-LPBF tensile samples, as shown in . The data show that milling reduced surface roughness (Ra) from 33.90 ± 5.51 µm to 20.71 ± 3.50 µm for the transverse samples. Milling was most effective for the longitudinal samples that were built with support structure, reducing the Ra value from 144.31 ± 13.54 µm to 20.71 ± 3.50 µm. As milling removed the stress concentrations due to surface roughness, the elongation increased.Additionally, undulations in the cross-sectional area due to the layer-by-layer fabrication process, as well as partially melted particles on the surface, of as-built samples result in the overestimation of continuous load bearing cross-sectional area, and therefore, the underestimation of the sample strength . This correction technique may be used for computing representative strength values for as-built samples, or for comparing strength data among studies in the literature with disparate surface finishes.In the present paper, the anisotropic microstructure in additively manufactured Ti-6Al-4V was quantitatively linked to the anisotropic ductility using samples manufactured with a continuous-wave laser PBF system, a pulsed laser PBF system, and data from literature. Additionally, a method for comparing strengths measured for samples with different surface roughness conditions was presented. The following primary conclusions can be drawn from this study:For fully dense, near defect-free Ti-6Al-4V fabricated using AM, the elongation to failure was determined to be anisotropic with prior-β grain aspect ratios greater than 6. In these samples, the elongation was higher in the transverse direction than longitudinal direction. The quantitative relationship proposed herein to link microstructure to anisotropic elongation can be used to predict the anisotropic elongation to failure behavior in additively manufactured Ti-6Al-4V with knowledge of only prior-β grain morphology.The variations in AM processing conditions studied here resulted in variations in grain morphology, and consequently mechanical properties. The P-LPBF samples contained prior-β grains with a high aspect ratio induced by a high G/R ratio. This led to anisotropic elongation. The CW-LPBF samples had nearly equiaxed grains due to a low G/R ratio, resulting in nearly isotropic elongation behavior.The finer grains in the CW-LPBF samples compared to the P-LPBF samples resulted in higher ultimate tensile strength, yield strength, and elongation in the CW-LPBF samples.Correcting for surface roughness in as-built P-LPBF samples through calculation of a representative load-bearing cross-sectional area can be used to compare strengths of as-built samples with rough surfaces to those of samples with machined surfaces reported in the literature.Temperature dependence of mechanical stressMechanical stress in atomic-layer deposition (ALD)-Al2O3 films was investigated at room temperature and during thermal cycling up to 870 °C. The films were generally under tensile stress. Thicker films (25–60 nm) showed a sharp stress increase at about 780–790 °C. X-ray diffraction (XRD)-, X-ray reflectance (XRR)- and X-ray photoelectron spectroscopy (XPS)-measurements indicate an irreversible phase transition from amorphous AlO(OH) to a mixture of different crystalline Al2O3-phases. Annealing at higher temperatures leads to a stress reduction as a result of diffusion and recovery processes. The stress behaviour of thinner films (<20 nm) during thermal cycling is quite different. Tensile stress increases with increasing temperature and decreases to nearly the same value during cooling down. The process is continuous and reversible.Temperature dependence of mechanical stressAluminium oxide is a potential high-k material for different semiconductor devices, for instance, as a dielectric layer in DRAM trench structures. Mechanical stress measurements in such films after deposition and especially during and after thermal treatment provide information about thermal mismatch between substrate and film, densification and recovery processes, phase transitions, etc. Because changes in mechanical stress can be influenced by changes in defect densities and vice versa; such investigation can be useful for the interpretation of electrical data Aluminium oxide films (thickness 3–60 nm) were deposited on 8 in.-Si(1 0 0)-wafers (with SiO2−- or SiNx-interfaces) by atomic-layer deposition (ALD) with trimethylaluminum (TMA) as metal-precursor and ozone or water as oxidation precursors at temperatures between 200 and 500 °C.Stress values were calculated following the well-known Stoney's law (Eq. where Es is the Youngs modulus, νs the Poisson ratio and ts is the substrate thickness. Wafer curvature radii R0 and Rf (before and after deposition) were determined with an Eichhorn + Haussmann (MX-208) at room temperature as well as with a Tencor Flexus (laser beam reflection) in situ from room temperature up to 870 °C. We measured the film thickness tf with TEM and ellipsometry. For the interpretation of the results X-ray diffraction (XRD), X-ray reflectance (XRR)- and X-ray photoelectron spectroscopy (XPS)-measurements were added.After deposition, the films were generally under tensile stress. For film thicknesses from 5 to 20 nm stress values increase slightly, whereas for higher thickness (up to 60 nm) they are nearly constant. We found decreasing tensile stress with increasing deposition temperature (500 MPa at 200 °C deposition temperature to 50 MPa at 500 °C). No significant influence of different SiO2- and SiNx-interfaces on the film stress could be observed.) show that the as-deposited films were completely amorphous and their density (XRR-measurements) was 3.04 ± 0.05 g/cm3, which is close to the value of 3.01–3.06 g/cm3 for Bohemit (α-AlO(OH)) The temperature cycling measurements were focused on the time, temperature and thickness dependence of the stress values. Thicker films (60 nm) exhibit nearly constant tensile stress during heating up and a sharp increase (+800 MPa) at temperatures slightly below 800 °C (, squares). This transition temperature is lower than described in After annealing, we detected a shrinkage of the film thickness by about 10% and found with XRR that the film density increases to (3.32 ± 0.05) g/cm3. This value is higher than the density of the low-temperature crystalline γ-Al2O3 (3.2 g/cm3), but lower than for other crystalline Al2O3 phases (θ, η, δ, κ: 3.4–3.76 g/cm3, see b) confirms the crystallization after annealing, but the single diffraction peaks could not be explained by a single crystalline phase. The results indicate a mixture of different crystalline Al2O3-phases (γ, θ, δ, κ). The average grain size was 10 nm for films thicker than 25 nm.The phase transition is also visible in a shift of the Al-2p binding energies (XPS) from 75.0 to 73.2 eV, but these values could not be assigned to particular aluminium oxide/hydroxide phases.Annealing at temperatures higher than 800 °C leads to stress reduction by diffusion and recovery processes (a, triangles). During cooling down the tensile stress increases nearly linearly and reaches values of about 1300 MPa at room temperature. This could be contributed to the thermal mismatch between the substrate and the film.Multiple temperature cycling procedures show that the jump of the stress at about 780 °C is irreversible. All stress values were on the top curve (a, rings) during further heating-up and cooling-down procedures.Decreasing film thickness changes the stress behaviour during thermal cycling (b–d). The tensile stress increases continuously during heating up to rather high values (>2500 MPa). For 25 nm films the sharp phase transition can be observed at about 800 °C and the behaviour is quite similar to a 30 nm Al2O3-film as discussed in The stress development during cooling down is completely different. The thinner the film is, the more similar are the stress values during heating up and cooling down. Thin films (<20 nm) show a reversible stress-temperature behaviour even after multiple cycles.XRD-measurements on thin Al2O3-films (tf

< 10 nm) c and d) up to 850 °C just reaches low crystallinity and not the above-mentioned irreversible crystalline state. Further structure investigations seem to be necessary to understand the stress behaviour.ALD-Al2O3 films were generally under tensile stress after deposition. Thicker films (25–60 nm) showed a sharp stress increase at about 780–790 °C. XRD-, XRR- and XPS-measurements indicate that an irreversible phase transition from an amorphous phase (possibly Bohemit (α-AlO(OH)) to a mixture of different, partly high-temperature crystalline Al2O3-phases (χ, η, δ, κ) takes place. The volume reduction as a result of the phase transition is the main reason for the increase in tensile stress. Annealing at temperatures above 800 °C leads to a stress reduction as a result of diffusion and recovery processes (grain growth and vacancy annihilation).The stress behaviour of thinner films (<20 nm) during thermal cycling is quite different. Tensile stress increases with increasing temperature and decreases to nearly the same values during cooling. The process is continuous and reversible.Steel-reinforced concrete-filled circular steel tubular (SRCFT) stub columnCompressive behavior of steel-reinforced concrete-filled circular steel tubular stub columnsSteel-reinforced concrete-filled circular steel tubular (SRCFT) stub columns is a novel type of composite columns, which have a great potential to be used as piers or columns in practical engineering. Hence, it is essential to comprehend the compressive performance of SRCFT stub columns and suitably predict the compressive strength for the engineering design and applications. This paper aims to investigate the compressive behavior of SRCFT stub columns through combined experimental and numerical studies. A total of 8 specimens were tested to investigate the compressive performance of SRCFT stub columns in detail, in terms of the axial load-strain curve, ultimate bearing capacity, ductility, strength-weight-ratio and strain ratio. The SRCFT stub columns exhibited better ductility than CFT stub columns. The inserted steel section can effectively prevent shear cracks in the core concrete from propagating quickly. Furthermore, finite element model was established and verified by comparing the experimental and FE results. Then, the complex composite action among the steel tube, core concrete and steel section was discussed and clarified comprehensively. In addition, the capacity of the energy dissipation for SRCFT stub columns was discussed. Finally, a novel simplified formula was proposed to predict the ultimate bearing capacity of SRCFT stub columns. The studies may provide a considerable reference for designing this type of structures in engineering practice.Steel-reinforced concrete-filled circular steel tubular (SRCFT) stub columnRatio of initial tangent modulus to secant modulus at peak stressInitial equibiaxial compressive yield stress of concreteUniaxial compressive strength of concreteInitial uniaxial compressive yield stress of concreteUltimate bearing capacity of SRCFT stub columns from calculated resultsUltimate bearing capacity of SRCFT stub columns from experimental resultsUltimate bearing capacity of SRCFT stub columns from FE resultsWall thickness of steel tube or steel sectionAxial compressive stress of core concreteAxial compressive stress of steel sectionRadial concrete stress of the confined areaAxial strain when the load attains of 75% the ultimate load in the pre-peak stageStrain when experimental bearing capacity is decreased to 85% of ultimate valueStrain corresponding with the peak compressive stress of concreteSteel-concrete composite structure such as concrete-filled steel tube (CFT) column have been widely used as modern structure members in building structures and bridges So far, extensive experimental and numerical studies However, with the increase of the bridges span and buildings height in the practice engineering, the cross area of composite column is often designed bigger to meet the higher requirement of bearing capacity, ductility and stiffness. For example, the diameter of CFT column in the first story of the Shenzhen Saige Plaza Building even reached 1600 mm. Such a large cross section may result in reduced the useful indoor space . Compared with previous studies as mentioned above, limited studies have been conducted on the performance of SRCFT column. Wang et al. With the advancement of computing techniques to conduct research more efficiently and economically, the finite element analysis is becoming more and more popular to simulate the behavior and predict the response of steel–concrete composite structural, because which can simulate the case that are difficult and/or complex. Chang et al. Therefore, the objective of this paper is to rationally investigate the compressive behavior of SRCFT stub column, and to develop a more concise and precise formula to calculate the ultimate bearing capacity accordingly. More specifically, based on the experimental and numerical research from our team In order to study the compressive performance of SRCFT stub columns, a total of 8 specimens were designed to test, including 4 circular concrete filled steel tubular (CFT) stub columns and 4 steel-reinforced concrete-filled circular steel tubular (SRCFT) stub columns, where the effect of concrete strength and steel section are considered in this study. The details of specimens are listed in , where D, t and L represent the diameter of the circular section, the wall thickness of steel tube and steel section and the height of the specimen, in mm, respectively. The circular steel tubes were all manufactured from Q235 steel plates. Butt welds were used according to the standard GB 50017-2003 For better observation and record of deformation and local failure of the steel tubes, red paint was sprayed on the outer surface of the steel tubes and 50 mm × 50 mm white grids were plotted on the painted surface. Before pouring concrete, the cover plate was welded to one end of the steel tube and then make the circular steel tube erect. The steel tube was adjusted carefully in order to the steel section lies in the core of the steel tube. Concrete was placed from the top of the specimens, and carefully vibrated using a vibrator to distribute the concrete evenly inside the tube. Finally the upper surface of concrete was smoothed for being at the same level with the steel tube. Meanwhile, the standard concrete cubes with a dimension of 150 mm were prepared and cured under the same condition as the concrete used in specimens. The concrete surface of the column specimens was polished with grinder after the concrete was hardened. And then a layer of epoxy resin binder was daubed to level the end surface. After that, the steel cover plates were welded to the end of the specimens to ensure that the load was applied to the steel tube, core concrete and steel section, simultaneously.Before tests, the mechanical testing, including the steel and concrete, were conducted to obtain the respective mechanical properties according to the corresponding standard methods. Tensile coupon tests were conducted to obtain the material properties of the steel tube and steel section used in the specimens in accordance with the standard GB/T228-2002 . Besides, the cubic compressive strength (fcu) of concrete was obtained from the testing of the concrete cubes according to the standard GB/T50081-2002 All compressive tests were conducted by hydraulic testing machine of 2000 T in National Engineering Laboratory for high-speed railway construction technology of Central South University. To accurately measure the deformation, four strain rosettes were glued on the external surface of the steel tube at the mid-height and two LVDTs were installed at the mid-height of the external surface, as shown in . Besides, the DH3818 static strain measurement system, electronic transducers and data acquisition system were employed to collect the axial load vs. strain curves and axial load-deformation curves, respectively.All specimens were tested under the monotonic static loading, and the load was applied on the top of the specimens. Firstly, the load is increased at a step of 1/20 of the expected ultimate load in the elastic stage. Secondly, the load was applied to the specimens by means of displacement control with an increment of 0.2 mm after the load reached about 60% of the expected ultimate bearing capacity. Each loading step took 3–5 min. After reaching the ultimate load, the specimen was loaded slowly and continuously at a step of 0.5 mm and maintained 5 min, and the data is recorded continuously. Finally, the tests were stopped when the axial strain reached 0.04 which was the largest strain of the specimens. The whole loading period for each specimen was about 2 h.Based on the test observation and obtained axial load vs. strain curves of all the specimens, as shown in , the specimens under axial loading were considered to experience three stages until failure: elastic stage, elastic–plastic stage and post-peak stage.Elastic stage: At the initial loading stage, all the specimens were in elastic phase, indicating that the axial load increased linearly with the increase of the displacement. The compressive stiffness of specimens in this stage was larger than that in others stage, consequently the axially elastic displacement was very small. In this stage, there is no evident confinement effect of the steel tube on the core concrete.Elastic-plastic stage: The steel tube firstly began to yield, and then the axial load vs. strain curves diverged significantly from its initial linearity when reaching about 60% of the ultimate load. At the same time, the outer steel tube produced the confinement effect on the core concrete. In this stage, the visible local buckling of the steel tube initially appeared near to the upper end of the specimens due to the end effect, and later on generated in the middle where local bucking developed sharply. When the imposed load reached the ultimate loading capacity, an apparent buckling was observed on the steel tube.Post-peak stage: The imposed load decayed sharply as the displacement continuously increased, as shown in . This is mainly attributed to the crush of the core concrete and the further bucking of the steel tube. Herein, it should be noted that the curves of FCST1and FCST2 specimens declined slower (with bigger ductility) than those of C1 and C2 specimens (with smaller ductility). The comparison results indicated that the steel section embedded in the core concrete can effectively improve the ductility of specimens. gives the failure modes of the typical tested specimens. As can be seen from , obvious difference of the failure modes between CFT stub columns (C1 and C2 specimens) and SRCFT stub columns (FCST1and FCST2 specimens) were observed.For C1 and C2 specimens, obvious local buckling was observed at different heights on the opposite sides of the steel tube, and there was a visible shear plane occurred between the local buckling, as shown in . The outer steel tube was cut open and removed from the core concrete after test. It should be notable that there were inclined shear rupture zones or/and even crushes in the core concrete, as shown in . The results indicated that the outer steel tube can’t effectively prevent the generation of shear sliding crack in the core concrete.For FCST1 and FCST2 specimens, several local buckling were observed along almost the same height, and there was no obvious existent shear failure plane along the specimen height. Meanwhile, the most obvious bulges almost formed a ring in the steel tube, as shown in . When the tests were stopped, the outer steel tube was removed from the core concrete. It can be found that the core concrete in the bulges zone were crushed, but the core concrete remained its entirety due to the confinement effect of the outer steel tube and steel section, as shown in . Therefore, it can be concluded that the inserted steel section can substantially prevent shear cracks in the core concrete from propagating rapidly, and change failure modes of composite columns, where the effect of the steel section may be seriously considered for columns in the future design.The presence of the steel section is a special character of SRCFT stub columns and a marked difference compared with common CFT stub columns. Hence, the role of the steel section is discussed in detail in this section. gives the experimental results for ultimate bearing capacity of all the specimens. The steel section is inserted into CFT stub columns, while other parameters are kept the same as mentioned above. Compared to C1 specimen, the average ultimate bearing capacity of FCST1 specimen was improved by 32%. Meanwhile, in comparison to C2 specimen, the average ultimate bearing capacity of FCST2 specimen was improved by 11%. Therefore, the steel section is an effective way to enhance the ultimate bearing capacity, especially for SRCFT stub columns with low-grade concrete.The concrete strength is another key parameter affecting the behavior of SRCFT stub columns. The concrete strength was the variable and other parameters were kept the same as mentioned above. Compared to C1 specimen, the average ultimate bearing capacity of C2 specimen was improved by 35% with a 46.1% increase of concrete strength. Besides, in comparison to FCST1 specimen, the average ultimate bearing capacity of FCST2 specimen was improved by 13% with a 46.1% increase of concrete strength. In summary, the ultimate bearing capacity increase with the increase of concrete strength.where ε0.85 is the axial strain when the load falls to 85% of the ultimate load, εb is equal to ε0.75/0.75. ε0.75 is the axial strain when the load attains of 75% the ultimate load in the pre-peak stage. shows comparison of the DI calculated by Eq. for all tested specimens, where a higher value (DI) indicates a slower process of load reduction after the ultimate state.Compared to C1 and C2 specimen, the average value (DI) of FCST1 and FCST2 specimen after inserting the steel section were improved by 156% and 130%, respectively. In addition, in comparison to C1 and FCST1specimen, the average value (DI) of C2 and FCST2 specimen were reduced by 64% and 58%, respectively, with the increase of the concrete strength from 39.3 MPa to 57.4 MPa.Therefore, it can be concluded that the inserted steel section can help to increase the ductility of specimens. Moreover, increasing concrete strength leads to a weaker ductility. Taking the ductility into consideration, the novel column with the steel section and high-grade concrete is suggested to be used in engineering.The concept of the strength-to-weight ratio (γ) is introduced to investigate the effect of key parameter on the compressive behavior of CFT and SRCFT stub columns in this section, which is defined as the ultimate bearing capacity (Nu) of columns divided by the weight (G) of columns. The large strength-to-weight ratio (γ) is, the larger the net floor space is. gives comparison of the strength-to-weight ratio (γ) for all tested specimens. From , it can be seen that the average value (γ) of FCST1 specimen was improved by 22% compared to C1 specimen, and the average value (γ) of FCST2 specimen was improved by 15% compared to C2 specimen. Besides, the average value (γ) of C2 specimen was improved by 35% compared to C1 specimen, and the average value (γ) of FCST2 specimen was improved by 15% compared to FCST1 specimen.In a summary, the ultimate bearing capacity of SRCFT stub columns is obviously greater than CFT stub columns under the same cross-sectional dimensions. What’s more, the SRCFT stub columns used in building could result in significant savings in material and increasing in net floor space., the measured steel ratio was nearly constant and approximately equivalent to Poisson’s ratio of steel in the early stage, indicating that the steel tube and core concrete worked independently, and almost no confinement effect was produced. And then, the steel ratio began to increase gradually and exceeded the Poisson’s ratio when the imposed load was up to 60% of the ultimate bearing capacity. The change in strain ratio indicated the steel tube produced a significant confinement effect on the core concrete.Finite element (FE) models are established using ABAQUS version 6.1 The loading plate is modeled as rigid bodies. A surface-based interaction with hard contact in the normal direction and the Coulomb friction coefficient of 0.5 in the tangential direction to the interface is used to simulate the interfacial behavior between steel tube and core concrete, in which the sliding formulation is finite sliding. A tie constraint may couple two separate surfaces so that no relative motion occurs between them. Therefore, the tie option is adopted for the constraint between the steel tube, core concrete and loading plate, core concrete and steel section to ensure that the load is applied to the specimen during the whole loading process.To model the descending stage of load-bearing capacity of specimens, load is applied by means of the specifying displacement at the upper loading end, where all degrees of freedom at both ends of the specimens are constrained except the axial displacement at the upper loading end. The structured meshing technique option is adopted to mesh to the FE models, which is easier to converge and achieve accurate simulation, as shown in The following stress–strain relationship of concrete under uniaxial compression suggested by Ding et al. y=Ax+(B-1)x21+(A-2)x+Bx2x⩽1xα(x-1)2+xx>1where y = σ/fc and x = ε/εc are the stress and strain ratios of the core concrete to the uniaxial compressive concrete respectively. σ and ε are the stress and strain of the core concrete. fc = 0.4fcu7/6 is the uniaxial compressive strength of concrete, where fcu is the compressive cubic strength of concrete. εc is the strain corresponding with the peak compressive stress of concrete, where εc = 383fcu7/18 × 10−6. The parameter A is the ratio of the initial tangent modulus to the secant modulus at peak stress and equals to 9.1fcu-4/9. B = 1.6(A-1)2 is a parameter that controls the decrease in the elastic modulus along the ascending branch of the axial stress–strain relationship. For a concrete-filled steel tubular stub column, parameter α can be taken as 0.15. More information of the concrete model could be referred in Ding et al. In this paper, the concrete damaged plasticity model available in the material library integrated in ABAQUS was employed to investigate the compressive behavior of SRCFT stub columns. The dilation angle (θ), the eccentricity (e), the ratio of initial equibiaxial compressive yield stress to initial uniaxial compressive yield stress (fb0/fc0), the ratio of the second stress invariant on the tensile meridian to that on the compressive meridian (K) and the viscosity parameter (v) need to be determined. According to the relative research results According to a great number of experimental studies on the material properties of steel, an elasto-plastic model, with consideration of Von Mises yield criteria, Prandtl-Reuss flow rule, and isotropic strain hardening, was used to describe the constitutive behavior of steel. The expression for the stress–strain relationship of the steel was described as follow Ding et al. σi=Esεiεi⩽εyfyεy<εi⩽εstfy+ζEs(εi-εst)fy+ζEs(εi-εst)εst<εi⩽εufuεi>εuwhere, σi and εi are the equivalent stress and strain of the steel. fy, and fu (=1.5 fy) are the yield strength and ultimate strength respectively. Es (=2.06 × 105 MPa) and Est (Est = ζEs) are the elastic modulus and strengthening modulus. εy, εst and εu are the yield strain, hardening strain, and ultimate strain of steel, which is described by εu = εst + 0.5 fs/(ζEs), εst = 12εb, εu = 120εb and ζ = 1/216.Both material and structural nonlinearities were considered and solved using the incremental-interactive method in ABAQUS.According to the material constitutive models, interaction, and boundary condition as mentioned above, the accuracy of numerical analysis results from established FE models are standardized against the corresponding experimental results, in terms of the axial load vs. strain curves and ultimate bearing capacity. As is well known, the concrete material model proposed by Han et al shows the comparisons of the axial load vs. strain curves between experimental and corresponding FE results, reflecting the deformed shapes of specimens at each stage. And then, previous experimental data that the overall predicted curves agree reasonable with the experimental results, especially for elastic stage and elastic–plastic stage. In addition, the accuracy of test curves were difficult to maintain after the steel tube yielded or/and the core concrete were crushed, while FE models were calculated in a perfectly ideal case. And therefore, slight differences in Post-peak stage are observed between the experimental and FE results. Secondly, the comparison of the ultimate bearing capacity between experimental and FE results is listed in . Good correspondence between experimental and FE results is achieved in general and the discrepancies are less than 10% for all the specimens. And last, it should be noted that the constitutive model of concrete proposed by Ding et al. To sum up, the FE modeling approach and material constitutive model To deeper understand the difference of the compressive performance between SRCFT stub columns and CFT stub columns, the composite action between the steel tube and core concrete, core concrete and steel section is analyzed comprehensively in this section. The following will take C1 and FCST1 specimens as examples to discuss the composite action between the steel tube, core concrete and steel section. gives the axial and transverse stress vs. strain curves for the steel tube and steel section at different points. According to Ding et al. (a), it should be noticed that the intersection of CFT stub columns appears earlier than that of SRCFT stub columns. That is to say that the confinement effect of the steel tube on the core concrete for SRCFT stub columns is slightly reduced due to the addition of the steel section. (b) demonstrates that the intersection of the axial and transverse stress vs. strain curves appears at point C of the steel section, while no intersection is observed at points A and B. The results clearly indicate that the largest confinement effect of the steel section on the core concrete is optimal at point C, followed by point B and point A. From another point of view, (c) gives the stress vs. axial strain curves of the steel section at different points. The stress of the steel section at point A, B and C is slightly higher than that of the constitutive curve of steel tube, indicating that the limited interaction between the steel section and core concrete is produced, but to different extents. The results from (d) gives axial stress vs. strain curves of the core concrete on the half height of specimens. It is obvious that the compressive strength of core concrete is much higher than the plain concrete, because of the confinement effect of the steel tube on the core concrete. For C1 and FCST1 specimens, the shapes of the axial stress vs. strain curve of the core concrete are quite similar, and yet the curve of FCST1 specimen declines slower after reaching the ultimate bearing capacity than that of C1 specimen, indicating that the inserted steel section mainly influences the post peak behavior of the core concrete, and its effect on the enhancement of concrete strength can be neglected.The axial load vs. strain curves of the specimens and different components are shown in (f), where the red point in the curves represents the peak point of different components, respectively. It can be seen that both the steel tube and steel section has yielded before reaching the ultimate bearing capacity. That is to say that different components work together under axial loading, whereas the ultimate bearing capacity of the components is reached successively at different time.Based on the validated FE modeling approach, numerical analysis are conducted to extensively investigate the effect of some key parameters on the compressive behavior of SRCFT stub columns, including the concrete strength covering from C40 to C100, steel strength ranging from 235 MPa to 420 MPa, steel ratio from 0.02 to 0.08, steel section, and confinement index. The following steel and concrete were paired for the specimens: Q235 steel were paired with C40 and C60 concrete, Q345 steel were paired with C60 and C80 concrete, and Q420 steel were paired with C80 and C100 concrete. The following will take the specimens with D = 300 as examples.(a) illustrates the axial load (N) vs. strain (ε) curves with different concrete strength, where the concrete strengths varies from C40 to C100, while other parameters are kept the same. It can be observed from (a) that with the increase of concrete strength, the ultimate bearing capacities are improved by 25.1% (from C40 to C60), 18.8% (from C60 to C80) and 15.5% (from C80 to C100), while the ductility decreases. In addition, there is a limited difference in the elastic stage. The Concrete strength contributes to increase the ultimate bearing capacity.(b) gives the axial load (N) vs. strain (ε) curves of SRCFTs with varied steel strength, where the steel strengths are 235 MPa, 345 MPa and 420 MPa, respectively, while other parameters are kept the same. From (b), it can be found that the ultimate bearing capacities are improved by 10.0% (from 235 MPa to 345 MPa) and 4.9% (from 345 MPa to 420 MPa) with the increase of the steel strength. And no obvious difference is observed at the initial stage. Therefore, the steel strength has moderate effect on the ultimate bearing capacity.The steel ratio (ρs) is a key factor for the behavior of SRCFTs according to extensive previous work, which is defined as the steel area (As) divided by the total cross-sectional area (Asc), as follows: ρs = As/Asc. (c) presents the axial load (N) vs. strain (ε) curves with various steel ratios, where the steel ratios are 0.02, 0.05 and 0.08, respectively. It can be seen from (c) that with the increase of steel ratio, the ultimate bearing capacities are improved by 20.7% (from 0.2 to 0.5) and 18.3% (from 0.5 to 0.8), and the corresponding compressive stiffness are improve by 10.9% and 10.5%, respectively. Therefore, the steel ratio has obvious effect on the behavior of SRCFTs under axial loading.(d) gives the axial load (N) vs. strain (ε) curves with different confinement index, where the confinement index ranges from 0.7 to 2.0, respectively. As can be seen From (d), the ultimate bearing capacities are obviously improved with the increase of the confinement index, and the curves after post-peak stage decay slowly and even rise gradually. In summary, the confinement index has significant effect on the compressive behavior of composite columns.(e) presents the axial load (N) vs. strain (ε) curves with varied strength of the steel section, where the strength of steel section are 235 MPa, 345 MPa and 420 MPa, respectively. From (e), it can be found that the ultimate bearing capacities are improved by 6.0% (from 235 MPa to 345 MPa) and 3.8% (from 345 MPa to 420 MPa) with the increase of the strength of steel section. Besides, there is no difference on compressive stiffness in the elastic stage. To sum up, the strength of steel section has limited effect on the ultimate bearing capacity.(f) gives the axial load (N) vs. strain (ε) curves with different area of the steel section, where the area are 4.1%, 6.6% and 9.2%, respectively, while other parameters are kept the same. As can been seen from (f), the ultimate bearing capacities are improved by 15.9% (from 4.1% to 6.6%) and 9.4% (from 6.6% to 9.2%) with the increase of the steel ratio, and the corresponding compressive stiffness are improve by 10.2% and 9.8%, respectively. It can be concluded that the area of steel section has a very moderate effect on the behavior of composite columns.The presence of steel section is a special character of SRCFT stub columns and a marked difference compared with common CFT stub columns. Therefore, the effect of the steel section on behavior of SRCFT stub columns is investigated in detail in this section.(a) shows the effect of b/h on the axial load vs. strain curve of SRCFT stub column. For the case with different value of b/h, the ultimate bearing capacity and the shapes of curves are almost the same. It can be found that the b/h has no obvious influence on the behavior of SRCFT stub columns.The area of the steel section and the value of b/h are kept the same for this section, while the wall-thickness (t) of the steel section ranges from 3.7 mm to 18.5 mm. (b) gives the effect of wall-thickness (t) on the axial load vs. strain curve of SRCFT stub column. For the cases with different wall-thickness, both ultimate bearing capacity and the shapes of the curves are quite similar. The results clearly indicate that the wall-thickness of the steel section has negligible effect on SRCFT stub columns.The I-shaped steel section takes the place of the X-shaped steel section inserted into the circular steel tube, where the area of the steel section and the value of b/h are kept the same, while the wall-thickness (t) of the steel section ranges from 3.7 mm to 18.5 mm. (c) presents the effect of t of I-shaped steel section on axial load vs. strain curve of SRCFT stub column. For the cases with different t of I-shaped steel section, the ultimate bearing capacity and the shapes of the curves are quite similar. The results reveal that the wall-thickness of I-shaped steel section has no significant effect on SRCFT stub columns. Hence, the above comparison results clearly show that the effect of cross-section shapes of the steel section on behavior of SRCFT stub columns can be negligible and not considered in this study.where Ei, Ev, Eke, Efd are the internal energy, viscous dissipation, kinetic energy, frictional dissipation, respectively; Ew is the external work. For composite columns under axial loading in this study, Ev = Eke = Efd = 0. Therefore, the equilibrium equation of energy can be simplified as:The equation of the internal energy of SRCFT and CFT stub columns under axial loading can be expressed as below:where Ese and Epd are the elastic strain energy and plastic dissipation; Ecd and Eae are the creep dissipation energy and artificial strain energy; Edmd, Eqb and Eee are the damage dissipation energy, energy lost to quiet boundaries and electrostatic energy. For composite columns under axial loading, Ecd = Edmd = Eqb = Eee = 0. The equation of internal energy can be simplified as follow: shows the influence of the steel section on the energy dissipation of stub columns and its individual components, respectively. From (c), it is observed that there is no difference of the plastic dissipation and internal energy of the core concrete between C1 and FCST1 specimen before the axial strain reaches 0.025 and 0.02 respectively, and then two curves begin to deviate at this point, the curves of the core concrete of FCST1 specimen increases faster than C1 specimen due to the confinement effect of the steel section on the core concrete. However, the plastic dissipation and internal energy of the core concrete between C1 and FCST1 specimen is the same.(b) clearly demonstrates that the elastic strain energy of the core concrete of FCST1 specimen is obviously greater than C1 specimen. But, the proportion of the elastic strain energy in the total internal energy is very limited. It can be seen from (d) that the total internal energy of FCST1 specimen is significantly higher than C1 specimen. Therefore, the results indicate that the total internal energy, plastic energy and elastic strain energy of the stub columns and its individual components are improved but to different extents due to the existence of the steel section embedded into the core concrete. gives the influence of the concrete strength on the energy dissipation of stub columns and its individual components, respectively. As can be seen from , all the plastic energy, elastic strain energy and internal energy of the core concrete for FCST2 specimen is improved with the increase of the concrete strength from 39.3 Mpa to 57.4 Mpa, while the plastic energy, elastic strain energy and internal energy of the steel tube and steel section almost remains the same. Besides, it can be found from (d) that the total internal energy of FCST2 specimen is obviously improved compared to FCST1 specimen. It can be included that concrete strength help to increase the energy dissipation of composite column, but lead to a weaker ductility.From the above analysis, it can be found that the steel section is almost under uniaxial loading, and the mechanical properties of SRCFT stub columns are consistent with those of CFT stub columns. After that, the SRCFT stub column can be regarded as a combination of CFT stub column and steel section. Therefore, the practical formula for the ultimate bearing capacity of SRCFT stub columns can be deduced based on the limit equilibrium method through combined the strength calculation formula of CFT stub columns When the ultimate bearing capacity (Nu) is imposed on the end of specimens, the core concrete in the circular section endures axial compressive stress (σL,c) and radial stress (σr,c), while the steel tube bears the axial compressive stress (σL,s) and transversal stress (σθ,s), the steel section only bear the compressive stress (σL,g). Based on the static equilibrium criterion, as shown in (a), the ultimate bearing capacity (Nu) of SRCFT stub columns in the mid-height region can be determined by:where Ac, As and Ag represent the cross-sectional area of the core concrete, steel tube and steel section, respectively.The axial stress of the steel section under uniaxial loading can be given by:Firstly, provided that the axial compressive stress (σL,c) of the core concrete is provided by Ding et al. where p is the coefficient of lateral pressure, p = 3.4 according to Ding et al. (b), the relationship between the radial concrete stress (σr,c) of the core concrete and transversal stress (σθ,s) of the steel tube can be expressed:where ρ is the steel ratio of SRCFT stub columns, ρ = As/Asc, Asc = Ac + As + Ag.According to the Von Mises yield criterion for the steel tube, the following equation can be obtained as below:where Φ is the confinement index and is determined by with Eq. (14), the resulting ultimate capacity of the stub column Nu can be expressed as:Nu=Acfc1+p-Φ′Φσr,cfc+Φ′2-3Φ′2Φ2σr,cfc2+Agfgwhere Φ’ is the nominal confinement index and is determined by, the maximum ultimate capacity of the stub column Nu can be expressed as:It should be noted that when steel section area (Ag) is taken as 0, the Eq. considering the effect of the steel section is not only suitable for predicting the ultimate bearing capacity of SRCFT stub columns, but also can be applied to calculate the compressive strength of CFT stub columns.Comparisons of the ultimate bearing capacity between the experimental (Nu,c), FE (Nu,FE) and corresponding predicted results (Nu, Exp) using different formulas are summarized in , where the ratios are values of experimental results divided by the FE and corresponding predicted values. The average ratios, coefficient of variation, maximum and minimum values are also listed in lists different formulas available in the literature and standards for predicting the compressive strength of SRCFT stub columns., it can be found that the average ratios of Nu, Exp /Nu, c are 1.06, 1.09, 0.92, 1.11 and 1.24, respectively. Besides, the statistical analysis reveals that the COV and max–min are 0.041 and 0.15, 0.063and 0.26, 0.074 and 0.26, 0.63 and 0.31, 0.060 and 0.30, respectively. When the average ratio is closer to 1, the predicted results is more accurate. In this paper, the predicted results using the Eq. ensured same accuracy level as the formula given by references is much simpler in applications compared with the other formulas by The comparison results clearly demonstrate that the proposed formula Eq. can more offer better predicted values of SRCFT stub columns. Hence, the proposed formula can be adopted as the basic formula to calculate the ultimate bearing capacity.This paper presented a combined experimental and theoretical study on the compressive behavior of SRCFT stub columns, in terms of complete curve analysis, failure mode, parametric studies, and composite action and so on. Based on the limited results from the current study, the following conclusions can be drawn:Based on the experimental phenomenon and measured axial load vs. strain curves for all specimen, the specimens generally is considered to experience three stages until failure, including elastic stage, elastic-plastic stage and post-peak stage. The failure mode of SRCFT stub column is mainly influenced by the inserted steel section, which can substantially prevent or delay shear cracks in the core concrete from propagating, and thus change failure modes. The SRCFT stub columns could show better ductility than CFT stub columns.Both concrete strength and steel section can affect the ultimate bearing capacity, ductility and strength-weight-ratio (γ) for SRCFT stub column, but to different extent. The overall parametric studies indicate that the ultimate baring capacity of SRCFT stub columns increase obviously with the increase of concrete strength, whereas the ductility decreases. Besides, the inserted steel section can help to enhance the energy reserve, in terms of the ultimate bearing capacity and ductility. What’s more, the strength-weight-ratio (γ) of SRCFT stub column is greater than CFT stub column, indicating that SRCFT stub columns used in practical engineering could reduce obviously material costs and increase more net floor space.The reasonable agreement between experimental and corresponding FE results is achieved generally, and therefore the FE modeling approach is adopted to further simulate the compressive behavior of SRCFT stub columns.After inserting the steel section, the confinement effect of the steel tube on the core concrete is slightly reduced, and yet the steel section provides limited more confinement effect on the core concrete. Both sides effect are minor and can be ignored. In addition, under the same steel ratio of the steel section, the crosses-section shape of the steel section has almost no influence on the compressive performance of SRCFT stub column.A simplified formula is derived to calculate the ultimate bearing capacity of SRCFT stub column based on the limit equilibrium method. The proposed formula can reasonably predict the ultimate bearing capacity of SRCFT and CFT stub column at the same time. What’s more, the proposed approach is much more concise and accurate to use in engineering practices. Therefore, this formula is recommended to be used for calculating the bearing capacity of SRCFT stub columns.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Effects of alkaline etching on the surface roughness of a fibre-reinforced epoxy compositeThe adhesion of metal coating to a fibre-reinforced epoxy composite is controlled by mechanical keying. Rough surfaces are prepared by chemical etching. Therefore mechanical properties could be degraded if the difference in reactivity of the composite components (fibres and matrix) is not taken into consideration. To eliminate this surface heterogeneity, a plating epoxy resin film is applied to the composite, which become chemically homogeneous. During the etching process, this plating film will protect the integrity of the composite. This study describes the influence of an alkaline etching solution on the surface geometric characteristics of two composites. It was found that the mean peak-to-valley height (Rz) is not sufficient to explain the surface modification. The technique of profilometric relocation is used for quantifying the etching effect on the surface micro-geometry.Fibre-reinforced epoxy composite has found and will always find diverse applications in a number of industries. However, radio and electronics, computer technology require metallized polymers. Whatever, the intended purpose of metallization may be, it is apparent that a metallized polymer will not perform its intended function unless the metal adheres tenaciously to the polymer substrate. Adhesion to untreated epoxy composite is poor and for applications requiring good adhesion, pretreatment is necessary. Many treatments have been reported to improve adhesion to polymers surfaces; these included corona and glow discharge, chemical attack with oxidant solutions and solvent etching.Sulphuric acid, chromic acid or alkaline permanganate etching solutions are widely used for conditioning of plastic surface for electroless plating The subject of this paper is the application of a profilometry technique in study and control of epoxy composites roughness during alkaline etching. Two coated epoxy composites, noted fibre-reinforced glass epoxy composite coated with an epoxy film ((G/E) + F) and fibre-reinforced glass epoxy composite coated with an epoxy powder coating ((G/E) + P), were investigated. The fibre-reinforced epoxy composites were damaged heterogeneously during the conditioning, due to the difference in reactivity of the matrix–fibre system. So this point at issue was avoided by the application of a non-conductive homogeneous coating onto the fibre-reinforced epoxy composite surface. In this configuration, etching affects only the coating roughness; the integrity of the composite was conserved.A methodology was developed to determine the only roughness induced during conditioning. The results are substantiated by 3D profilometry before and after etching to analyse specific roughness induced. It has been shown that etching immersion time and non-conductors composition affect strongly the topography of the surface.Fibre-reinforced epoxy resin (glass epoxy composite: G/E) cast into thick sheets, supplied by Dassault Electronique (now Thomson CSF Detexis), France, were used. The mechanical characteristics and surface properties of this composite were shown in Two interfaces were tested, noted “film” (G/E + F) and “powder” (G/E + P). Although the two interfaces were in epoxy resin, their application processes were different and confidential. The “film” was put during composite elaboration; the “powder” coating was deposited by fluidisation after composite elaboration.Samples were previously degreased in a soak cleaning (5 min at , etched in 100 ml of permanganate alkaline solution at 80 °C then immersed into oxidant sulphuric acid solution (2% vol H2SO4

+ 2% vol H2O2) for 5 min at room temperature and rinsed in deionised water. of sodium hydroxide was used. Etching times were selected according to the objectives of the experiments. The temperature of the solution was 80 °C in all cases.A conventional stylus (Federal-diameter=2.5 μm, load=100 mg) was used and the resulting electrical signal was converted to a 12 bit digital signal (4056 levels) and this constituted the input signal to a microcomputer. Two stepper motors were used to provide displacement in two perpendicular directions with a step of 20 μm. Relocation was chosen for etching induced surface topography modifications as a powerful tool to study surface asperities creating of a pre-selected characteristic surface area.In order to appreciate the sole roughness caused by the etching (characteristic roughness), the difference between the micro-geometries of the initial (or untreated) and final (or etched) composite surfaces was calculated.Each sample was controlled by the following procedure:Characterisation of the untreated interface (before surface conditioning) using profilometer; definition and localisation of the test area.Neutralisation. This step was followed by a thorough water rinse and drying.Characterisation by profilometry, of the relocalized test area, after chemical treatments. was represented by 200 traverses of 200 data points per traverse. Two different aspects of this scanning stylus profilometry were used in the present study: three-dimensional (3D) surface relocation for first-order visual analysis and statistical analysis for overall quantitative study.3D surface relocation provided a comparison of several 3D surfaces measured during an etching process at exactly the same place on the surface.In order to appreciate the differences between the micro-geometries of the composite surfaces, the statistical parameter Rz, mean peak-to-valley height, was used. Three other parameters were also determined for a complete description of the surface:the “hole proportion” (%), defined as the holes area over the entire sample,the surface mean of the holes (valleys),the number of holes on the etched surface. showed the 3D relocation of the two untreated surfaces (G/E) + F and (G/E) + P topographies. Pronounced surface roughness was observed for the epoxy composite coated with powder due to the elaboration process (fluidisation).The object of this work was the characterisation of the roughness of the material induced by the etching. This necessitated several steps:acquisition of the image obtained before and after etching,comparison of the two 3D-images with the software Toposurf,subtraction of the two images obtained (treated and untreated surfaces) to obtain the sole roughness induced by the alkaline etching.The influence of duration of alkaline plating on the behaviour of the two interfaces, “film” and “powder”, was studied by measuring the surface topography.For each duration of immersion, the mean values of the parameter Rz were calculated (). The effects of etch duration on the parameter Rz are shown in . Both composites underwent progressively more severe roughening during the first 20 min, despite the difference of their initial roughness. Their “etching kinetics” were similar and the slopes of the first part of the curves were, respectively, for the film. The composite powder (G/E + P) showed a -Rz increase after 20 min etch but no further degradation for a more prolonged attack. The behaviour of the film composite (G/E + F) was different. The film roughness, after 30 min etching, was more important than the powder one according to the value of RzEven if the parameter Rz (mean peak-to-valley height) was indicative of the surface roughness amplitude, three other parameters were also determined for a complete description of the surface it means to understand the action of the etchant:the “porosity” (%), defined as the holes area over the entire sample,the surface mean of the holes (valleys),the number of holes on the etched surface.

The action of the alkaline etch was achieved by the combinations of these three parameters defined as the “porosity at constant surface” and “porosity with constant holes number”. For example, they allowed to know if an increase of the roughness was due to an increase of the number of holes or an enlargement of the existing holes. The sum of the two curves “porosity at constant surface” vs time and “porosity with constant holes number” vs time corresponded theoretically to the curve “porosity” vs time. showed clearly the difference of the composites behaviour during alkaline etching. For the powder, an increase of the etching duration induced an increase of the number of holes even though an opening of the holes was observed for the film. illustrated these observations. Remarkable changes in surface topography were observed on etched film and powder epoxy composites. The film surface has been extensively etched producing very large, relatively deep pits as compared to the powder surface. showed the SEM micrographs of the untreated and permanganate alkaline treated (at 80 °C) epoxy composites. They confirmed qualitatively the results obtained by profilometric method.The technique of profilometric relocation was used to understand the etching effect, and based on this, a preferential attack in the valleys (holes) created was observed for the film composite (GE/F) and an enlargement of number of the valleys (holes) for the powder composite.The use of the parameter Rz, mean peak-to-valley height, as a guide to surface roughness was found not to be sufficient and porosity parameters were necessary to explain the topography.The present methodology, subtraction of the initial roughness, allowed the determination of the sole roughness induced by the etching. It could be used for other applications such as tribology, polishing, plating, etc.Core-shell rubber nanoparticle reinforcement and processing of high toughness fast-curing epoxy compositesTo simultaneously address the lower toughness and the build-up of internal heat for fast-curing epoxy matrices, the influence of nominal 100 nm and 300 nm core-shell rubber (CSR) particles on the properties and rheo-kinetics were studied. The fracture energy was enhanced by a factor of 14.5, up to 2572 ± 84 J m−2 with 14.5 wt% of the nominal 300 nm diameter CSR particles, with evidence of cavitation and plastic void growth of the rubber core combined with shear band yielding of the epoxy matrix. These toughening mechanisms were modelled with an approximately linear increase up to 10 wt% for both particle types. At higher concentrations, deviation between the measured and modelled data was observed due to insufficient epoxy to dissipate additional energy. The CSR particles were not filtered out or damaged during the manufacturing of composites and reduced the total heat of reaction with a linear correlation, demonstrating a multi-functionality of simultaneous toughening and reduction of the exothermic peak.The highly cross-linked structure of epoxy polymers results in high modulus, high strength, low creep and good performance at temperatures below the glass transition. However, this molecular structure also leads to a poor resistance to crack initiation and propagation. Hence, epoxy polymers typically have a low fracture energy (or toughness), reducing the durability and damage resistance of composite materials.The fracture energy can be increased by the addition of a second phase, such as rubber Adding rigid fillers, such as silica nanoparticles, can increase the stiffness and fracture energy Traditionally the toughness of epoxies has been increased using reactive liquid rubbers such as carboxyl-terminated butadiene acrylonitrile (CTBN). These rubbers are initially soluble in the uncured resin, then phase-separate during curing to form rubbery particles. However, it is well known that such phase-separation depends on the curing conditions, e.g. To address these problems, the use of pre-formed core-shell rubber (CSR) particles has been investigated Only a few works have addressed toughening of fast-curing epoxies, i.e. those with curing times of a few minutes In the present work, the epoxy has a curing time of 5 min at 100 °C. The strong exothermic reaction during this short cure may lead to a significant overshoot of the resin temperature during cure The aim of this study was to investigate the influence of two types of core-shell rubber particles on the fracture properties and processability of a fast-curing epoxy for fibre composites. The kinetic and rheological properties were measured and carbon fibre reinforced composite plates were manufactured using compression resin transfer moulding (CRTM) A diglycidyl ether of bisphenol A (DGEBA) epoxy resin with an epoxy equivalent weight (EEW) of 181.5 g eq−1, XB 3585 from Huntsman Advanced Materials, Switzerland, was used. The curing agent was a mixture of diethylenetriamine and 4,4′-isopropylidenediphenol, XB 3458 from Huntsman Advanced Materials, Switzerland, which was used at a stoichiometric ratio of 100:19 by weight of epoxy to hardener.Two different core-shell rubber (CSR) particles, MX156 and MX960, were supplied by Kaneka, Belgium, pre-dispersed at 25 wt% in a DGEBA resin with an EEW of 243 g eq−1. The particles are distinguished by the core material and size. The MX156 has a polybutadiene core and a nominal particle diameter of 100 nm, while the MX960 has a siloxane core and a nominal particle diameter of 300 nm Carbon fibre non-crimp triaxial [45/0/-45] preform fabrics (759 g m−2, Toray T620SC 24k fibres) from Saertex, Germany, were used as a layup of [(45/0/-45/-45/0/45)2]s as reinforcement in the manufactured fibre composites.Bulk epoxy plates which were 6 mm thick were cast into 8 mm thick, preheated and release agent coated aluminium alloy moulds. The plates were cured in an oven at 80 °C for 12 min. This slightly slower cure cycle was used to prevent the strong exothermic reaction and decomposition, which was found to occur for 6 mm thick plates manufactured with the unmodified epoxy at higher temperatures, e.g. 100 °C CRTM, a through-thickness impregnation process, was used to manufacture the carbon fibre composite plates. The fibre preform was placed into the open tool (preheated to 80 °C to be consistent with the bulk epoxy plates), and the resin was dosed on top of the preform. Fabric compression and simultaneous impregnation took place via the velocity driven (0.5 mm s−1) closing of the tool, followed by a final dwell pressure of 20 bar to produce 4 mm thick plates with a fibre volume fraction of about 60%.Plane strain compression (PSC) tests were performed to determine the bulk compressive properties as proposed by Williams and Ford Three samples were tested for each formulation, of which one sample was interrupted in the strain-softening region, sectioned to a thickness of 100 μm and polished. This sample was placed between crossed polarisers and examined using transmission optical microscopy to confirm that shear band yielding could be observed.Single edge notched bending (SENB) tests were performed according to ASTM A Carl Zeiss Ultra Plus field-emission gun scanning electron microscope (FEGSEM) was used to obtain high-resolution images of the fracture surfaces using the in-lens detector. Samples approximately 3 mm thick were cut from the fracture test samples using a Struers Accutom-5 precision cutter equipped with a saw blade. The samples were then sputter-coated with a 5 nm thick layer of gold/palladium (Au/Pd) using a Cressington 208HR high resolution sputter coater to prevent charging. The thickness was measured during the sputter coating process with a Cressington MTM-20 high resolution film thickness controller. A typical accelerating voltage of 5 kV and working distance of 6 mm were used for the microscopic observation.The glass transition temperature, Tg, of the cured resin and the total heat of reaction, ΔHtot, were measured with a Mettler DSC 1 using 5 mg of resin in hermetic aluminium pans. A temperature range of 25 °C–150 °C was used to determine the Tg with a heating rate of 10 °C/min. To determine ΔHtot, a temperature range of −50 °C to 220 °C was used. Each formulation was measured at least twice. All measurements were conducted using a nitrogen sample purge flow to reduce oxidation of the resin.A PAAR Physica MCR 302 plate-plate rheometer was used to measure viscosity development in the unmodified and the CSR-modified epoxies with a strain of 1% and an angular frequency of 10 s−1. Disposable 25 mm diameter aluminium plates were used with the plate gap set to 1 mm for measurements at 25 °C, 40 °C and 50 °C.The morphology of the CSR-modified epoxy polymers at different concentrations is shown in . For both types of CSR particles, visual inspection showed that there was no agglomeration of the rubber particles, even at higher concentrations. This was further verified by quantifying the dispersion using the area disorder approach (a). The mean value of the area disorder of the Delaunay network, AD, represents the dispersion type where 2N is the number of Delaunay triangles, sΩ is the standard deviation of the Delaunay triangles' areas and L2 is the total area of the micrograph (L is the side length of the square AFM micrographs).The values of area disorder lie on the line corresponding to a “random-like” dispersion, which can be interpreted as the particles being dispersed in a manner that is indistinguishable from random.A mean core diameter of 48 ± 6 nm was measured for the MX156 CSR particles, and 110 ± 4 nm was measured for the MX960 CSR particles (see ) from the AFM images, which was significantly lower than the nominal diameters given in the respective manufacturer datasheets (100 nm and 300 nm).The volume fractions and mean diameters measured from the AFM phase images were typically lower than those expected from the known quantities of particles added. This is because the PMMA shell has a similar value of modulus and hardness to the epoxy matrix, hence it would not be possible to distinguish the shell from the epoxy matrix. This means that the AFM phase images only show the core of the rubber particles, thus reducing the apparent volume fraction and the mean diameter. The PMMA shell wall thickness was determined to be approximately 10–20 nm using transmission electron micrographs from the manufacturer. The discrepancy between the measured and expected mean diameter is also due to the microtoming process which cuts through the particle at random locations. Assuming the particle size is monodisperse, the actual diameter can be estimated from the diameter at the planed surface by multiplying by 4/π The measured volume fractions of CSR particles measured from the bulk plates were found to differ from the calculated values after mixing (see ), especially at higher concentrations, i.e. 10 wt% and above. This deviation may result from the presence of the PMMA shell.The volume fractions of the CSR particles in the carbon fibre composite plates were analysed at the top, middle and bottom with respect to the composite plate thickness, which corresponds to the impregnation direction for CRTM, as shown in . Samples were taken by fracturing the composite material between plies and studying the interlaminar region. A maximum variation of 1 wt% was measured between the top and bottom of the plates, which indicates that neither the MX156 nor the MX960 particles were readily filtered or damaged during the infusion process when manufacturing the carbon fibre composite plates.The different epoxy formulations exhibit similar behaviours in the compressive true stress vs true strain curves, as shown in . The compressive modulus decreases linearly with the volume fraction of CSR, as observed in a previous study as well For both types of CSR-modified epoxies, the strain softening region becomes less well defined (i.e. there is a smaller drop in the stress with increasing strain) with increasing particle concentration, as shown in . This provides an indication that the shear band yielding process has changed for higher amounts of CSR particles. To understand these changes, one sample of each formulation was compressed to the strain softening limit (i.e. the minimum stress), then sectioned and polished to a thickness of 100 μm. The samples were placed between crossed polarisers and examined using transmission optical microscopy, as shown in . Note that if no shear bands are seen, then the plane of the polarised light is not rotated and the image would appear black, e.g. The MX960 CSR particles are larger (140 nm), and thus the bulk samples were opaque at higher concentrations of 20 wt%, even when sectioned to smaller thicknesses. Hence, the images in for the MX960 modified epoxies were not as clear as for the MX156. However, the visible trend for this formulation does not appear to change with CSR concentration, i.e. qualitatively the same amount of shear band yielding was observed for all of the MX960 modified epoxy polymers.The measured values of the fracture energy are summarised in . A mean fracture energy of 177 ± 35 J m−2 was measured for the unmodified epoxy. The addition of only 4.5 wt% of MX156 CSR particles significantly increases the fracture energy up to 1126 J m−2, corresponding to an increase of 540%. This increase in fracture energy for the MX156 is linear up to about 9.1 wt%, where values of fracture energy up to 1889 J m−2 were measured. At CSR contents above 9.1 wt%, the relative increase in fracture energy was less, and the maximum value of Gc was measured to be 2464 J m−2. The fracture energy values for the MX960 CSR particle modified epoxy are similar at low weight contents, i.e. 1025 J m−2 with 4.5 wt% CSR, and increase linearly up to about 11.5 wt% with a value of 2410 J m−2. The relative increase in fracture energy was again less at higher concentrations with a value of 2572 J m−2 being measured at 14.5 wt%.The fracture surfaces of the unmodified epoxy were found to be smooth and glassy, as is typical for brittle homogeneous thermoset polymers, see (a). With the addition of the CSR particles, the fracture surfaces appear much rougher, as shown in (b–e). At higher resolutions, the individual rubber particles can be observed to have cavitated, followed by plastic void growth of the epoxy matrix. These voids were clearly visible within the process zone for all of the CSR-modified epoxy formulations. This correlates well with the high measured values of the fracture energy. There was no evidence that the PMMA shell debonds from the epoxy matrix, or that the rubber debonds from the PMMA because no debonded particles are visible on the fracture surfaces. By swelling with solvent, butadiene rubber was confirmed to be present in the voids. This shows that the epoxy matrix is well bonded to the PMMA shell, which in turn, remains in contact with the butadiene core, and that the rubber cavitates.The plastic void growth process increased the size of the voids, compared to the original rubber core size measured from the AFM micrographs, as shown in (b). There is a clear trend of decreasing void diameter with increasing rubber content. The larger MX960 CSR particles formed relatively large voids, with a wt% dependent diameter of 140%–220% of the original size, corresponding to a maximum void diameter of 240 ± 10 nm as shown in the tabulated data in . In comparison, the MX156 CSR particles only show a maximum void size of 150% of the original core size for the 4.5 wt% modified epoxy, which corresponds to a mean diameter of 63 ± 2 nm. A decrease in diameter was measured for the 20 wt% MX156 modified epoxy. However, this is most likely due to standard errors in image processing as the highly toughened resins also have significant height changes which can affect the measurements.It should also be noted that at the highest concentrations of rubber used, the rubber particles/voids can be very close to each other, often with only a thin ligament of epoxy between them, as shown in An analytical method to calculate the toughening contribution of micron-sized CTBN particles based on physically-observed mechanisms was first presented by Huang and Kinloch where Gcu is the fracture energy of the unmodified epoxy and Ψ is the overall toughening contribution due to the particles. The same analysis can be used for the CSR particles in this work. The two major toughening mechanisms observed experimentally are (i) plastic shear band yielding, ΔGs, and (ii) cavitation of the CSR particles and subsequent plastic void growth of the epoxy, ΔGv, where ΔGv can be seen as the dominant toughening mechanism The toughening contribution due to shear band yielding in the epoxy can be expressed as where vf is the volume fraction of CSR particles, σyc is the plane strain compressive true yield stress, ɛfu is the true fracture strain of the unmodified epoxy and F′(ry) is a term taken from F′(ry)=ry[(4π3vf)13(1−rpry)3−85(1−rpry)(rpry)52−1635(rpry)72−2(1−rpry)2+1635]where ry is the increased plastic zone size due to stress concentrations in the epoxy and can be calculated using:where μm is a material constant which allows for the pressure-dependency of the yield stress, which was found to be in the range of 0.175–0.225 by Sultan and McGarry where Kcu is the fracture toughness and σyt is the tensile true yield strength of the unmodified epoxy.The toughening contribution of the plastic void growth, ΔGv, was calculated according to where μm is the same constant as in Eq. , and vfv and vfp are the volume fraction of voids and CSR particles, respectively. The maximum hoop strain that a shell void could sustain is (1+ɛfu)rp, and this value was used to calculate vfv. The von Mises stress concentration factor for voids, Kv was taken from Huang and Kinloch The overall toughening contribution, Ψ can then be expressed in the following form by combining Eqs. Ψ=0.5vfσycγfuF′(ry)+(1−μm23)(vfv−vf)σycrpzuKv2The values of the parameters used for the modelling are given in . The fracture surfaces showed that all of the CSR particles cavitated, as can be seen in , hence 100% of the particles were assumed in the modelling to have initiated shear bands and allowed plastic void growth of the epoxy matrix to the maximum extent.The value of Tg for the uncured epoxy, Tg0, was measured to be −28.0 ± 1 °C, and for the cured epoxy, Tg∞, to be 117 ± 1 °C with no change for the CSR-modified epoxy. The value of the cured Tg is slightly lower compared to a previous result The total heat of reaction, ΔHtot, of the unmodified epoxy, obtained from dynamic measurements with a heating rate of 10 °C/min as shown in , was determined to be 513 ± 3 J g−1, which shows a slight deviation compared to previous measurements (494 ± 4 J g−1) Isothermal measurements from room temperature to 50 °C were conducted to study the influence of the core-shell rubbers on the initial viscosity, i.e. the viscosity at which impregnation would be conducted (see ). Roughly 30–60 s of rheometry data were lost between the time when the resin first touched the preheated plates and start of the measurement due to the closing motion of the upper plate.The initial viscosity increased exponentially with increasing wt% of the core-shell rubbers. A similar increase was measured for both CSRs at low wt%, e.g. with the addition of 4.5 wt% the viscosity increased by a factor of 1.34 ± 0.03 with the MX960 compared to 1.38 ± 0.03 with the addition of MX156 for a temperature range between 25 and 50 °C. A more significant increase was measured for the MX156 in the case of high wt%. Hence, particle size effects seem to be more significant with higher wt% of the CSR particles.A particle dependent exponential term can be added to the Arrhenius law previously shown in where T is the temperature in Kelvin, R is the universal gas constant, A1, E1, C1 and D1 are constants given in , and CSRwt% denotes the wt% of the CSR particles. The values of A1 and E1 were taken from can be used to calculate the initial viscosity during infusion, and hence can be used to predict the filling time of a composite part when using a liquid composite moulding processes for the whole range of CSR wt%.Based on the analytical description of the particle position from microscopy, the CSR particles remained well dispersed in the matrix with a random-like distribution. The Tg remained unaffected with a value of 117 ± 1 °C and a linear decrease of the total heat of reaction was observed, resulting from the decreased exothermic mass (epoxy and curing agent). No influence on the curing reaction was observed and the particles aid reduction of the exothermic temperature peak during cure.There were no significant differences in fracture energy between the MX156 and MX960 CSR-modified epoxy polymers up to about 9.1 wt%, with a linear increase in toughness for both particles. Higher concentrations lead to a more gradual increase in the fracture energy. Furthermore, it is known that the bulk polymer fracture energy shows efficient toughness transfer to that of a fibre composite up to a point, e.g. Refs. However, the fracture behaviour at higher wt% loadings was not described accurately by the model, as the increase in the fracture energy became more gradual. This difference is believed to be a result of stress overlaps around particles and the lack of sufficient epoxy to plastically deform, when the particles are close together, resulting in a saturation of the fracture energy enhancement. SEM images confirmed that only a thin ligament of epoxy remains between rubber particles/voids at higher concentrations, as shown in (b) and (d). The more gradual increase in fracture energy was also reflected in the measured void diameters, where the void diameter continually decreases with the weight percentage of CSR particles. This shows that there is a lower energy contribution from the void growth of each cavitated particle due to the space constraint which means that there is a smaller volume of epoxy which is able to deform around each particle.Compression tests showed a clear strain-softening region, suggesting that shear band yielding occurs, which is a major toughening mechanism due to the large amounts of plastic deformation that occur. Indeed, shear bands were observed in micrographs taken from polished samples using cross-polarised light. A further toughening mechanism, rubber core cavitation and plastic void growth was observed from SEM images of the fracture surface.The addition of CSR particles resulted in an exponential increase of the initial viscosity, and the particle size effect on the viscosity was found to be more significant with increasing wt%. At 4.5 wt%, the influence is relatively small with an increase in viscosity by factors of 1.38 and 1.34 for the MX156 and MX960 particles respectively. Interestingly, a larger increase was measured with the addition of MX156 with smaller particle size at higher particle content, i.e. an increase by a factor of 2.3 with 10 wt%, compared to by 1.9 with MX960.A fast-curing epoxy was modified with the addition of two core-shell rubbers (CSR), with measured particle core diameters of 61 nm and 140 nm, to investigate the mechanical and rheo-kinetic properties, size effects and filtration during infusion. Both particles remained well dispersed with no evidence of filtration in the fibre composite, and with a constant Tg, showed no influence on the curing reaction. The total heat of reaction was identical for both particle types with a linear decrease with increasing wt% of CSR. This helps to reduce the strong exothermic reaction exhibited by the fast-curing resin that was studied.No significant effect of particle size was found for the fracture energy up to 10 wt% CSR. Extremely tough parts were produced corresponding to a maximum increase in fracture energy by a factor of 14.5. Shear band yielding and plastic void growth was observed for both CSR particles, which at higher particle content, plateaued due to insufficient epoxy to dissipate energy.Experimental measurements showed that very tough epoxy polymers can be fully cured in a few minutes. The measured viscosity values for 4.5 wt% CSR increase by a factor of only of 1.38 and 1.34 for the MX156 and MX960 respectively. Simultaneously, a significant increase in the fracture energy was measured by a factor of 6.36 and 5.79 respectively. In comparison, a similar increase of viscosity was measured with the addition of 10 wt% of silica nanoparticles, but without a significant toughening effect can be used to find the optimal balance between toughness and infusion time by predicting the fracture energy and initial viscosity of the resin for different wt%.During manufacture using compression resin transfer moulding, none of the CSR particles were filtered out during infusion in the through-thickness direction when using a pressure of up to 20 bar, indicating that these particles are suitable to be used to manufacture fibre composites. We could demonstrate, that the combination of these nanometre scaled CSR particles with a fast-curing epoxy represents an attractive approach to produce very tough composite parts with cycle times of a few minutes.Modeling of the plastic deformation of nanostructured materials with grain size gradient► Dislocation-density-based model is developed for surface-nanocrystallized materials. ► The stress-driven grain growth observed in experiments is incorporated in the model. ► The effect of thickness fraction of the grain size gradient region is considered. ► This study can provide some guidelines for optimization of mechanical properties.Many studies have shown that the outstanding mechanical properties of surface nano-crystallized (SNC) materials were primarily attributed to the grain size gradient (GSG) region on their surface in which the grain size was ranging from tens of nanometers to tens of micrometers. In the present study, a dislocation density-based theoretical model was developed to investigate the mechanical behavior of the mentioned SNC materials. The constitutive behaviors of metallic materials with grain size ranging from tens of nanometers to tens of micrometers were established first. Note that an additional dislocation dynamic recovery term, which is grain size dependent, was included in the present model to account for the decrease of work hardening due to grain refinement. In addition, the stress-driven strain growth observed in experiments has also been incorporated into the proposed model. The proposed quantitative continuum plasticity model was capable of investigating the role of GSG in tuning the strength, ductility and work hardening rate of SNC materials. Our theoretical predictions were in good agreement with the existing experimental results. Furthermore, it has been found that the thickness fraction of the GSG layer and grain growth have significant influences on the strength and ductility of SNC materials. Therefore, the proposed model can be employed to optimize the mechanical behavior in SNC materials by controlling the thickness fraction and grain size in the GSG region and grain growth.Extensive investigations over the past few decades have shown that nanostructured and ultrafine-grained metals/alloys possess remarkable physical and mechanical properties as compared with their coarse-grained counterparts, especially in terms of strength and hardness (). However, the superior mechanical strength of nanostructured and ultrafine-grained materials is attained at the expense of their ductility, which is usually limited to a few percent of uniform elongation; and the extent that work hardening can be performed in these materials is also very limited before catastrophic failure occurs (). As a result, fabrication of materials with both high strength and ductility has attracted intensive scientific interests over the past few decades (). Several well-established synthesization strategies have been proposed to achieve this goal, which include production of (i) metals with bimodal grain size distribution (); (ii) metals with coherent nanoscale twin boundaries (); and (iii) metals with a grain size gradient region on their surface, in which the grain size varies from tens of nanometers to tens of micrometers, and a coarse grained (CG) interior with grain size of tens of micrometers (). Metals consisting of a GSG region on the surface and a CG interior were called surface nanocrystallized (SNC) materials in our previous investigations (). The first two metal types are synthesized by sophisticated methodologies, i.e., thermomechanical approach for bimodal copper () and pulsed electrodeposition for nanotwinned copper (). Whereas, the third type is fabricated by various simple and flexible surface severe plastic deformation techniques viable for engineering applications. The said techniques include shot peening (), surface nano-crystallization and hardening (SNH) (), surface mechanical attrition treatment () and surface mechanical grinding treatment (SNC materials have attracted growing attention of many researchers because of their good balance of strength and ductility (), which is preferable in designing lighter, stronger and simultaneously tougher structures. For example, the 0.2% offset yield strength of a nickel-based alloy was increased by 65% and 84% over that of the CG core via 30 min and 180 min SNH processing, respectively, while its tensile elongation was only decreased from about 60% of that of the CG core to 40% and 30%, respectively ( reported that the yield strength of SNC copper was twice that of its CG counterpart, while its tensile uniform elongation could reach as high as 31 ± 2%, which was similar to that of the CG sample, i.e., 32 ± 2%. These exceptional mechanical properties were attributed to a transition of plastic deformation mechanism from mechanical driven grain boundary migration to conventional dislocation slip, due to the observed grain growth on the surface layer. However, the roles that grain size gradient and grain growth play in the SNC copper, which possesses much higher strength and comparable ductility with respect to its CG counterpart, are unclear.To-date, the works done on SNC materials were mainly focused on several aspects: (i) grain refinement mechanism in the surface layer of various metals and alloys, including pure copper (); (ii) mechanical behavior and properties such as tension (); (iii) major factors that affected mechanical properties (); (iv) optimizing parameters for achieving favorable properties based on surface nanocrystallization techniques (); and (v) characterization of the plastic properties at metal surface (). Most of these works were experimental studies (e.g. ); and there were only a few theoretical modeling studies. Therefore, a deformation mechanism-based constitutive model must be developed to quantitatively predict the plastic deformation of SNC materials with reasonable accuracy. Moreover, some interesting studies were carried out to investigate the ductility of a thin metal film mounted on substrate (). The results obtained show that the polymer or elastomer substrate was capable to increase significantly the limited tensile ductility of a thin metal film. In addition, found that the ductility of metals in the form of sheet or round bar can be enhanced by cladding a ductile layer or ring on it. These results inspired us to investigate the role of GSG in regulating the ductility of SNC materials.The objective of this manuscript is to develop a theoretical model to study the role of the GSG in tuning the strength and ductility of SNC metals/alloys. In the proposed model, the SNC material can be viewed as a multi-phase composite structure comprising various phases of different grain sizes varying from tens of nanometers to tens of micrometers. The above-mentioned range of grain size variation in SNC materials have been determined in experimental studies (). A dislocation-density-related theoretical model is developed based on the Kocks–Mecking–Estrin (KME) model () to identify the elastic-plastic constitutive behavior in the phases with grain size varying from tens of nanometers to several hundred nanometers (called nanocrystalline (NC) phases) and that in phases with grain size varying from submicrometer to tens of micrometers (called microcrystalline (MC) phases). The overall stress-strain response of SNC materials can be derived by combining the constitutive relations of all phases using the rule of mixtures (ROMs) of Voigt model. The proposed model will be employed to predict the mechanical properties of SNC copper under unidirectional loading. Grain growth in the GSG region of SNC materials is also accounted for in the present model.For modeling of SNC materials, the first step is to establish the constitutive framework of the nanostructured materials with grain size gradient at the material surface. The SNC material can be postulated as a multi-phase composite structure with all phases having different grain sizes, as observed in experiments. For convenience, we assume that there are n phases with various grain sizes in the GSG region, as shown in . The phase number n is to be ascertained to ensure that it is sufficiently large to describe the continuous variation of grain size and, thus, the plastic properties in the GSG region. Due to symmetry, only a quadrant of the object needs to be modeled. The initial thickness of the GSG region, which consists of n phases, and the CG core are denoted by hg and hc, respectively. For the sake of describing the overall mechanical properties of SNC materials using the rule of mixtures, the constitutive behavior of each and every phase of the structure must be established first. In view of the good capability and reliability of the dislocation-density-based constitutive modeling for predicting the mechanical response of coarse-grained or fine-grained metallic materials subjected to uniaxial loading (), a typical dislocation-density-based-model, i.e., the KME model (), is selected for modification to describe the stress-strain response for each of the phases in SNC materials. Considering the fact that the grain size in the GSG region varies from tens of nanometers to tens of micrometers, two different regions must be modeled separately, one for the phases with grain size ranging from tens of nanometers to hundreds of nanometers, i.e., NC phases, and the other for those with grain size ranging from submicrometer to tens of micrometers, i.e., MC phases. A brief summary of the general framework of the KME model with flow stresses modified is given below:The total strain rate ε˙ij can be decomposed into its elastic and plastic parts, ε˙ije and ε˙ijp, respectively:The elastic part obeys Hooke’s law and can be expressed aswhere σij is the stress tensor and a dot denotes differentiation with respect to time t; Sijkl is the elastic compliance tensor given bywhere E, ν and δij are Young’s modulus, Poisson’s ratio, and Kronecker’s delta, respectively. The plastic part of the strain rate can be obtained based on the J2 deformation theory as follows:where sij

=

σij

-

σkkδij/3, ε˙p=2ε˙ijpε˙ijp/3 and σ=3sijsij/2 are the deviatoric stress, the equivalent von Mises plastic strain rate and the equivalent von Mises stress, respectively. A kinetic equation is introduced to relate the equivalent plastic strain rate and equivalent stress as follow:where ε˙0∗, σf and m are the material parameter, flow stress and rate-sensitivity exponent, respectively. Note that in the present work the focus is primarily on the stress-strain response of metals with various grain sizes ranging from nanometer to micrometer scale. Despite the fact that nanocrystalline materials are strongly rate dependent (), for simplicity the rate dependence is not rigorously accounted for in the present constitutive model. To avoid introducing strain rate sensitivity, ε˙0∗ in Eq. is replaced by the equivalent strain rate ε˙=2ε˙ij′ε˙ij′/3, where ε˙ij′=ε˙ij-ε˙kkδij/3 is the deviatoric strain rate, as adopted by in their proposal of a conventional theory for mechanism-based strain gradient plasticity. Eqs. are then established to describe the triaxial constitutive relation for all phases of different grain sizes in SNC materials. In the present study, these equations are numerically integrated using a fourth-order Runge–Kutta method. The flow stress σf for nanocrystalline and microcrystalline phases will be discussed in the following subsections.Many molecular dynamics (MD) studies have illustrated that grain-boundary mediated deformation mechanisms (e.g. grain-boundary sliding ()) played an increasingly important role as compared with that of intragrain dislocation mediated plasticity with decreasing grain size in nano/ultrafine grained materials. For example, found that plastic deformation was dominated by grain-boundary sliding for copper with grain size smaller than 8 nm, at which the strength of NC Cu approaches maximum. has also shown that a transition of deformation mechanism from dislocation-mediated plasticity to grain-boundary sliding occurred in Cu at 10–15 nm, which corresponds to softening with decreasing grain size. Moreover, many experimental investigations have shown that in the case of Cu, Hall–Petch strengthening is valid down to a grain size of 10 nm ( have also pointed out that grain-boundary sliding did not dominate in nanocrystalline Cu of 10 nm grain size. One may infer from these findings that dislocation slips still play an important role at a grain size as small as 10 nm (). In the case of SNC materials, the average grain size at the topmost surface is about 20 nm (); thus, deformation mechanisms such as grain boundary sliding and diffusion, which dominates at grain size below 10 nm needs not be included in the present model. In addition, many experimental observations () have illustrated that grain boundaries acted both as dislocation sources and sinks, and that grain-interior plasticity is accommodated by dislocations emitted from grain boundary. Therefore, the dislocation-density-based KME model is modified in the present study to describe the constitutive behavior of nano/ultrafine grained materials. The flow stress in the NC phases can be expressed as:where M, α, b, μ, ρ are the Taylor factor, Taylor constant, magnitude of Burgers vector, shear modulus and dislocation density, respectively; σ0 is the lattice friction stress. The second term on the right hand side of Eq. describes the Taylor-type relationship between the flow stress and dislocation density; the last term σGB is a contribution to flow stress arising from grain boundary, as adopted by to study the strain rate sensitivity and activation volume of nanocrystalline Cu, and by to investigate the elasto-viscoplastic behavior of nanocrystalline materials. This term can be expressed as follow:where kHP is a constant Hall–Petch slope, and d denotes grain size.The evolution of dislocation density ρ with respect to plastic strain can be expressed as:where k

=

k3/(bd), k1

=

ψ/b, k2=k20(ε˙p/ε˙0)-1/n0, ψ is a proportionality factor, k20 and ε˙0 are material constants, k3 is a geometric factor related to the grain shape and proportion of dislocations arriving at the grain boundaries, ke is an additional dynamic recovery factor and n0 is inversely proportional to temperature. The first and second terms on the right-hand side of the equation are associated with the athermal storage of dislocations, and the third is related to the annihilation of dislocation during dynamic recovery which is independent of grain size. To describe the grain size dependent work hardening in nano/ultrafine grained materials, the last term is included in our model to account for the additional dislocation dynamic recovery at grain boundaries due to the decreased grain size in nano/ultrafine grained materials. Experiments have demonstrated that the dynamic recovery rate increased with decreasing grain size below a critical value, at which the mean free path of dislocation was no longer governed by the dislocation structure, but by the grain boundaries (). The critical grain sizes were determined as approximately 300 nm, 1 μm and 700 nm for aluminum (), respectively. The additional dynamic recovery term was widely used to investigate the plastic behavior of nano/ultrafine grained materials. For example, this term was used by to study the behavior of grain boundaries during superplastic deformation and by to explain the lack of accumulation of dislocations in ultrafine copper grain. also successfully used this term in his model to investigate the plasticity, strength, inverse Hall–Petch relation and strain rate sensitivity of nanocrystalline and nanotwinned materials. In view of the above work, this term can be expressed as follows:where de is a reference grain size corresponding to the critical grain size at which enhanced dynamic recovery occurs, and d is the grain size. Note that ke is a grain size dependent term, which can be used to account for the decrease of work hardening with respect to the refinement of grains in nano/ultrafine grained materials.In microcrystalline materials with grain size ranging from submicrometer to tens of micrometers, the primary deformation mechanism is the intragrain dislocation-mediated interaction. An enhancement of dynamic recovery also occurs similar to that in the nano/ultrafine grained materials when the grain size is decreased to submicrometer level. The flow stress in the microcrystalline phases can be expressed as:where σGB is a term related to grain boundary which has the same expression as that for nanocrystalline materials (refer to Eq. in their microstructure-based strain hardening model to describe the isotropic hardening of fine-grained dual phase steels with grain size of several micrometers; and σb is the back stress. Two categories of work hardening are described in Eq. , i.e., isotropic and kinematic strain hardening. The former is described by the Taylor-type relationship between the flow stress and dislocation density, i.e., the second term in Eq. ; whereas, the latter is originated from accumulation of dislocations at the grain boundaries, giving rise to the back stress σb. To-date, many studies have illustrated the contribution of back stresses to the kinematic strain hardening in micro-sized and coarse-grained metals/alloys such as copper ( have successfully developed a unified constitutive model by defining different evolution rules of short- and long-range back stresses to simulate uniaxial tensile and strain rate jump tests, short term creep tests with stress jump, and uniaxial ratcheting tests of lead free soldering alloys. have also established a crystal-plasticity-based model to investigate the effect of back stresses on kinematic strain hardening of FCC polycrystals. In addition, the contribution of back stresses on strain hardening has also been taken into account to analyze the stress-strain response of nanotwinned copper (). The back stresses can be expressed in the following simplest form:where N is the number of dislocations accumulating at the grain boundaries (GBs), which obeys the following evolution law (where ɛp, ξ and N∗ are the plastic strain, the mean spacing between slip bands and the maximum number of dislocation loops at the GBs, respectively.Note that our microcrystalline model should be applicable to as low as submicrometer level, at which enhanced dynamic recovery occurs as stated above; an additional term keρ should also be included in the evolution equation of dislocation density, which would have the same form as that of Eq. . The value of de may be different from that in a nanocrystalline model, which can be determined through experiments.The focus of the present study is on uniaxial tensioning of SNC materials because there is existing experimental data for comparison. Since a SNC material is subjected to uniaxial uniform strain through its thickness in experiments, the rule of mixtures (ROMs) of Voigt model (), in which equal strain is assumed, can be employed to accurately determine the overall stress-strain response in the material. Therefore, the tensile stress σxx∗ induced in a SNC material subjected to uniaxial strain εxx∗, as schematically shown in where k, c and n denote the kth phase, CG core and number of phases in the GSG region, respectively; σxxk, σxxc are the stresses applied on phase k and CG core, respectively; hk

=

hg/n, hc, h are the thickness of each phase in the GSG region, CG core and the entire SNC material, respectively. Note that if the applied load is more complex than uniaxial tension, the following methods may be adopted for solving the problem. In view of the fact that there is a grain size gradient on the surfaces of a SNC material, it can be treated as a kind of functionally graded material (FGM), which is characterized by a continuous variation of property in its composition, microstructure or crystal structure from one surface of the material to the other (). In the case of small deformation under complex loading, the three-dimensional semi-analytical framework developed under small strain assumption by for investigating the themo-elastoplastic behavior of a FGM system (a Ni–Al2O3 system) can be employed to calculate the stress and strain distribution through thickness, which can in turn be used to determine the overall stress-strain response of a SNC material. As for large deformation under complex loading, the stress-strain data for phases with various grain sizes generated by the developed theoretical model for NC and MC phases can be used as input data for finite element simulations.To validate the results obtained by the above simple ROMs of Voigt model, a quadrant 2-D finite element model (FEM) was developed to simulate a symmetric SNC material subjected to uniform uniaxial displacement loading using ABAQUS v6.8. 34480 three-noded triangular elements were used for the GSG region and 17240 four-noded quadrilateral elements for the CG core in modeling the SNC Cu. The stress-strain data generated by the developed theoretical model for phases with various grain sizes were used as input data for the corresponding phases in this 2-D FEM. The plastic deformation of each and every phase in the GSG region and CG core was assumed to be adhering to the J2 deformation theory. The resultant stress distribution through thickness and the applied displacement were extracted from ABAQUS to calculate the overall stress-strain relations for SNC Cu using Eq. In order to calculate the uniform elongation and ultimate strength of SNC materials, a Considère criterion () is implemented in our model as follow:It is worth noting that this criterion has been widely used in the theoretical and experimental studies performed in materials science ( employed this criterion to evaluate the ductility of particle-reinforced metal-matrix composites. also implemented this criterion in their physically-based strain hardening model to investigate the optimization of strength and ductility of ultrafine-grained dual phase steels. As for yield strength, it is commonly defined as 0.2% offset stress.), the common logarithm of grain size d in the GSG region is assumed to vary linearly with the distance z from surface as follow:where d0

=

d1, kd

=

lg(dc/d1)/hg, and d1, dc are the grain size of the topmost phase and CG core, respectively.According to the experimental investigation carried out by , the grain growth in the GSG region played an important role in the plastic deformation of SNC copper for it possessed much higher yield strength and comparable ductility compared with its CG counterpart. Therefore, grain growth is also accounted for in our model. As suggested by , the grain growth in the GSG region of a SNC Cu is a mechanically driven grain boundary migration. In F of their article, the grains in the GSG region first grew linearly with respect to the true strain and they stop growing when the strain approached a sufficiently large value. The grains with an average size of 20 nm at the depth of 0–20 μm in SNC Cu grew to around 400 nm, while grains of average size 60 nm at the depth of 20–40 μm grew to around 560 nm. In our model, the grain growth effect is taken into account in a phenomenological manner based on experimental observations, i.e., an approximately linear relation between grain size and strain during deformation. The size of the grains in the GSG region is thus assumed to vary in the following manner in our model:dk=dk0,εp<ε0pdk0+kg(εp-ε0p),ε0p⩽εp<εspdks,εp⩾εspwhere kg=(dks-dk0)/(εsp-ε0p); ɛp is the plastic strain; ε0p and εsp correspond to the plastic strain values at which the grain begins and stops growing based on experimental observations, respectively; dk0 and dks are the grain sizes of phase k (k∈[1,n]) in the GSG region before and after grain growth, respectively, which can be determined from the experimental results based on the assumption that the common logarithm of grain size in the GSG region after growth obeys a linear relation with respect to the distance from surface. In this study, two cases of grain growth will be considered: (i) all grains in the GSG region grow during deformation, called gg I; (ii) only grains of size less than 500 nm in the GSG region grow during deformation, called gg II. Based on the experimental observation, the second case appeared to be more realistic. To establish the stress-strain relations of SNC materials with grain growth for each phase, Eq. , which relates flow stress to grain boundary, the evolution equation of dislocation density given by Eq. . As deformation proceeds, the grain size in the GSG region increases, which in turn decreases the flow stress of each phase and finally affecting the overall stress-strain response of SNC materials.The proposed model is used to predict the mechanical behavior of SNC copper for which a larger number of experimental data is available. Note that the proposed NC sub-model is applicable to nano/ultrafine grained materials with grain size varying from tens of nanometers to several hundred nanometers while the MC sub-model is for micrometer-sized materials with grain size ranging from sub-micrometer to tens of micrometers. There is no clear boundary between these two sub-models. The experimental data of 500 nm copper () is used to simultaneously determine the material parameters in the NC and MC sub-models. First of all, the model is validated by comparing its predictions with a set of experimental data. A set of parametric studies is then performed to investigate the effect of phase number, thickness fraction of the GSG region, grain size of the topmost phase and grain growth on the mechanical response of SNC materials. The material parameters used in simulations are provided in for SNC copper. These parameters are mostly determined from either experiments or the existing literature as follows: (i) The elastic modulus for all phases in the SNC copper is assumed to be the same and its value is obtained from experimental measurements (); (ii) The Hall–Petch slope and frictional stress are taken from the Hall–Petch relation for the yield strength of copper given by ; note that the former has been used by in their multi-scale modeling of the mechanical behavior of nanocrystalline copper; (iii) The dynamic recovery constant n0 is taken from for nanocrystalline copper; the value of another dynamic recovery constant k20 and the proportionality factor ψ are determined by fitting the predictions of NC and MC sub-models to the experimental data for 500 nm copper and 25 μm copper, respectively; (iv) The parameters for back stress and reference grain size de in the enhanced dynamic recovery term in MC copper are then determined by fitting the mentioned prediction to the experimental results of 500 nm copper, while the reference grain size de in NC copper can be determined by fitting the prediction to the experimental results of 200 nm copper; and the value of geometric factor k3 for NC and MC copper are taken from the original KME model and , respectively. The materials data for SNC copper are mainly adopted from the experimental measurements by In order to study the overall stress-strain response of a SNC copper, the constitutive behavior of the simulated NC and MC copper must be validated first. compares the predicted true stress-strain curves with the corresponding experimental results for NC copper with average grain size of 30 nm, 200 nm and 500 nm. By fitting the elastic region of the experimental stress-strain curve using the least squares method, the elastic modulus of 30 nm and 200 nm grain size copper are determined as 23 GPa and 36.85 GPa, respectively. The predicted stress-strain curves for MC copper with average grain size of 500 nm, 1 μm, 5 μm and coarse grained copper are compared with the corresponding experimental results presented in that our predictions are in good agreement with the experimental results for both NC and MC copper in terms of strength and work hardening. Note that the relation between the flow stress and the grain size predicted by the original KME model follows the classical Hall–Petch (HP) relation. However, the KME model can only be used to predict the plastic behavior of face centered cubic (FCC) microsized metals (). The additional grain size dependent dynamic recovery term introduced in the present model in the evolution of dislocation density in both NC and MC coppers can successfully account for the decreasing work hardening as grain size reduces. Since the additional dynamic recovery term depends on grain size, the flow stress does not follow the HP relation because of the decreasing work hardening and corresponding decreasing flow stress due to grain refinement. Thus, the model devised can be used to predict the grain size dependent stress-strain response of copper with various grain sizes ranging from namometer to micrometer scale.The overall mechanical behavior of SNC copper can now be studied using the model devised. To make comparison with the experimental data for SNC copper obtained by , in which grain growth played a vital role during deformation, grain growth is also accounted for in the proposed model based on the approach described in Section . According to experimental observations, the thickness of GSG was around 700 μm and that of the entire SNC sample was 3 mm, resulting in a thickness fraction of f

= 1/4.3 for the GSG region. The value of phase number n was ascertained as twenty to ensure that the mechanical response of SNC material predicted is independent of phase number, while those of parameter ε0p and εsp will be obtained by fitting the model predictions with the experimental data. shows the predicted and experimental stress-strain curves of a SNC copper with grain growth; two cases of grain growth are included, i.e., gg I and gg II, as described in Section . It is obvious that the predicted yield strength and strain hardening are in good agreement with the corresponding experimental data for both of the two cases of grain growth even though there are some discrepancies. The predicted true stress-strain relations of SNC copper without grain growth is also presented in for comparison, which indicates that grain growth can significantly influence the stress-strain response of SNC copper.To validate our theoretical predictions for SNC Cu, a 1/4 symmetric 2-D finite element model with the same geometric and material parameters as those used in theoretical analysis was developed, as described in Section . The FEM predictions for SNC Cu without grain growth are also presented in , which shows that the theoretical results are in excellent agreement with the FEM simulation, which validates the capability of the proposed theoretical model for prediction of the plastic deformation in SNC materials. As for SNC Cu with grain growth, since such growth will lead to strain softening in nanocrytalline phases in the GSG region during deformation, the present finite element modeling approach that employs the existing material models in ABAQUS is unable to predict the deformation behavior of the material. A user-defined material should be established in ABAQUS to achieve the goal of simulating materials with grain growth.To achieve a comprehensive understanding of our model, the grain size dependence of dislocation density evolution with respect to plastic strain in NC phases and the contribution of back stress in MC phases are investigated in this section. (a) shows the relation between dislocation density and plastic strain in NC copper with grain sizes ranging from 20 nm to 500 nm, which is calculated using Eq. (a) shows that with decreasing grain size, the maximum dislocation density is firstly increased to 8.3 × 1014/m2 at 200 nm grain size, and then decreased to 2.3 × 1014/m2 at 20 nm grain size. The reduction of dislocation density is caused by the significantly enhanced dynamic recovery as the grain size is decreased to well below 100 nm. (b) presents the true stress-strain curves of NC copper with various grain sizes, which shows that the strength increases, while the strain hardening capability decreases, with decreasing grain size, as observed in experiments (). Note that SNC copper with grain size smaller than 50nm displays a near-perfect elastoplastic behavior, which is in agreement with experimental results ( compares the true stress-strain responses with and without back stress for MC Cu with different grain sizes. The results obtained show that with decreasing grain size of MC copper, the back stress plays an increasingly important role in influencing the total true stress, as predicted by Eq. The influence of grain growth will be studied by modifying the proposed model to incorporate such growth in the investigation of SNC materials with a GSG region of five different thickness fractions, i.e., f

= 1/10, 1/8, 1/6, 1/4.3, 1/2. First of all, the effect of grain growth on the dislocation density in crystals with grain sizes of 20 nm, 30 nm, 55 nm, 100 nm and 300 nm in SNC Cu and the corresponding stress-strain responses are investigated, as shown in , based on the second case of grain growth, i.e., gg II. For clarity, the curve for 300 nm grains is not shown in (a). The results obtained show that the dislocation density increases for grains of size smaller than 55 nm, while it decreases for those larger than 55 nm, during deformation. Specifically, in the case of 20 nm grain size, the maximum dislocation density increases considerably from 2.3 × 1014/m2 to 6.0 × 1014/m2 upon completion of grain growth; while the dislocation density in 30 nm grains is also increased, it is decreased in 100 nm and 300 nm grains upon completion of grain growth. The considerably enhancement of dislocation activity in grains of initial size smaller than 55 nm may increase the capability for work hardening in SNC Cu with grain growth. (b) shows that the true stresses of all crystals with grain growth decrease with growing grain size during deformation. Note that the smaller the grain size, the larger the decrease of true stress would be. presents the stress-strain relations, uniform elongation and yield/ultimate strength for various GSG thickness fractions in SNC copper with case 1 grain growth, i.e., gg I, accounted for; the corresponding values for SNC copper without grain growth are also included for comparison. Note that n is set as 50 in the simulation studies to ensure that the results obtained are independent of the phase number. The results for SNC Cu without grain growth show that there is an approximately linear relation between the uniform elongation, as well as between the yield/ultimate strength, and the thickness fraction of GSG, as shown in (b) and (c), respectively. Note that a similar linear relation between the uniform elongation of sheet metals and the thickness fraction of cladding layer was also found by (a) shows that grain growth has considerable impact on the stress-strain response of SNC copper with different thickness fractions of the GSG layer. For clarity, only the curves for three different fs are presented. It can be clearly seen from (b) that grain growth only leads to a slight increase in the uniform elongation of SNC copper. It is worth noting that in the case of f

= 1/4.3, the uniform elongation even with grain growth considered is still significantly lower than that of CG copper. This discrepancy is probably due to the fact that the present model is strain rate independent, whereas, in reality the process of grain growth could be strain rate dependent (). Thus, the uniform elongation should be calculated using Hart’s criterion (), which takes into consideration the strain rate sensitivity, instead of Considère criterion. Another possible reason is that grain boundary (GB) migration was not incorporated into the present model. Bobylev et al. (2010) has demonstrated that in nanocrystalline (NC) solids, GB migration with GB sliding incorporated is more energetically favorable than pure GB sliding and, thus, the intrinsic ductility is enhanced considerably in the former case than the latter. This is to say GB migration may play an important role in enhancing the ductility of SNC Cu with grain growth. The authors would make an attempt in future to include GB migration in the proposed model for enhancing its capability to predict the ductility of SNC Cu. (c) shows that grain growth has no effect on the yield strength but it leads to a slight decrease in ultimate strength. The ultimate strength of SNC Cu with grain growth seems to be independent on the GSG thickness, which can be explained as follow. Upon completion of grain growth, the 20 nm grain in the topmost phase of the GSG region has grown to 400 nm for all GSG thickness fractions. Therefore, the size of all grains in the GSG region of SNC Cu lies in the range of 400 nm–25 μm, which leads to a small difference in terms of maximum flow stress between various phases in the GSG region and the CG core, i.e., 71 MPa. As a result, the GSG thickness has insignificant influence on the ultimate strength of SNC Cu with grain growth. Actually, the magnitude of difference for ultimate strength between different thickness fractions of GSG (i.e., f varies from 0.1 to 0.5) is less than 1 MPa, as shown in (c). These results indicate that the grain growth in GSG region is of vital importance for SNC copper to possess much higher yield strength as well as high ductility compared to its CG counterpart.A dislocation-density-based model has been successfully developed to investigate the mechanical behavior of nanostructured materials with grain size gradient. The constitutive behaviors of metallic materials with grain size ranging from tens of nanometers to tens of micrometers were established first. An additional grain size dependent dislocation dynamic recovery term and a term attributed to grain boundaries have been added to the evolution equation of dislocation density and the expression of flow stress, respectively, for both for the nanocrystalline and microcrystalline sub-models. For microcrystalline materials, the contribution of back stress to the flow stress was also considered. Grain growth has also been accounted for in the present model in view of the fact that it plays a vital role in the mechanical behavior of SNC materials. It is important to note that the results obtained from the proposed model are in good agreement with experimental measurements in terms of strength and work hardening for SNC copper with grain growth. In addition, it has been found that the thickness fraction of GSG and grain growth have significant effects on the stress-strain response of SNC materials. The results obtained from the proposed model will shed some light on optimizing the strength and ductility of SNC materials by manipulating the thickness fraction of GSG layer and grain growth.Analyses of Young's modulus and thermal expansion coefficient of sintered porous alumina compactsThis paper reports the derivation of Young's modulus (E) and thermal expansion coefficient (TEC, β) for a sintered porous structure with open pores. The theoretical E is affected by the number of grains and the grain boundary area in sintered ceramics. The measured E–porosity relationship for porous alumina compacts were compared with the theoretical E values that were derived for the present open-pore structure and also for the dispersed (closed)-pore structure treated previously. With decreasing porosity (50% → 10%), the scattered E values showed a gradually increasing tendency, which were located between two theoretical curves for the open-pore structure. The sudden increase of E values in the porosity range from 10% to 0% was well explained by the theoretical dependence of E on porosity for the open- or closed-pore structure. The β values for the porous alumina structures were independent of porosity and close to the β values reported for fully dense alumina compacts. This result was in accordance with the theoretical β–porosity relationships for the open-pore and closed-pore structures.Porous ceramics have been widely used as exhaust gas filters, suspension filters, catalyst supports, thermal insulators and electrodes in fuel cells. The mechanical and thermal properties of porous ceramics have great influence on the above functions and durability. To use porous ceramics safely under a given stress or in a heated atmosphere, the relationship between the porous structure and the mechanical or thermal property should be quantitatively understood. In our previous papers . (3) Theoretical modelling of neck growth during sintering explains well the measured compressive strength as a function of shrinkage. (4) The newly developed mixing rules of Young's modulus (E) and thermal conductivity for a solid material with simple cubic particulate inclusion () are applicable to the properties of porous ceramics when E1(inclusion) is treated as E1(pore) = 0 GPa or κ1(gas) is substituted for κ1(inclusion). The mixing rules are discussed in On the basis of the above successful analyses of the properties of porous ceramics, the present paper focussed on the analyses of the Young's modulus and thermal expansion coefficient (β) of porous alumina ceramics with a large amount of open pores. The modelling of E and κ of porous ceramics () has been previously developed for the structure with dispersed (closed) pores. With increasing porosity, the structure of pores changes from closed pores to open pores. In our previous experiment ). However, the effective E value for the structure with a continuous pore phase (open-pore structure) results in 0 GPa because E2(continuous phase) = E(pores) = 0 GPa. Therefore, a new model is needed to explain the Young's modulus for porous ceramics with open pores.). The mixing rule of the effective β for composites (a) shows the model structure with a simple cubic inclusion (parallel structure) and 2 (series structure). The Ec and βc for E1 = 0 GPa and β1 = 0 m/mK represent the effective Young's modulus (Ep) and thermal expansion coefficient (βp) for porous ceramics. shows Ep and βp for the parallel and series structures with dispersed (closed) pores. Eqs. for the parallel and series structures, respectively, provide the different porosity dependences of Ep. However, both equations result in Ep = E2 GPa (for continuous solid phase 2) at Vp = 0% and Ep = 0 GPa at Vp = 100%. On the other hand, the βp in the parallel structure model is independent of porosity and equal to β2. The βp in the series structure is a function of porosity, which gives a minimum value at Vp = 29.6% and results in βp = β2 at Vp = 0% and 100%. shows the scheme of a porous structure with continuous (open) pores and connected grains. The apparent compressive fracture strength (σp) of the porous structure has been analysed in our previous papers has been derived for σp of the porous structure.σp=σ0πy2(NV)2/3=σ03.0462p−p2(4−3np2+np3)2/3D2/3where p (= h/r), n, and D represent the ratio of the distance shrunk (h) between two particles to the particle radius (r), the coordination number of the particles and the relative density (ratio of bulk density to theoretical density), respectively. The σ0 value corresponds to the strength needed to fracture the several grain boundaries three-dimensionally surrounding one grain at D = 1 is a function of p value and expressed by Eqs. where D0 and q represent the relative density before sintering (green compact) and the linear shrinkage (h/r0) based on the radius of starting particles (r0), respectively. Once D0 and q are measured for the porous alumina ceramics before and after sintering, the combination of Eqs. gives σp through the following calculation step: q → D (Eq. The above analysis of σp for the porous structure was applied to derive the Young's modulus for the porous structure with continuous (open) pores shown in shows the schematic compressive stress (σ)–strain (ε) relationship. The σ2 and σp in indicate the compressive strength for fully dense ceramics and porous ceramics fractured at ε2 and εp, respectively. The slope of the straight line corresponds to the Young's modulus for dense ceramics (E2) or porous ceramics (Ep). The σ2 at D = 1 in Eq. where the F value corresponds to f(p) at D = 1 and is calculated through the following step: D = 1 → q (Eq. ) → f(p) (= 3.046(2p − p2)D2/3/(4 − 3np2 + np3)2/3) → F. In this experiment with the supplied alumina particles, the F value was calculated to be 0.3771 for D0 = 0.5986, q = 0.1572, p = 0.1469 and n = 12 (for a random close-packed structure). Therefore, the σp for porous ceramics (Eq. is transformed according to the two cases of strain at fracture in , case 1: εp = ε2 and case 2: εp ≠ ε2. In case 1, the σp is expressed by Eq. σp=σ2f(p)F=(σ2ε2)f(p)Fε2=(E2f(p)F)ε2=Epεp provides the following relationship of Young's moduli between dense (E2) and porous (Ep) ceramics.σp=σ2f(p)F=(σ2ε2)f(p)ε2F=(E2f(p))ε2F=Epεp leads to the relationships given by Eqs. As derived above, the Young's modulus for the porous structure shown in is a function of f(p) and expressed by Eqs. A linear TEC (β) of a material with length L, which is heated uniformly at the atmospheric pressure (P), is defined by Eq. where L1 is the length at room temperature and ε is the strain. On the other hand, the Young's modulus in represents the slope of σ(applied stress)–ε relation by Eq. The stress (σ(porous)) applied to the porous structure is expressed by Eq. σ(porous)=E2Jε(porous)(J=f(p)Forf(p),Ep=E2J)β(porous)=1E(porous)(∂σ(porous)∂T)P=1E2J[J∂(E2ε)∂T]=1E2(∂σ2∂T)=β2That is, the TEC value (β(porous)) for the porous structure is equal to the TEC value (β2) of the fully dense structure. The derived Ep and βp values for the porous structure with continuous (open) pores are summarised in . Interestingly, the calculated result yields the same conclusion of βp = β2 for closed pores in the parallel structure (The following alpha-alumina powder was employed to make porous alumina compacts with 52.2–97.0% relative density: AKP50 powder, Al2O3 purity > 99.99 mass%, specific surface area S0 = 9.61 m2/g, equivalent diameter d0 = 156 nm, isoelectric point pH 6.37, Sumitomo Chemical Co. Ltd., Tokyo, Japan. The alumina powder was dispersed at a solid content of 30 vol% in double-distilled water adjusted to pH 3 with a 1 M HNO3 solution. The alumina suspension was stirred for 24 h and then transferred into a gypsum mould for 7 days. The consolidated alumina compacts were sintered at 1073–1873 K in air for 1 or 4 h. According to the Japanese Industrial Standard R1634 where dw is the density of water. Thus the relative density (D), open porosity (Vp(open)) and closed porosity (Vp(closed)) are expressed by Eqs. The sintered porous alumina compact was cut into a rectangular prism 3–5 mm in width, 3–5 mm in height and 3–5 mm in length. The polished alumina sample was then sandwiched between two three-layer laminates comprising two copper plates (20 × 20 × 1 mm3) with a sintered SiC plate (20 × 18 × 7 mm3) between them. The sample was then compressed at a crosshead speed of 0.5 mm/min while the strain along the compressive direction was measured using a strain gauge attached to the sample. The detailed configuration for the compressive test has been reported elsewhere The TECs of the sintered alumina of a rectangular prism 3–5 mm in length, 3–5 mm in width and 10–20 mm in height were measured using thermomechanical analysis equipment (TMA, model 8310, Rigaku Co., Japan). The sample was heated once from 298 to 673 K in air and cooled to 373 K at heating and cooling rates of 5 K/min. The expanded length of the alumina was measured upon the subsequent heating from 373 to 1073 K at a heating rate of 10 K/min. The length measurement was calibrated against a blank test and dense α-alumina reference data. For each sintering condition, three alumina samples were tested. The calibrated thermal expansion of the alumina was approximated by Eq. where L1 is the initial length at room temperature, L is the length at a high temperature, c1, c2, c3 and c4 are the experimental constants, and T is the absolute heating temperature. The linear thermal expansion coefficient (TEC, β) based on the sample length at room temperature was determined by differentiating Eq. shows the porosities of alumina compacts sintered at 1073–1873 K for 1 h in air. Most of the pores formed in the alumina sintered below 1273 K were open pores and disappeared at higher sintering temperatures. The increase in the porosity of open pores at 1073–1173 K is not well understood at present. As seen in , some open pores changed to closed pores above 1473 K, and the formed closed pores disappeared with increasing heating temperature. shows the influence of heating time on the porosities of alumina compacts sintered at 1173–1473 K. The open pores formed at 1173 K were relatively stable after heating for 4 h but easily changed to closed pores after heating for 4 h at 1273 K. The amount of the formed closed pores decreased at 1473 K but the remaining closed pores were stable upon heating for 4 h; that is, the alumina compacts sintered in air at ambient pressure contained nearly 100% open pores or a mixture of open and closed pores, depending on the heating temperature and heating time. On the other hand, α-alumina powder with a median size of 550 nm was densified above 99% relative density by hot-pressing at 1773 K under a pressure of 39 MPa for 2 h in an Ar atmosphere shows a typical compressive stress (σ)–strain (ε) relationship for the sintered porous alumina with relative densities of 52.7–95.9%. Each sample exhibited good linearity of the σ–ε plot. The reproducibility of the measurement was high and no crack formation was observed in the alumina samples after several repeats of the loading and unloading experiments. As seen in , the slope (Young's modulus) of the straight line became larger with increasing relative density (D). The Young's modulus for dense alumina (D = 0.9907) hot-pressed at 1773 K was 282.76 GPa (E2) in our previous paper summarises the ratio of Young's moduli between the porous (Ep) and dense (E2) alumina sintered at 1073–1873 K for 1–4 h. The alumina compacts contained almost 100% open pores in the porosity range larger than 40%. In the porosity range smaller than 40%, mixed pores (open pores and closed pores) were formed as seen in . The Ep/E2 ratios were scattered at similar porosity, indicating that the Ep value is very sensitive to the porous microstructure. Scattered Ep/E2 ratios at 50–10% porosity were observed between the two curves produced by Eqs. for the porous structures with open pores. It is apparent that the Young's moduli by Eqs. derived for porous structures with dispersed (closed) pores () give E values higher than the measured Ep values or the calculated Ep values for open-pore structures. When the porosity decreased less than 10%, the measured Ep/E2 ratios suddenly jumped and approached unity. In this porosity range, the theoretical Ep/E2 curve by Eq. approaches the theoretical Ep/E2 curve by Eq. for the closed-pore structure, indicating no significant influence of pore structure on the theoretical Young's moduli for the alumina compacts with 10–0% porosity. The measured Ep/E2 ratios were observed around the three theoretical curves produced by Eqs. , showing the decreased sensitivity of Young's modulus to the pore structure. As mentioned above, the scattered Young's moduli at 50–10% porosity are well related to the development of grain boundary area for the open-pore structure. With decreasing porosity at 10–0% porosity, the measured Young's moduli increased along the theoretical Young's modulus curve for the closed-pore or open-pore structure. shows the relationship between the thermal expansion and heating temperature for the alumina compacts with the porosities of 2.3–39.6%. The thermal expansions were measured on the basis of the lengths of porous alumina at room temperature (L1). As seen in , the thermal expansions were independent of the porosity of sintered alumina. shows the TEC (β) values at 600–1000 K for the sintered alumina as a function of porosity. The measured β values, which increased with increasing heating temperature, were almost independent of porosity and close to the reported β values at 0% porosity . The parallel structure model for dispersed (closed) pores in (b) and the porous structure model with continuous (open) pores in , respectively, which are independent of porosity. On the other hand, Eq. for the series structure model with dispersed (closed) pores in (c) shows a minimum β value at 29.6% porosity. Comparison of the measured and calculated β values suggests no dependency of β values on porosity based on the open-pore structure at 50–10% porosity and the parallel structure model of closed pores at 10–0% porosity. The above discussion suggests that the theoretical Ep/E2 curve is invalid for the series structure with dispersed pores (Eq. The Young's modulus (E) and thermal expansion coefficient (TEC, β) for a porous structure with continuous (open) pores were theoretically derived on the basis of a previously derived equation for the compressive strength, which is dominated by the number of grains and the grain boundary area in a sintered porous compact.The measured ratios of Ep (for porous alumina compacts)/E2 (for fully dense alumina compacts) were scattered at similar porosity, indicating that the Ep value is very sensitive to the porous microstructure.The scattered data on Ep/E2 ratios for porous alumina compacts increased with decreasing porosity (50% → 10% porosity), which were located between the two theoretical Ep–porosity curves calculated for the open-pore structure.When the porosity decreased by less than 10%, the measured Ep/E2 ratios suddenly jumped and approached unity. This tendency was well explained by the Ep–porosity curves developed newly for the continuous (open)-pore structure and also for the dispersed (closed)-pore structure treated previously.The measured β values for the porous alumina compacts, which increased with increasing heating temperature, were independent of porosity and close to the reported β values at 0% porosity.The above result for the β values of porous alumina compacts is in accordance with the theoretical prediction of the β–porosity relationship for closed- or open-pore structures.Effects of molecular weight on poly(ω-pentadecalactone) mechanical and thermal propertiesA series of poly(ω-pentadecalactone) (PPDL) samples, synthesized by lipase catalysis, were prepared by systematic variation of reaction time and water content. These samples possessed weight-average molecular weights (Mw), determined by multi-angle laser light scattering (MALLS), from 2.5 × 104 to 48.1 × 104. Cold-drawing tensile tests at room temperature of PPDL samples with Mw between 4.5 × 104 and 8.1 × 104 showed a brittle-to-ductile transition. For PPDL with Mw of 8.1 × 104, inter-fibrillar slippage dominates during deformation until fracture. Increasing Mw above 18.9 × 104 resulted in enhanced entanglement network strength and strain-hardening. The high Mw samples also exhibited tough properties with elongation at break about 650% and tensile strength about 60.8 MPa, comparable to linear high density polyethylene (HDPE). Relationships among molecular weight, Young's modulus, stress, strain at yield, melting and crystallization enthalpy (by differential scanning calorimetry, DSC) and crystallinity (from wide-angle X-ray diffraction, WAXD) were correlated for PPDL samples. Similarities and differences of linear HDPE and PPDL molecular weight dependence on their mechanical and thermal properties were also compared.Polyethylene is the most widely used commodity polymer. It is found in many consumer products, such as milk jugs, detergent bottles, margarine tubs, garbage containers, water pipes, just to name a few. Poly(ω-pentadecalactone) (PPDL) is a new type of thermoplastic that can be synthesized by lipase catalysis ). Polyethylene (PE) cannot be easily decomposed into small molecules after usage. To achieve extensive degradation of the PE carbon backbone, treatment of PE with strong oxidized agents such as nitric acid Chemical catalysts such as potassium alkoxides The current study aimed to investigate the effect of PPDL molecular weight on its mechanical, thermal and rheological properties. The chosen synthetic methods enabled the preparation of PPDL with Mw values up to 48.1 × 104 and PDI values close to 2.0. Films were prepared by press-molding at 130 °C and tensile testing was performed on these samples. Based on the shape of stress–strain curves, a brittle-to-ductile transition along with maximum elongation at break was observed. Since the chemical structure of PPDL is similar to polyethylene (PE), its thermal and mechanical properties were compared with those of a commercially obtained PE sample Samples of ω-pentadecalactone (PDL, 98%) and anhydrous p-xylene (>99%) were purchased from Aldrich Chemical Co. and were used as received. Chloroform was purchased from PHARMCO-AAPER Inc. (>99.9%). Anhydrous toluene (98%), purchased from Aldrich Chemical Co., was dried over sodium and then was distilled under nitrogen. Novozym 435 (specific activity 10,000 PLU/g) was a gift from Novozymes (Bagsvaerd, Denmark) and consists of Candida antarctica Lipase B (CALB) physically adsorbed within the macroporous resin Lewatit VPOC 1600 (poly[methyl methacrylateco-butyl methacrylate], supplied by Bayer).Pentadecalactone (PDL, 40 g) was polymerized to prepare PPDL samples 1–3 at 70 °C, with magnetic stirring, in toluene (monomer:toluene = 1:2 wt/v), using Novozym 435 as the catalyst for predetermined reaction times (see The method used was a variation of that described by de Geus Both 1H and 13C NMR spectra were recorded at room temperature on a DPX300 spectrometer (Bruker Instruments, Inc.) at 300 MHz in chloroform-d. Chemical shifts (in parts per million) for 1H and 13C NMR spectra were referenced relative to tetramethylsilane as an internal reference at 0.00. All synthesized PPDL materials described herein had identical NMR spectra with signals and assignments as follows: –[C(O)–CH2b–CH2c–(CH2d–CH2d–)5–CH2c–CH2a–O]– 1H NMR (CDCl3, δ): 4.01 (t, J 6.5 Hz, CH2aO); 3.58 (t, J 6.5 Hz, CH2OH); 2.24 (t, J 7.5 Hz, CH2bCO); 1.59, 1.22 (brs, CH2c,d) ppm. 13C NMR (CDCl3, δ): 173.9 (COCH2), 64.4(CH2aO), 34.4(OCOCH2b), 29.6–29.1, 28.6, 25.9, 25.0 (all other carbons) ppm.To determine the absolute molecular weight of PPDL samples, a Wyatt HELLOS multi-angle light scattering detector and a Wyatt Optilab rEx differential refractive index detector were used. These two detectors were connected by a steel tube with inner diameter 0.5 mm and a CRZEL syringe pump. The solution concentration ranged from 5 × 10−4 to 5 × 10−3

g/mL. Astra V software was used to acquire and process the data according to Zimm plots.The relative molecular weight of PPDL samples were determined by gel permeation chromatography (GPC) using a Waters HPLC system equipped with a model 510 pump, model 717 autosampler, and model 410 refractive index detector with 500, 103, 104, and 105

Å Ultrastyragel columns in series. Waters Empower GPC software (Version 3, Viscotek Corp.) was used for data analysis. Chloroform was used as eluent at a flow rate of 1.0 mL/min. Sample concentrations and injection volumes were 0.2% w/v and 100 μL, respectively. The number-average molecular weight (Mn) and weight-average molecular weight (Mw) were determined based on a calibration curve generated by narrow molecular weight polystyrene standards (Aldrich Chemical Co). GPC and light scattering determined values of Mw were similar (within 12%). For PPDL molecular weight values reported herein, light scattering Mw and GPC Mw/Mn values were used (see To separate low molecular weight material in sample 5, corresponding to GPC peaks observed at longer retention times, the sample was fractionated as follows. A solution containing 20 mL of chloroform and 0.2 g of Sample 5 was prepared. The mixture was maintained at 20 °C for 24 h with continual mixing by magnetic stirring. The ‘insoluble’ PPDL fraction was separated by filtration and solvent was removed in a vacuum oven at 40 °C for 7 days, giving 0.17 g of ‘insoluble’ solid sample. Solvent removal from the ‘soluble’ PPDL fraction was carried out by rotor-evaporation and subsequent drying in a vacuum oven, where the recovered ‘soluble’ PPDL sample was 0.03 g.Dumbbell shaped sample bars with dimensions of 20.0 mm (length) × 4.0 mm (neck width) × 1.5 mm (thickness) were prepared by press-molding at 130 °C and subsequent quenching at ambient temperature. An Instron 5542 tensile testing machine with a 500 N load cell was used for mechanical study (the crosshead speed was 3 mm/min and the test temperature was 25 °C). The Merlin software was used to collect and analyze the tensile results (stress was calculated according to the initial cross-section area). The values of tensile strength, Young's modulus, elongation at yield and break, stress at yield were obtained by averaging the data obtained from more than 4 specimens.PPDL samples were molded into rectangular bars with dimensions of 30 mm (length) × 5 mm (width) × 1.5 mm (thickness). DMA measurements were performed in single-cantilever bending mode using a dynamic mechanical thermal analyzer (DMTA) (Tritec 2000 DMA, Triton Technology Company). Measurements were performed from 30 °C to 95 °C at a heating rate of 2 °C/min and frequency of 1 Hz. Two identical specimens with the same molecular weight were evaluated and results reported were mean values. The Triton Technology DMA Software was used to acquire and process the data.DSC measurements were performed using a differential scanning calorimeter (Model 2920, TA Instruments). Temperature calibration was carried out using an indium standard. Measurements were performed under a nitrogen atmosphere at a flow rate of 50 mL/min. Typical parameters for experimental measurements are as follows: i) sample cooled to 10 °C, ii) heated to 150 °C at 10 °C/min, iii) held at 150 °C for 3 min and then iv) cooled to 10 °C at 10 °C/min. When temperature reached 10 °C, the sample was heated again to 200 °C at 10 °C/min. The melting temperature, melting enthalpy, crystallization peak temperature, and crystallization enthalpy were analyzed by using the TA Universal Analysis software.Wide-angle X-ray diffraction (WAXD) and small-angle X-ray scattering (SAXS) experiments using the dumbbell shaped bars for tensile testing were carried out at the X27C beam line at the National Synchrotron Light Source (NSLS), Brookhaven National Laboratory (BNL). The wavelength of synchrotron radiation was 1.371 Å. A three-pinhole collimation system was used to reduce beam size to 0.6 mm in diameter. Two-dimensional (2D) WAXD and SAXS patterns were collected using a MAR CCD X-ray detector (MAR-USA), which had a resolution of 1024 × 1024 pixels (pixel size = 158.44 μm). The typical image acquisition time was 30 s for each data frame. Sample-to-detector distance was 1923.7 mm for SAXS (calibrated by a silver behenate, AgBe, standard) and 116.4 mm for WAXD (calibrated by an aluminum oxide, Al2O3, standard). All X-ray images were corrected for background scattering, air scattering and beam fluctuations. The obtained 2D SAXS and WAXD patterns were analyzed using the POLAR software to obtain one-dimensional SAXS and WAXD profiles. One-dimensional WAXD profiles were then processed by the linear square method to obtain crystallinity.Previous work by our laboratory demonstrated that polymerization of PDL for 2 h at 70 °C in dry toluene (PDL to toluene 1:2 wt/vol) with 10 w/w-% monomer-to-catalyst gave PPDL (without fractionation) with Mn 7900 g/mol (GPC relative to polystyrene) ). By increasing the reaction time from 8 to 16 and 26, samples 4, 5 and 6 were prepared having Mw values (×10−4, determined by light scattering) of 18.8, 28.0 and 48.1 g/mol, respectively. For the above, PPDL was synthesized in quantities up to 40 g and yields from 75 to 80% after precipitation.Dumbbell shaped sample bars were prepared (see , above) for tensile testing. Sample 1 (Mw 2.5 × 104) was found to be too brittle to be tested. Stress–strain curves for Samples 2–6 are illustrated in f displays four regions according to slope change in the stress–strain curve. Manson et al. . Inspection of PPDL samples before stretching and after fracture showed that the volume of the bar subjected to high strain was found to be about two times that of the original volume (i.e. 50% polymer and 50% void formations). This observation is in agreement with that by Ward was estimated by the load divided by the initial cross-sectional area, for samples 4, 5, and 6, the true stress values at break were recalculated using the value of stress at break (in ) multiplied by a factor (1 + strain)/2. For samples at lower strains (samples 2, 3), the true stress values at break were recalculated by the value of stress at break multiplied by a factor (1 + strain), i.e., neglecting the effect of volume increase at lower strains. The recalculated values of true stress at break are listed in . Variations of Young's modulus, true stress at break, elongation at break, and strain and stress at yield as a function of PPDL molecular weight are shown in reveal unambiguously that they are dependent on PPDL molecular weight. a illustrates that, for sample 2 (Mw

= 4.5 × 104), stress first increased rapidly with strain but then the slope of stress–strain curve began to decrease. When strain reached 4.5 ± 0.8%, the sample fractured (the fracture was homogeneous as is typical for brittle fracture), while no neck was observed before fracture. As Mw increased to 8.1 × 104, the shape of the stress–strain curve (b) deviated substantially from that of sample 2. For sample 3, necking and plastic flow were observed during stretching; stress first reached the maximum value and then decreased, followed by stable propagation before fracture. When the sample bar length reached 237 ± 25% of the initial length, the sample bar fractured. Thus, with an increase in the molecular weight of about 2 times from sample 2 to sample 3, the elongation at break increases almost 30 times and the fracture fashion of PPDL changes from brittle-to-ductile. In addition, a stress-whitening phenomenon was observed during stretching of sample 3, which is quite different from sample 2. Interestingly, for PPDL of Mw

= 18.9 × 104 (sample 4), the elongation at break increased to 650 ± 30% while the strain-hardening phenomenon was also observed (c). With further increases in Mw, samples exhibited a similar stress–strain curve shape as that of sample 4 and values of elongation at break remained almost constant (e.g. 700 ± 70% for Mw

= 28 × 104 and 580 ± 30% for Mw

= 48.1 × 104). This suggests that for PPDL, it is not necessary to synthesize polymers with molecular weights above Mw

= 18.9 × 104 to attain higher draw ratios since the elongation at break remains almost constant above Mw

= 18.9 × 104.The plot of true stress at break (tensile strength) versus Mw (b) showed a similar trend as that of elongation at break versus Mw. The stress at break first increased from 21.3 MPa to 60.8 MPa as Mw increased from 4.5 × 104 to 18.9 × 104 and then remained almost constant at high Mw. The curve of Young's modulus versus Mw (c) exhibited a minimum at Mw

= 28.0 × 104. Young's modulus first decreased from 690 MPa to 290 MPa with increasing Mw, and then increased to 390 MPa. However, the corresponding curve of strain at yield versus Mw first increased with Mw from 12.0 to 20% and then decreased slightly to 17.4%. The trend of stress at yield versus Mw is almost opposite to that of strain at yield versus Mw, since the stress first decreased from 24.1 to 13.3 MPa, and then increased to 18.2 MPa with increasing Mw.Thermal analysis of several samples with different molecular weight was carried out after processing samples into dumbbell shaped bars prior to tensile testing. displays DSC curves from first heating (c) scans of PPDL samples (1–6). The trend of peak melting temperature with Mw in heating is similar to that in cooling curves. Values of melting enthalpy (ME) and melting temperature (MT) during first and second heating scans, and values of crystallization enthalpy (CE) and crystallization peak temperature (CPT) during cooling scans are summarized in . Since the trends of change in MT and ME during the first heating scan are similar to those during the second heating scan, only values of ME and MT in first heating and values of CE and CPT in cooling are plotted in a indicates that ME first increased from 134.0 J/g to 164.4 J/g and then decreased afterward with increasing Mw. When Mw was 28.0 × 104, ME reached a minimum value (101.6 J/g); when Mw was 48.1 × 104, ME climbed back up to 115.3 J/g. Similarly, MT first increased from 97 °C to 103.2 °C, and then decreased to 91.7 °C, and again increased to 99.4 °C with increase of Mw. Upon crystallization of PPDL during cooling, the changes of CE and CPT with Mw were similar to those of ME and MT (b). Generally, the value of CE was smaller than that of ME in the same sample, possibly due to recrystallization upon heating To compare DSC results, wide-angle X-ray diffraction (WAXD) experiments were performed on the same dumbbell shaped sample bars prior to tensile testing. Fractions of crystalline phases plotted in were obtained by peak deconvolution of integrated WAXD profiles. The trend of crystallinity change with Mw was similar to the variation of ME and CE with Mw. To compare the DSC and X-ray results, the values of crystallization enthalpy from cooling (instead of the melting enthalpy) were used in order to eliminate the effect of recrystallization during DSC scanning. The equilibrium melting enthalpy of PPDL had been estimated to be 264 J/g and 233 J/g, respectively, by Wunderlich a illustrates curves of storage modulus vs. temperature for PPDL samples with different molecular weights. For each sample, the storage modulus decreased with increasing temperature and no distinct transitions were observed. The results in samples 3 and 4 exhibited an intersection at 85 °C, while samples 5 and 6 exhibited an intersection at 56 °C. Storage moduli at 40–90 °C for five samples are listed in b. This figure indicates that the trend of storage modulus changes with Mw at 40 °C was similar to that of the Young's modulus with Mw (c). Interestingly, as the temperature increased from 40 to 60 °C, the discrepancy between samples 5 and 6 decreased. For temperatures above 60 °C, the storage moduli of samples 5 and 6 are the same. This implies that above 60 °C, the molecular networks formed for samples 5 and 6 have the same elastic rigidity.This section considers the effects of Mw on PPDL mechanical, thermal and crystallization properties and compares these to those of linear PE.The effect of molecular weight of linear HDPE on its tensile properties was studied by Ward et al. c, Young's modulus first decreases to a minimum (at sample 5, Mw 28.0 × 10−4) and then increases. Without sample 5, behavior of PPDL would be similar to linear PE. Perhaps this can be rationalized as follows. Several other studies have indicated that the degree of crystallinity is the primary factor affecting the Young's modulus of semi-crystalline polymers ). These low Mw fractions would be expected to decrease crystallinity and lower the Young's modulus. To verify this hypothesis, the following experiment was performed. Sample 5 was first immersed in chloroform for 24 h at 20 °C to solubilize a large part of low Mw fractions. The remaining insoluble material after thorough removal of solvent was subjected to GPC and DSC analyses. The GPC chromatogram displayed in indicates that a substantial part of low molecular weight fractions were removed by solubilization in chloroform. DSC results indicate that, after solvent extraction, the melting peak of sample 5 shifted to a higher temperature during both first and second heating scans (). The melting temperature, melting enthalpy for the first and second heating, the cooling crystallization peak, and cooling crystallization enthalpy are listed in for comparison with data for non-fractionated sample 5. Indeed, the value of melting enthalpy increases after extraction and falls between those of samples 4 and 6. Thus, it appears that the crystallinity and Young's modulus of PPDL have a similar molecular weight dependence as linear PE based on studies by Ward et al. Since crystallinity plays a major role to influence Young's modulus, their relationship was investigated. The crystallinity was calculated by dividing the cooling crystallization enthalpy with the equilibrium enthalpy value of 264 J/g a. The data set can be represented by a straight line without the extrapolation to the origin. This behavior was also observed by Mandelkern et al. , which will be used to explain the different observed properties in the samples.The plot of yield stress versus crystallinity is shown in b, where the behavior can be rationalized by the occurrence of one or both of the following phenomena where τy is the shear yield stress, Lc is the crystal lamellar thickness, κ is a function of the crystal shear modulus, b is the Burgers vector having the same value as the PPDL unit cell c-axis, ΔGc is the critical activation energy for dislocation growth, with values in the range between 40 and 80 κT. To correlate the relationship between yield stress and crystallite thickness, SAXS experiments were carried out where the results (the Lorentz-corrected SAXS profiles) are shown in . These profiles exhibit two peaks at very low q values. To obtain the long period L and crystallite thickness Lc, the one-dimensional correlation function K(r) is calculated by the following equation where I(q) is the scattering intensity and r is the distance for which the electron density correlation is measured. The long periods for samples 3, 4, 5 and 6 were 17.5 nm, 18.9 nm, 19.3 nm and 18.1 nm, respectively. Assuming the crystal thickness should be thicker than the amorphous regions in the two-phase model, the estimated crystalline thickness values are 10.7 nm, 12.2 nm, 13.4 nm, 11.1 nm, respectively. The close values of the crystalline thickness lead us to conclude that the screw dislocation theory is not appropriate to explain yield stress data.a). Interestingly, the corresponding true stress at break also reaches an asymptotic value of about 60.0 MPa at Mw

= 18.9 × 104 (b). One possible explanation for the discrepancy is as follows. Since the PPDL chain consists of C–C and C(O)–O (or–O–C–) bonds along the backbone, where the C–C bond has larger bond energy than C(O)–O (or –O–C–) bonds. Hence, the stress distribution should be heterogeneous along the chain during stretching. In high Mw samples (>18.9 × 104), increased Mw would lead to enhanced strength of the entanglement network in the molten melt as well as in the inter-lamellar amorphous region when solidified. However, as the C–O bond is weaker than the C–C bond, the former probably breaks at a lower stress under deformation, thus limiting the ultimate strength of PPDL. This hypothesis is consistent with the observation of constant stress as well as the constant elongation at break value above Mw

= 18.9 × 104 under deformation. This may also explain why PE has a much higher draw ratio (18 or higher) than PPDL of the same or higher Mw, as the PE chain has only carbon–carbon bonds The brittle-to-ductile transition in semi-crystalline polymers can be affected by many factors such as deformation rate and crystallinity ) is in agreement with the Mw range observed for PE. In addition, quenched PPDL sample 3 exhibited ductile behavior, while the annealed sample with higher crystallinity revealed brittle behavior. This finding agrees with Mandelkern's experiments on linear fractioned PE with Mw

= 7.0 × 104Finally, we caution that when assessing the mechanical properties of PPDL samples of different molecular weights, in addition to crystallinity, crystallite thickness, inter-lamellar thickness and density of physical entanglements in the amorphous region, the fraction of the interfacial region and supermolecular morphology should also play an important role Synthesis of large-scale PPDL samples was performed by lipase catalysis. Variation of PPDL Mw from 2.5 to 48.1 × 104 was achieved by manipulating reaction variables including reaction water content, method of mixing and reaction time. Tensile testing, DSC, X-ray, DMA and GPC were used to investigate the effect of molecular weight on mechanical, thermal and crystalline material properties. Cold-drawing tensile tests at room temperature revealed a brittle-to-ductile transition for PPDL samples with Mw values between 4.5 × 104 and 8.1 × 104. For PPDL with Mw

= 8.1 × 104, the entanglement network strength in non-crystalline regions is not sufficiently high to transmit stress during stretching, whereby inter-fibrillar slippage dominates until fracture. As PPDL Mw is increased from 18.9 × 104 and above, the entanglement network strength is greatly enhanced and strain-hardening takes place at high strains prior to failure. Furthermore, the elongation at break and tensile strength (i.e., true stress at break) reach asymptotic values of 650% and 60.8 MPa, respectively. The trends for changes in Young's modulus, melting enthalpy (from DSC) and crystallinity (from WAXD) as a function of molecular weight are all similar. Abnormality of one sample (5) was explained by its contamination with high levels of a lower molecular weight fraction. Storage modulus (from DMA) revealed similar molecular weight dependence trends at temperatures below 60 °C. However, above 60 °C, storage moduli of higher molecular weight samples becomes indistinguishable, indicating that the crystalline network structures of these samples at higher temperatures are also similar. Overall comparisons between PPDL and linear high density polyethylene (HDPE) mechanical properties showed similar trends in Young's modulus with molecular weight, but differing trends with respect to elongation at break and true stress at break as a function of molecular weight. The latter differences are explained by the presence of a low but persistent density of C–O bonds in ester links of PPDL that are not present in PE. We therefore conclude that PPDL-like polyesters have excellent potential to function in similar ways to PE. This bodes well to the potential further development of similar or related polymers for commercial use. Indeed, our laboratory has developed biocatalytic methods using an engineered Candida tropicalis strain to convert fatty acids, such as tetradecanoic acid, in volumetric yields of up to 160 g/L to their corresponding ω-hydroxyfatty acid (e.g. ω-hydroxytetradecanoic acid) (manuscript in review). Subsequent conversion of ω-hydroxyfatty acids to polyesters by condensation polymerization will provide a low-cost scalable route to biobased PPDL-like materials that can function in similar ways to PE.Experimental investigation of strength, stiffness and drift capacity of rubble stone masonry wallsThere is limited available research on rubble stone masonry walls, which are vulnerable under seismic loading. This paper presents an experimental campaign of cyclic shear compression tests on six large-scale walls of this topology. The effect of the axial load and shear span ratio on the wall behaviour, notably on the wall stiffness, strength, and drift capacity, was investigated. It was found that the drift at crack onset is only half of that in previous campaigns on stone masonry walls, likely because one face of each wall was plastered, making the damage more visible. Additionally, splitting cracks opening between the wall leaves appear to play a key role in the collapse mechanism. Finally, testing the walls up to the loss of their axial-load-bearing capacity provides new input for the collapse risk analysis of stone masonry buildings.Stone blocks were often used in historical buildings due to their aesthetic beauty, low cost, durability, and the availability of natural stone Previous earthquakes highlighted the high vulnerability of stone masonry works Performance-based assessment methods for masonry buildings require the drift capacity of the walls at one or several limit states, where “drift” is the relative horizontal displacement between the top and bottom of a wall divided by its height. Drift capacity models for masonry walls in general and stone masonry walls, in particular, are still under development. Available models in current codes have not been tailored for stone masonry walls, and the effect of this wall typology has not yet been included An analysis of the stone masonry database shows the tests available in the database published in Vanin et al. , limited data is available on walls of typology A that are tested at a low axial load ratio (i.e., the ratio of the applied vertical stress to the compression strength of masonry).In this study, the influence of the axial load ratio and the shear span ratio on the in-plane stiffness, strength, and drift capacity of walls of typology A is investigated. To do this, six quasi-static cyclic shear compression tests on large-scale rubble stone masonry walls and six material tests on rubble stone masonry wallettes were carried out at the École Polytechnique Fédérale de Lausanne (EPFL). The primary aim of this study is to generate new experimental data on the cyclic behaviour of rubble stone masonry walls to contribute to the available database and to assess to what extent the available models for stiffness, strength, and drift capacity can predict this new data. In addition, this test series investigated the drift capacity at axial load failure, i.e., the point where the walls were no longer able to support the constant axial load that was applied during the test, putting this among few campaigns on rubble stone masonry walls where the walls are tested to this point. This is a much-needed input parameter for the seismic risk estimation of stone masonry buildings. describes the experimental approach, construction material and instrumentation; discusses the main findings, such as the influence of the axial load ratio and shear span ratio on the failure mode, stiffness, strength and drift capacity, including the drift at axial load failure. The final section summarizes the key findings of the study.This section describes the test units, test set-up, loading protocol and instrumentation for the shear-compression tests. In addition, it presents the results of the material tests on mortar, plaster, and masonry.The experimental campaign involved shear compression tests on six large-scale walls (specimen label: RS) of the dimensions 1600 mm × 1600 mm × 400 mm (H × L × t). In addition, diagonal compression tests on three wallettes (RSD) and simple compression tests on three wallettes (RSC) were carried out. The size of the wallettes used for the diagonal compression tests was 900 mm × 900 mm × 400 mm. Two wallettes with the dimensions of 900 mm × 900 mm × 400 mm and one with the dimensions of 900 mm × 800 mm × 400 mm were used for the simple compression tests.The walls were constructed by two experienced masons using uncut limestone blocks with dimensions between 10 and 30 cm and pebbles with a maximum size of around 10 cm. At each layer, the stones of the two outer layers were placed first, and then the area between the leaves was filled with stone chips and pebbles (). To represent old building material characteristics a and b show the constructed large-scale walls and a section of one of the walls during construction. c shows a wall with the applied plaster. A strip of around 1–2 cm in height at the top and bottom of the wall was left unplastered to avoid direct loading onto the plaster layer during testing.During construction, material samples were taken from each batch of mortar and plaster, and prisms of dimensions 160 mm × 160 mm × 40 mm were cast. To characterize the mortar and plaster, three-point bending tests on the mortar prisms and compression tests on the halves resulting from the three-point bending tests were performed according to EN 1015-Part 11 , the results of material tests on the mortar and plaster samples are given. The terms ftmo, fcmo and ftp, fcp are the tensile and compression strength of the mortar and plaster, respectively.To determine the mechanical properties of masonry as a composite, simple compression tests (i.e., with no confinements at the top and bottom of the specimens) and diagonal compression tests were conducted on stone masonry wallettes according to the described procedures in EN 1015-Part 1 summarizes the Young’s modulus and compression strength (fc) of the masonry. The obtained compression strength (0.76 MPa) is smaller than the lower bound of the range (1–1.8 MPa) suggested by the Italian code To obtain the tensile strength of masonry, diagonal-compression tests were performed on square wallettes. The test setup for diagonal compressions tests () consisted of two V-shape steel shoes placed at the two corners of wallettes. The wallettes were rotated by 45 degrees and placed on the top of a steel beam. The force was applied vertically using a hydraulic actuator connected to the bottom steel beam, while the top beam was fixed. The deformation of wallettes was measured by taking images with two sets of a stereo camera at both sides of specimens and using the DIC method. In this test series, the applied load was increased monotonically. The tensile strength of the masonry was calculated as follows:where Pmax is the maximum applied vertical force, A=(L+H)t/2 is the net area of the specimen and α is the coefficient correcting for the stress state. Note that the interpretation of the diagonal compression test depends on the assumed stress state summarizes the computed tensile strength using various values for α that were suggested in the literature. For stone masonry walls of typology A, the tensile strengths included in the Italian code depicts the set-up for the shear compression tests. To apply the pre-compression load and the boundary condition, three servo-hydraulic actuators with a force capacity of ±1000 kN and a displacement capacity of ±500 mm were used. The forces were measured by load cells. The tests were performed in two phases. In the first phase, the axial load was applied through three loading/reloading cycles using two vertical actuators. In the second phase, horizontal displacements were applied following a loading protocol () in which two cycles were performed for each drift target level while the axial load was kept constant. To keep the height of the zero-moment constant along the moment profile, the two vertical actuators were force-controlled and were coupled to the force applied by the horizontal actuator, as described by Godio et al. In this study, the axial load ratio applied at the top of the wall (σv,top/fc, where σv,top is the vertical stress applied at the top of the wall) and accordingly at the bottom of the wall (σv,bot/fc, where σv,bot is the vertical stress at the bottom of the wall), and the height of zero-moment to height of the wall ratio (H0/H) were varied for different tests as indicated in . The three axial load ratios correspond to average axial stresses between 0.06 and 0.19 MPa. This range was selected to increase the number of specimens tested under low and moderate levels of axial load ratios in the current database (see c). For the H0/H ratio, a value of 0.5 corresponds to a double-bending boundary condition. For a value of 1, the height of zero moment is at the top of the wall (i.e. lower edge of the steel plate, see ). For a value of 1.5, the height of zero moment is above the wall, i.e., the top and bottom moment have the same sign. A similar test design was used Petry and Beyer In this testing campaign, two types of measuring systems were used: (i) classical hard-wired instruments to measure the forces and selected displacements and (ii) an optical measurement system that used digital image correlation (DIC) to measure the 3D displacement field on both sides of the wall.The hard-wired instrumentation consisted of 14 linear variable differential transformers (LVDTs) denoted as “lv” and arranged as depicted in . Four LVDTs (lv1-lv4) were used to monitor the loading beam movement during the tests. Two LVDTs (lv1 and lv2) measured the top and bottom horizontal displacement of the loading beam in regards to a reference column (). Two further LVDTs (lv3 and lv4) at the two extremities of the beam were used to compute the rotation of the loading beam. These four LVDTs (lv1-lv4) were only used during the testing phase. LVDTs lv5-lv7 and lv12-lv14 were placed on the north and south sides of the walls to measure the separation of the leaves. The LVDTs were mounted approximately 400, 800 and 1200 mm above the wall base with a base length of around 25 cm. Additionally, to obtain local deformations of the compressed zones, four LVDTs (lv8-lv11) measured the vertical displacements on the north face of the walls. The forces applied by the three actuators were measured using load cells.The optical measurements consisted of two stereo-camera systems placed on both sides of the wall that acquired high-resolution grey-scale images. The digital image correlation measurements followed the procedures suggested by the International Digital Image Correlation Society ). Additionally, some speckled patches were placed on the foundation, plate and loading beam to track their displacements (). The cameras were calibrated by taking images before the test using a standard calibration target plate. To compute the 3D displacement field, the software VIC-3D version 8.2.4 were derived from the DIC measurements.The cyclic shear force-horizontal displacement/drift hystereses of the six tested walls are plotted in . The horizontal displacement was computed as the average value of the horizontal displacements of the speckle patterns on the steel plate ( also depicts the envelope and bilinear curves, together with the hystereses. The envelope curves are derived by connecting data points corresponding to displacements that the wall experiences for the first time throughout the applied loading history The effective stiffness is computed as the secant stiffness at 70% of the peak force VP; the ultimate drift is the drift at which the shear resistance has dropped by 20% VP; and the ultimate force is defined such that the area under the bilinear curve is equal to the area under the envelope curve. The mentioned procedure was applied for all walls except RS5, as there was not a 20% drop of force in the negative direction. For this wall, the maximum drift that the wall reached was considered to be the ultimate drift. For RS5, the test was stopped at this point because the sudden occurrence of a diagonal crack and the sudden increase in vertical displacement indicated that wall collapse was imminent (see The evaluation of the progressive damage and the failure mode of the walls was based on a visual inspection during the tests, the analysis of the photographs that were taken and conclusions inferred from the shape of the hystereses of the walls (). Two types of damage were considered in the evaluation. The first was the in-plane damage, i.e., damage that manifested itself through cracks in the masonry (east) and plaster (west) sides of the wall. The second was the development of out-of-plane deformation caused by an opening between the masonry leaves or by the detachment of the plaster from the masonry.Because the crack initiation was more visible on the plastered face (west), the observations presented here are mainly based on the crack pattern of the plastered face. However, at near collapse, the cracks also became apparent on the masonry side. show the damage patterns at the last drift level before the axial load failure, i.e., before the wall could no longer carry the applied vertical load.The wall RS1 was subjected to a double-bending moment profile and the lowest of the three axial load ratios (σv,top/fc = 8%). The first visible cracks, which appeared at a small drift demand (0.04%), were a horizontal crack at the interface between the wall and the foundation and a horizontal crack below the top row of stones. A diagonal crack started to form as the drift amplitude increased (drift = 0.15%). The diagonal cracks started at the centre of the wall and propagated towards the corners, suggesting that the behaviour changed from flexure-dominated to shear-dominated. This is supported by the loops of the hysteresis curves, which are narrow at small drift levels, confirming the flexure-dominated response. With increasing drift demands (from approximately 0.15% onwards), the loops widened and the behaviour was increasingly dominated by shear deformations.The walls RS2 and RS3 were, as RS1, subjected to a double-bending moment profile, though with a higher axial load ratio, i.e., 25% and 17%, respectively. These two walls developed a very similar crack pattern, with first visible cracks appearing at the centre of the walls and propagating diagonally towards the corners at a 0.04% and 0.08% drift demand for RS2 and RS3, respectively. The high energy dissipation and high stiffness degradation observed in the post-peak regime of the hysteresis curves point towards a pure shear failure mode.The wall RS4 was tested as a cantilever under an axial load ratio of 25%. Its cracking sequence was similar to that of RS1: First, a horizontal crack was observed along the interface between the wall and foundation (drift = 0.09%), and then diagonal shear cracks followed.The wall RS5 was the only one tested with a shear span ratio of H0/H=1.5, and it was subjected to an axial load ratio of 25%. Like for walls RS1 and RS4, the first crack appeared at the interface between the wall and the foundation. For each loading direction, a horizontal crack developed at mid-height of the wall that changed to reach diagonally from the wall edge to the left and right bottom corners. At a drift demand of 1.39%, a sudden diagonal crack appeared, and the previous horizontal cracks closed, which marked the transition from a flexure-dominated to a shear-dominated response. This transition is also reflected in the hysteretic response, which is characterized by a nearly constant peak force over a large drift range. The appearance of the diagonal crack led to a sudden drop in the horizontal force in the positive direction (). At this point, the vertical displacement also suddenly increased, and the test was therefore stopped.The wall RS6 was tested as a cantilever with an intermediate axial load ratio of 17%. The first visible cracks were similar to those observed for RS1, RS4 and RS5 (horizontal cracks at the interface between the wall and foundation). As for RS5, two horizontal cracks appeared at mid-height at the extreme borders of the wall at a drift demand of 0.04%. At the final load steps (after 0.25%), a shear diagonal crack developed in the centre of the wall that extended towards the corners. This is also reflected in the hysteretic response, which began as a nearly elastic nonlinear response that indicated a rocking behaviour. At a drift of 0.25%, the hysteretic loops became fatter, indicating the onset of a shear failure.In this campaign, out-of-plane deformations due to either splitting cracks between the two masonry leaves (b) were observed. The deformations due to splitting cracks opening between the masonry leaves are reported in other experimental campaigns performed on multi-leaf stone masonry walls The opening between the masonry leaves was recorded by LVDTs lv5-7 (north face) and LVDTs lv12-14 (south face), see . Out-of-plane deformations of the east and west faces were also optically recorded. As an illustration, the displacements measured by the LVDTs placed on the north and south cross sections of wall RS2 are plotted in . The opening between the leaves accumulated over the cycles and increased significantly after the 20% drop in horizontal force in the positive direction. summarizes the opening of the leaves at different levels of force in the post-peak regime for all the tested walls. In these tests, as the axial load increased, the opening of the leaves increased. Moreover, in general, the higher the shear span ratio, the larger the opening became. Assuming a no-tension elastic material law for a masonry wall, the compressed length at the base of the wall can be expressed as lc=3(l/2-VH0/N)Using the results of both simple and shear compression tests, the following values were computed for the wall stiffness (see ): the initial stiffness of the walls from the cyclic response (Kinitexp) defined as the secant stiffness at 15% VPa showing the plot of the effective stiffness vs. axial load ratio at the base of the wall, the effective stiffness of the walls generally increased with the axial load ratio except for walls RS1 and RS3. In the literature, some experimental campaigns such as Vasconcelos b illustrates the ratio of the effective to initial stiffness. The mean of Keff/Kinitexp is 0.59. This is 18% and 9% higher than 0.50, the value recommended by Eurocode 8 - Part 3 The ratio of Keff/KinitTimoshenko vs. axial load ratio is depicted in c. The mean of the Keff/KinitTimoshenko values is 0.49, which is quite close to 0.50 shows one standard deviation above and below the mean. Comparing b and c, one can see that using Young’s modulus as determined from the compression tests to calculate the elastic stiffness of the walls resulted in a less robust estimation of effective stiffness because the effect of axial load ratio on the E-modulus was not considered. Moreover, the elastic stiffness obtained using theE-modulus from the compression tests overestimated the actual initial (elastic) stiffness of the walls (i.e., on average Kinitexp/KinitTimoshenko<1, as the mean of Keff/Kinitexp=0.59 is higher than the mean of Keff/KinitTimoshenko=0.49). This finding is not in accordance with what Vanin et al. The E-modulus was also back-calculated from the second phase of the shear compression test, i.e., the application of the horizontal load, using a Timoshenko-beam model The E-modulus was calculated by setting G/E=0.33. The blue and black lines represent the average of the E-modulus value from the simple compression (1191 MPa) and shear compression (1037 MPa) tests, respectively. The obtained Young’s modulus estimates were rather close, which suggests that obtaining the initial stiffness by performing three loading/reloading cycles in the simple compression tests results in a robust approximation of the Young’s modulus. These conclusions are, of course, limited to the axial load ratio range covered in this test series, i.e., σv,top/fc between 8 and 25%.a compares the peak shear forces (Vp) obtained in the positive and negative loading directions. The similarity in these values confirms the quality of the testing set-up. In b, the average peak shear force is plotted against the axial load ratio. The peak shear force increases as the axial load ratio increases and decreases as the shear span increases. These trends are well known in the literature and are captured by standard shear force capacity models for masonry walls c. The mean of Vu/Vp is almost 0.94, which is close to the value found by Tomaževič The Turnšek–Čačovič criterion is expressed as:where ft is the tensile strength of masonry, σv,mid is the vertical stress at mid-height of the wall and b=H/La), the tensile strength was estimated to be 0.036 MPa, which is only 8% lower than the value suggested by the Italian code The Coulomb criterion (Eurocode 8 - Part 3 in which c and μ are the characteristic cohesion and friction coefficients of masonry, respectively. By fitting the Coulomb criterion to all data points except RS5 and setting VMC=VP (see a), c and μ were calculated to be 0.036 MPa and 0.34, respectively. The cohesion was estimated by c=2μft, with μ=0.4 (from EC8 For each of the walls, drifts were determined at six limit states defined in Vanin et al. ), which are: δcr, drift at the onset of cracking; δy, drift at yield, defined in ; δp, drift at peak force; δSD, drift at significant damage limit, defined as δSD=min(3δcr,δp); δu, ultimate drift, defined in ; and δc∗, drift at 50% drop in force. Additionally, the drift at collapse δc is defined and discussed in , the drift values obtained for RS1–6 are plotted against the axial load ratio and compared to the models proposed by Vanin et al. Because horizontal cracks appeared at the early stage of horizontal loading in all walls except RS2 and RS4, the drift at the onset of cracking (δcr) increased with the axial load ratio, since a higher axial load ratio delays the decompression of the section. (drift at plaster cracking) shows that the first cracks were observed at about half the drift value reported by Vanin et al. , which summarizes the drift limits, reports the drifts at the appearance of two types of cracks, horizontal (flexural) and diagonal (shear). In (drift at plaster cracking), the smaller of the two values is plotted.The overall correlation between the drift at yield, at peak force, at ultimate limit state and at a 50% drop in force and the axial load ratio was negative, which was expected and also observed in other test campaigns Keeping the axial load ratio constant for walls RS3-RS6 and RS2-RS4-RS5 while changing the shear span ratio shows that the ultimate drift increases with an increasing shear span ratio (), which can be the result of additional flexural cracks. The same trend was observed in other test campaigns, such as tests on brick masonry walls with different boundary conditions by Petry and Beyer , it seems that the formula suggested in Vanin et al. In this test series, the walls were loaded up to the loss of their axial load-bearing capacity. Data about this parameter is scarce for any masonry typology, with walls tested up to failure in only a few examples, including a test series on single-leaf stone masonry walls with dressed rectangular stones by Godio et al. ] used the drift at 50% drop in force as a proxy for the drift at collapse. Despite this approximation, such data was only available for 7 of the 67 walls contained in the database, as most of the campaigns reported in the literature were stopped at a 20% drop in force.The axial load failure is herein defined as the point in the test at which a sudden drop of axial force is recorded and the wall can no longer support the axial load to which it was subjected. This point is illustrated in , where the shear vs. axial force and the vertical displacement vs. drift are plotted, respectively. The drift corresponding to this situation, called drift at collapse δc, is defined as the maximum absolute drift that the wall can resist up to this point. In , the data points recorded before and after the occurrence of axial load failure are depicted as black and red, respectively. The values of δc for walls RS1–RS6 are summarized in Very little literature data is available for the drift at axial load failure (δc), though much more is available for the ultimate drift (δu), making the ratio between the two drifts of particular interest for studies on the collapse risk of buildings. This ratio would provide an indication of the reserve displacement capacity between the horizontal and axial load failure of a pier. a plots the ratio δc/δu against the axial load ratio for the three test series in which testing was continued up to axial load failure. With an increasing axial load ratio, the δc/δu ratio decreases. Because δu also decreases with an increasing axial load ratio, the difference between the ultimate drift and the one at axial load failure decreases with an increasing axial load ratio. a shows that, for a given axial load ratio, these ratios are higher for rubble stone masonry (current tests) than for stone masonry walls with well-dressed stones that have a regular texture (tests by Godio et al. b depicts the δc∗/δcratio, showing that this ratio varies between 0.46 and 1.0 and tends to increase with an increasing axial load ratio. The parameter δc∗ therefore appears to be a biased proxy of the drift at collapse—for high axial load ratios (σv,top/fc>20%), it estimates the drift at collapse well, whereas for low axial load ratios, it significantly underestimates it. This finding applies for stone masonry walls. For modern brick masonry walls, the available data suggests that δc∗ leads to a good approximation of the actual collapse drift.Six quasi-static shear-compression tests on rubble stone masonry walls with plaster were conducted at the École Polytechnique Fédérale de Lausanne (EPFL). The tests were designed to determine the influence of static boundary conditions, herein idealized in terms of the shear span ratio and the axial load ratio, on the in-plane behaviour of rubble stone masonry walls. This is especially relevant to the stiffness, force capacity and drift at different limit states, which are the key parameters characterizing a wall behaviour in performance-based seismic assessment methods. The focus was on the drift capacity at the loss of axial load bearing capacity, shortly denoted as axial load failure, for which the available literature data is particularly scarce. The main findings of the paper are summarized as follows:The Young’s modulus of the masonry walls back-calculated from the shear-compression test using a Timoshenko beam model (1037 MPa) was quite close to the value obtained from the simple compression test (1191 MPa), which supports the practice of deriving the Young’s modulus from simple compression tests. The obtained values from both approaches were close to the upper bound of the range (690–1050 MPa) recommended by the Italian code The effective stiffness of the walls tended to increase with increasing axial load, which is in agreement with the findings in Vanin et al. The tests of the present campaign confirmed that both codified models for the force capacity of the walls, i.e. Turnšek–Čačovič and Coulomb, could estimate the peak force on a different range of axial loads and shear spans rather well when the correct tensile strength, cohesion and coefficient of friction of rubble stone masonry were assumed. These properties were back-calculated by fitting the models to the experimental data. A comparison with code values showed that the tensile strength of 0.039 MPa suggested by the Italian code In addition to the in-plane cracking typically observed in walls tested under shear compression, i.e., shear tension and flexural cracks, bulging of the wall leaves was observed. These out-of-plane deformations continuously increased during the test and became significant in the post peak regime. An increase in the axial load and/or the shear span ratio led to a larger opening between the wall leaves. The largest opening was therefore observed for the test unit RS5, which was subjected to the largest axial load ratio (25%) and the largest shear span ratio (H0/H=1.5).The tests confirmed previous findings regarding the influence of the axial load ratio and the shear span ratio on the drifts at various limit states as the drift at yield, at peak force, at ultimate state and at 50% drop in force decreased with an increasing axial load ratio and increased with an increasing shear span. The obtained drift capacities were compared to the empirical drift capacity models proposed by Vanin et al. Next to the general objective of providing more data on the in-plane behaviour of rubble stone masonry walls, this test series was designed to yield information on the deformation capacity at collapse. The tests were therefore continued up to the point when the walls could no longer sustain the applied axial load. A comparison of this data with the very few other series that had been tested up to this point led to the following conclusions: As at the other drift limits, the drift at collapse decreases with an increasing axial load ratio. The drift at a 50% drop in force, which was considered a proxy of the drift at collapse in a previous study, yields a biased estimate for the drift at collapse because the ratio of δc∗/δcvaries with the axial load ratio. For low axial load ratios, δc∗ was found to be only about half the drift at collapse. The delta in drift δc-δu decreases strongly with axial load ratio, i.e., for walls subjected to high axial load ratios, the collapse occurs soon after horizontal load failure (δu), while for low axial load ratios, a considerable margin between the two limit states exists. This margin is much larger for stone masonry than for modern brick masonry, where brittle unit failure occurs.In this testing campaign, one single specimen was tested per a combination of axial load ratio and shear span ratio. More experimental tests are certainly required to have an estimate of the confidence interval related to the obtained data points. The variability of experimental test results with regard to stiffness, strength and drift capacity of stone masonry elements was evaluated in Vanin et al. Amir Rezaie: Methodology, Validation, Formal analysis, Investigation, Data curation, Writing - original draft, Writing - review & editing, Visualization. Michele Godio: Methodology, Investigation, Writing - original draft, Writing - review & editing, Visualization, Supervision. Katrin Beyer: Conceptualization, Methodology, Investigation, Resources, Writing - original draft, Writing - review & editing, Supervision, Project administration, Funding acquisition.The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.An analysis of the causes of a BWE counterweight boom support fractureA bucket wheel excavator (BWE) collapsed in a brown coal mine. As a result of a tie rod fracture the counterweight boom, the discharge boom and several other components of the other assemblies underwent plastic deformation. This paper presents the results of computer simulations of the collapse. A finite element method analysis of the counterweight boom tie rod showed stress concentrations exceeding the allowable level. Also material tests of the fracture surface were carried out to identify the causes of the collapse. Macroscopic and microscopic images of areas within the fracture were obtained. Measurements of hardness and microhardness in the vicinity of the weld were performed. The FEM analyses and material tests showed that the causes of the collapse were design and welding faults.At the beginning of 2008 a bucket wheel excavator KWK-1400 collapsed in the Turów Brown Coal Mine in Bogatynia (Poland). While the excavator was digging overburden and the bucket wheel boom was in its low position the left tie rod in the counterweight boom support broke (). BWE breakdowns are due to structural, manufacturing or operating faults In order to run a simulation of the collapse a special model was created in ABAQUS/Explicite. The software enables advanced dynamic analyses which take into account plastic deformations and contacts. Because of the high complexity the excavator’s load-carrying structure the calculation model was built using beam elements (The simulation showed that some beams in the bucket wheel boom might have undergone permanent deformation (). One should expect plastic deformation also in the part of the mast structure, on which the counterweight rested.One of the aims of FEM computations was to identify the spatial stress field generated by the excavator operational loads and to determine its effect on the life of the tie rods. Most of the structural members were modelled using beam elements. One of the two tie rods of pulley mast was modelled using shell elements (). As compared with beam models, shell models yield more precise results since they take into account the ways in which the structural members are joined together and their influence on the stress field (e.g. in the vicinity of welded joints). Although beam models do not take into account the stiffness of structural joints, the obtained results are closer to the results of analytical calculations than the ones yielded by shell models.The effort was calculated for the loads specified in DIN 22261. Static and changing loads, defined respectively as H1a and H1b in the standard, were analyzed. The source of changing loads are dynamic interactions, body forces produced by operating motions, tilts, output loads, digging forces, etc. The results of the strength calculations confirmed that stress concentrations occur in the fracture location for H1a () and the standard allowable stress (based on the welded joint notch class, the level of mean stress and the stress change range) for changing loads H1b is exceeded. The mean stress at the stress concentration site was σm

= 302 MPa for stress change amplitude σa

= 31 MPa. The allowable stress for this case is σdop

= 315 MPa, i.e. it is exceeded in cycles of changes from σmin to σmax since σmax

=

σm

+

σa

= 333 MPa.Additional calculations, consisting in the simulation of a fracture extending over 50% of the tie rod cross section, were carried out. The results are shown in . It was found that at this degree of cross section weakening the tie rod would immediately fail. Such a failure () and to the introduction of a notch by the weld (the former depends on the weld class). One should note that values obtained analytically (e.g. using an analytical tie rod model in the form of a truss) would not take into account the stress field revealed by the FEM calculations. When the machine was being designed (the early 1980s) there was no possibility of verifying the suitability of the joint (responsible for the failure of the tie rod) through computations.Fragments were taken from the fracture area for macroscopic and microscopic examinations (The macroscopic examinations of the external surfaces were photographically documented using a Nicon Coolpix 990 camera. The macroscopic images show fatigue lines and material delaminations in the immediate fracture zone. The fatigue zone extended over the whole cross section of tie 1 and 2. The immediate fracture zone occurred in ties 3 and 4. The fatigue lines propagating in the fatigue zone of tie 1 and 2 indicated that the focus of the fracture was located in the vicinity of the welded joint between ties 1 and 2 – site A in . A magnified image of site A is shown in a. The arrows point to the probable focus of the fatigue crack from which the fracture propagated. This was the place of direct interaction between the fillet weld joining ties 1 and 2 and the welds joining tie 2 with the brace. A magnified fragment of the fracture at site B, with marked fatigue zone and immediate fracture zone and visible material delaminations in the immediate fracture zone, is shown in b. The morphology of the fracture fatigue zone at site A in tie 1 observed under a scanning microscope is shown in a. The morphology of the immediate fracture zone at site B in tie 3, with visible plastic deformations and material delaminations, is shown in The microscopic examinations were carried out using an optical microscope NEOPHOT 32 and a Visitron Systems digital camera. The microstructure was examined in detail in the area of the joint between ties 1 and 2 in the places shown in . The material of tie 1 showed a ferritic–pearlitic structure (a), but in the cross section the microstructure was highly heterogeneous (A pearlitic–ferritic structure with Widmannstätten structure features (), i.e. brittle areas which had initiated the fatigue crack, was found in the vicinity of the welds. The segregations could have been the result of welding errors, e.g. too rapid cooling.The microscopic examinations were performed in cross section A–A (marked in a) running near the fatigue fracture focus. The plane of the microsection was perpendicular to the surface of the fracture and to the weld joining tie 2 with the brace while the microsection ran longitudinally to the weld between tie 1 and 2. In the microsection area there were the influences of the welds between: ties 1 and 2 and between tie 2 and the brace or the interactions between the heat-affected zones (HAZ). The crack in tie 2 probably extended into the HAZ of the weld between tie 2 and the brace.The microscopic examinations revealed a very narrow area which might indicate the presence of a HAZ belonging to the end (or beginning) of the weld between the tie and the brace.The HAZ partially extended onto the weld area between ties 1 and 2. The area is indicated by an arrow in . A pearlitic–ferritic structure showing Widmannstätten structure features was found in the area, which proves that the latter had undergone fusion or overheating. Darker etching “islands” were observed in this area, which might indicate the presence of martensite or bainite. Vickers microhardness measurements under 50 G were carried out to identify the martensitic areas. shows a magnified image of the microstructure in the place indicated by the arrow in The measurements showed that the allowable hardness of 350 HV was reached and even exceeded in many places. Even though the investigated area was small and the measurements were performed under a small load, hardness above 350 HV proves that the material had locally undergone hardening in the HAZ. The results of the microhardness measurements are shown in A gravimetric analysis of the chemical composition was carried out. After the paint coats and corrosion products had been removed, chips for the analysis were taken from the cross section of tie 1, 2 and 3. The analysis showed that as regards chemical composition the analyzed steels correspond to low-alloy constructional steels of grades: S355JO, S355J2G3, S355J2G4, S355K2G3, S355K2G4 (in accordance with PN-EN 10025:2002). The grade of the steel agreed with that specified in the tie rod working documentation.The tests showed that as regards its strength properties the material of ties 1 and 3 meets the PN-EN 10025:2002 requirements for steel S355J2G3. The material of tie 2 does not meet the standard requirements as regards its yield point – steel with a yield point (Re

= 322 MPa) lower than required for steel S355J2G3 (Re

= 355 MPa) was used. As a result, the cross section was by about 10% weaker as regards temporary strength and the allowable fatigue analysis stress amplitudes were lower.A fatigue fracture of the tie rod near the end eye connection with the counterweight () was the direct cause of the collapse of the BWE. The character of the tie rod fracture points to longitudinal (tensile) loads, but also to cyclic bending (). The fracture was caused by an additional pulsating bending moment () acting in the place where the angle brace joined the tie rod (). Also the FEM strength analysis of this joint in the tie rod showed that stress concentrations (exceeding the lower yield point) occurred at the fracture site (). The loads were slight since the tie rod fractured only after 28 years of service. The indirect cause was a structural fault in the tie rod’s joint, compounded by the improper welding of the plates. The designer who designed the bracing along the whole length of the tie rod in order to reduce and prevent the free vibration (whipping) of the tie rod did not foresee that this could lead to a disaster.The methodology used to investigate the causes of a serious failure of a bucket wheel excavator was described. Modern computer methods (FEM) and detailed load-bearing structure models were employed. Static and fatigue strength calculations showed that the analyzed cross section does not meet the requirements set in the current standards.The material tests showed technological faults in the making of the welded joints and the poor properties of the steel. They also revealed a pearlitic–ferritic structure with Widmannstätten structure features and local martensite or bainite segregations, i.e. brittle areas, from which the fatigue failure originated. The martensite and bainite segregated in the weld region are the consequence of welding errors, e.g. too rapid cooling.The collapse of the KWK-1400 bucket wheel excavator in the Turów mine (Poland) and a similar collapse of the SchRs 1760/5 × 32 excavator in the Kolubra mine (Serbia) Also other surface mining machines (with a similar counterweight boom structure) are being subjected to strength analyses with regard to the current standard requirements, using state-of-the-art computer simulations.Effect of microstructure and addition of alloying elements on hydriding kinetics of Zr–Nb-based alloysThe addition of Al, Cu and Mn in Zr–Nb-based alloys was evaluated through corrosion and hydrogen absorption tests in cold rolled and annealed conditions. For the studied alloys, the heat treatment resulted in an α-Zr matrix with β-Nb precipitates and X-ray diffraction (XRD) confirmed the formation of hydrides after hydrogenation in all conditions evaluated. It was observed that the alloy with Al addition absorbed the largest amount of hydrogen and the alloy containing Cu took more time to begin the absorption. The alloy with Mn absorbed little hydrogen but the onset of absorption occurred in a short time. The results indicate that the alloy with Cu addition is the most promising and the alloy with Al addition is inadequate for nuclear application.Zirconium-based alloys are used for the manufacture of nuclear reactor fuel rods due to their properties of low neutron absorption cross-section, high mechanical strength and high resistance to corrosion at high temperature and pressure The security requirements and the high operating temperatures to increase the efficiency of these components contribute to the development of new zirconium alloys with improved properties. During the fabrication of the structural components for nuclear reactors, which involves a complex thermo-mechanical processing, the microstructure determines the long and short-term properties of these components Another aspect that deserves attention is the changes that these alloys suffer during operation. In Pressurized Water Reactor (PWR), the integrity of zirconium alloys can be limited by hydrogen absorption, due to low solubility of hydrogen in the α-Zr phase which leads to the hydride formation, reducing ductility and fracture toughness of the Zr alloys Therefore, when developing new materials it is desirable that the added elements reduce the deleterious effects of hydrogen and increase the mechanical strength without adding significant absorption of neutrons. Thus, the selection of alloying elements is limited to a few elements at low percentages.The most effective way to reduce or delay the absorption of hydrogen and the subsequent hydride formation is controlling the manufacturing processes. The most important commercial Zr-based alloys, including Zr–1Nb, are totally or partially recrystallized and are composed of an α-Zr matrix containing Nb in solid solution and fine β-Nb precipitates. This configuration allows the combination of high creep resistance The development of new zirconium alloys for nuclear application should aim to optimize properties within specific ranges of composition, paying attention to microstructural changes such as morphology, size and distribution of grain, crystallographic texture, nature and distribution of precipitates.Depending on the service requirements, the components are thermomechanically processed in different ways to achieve the best combination of properties. The microstructural control of these alloys through the appropriate selection of process parameters is currently one of the major challenges in industry. The microstructural changes that these materials suffer throughout their service life determine the operating life of these components.This paper aims to develop Zr–Nb-based alloys that have the potential for nuclear application. The performance of the alloys with the proposed additions was evaluated through mechanical processing and heat treatment to adjust the microstructure. The effect of hydrogen, its interaction with the microstructure and hydride-forming conditions were determined by absorption kinetics assays, gaseous hydrogenation and X-ray diffraction. Tests to evaluate the corrosion resistance were also conducted.Zirconium alloys with compositions Zr(0.8Nb–0.2M), where M = Al, Cu or Mn, were melted in an arc furnace Bühler AM Vario-400 under Ar atmosphere at 200 mbar. To homogenize, the specimens were turned over and melted more than four times, resulting in ingot of 60 g measuring 100 mm length, 14 mm width and 10 mm thickness.The alloys were cold rolled to obtain sheets 2 mm thick and about 35 cm long through slight deformation passes. To reduce surface oxidation, the samples were encapsulated in glass vacuum and then heat treatments were performed at 750 °C for 2 and 4 h with air-cooling. The samples remained encapsulated until ambient temperature. To evaluate the effect of annealing on the microstructure and determine the ideal heat treatment conditions, micro hardness tests were carried out in the cold rolled and annealed alloys. The tests were performed in a Zwick/Roell micro durometer, Indentec ZHμ-MkII-M. The parameters used for the measurements were 200 g, 15 s.Hydrogen absorption kinetics measurements were used to determine the maximum amount of hydrogen that can be absorbed by the alloy at temperature and pressure close to nuclear reactor operating condition. With this test, it is possible to determine the weight percentage of hydrogen absorbed and the time that absorption occurs. The kinetics tests were performed using an automatic Sievert's type apparatus, PCT-Pro 2000 Hy Energy at 320 °C under 10 bar of hydrogen gas atmosphere. In order to reduce the surface oxidation, the specimens were polished up to 1200 grit emery paper to obtain a surface free from contamination. Subsequently, they were cleaned in acetone, dried and immediately placed on the equipment.To evaluate hydride formation, gas hydrogenation was performed for 48 h at a temperature of 320 °C under 10 bar H2 in the same equipment used to carry the hydrogen absorption kinetics tests on cold rolled and heat treated samples for 750 °C for 4 h.STEM images for heat treated alloys were obtained with an FEI Titan G2 80-200 instrument at 200 kV, coupled to X-ray mapping with Bruker ChemSTEM technology. The samples were cut and then electropolished in dual jet Struers Tenupol 5 device using a solution of 9 vol.% HClO4 and 91 vol.% C2H4O2 at 26 V and 20 °C.The phase structures of the samples before and after hydrogen absorption were identified at room temperature by a Shimadzu XRD-6000 diffractometer using Cu-Kα radiation.Potentiodynamic polarization tests using as electrolyte 0.1 M LiOH were performed on the samples, whose area was 1 cm2, cold rolled and heat treated at 750 °C for 2 h to evaluate the corrosion resistance of the alloys at room temperature. The experiment was carried out using a potentiostat AUTOLAB model PGSTAT 100 in the range of −2 to 2 VSCE. The saturated calomel electrode (Pt, Hg/Hg2Cl2/KClsat – SCE) was used as a reference to measure the sample and the platinum counter electrode was used to promote current flow between the sample and the counter electrode.Optical microscopy allowed the observation of the orientation and distribution of hydrides formed after hydrogenation. Sample preparation consisted of SiC grinding with sandpapers: 100, 220, 320, 400, 1200 and polishing cloth using OP-CHEM 200 mm from Struers with a 10% solution of oxalic acid and OP-FELT cloth 200 mm from Struers with OPS solution (Colloidal Silica 0.04 microns). The microscope used to obtain the images was the Olympus BX60M.Cold rolling was performed to obtain plates with up to 2 mm, however to obtain the final thickness, 73 passes were required for the Zr–0.8Nb–0.2Al alloy, 40 passes for the Zr–0.8Nb–0.2Cu alloy and 34 passes for the Zr–Nb–Mn alloy. All alloys exhibited cracking at the edges and some surface cracks, however, the Zr–0.8Nb–0.2Al alloy underwent a complete break after in 44 passes. shows the final appearance of the plates obtained after cold rolling.To verify if the rupture that occurred in the Zr–0.8Nb–0.2Al alloy was a unique case generated by the segregation of an element, for example, three more samples were melted and the same behavior was observed. The micro hardness tests performed in the three alloys after cold rolling indicate that Zr–0.8Nb–0.2Al has higher hardness, as shown in Recrystallization processes are suitable for nuclear applications because they enable stress relief that should restore ductility after cold work. Depending on the annealing time and temperature, degree of cold work and existence of second-phase particles, it is possible to adjust the microstructure to improve the mechanical properties and corrosion resistance. To determine the appropriate heat treatment that leads to a microstructure similar to Zr–1Nb commercial alloy, heat treatments were performed at different times and temperatures and the micro hardness was measured. Subject to the availability of samples, this assessment was made only for Zr–0.8Nb–0.2Mn alloy. The goal was to obtain the parameters of a heat treatment where precipitation occurs. Recrystallization of a material can be interpreted by hardness measurements. The results show a decrease in hardness value, due to recrystallization, until the heat treatment of 700 °C for 1 h. At higher temperatures and longer times, the hardness increased again, as shown in . This indicates the occurrence of aging where precipitates are formed and grow concomitant with the recrystallization process and was also observed for other Zr-based alloys shows TEM micrographs for the Zr–0.8Nb–0.2Al alloy after heat treatment at 750 °C for 2 h. Spherical grains measuring around 2 μm can be observed, as well as the presence of spherical and elongated precipitates, between 400 and 600 nm, in intra and intergranular domains. The EDX analysis, , of the precipitates indicates the presence of Nb and Al. The electron diffraction confirms the formation of β-Nb precipitate and the EDX mapping, , presents Nb segregation at grain boundaries, mainly at triple grain boundaries.The overview of the microstructure of the Zr–0.8Nb–0.2Cu alloy after heat treatment at 750 °C for 2 h is shown in . Equi-axed grains are observed with average size of 4 μm and the presence of precipitates between 400 and 900 nm in intra and intergranular domains. The EDX analysis, , indicates the presence of Nb in spherical and rod particles and in triple points. EDX mapping, shows the Cu-rich precipitates associated with Nb-rich precipitates.The microstructure found for the Zr–0.8Nb–0.2Mn alloy after heat treatment at 750 °C for 2 h, , is composed of equi-axed grains, about 5 μm, and spherical and elongated precipitates between 500 and 1000 nm in intra and intergranular domains. In this alloy, the precipitates are slightly larger and more spherical than those found in the alloy containing Cu. presents the electron diffraction pattern of the β-Nb elongated precipitate and EDX elemental mapping, , shows that the precipitates are Nb-rich, but segregation of Nb in triple points and an alignment of precipitates is not observed, different from the alloy containing Al.In all the cases it is possible to identify the presence of alloying elements by EDX analysis. The microstructure of all the alloys consists of recrystallized grains, measuring between 400 and 1000 μm, with the presence of inter and intragranular β-Nb precipitates. Zr–0.8Nb–0.2Al alloy presents smaller grains and precipitates sizes than the alloys with Cu and Mn additions. The segregation of Nb in triple points and aligned precipitates was observed for Zr–0.8Nb–0.2Al and Zr–0.8Nb–0.2Cu alloys. These features were not observed in the alloy containing Mn, but the large grain size of the sample made it difficult to obtain low magnification image for the whole grain. The presence of aligned precipitates in Zr–Nb-based alloys has been extensively studied Multiple fields were scanned to perform the mapping of the elements present in the alloy but no segregation of Zr, Al and Mn was noted in metallurgical sites such as grain boundaries or precipitates and it was not possible to differentiate these elements on the EDX elemental mapping. Segregation of the alloying addition was identified only in the Zr–0.8Nb–0.2Cu alloy. The presence of precipitates containing Cu at different annealing conditions was also reported by Kim et al. shows the polarization curves for the three alloys in cold rolled condition. The anodic polarization curves present a continuous and slight increase of the current density as the potential is raised from the corrosion potential. In the alloys with the addition of Cu and Mn, a range of passivity is observed in current densities varying from approximately 9 × 10−7 to 2 × 10−6 A cm−2, while the alloy with addition of Al has a little range of passivity from 5 × 10−7 to 1.5 × 10−6 A cm−2. The cathodic curves are similar for the three alloys. The alloy of lowest corrosion current is the one containing Cu. The noise in the region close to the open circuit potential can be associated with the accumulation of bubbles on the surface of the sample as consequence of hydrogen generation during the execution of the cathodic curve.In the samples treated at 750 °C for 2 h, , the cathodic curve has the same behavior for the three alloys. The anodic curves of Zr–0.8Nb–0.2Al and Zr–0.8Nb–0.2Mn alloys present a similar range of passivity from 1 × 10−6 to 2 × 10−6 A cm−2 but it is possible to observe a different behavior of the alloy with addition of Cu, which forms a plateau at 0.7 V. This threshold can be linked to LiOH solution attack of the base used to isolate the edges of the sample, assessed visually after the test, since the formation of either film or pitting that could justify this behavior was not verified. This same problem was also detected in the cold rolled alloy Zr–0.8Nb–0.2Al, but the noise generated was more discreet. presents the corrosion current and the open circuit potential obtained for all the conditions studied. Whereas, the lower the icorr, the greater the corrosion resistance of the material, we can evaluate the performance of the alloys in the different conditions evaluated.Of all cold rolled alloys, Zr–0.8Nb–0.2Cu alloy has lower icorr, while in the annealed condition; the alloy containing Al has a better performance. In assessing the same alloy in different conditions, the performance against corrosion improves in the annealed alloys, except for the alloy containing Mn, which practically maintains the value of icorr. Compared to the other two alloys studied, Zr–0.8Nb–0.2Cu alloy has the best performance in terms of corrosion resistance. Kim and Jeong The corrosion resistance of Zr–Nb alloys in aqueous media is sensitive to the microstructure which in turn is influenced by Nb concentration and heat treatment. The structural changes produced by variation in Nb concentration and heat treatment have been investigated by many researchers Regarding the corrosion resistance of different microstructures, heat treated samples have better corrosion resistance than cold rolled samples. This behavior can be explained by the existence of Nb-rich precipitates finely dispersed in the Zr-matrix, as presented in , inhibiting the formation of oxide and increasing the corrosion resistance.The same behavior was observed by Choo et al. The hydrogen absorption kinetics of the cold rolled samples is shown in . The alloy containing Al absorbed a quantity close to the terminal solubility of hydrogen on Zr (1.9 wt.%) leading to complete hydride transformation and its collapse. For other alloys and conditions, the samples remained intact (with their original format). Regarding the time until the beginning of the absorption, hydrogen absorption of the alloy containing Cu only started after 31 h. The alloy containing Mn began to absorb after 5 h of test, however, absorbed less hydrogen.In relation to the absorption kinetics for the condition heat treated at 750 °C for 2 h, , the highest absorption was obtained for the Zr–0.8Nb–0.2Mn alloy and the lowest for the alloy containing Al. Regarding the time elapsed until the beginning of the absorption, the Zr–0.8Nb–0.2Al and Zr–0.8Nb–0.2Mn alloys showed very similar values (around 10 h), while the alloy containing Cu, showed slower absorption kinetics, about 30 h for the start and reached its maximum at 45 h. The hydrogen absorption kinetics was different in the annealed samples. In the alloy containing Al, hydrogen absorption was reduced and the time needed to initiate the absorption was shorter. For the alloy containing Mn, the absorption time remained invariant and the amount of hydrogen absorbed increased.In samples heat treated at 750 °C for 4 h, , the absorption kinetics occurred at similar times for the three alloys, between 12 and 15 h, however, the sample containing Al showed the highest hydrogen absorption capacity and the lowest hydrogen uptake was of the Zr–0.8Nb–0.2Cu alloy (0.6 wt.%H). shows the results of the absorption kinetics grouped by alloy. In general, the best performance can be attributed to the Zr–0.8Nb–0.2Cu alloy which took more time to start hydrogen absorption. The Zr–0.8Nb–0.2Al alloy had the worst performance by absorbing more hydrogen in less time, and having collapsed in the cold rolled condition.The absorption kinetics results can be evaluated under two aspects: the influence of the addition of an alloying element and the thermomechanical process. The most significant results of the added element can be seen in the cold rolled samples. Zr–0.8Nb–0.2Al alloy presented the highest hydrogen absorption, becoming completely hydride and taking the sample to collapse. Several factors may have contributed to this result. Aluminum is an α-stabilizing element, while copper and manganese are β-stabilizers. Moreover, the atomic radius difference between Al and Zr is higher compared to that of the other added elements, which creates higher distortion in the network, leading to an increased hardening after mechanical processing, as seen in the micro hardness tests after the cold treatment.The cold rolled condition increases the dislocation density and, depending of heat treatment can result in a fine grain structure with a large amount of grain boundaries distributed along the as-rolled orientation. Cox A hardened microstructure associated with the effect of the increased hydrogen solubility in Zr, characteristic of α-stabilizing elements Regarding the thermal treatments, different behaviors were obtained for each proposed addition. The annealed sample exhibits an equi-axed and recrystallized grain with fine Nb-rich precipitates. The TEM images show that Zr–0.8Nb–0.2Al alloy has smaller grain and precipitate sizes, followed by alloys containing Cu and Mn. In Zr–0.8Nb–0.2Al alloy, the amount of hydrogen absorbed was almost half but the onset time of absorption was shorter when compared to the results of the cold rolled condition. Heat treatment time had almost no effect on absorption kinetics. This is due to the fact that the heat treatment for recrystallization of a hardened material promotes more nucleation and less growth and therefore an increase in thermal treatment time has little influence on the absorption process.Hydrogen absorption in Zr–0.8Nb–0.2Mn alloy did not significantly change under the three conditions examined, however, the onset time of hydrogen absorption increased, as the heat treatment time increased, condition desired in materials for nuclear application. From these curves, it is clear that cold rolling drastically improves activation kinetics. The same effect was observed by Dupim et al. Hydrogen absorption kinetics of Zr–0.8Nb–0.2Cu alloy was almost maintained for cold rolled and 2 h heat treatment conditions. After cold rolling, this alloy presented the lowest hardness and the microstructure after heat treatment for 2 h showed equi-axed grains and precipitates containing high amount of alloying element. Therefore, hydrogen solubility increased slightly with heat treatment, increasing the amount of absorbed hydrogen. In contrast, annealing at 750 °C for 4 h produced larger increase in precipitation, which acts as a barrier to hydrogen diffusion. The combined action of precipitation and recrystallization reduced hydrogen absorption by the samples. This suggests that the precipitates act as inhibitors of hydride formation, thereby reducing the solubility of hydrogen in the alloy. The significant reduction in onset absorption time can be attributed to the diffusion of the alloying element in the precipitates.Another result that corroborates the one obtained in this study refers to the absorption kinetics in hydrogen permeation of Zr–1Nb and Zr–1Nb–1Sn–0.1Fe alloys that have very different behavior in terms of absorption and hydride generation Although the absorption kinetics does not allow inferences on the diffusion of hydrogen itself, since it is not possible to separate the portion corresponding to the diffusion and the fraction of the hydride formation, this test is of great importance to material performance evaluation in degrading environments, since it is not always possible to evaluate the distribution of materials in cases where the permeation of hydrogen does not occur.The XRD analyses indicate that all samples analyzed independent of the condition, have an α-Zr matrix. After hydrogenation, hydride formation occurred. shows the diffraction pattern Zr–0.8Nb–0.2Al alloy in different analyzed conditions as an example.After hydrogenation, in the cold rolled condition, , the diffraction indicates the existence of an α-Zr matrix. After hydrogenation, , the entire sample was transformed into ZrH, which explains why the sample collapsed during the test. For the heat treated conditions, for 2 and 4 h, In relation to hydrides formation and morphological appearance of this phase in the studied conditions, the samples absorbed total hydrogen capacity, as shown in the absorption kinetics curves. In all samples, regardless of the conditions examined, there was a hydride formation. In the cold rolled samples, the hydrides are oriented in the rolling direction. This preferred orientation is associated to the texture obtained after this process. In the annealed samples, hydrides lose the preferred alignment. This effect is associated with the decrease in the texture intensity, resulting from the thermal treatment performed. The relationship between the orientation of the crystallographic texture and hydrides was already described shows the optical microscopy images of the cross-section of the cold rolled and annealed Zr–0.8Nb–0.2Mn alloy after the hydrogenation at 320 °C for 48 h.The degree of alloy embrittlement by hydride formation depends on the concentration of hydrogen, the hydride volume fraction, its size, distribution and orientation. Regarding the morphology, distribution and amount of hydrides, it is observed that the cold rolled alloy has elongated and oriented hydrides while heat treated samples have fewer hydrides but there is a loss of preferential orientation.The same hydride configuration was observed in Zr–1Nb alloys hydrogenated under similar conditions The main conclusions on the interaction of hydrogen with the microstructure of Zr and Zr alloys at different thermo-mechanical conditions are:The precipitation of intermetallic compounds in annealed specimens was observed. This is due to the segregation of alloying elements which provide the formation of precipitates. The precipitation cannot be observed by XRD due to the low volume fraction of the precipitates.The hydrogen absorption kinetics indicates that the alloy containing Al has the highest amount of absorbed hydrogen when cold rolled. For the heat treated at 750 °C for 4 h condition, the time to onset of absorption becomes invariant for the three alloys.Compared to Zr–1Nb, alloys containing Cu and Mn additions achieved similar results with respect to corrosion and gas hydrogenation.The result set shows that the alloy best suited to continue the study of development of a new alloy for nuclear application is the Zr–0.8Nb–0.2Cu, which presented the highest corrosion resistance and took longer to begin hydrogen absorption. The Zr–0.8Nb–0.2Al alloy is considered inadequate for this application, since it presented breakage during the lamination process and collapsed in the hydrogen absorption kinetic tests.Cyclic response of concrete members with bond-damaged zones repaired using concrete confinementThis paper presents the results of experimental investigation undertaken for evaluating the cyclic response of concrete members which have already experienced structural damage and total loss of load resistance due to splitting bond failure of the tensile reinforcement, and then repaired for upgrading their bond strength and flexural capacity. The original (intact) specimens consisted of beams reinforced with identical top and bottom spliced reinforcement and subjected to inelastic cyclic load reversals until total bond degradation and complete loss of flexural strength. The repair procedure consisted of removing the deteriorated concrete within the damaged splice zone, adding concrete confinement and casting new concrete. Three types of concrete confinement were investigated, namely, internal confinement by steel ties or wire mesh reinforcement, and external confinement by FRP laminates. It was found that repairing the bond-damaged zone through concrete confinement leads to substantial regain of flexural stiffness and strength up to or exceeding those for the original specimens, reduces the structural damage, and results in considerable improvement of the energy absorption and dissipation capacity under cyclic loading. The experimental results were discussed, and comparison between the experimental data and analytical predictions is undertaken.cylindrical concrete compression strengthWhen designing vertical concrete members such as bridge piers or columns under gravity load, it is a common practice to splice the longitudinal reinforcement at the base of the members. The corresponding reinforcement may either be spliced with starter bars projecting from the foundation, or with the column reinforcement extending from the lower stories (). While this reinforcement detailing is allowed by the codes of practice for gravity loads, it is critically inadequate for lateral earthquake load, particularly because the region where the reinforcement is spliced coincides with the region where plastic hinging is expected to develop. That is, in the event of strong earthquake ground motion, it is likely that splitting bond failure of the spliced reinforcement will take place at this critical region before the structure is able to mobilize sufficient flexural strength or deformation capacity demanded by the earthquake, causing considerable loss of load resistance, and possibly collapse of the structure. In fact most failures of concrete bridges or buildings following major earthquakes around the world are attributed predominantly to substandard detailing and inferior bond performance of the steel reinforcement at the critical regions of the structures where plastic hinging is likely to develop.Earlier analytical and experimental studies This paper presents the results of experimental investigation undertaken for evaluating the cyclic response of concrete members that have already experienced structural damage due to splitting bond failure of their tensile reinforcement and then repaired for restoring the bond strength and upgrading the flexural capacity of the members. The original specimens consisted of beams reinforced with identical top and bottom reinforcement which were spliced at midspan. The specimens were subjected to cyclic loading with large load excursions into the post-splitting range until total loss of load resistance. The specimens were then repaired by removing the damaged concrete within the splice zone, providing concrete confinement, and casting new concrete around the spliced reinforcement. Three confinement systems were evaluated, namely, internal confinement by steel ties, internal confinement by wire mesh reinforcement, and external confinement using FRP laminates.Two distinctive modes of bond failure characterize the bond strength between steel bars and concrete, namely, pullout bond failure and splitting bond failure One of the most well-known expressions in North American practice for evaluating the average bond strength of plain unconfined concrete, and upon which the ACI Building code in which the average bond strength Up and fc′ are in MPa, Ld is the development (or splice) length, and cm is the smaller of side cover cs, bottom cover cb, or 1/2 the clear distance between the bars.Based on reevaluation of test data, Zuo and Darwin Upfc′4=0.23+0.46cmdb+14.10dbLd0.1cMcm+0.9in which cm is the minimum and cM is the maximum value (cM/cm

< 3.5) of cs and cb where cs is the smaller of side cover or 1/2 clear distance between bars +6.0 mm.To account for the effect of confinement by ordinary transverse steel within the development/splice region, the following expressions were proposed by Orangun et al. in which Utr (MPa) is the incremental increase in average bond strength; Atr (mm2) is the total cross-sectional area per spacing s (mm) of all transverse reinforcement crossing the potential plain of splitting through the reinforcement being developed; ns is the number of spliced or developed bars and fyt is the yield stress (MPa) of transverse reinforcement; tr

= 9.6Rr

+ 0.28 and td (SI system) = 0.03db (mm) + 0.22, where Rr

⩽ 0.14 is the relative rib area of the developed or spliced bar (equal to ratio of projected rib area normal to the bar axis to the product of nominal bar perimeter and center-to-center spacing of ribs). Note that the total bond strength U for confined concrete is obtained by adding the value of Utr to the value of Up for plain concrete obtained using Eq. Using the equation proposed in the CEB-FIP model code U=1.64fc′-2.75102/3(1.15-0.1cm/db)1-K∑Atr-∑AtrmAbin which fc′ is the design concrete strength (MPa); K is equal to 0.1 for a bar confined at a corner bend of a stirrup or tie and is equal to 0.05 for a bar confined with a tie and equal to 0 for unconfined concrete; ∑Atr is the total area of transverse reinforcement along the development/splice length and ∑Atrm is equal to 0.25Ab for beams and 0 for slabs; Ab is the area of the largest bar being developed.in which nf is the number of FRP wraps and tf is the thickness per one wrap; Ef is the modulus of elasticity of the FRP material. Eq. is applicable for both NSC and HSC. To obtain the total bond strength U of FRP confined concrete, it was proposed The original specimens used in this investigation consisted of beams having 2.0 m span length and 240 mm wide by 300 mm deep rectangular cross-section (see ). The specimens were reinforced with four identical 20 mm or 25 mm diameter bars placed symmetrically relative to the cross-section. The bars were spliced at midspan with a splice length of 20db (db

= bar diameter). The splice zone was reinforced with different volume fraction of steel fibers Vf varying between 0.0% (control specimen) and 1.5% with the objective of evaluating the effect of fiber reinforcement on the bond performance of spliced bars under cyclic loading. In order to provide uniform concrete confinement around the spliced bars and to control the concrete cover parameters, the number of splices ns (two), side cover cs, bottom cover cb, and 1/2 clear spacing between the spliced bars were kept identical in all specimens. The section dimensions, the number of splices and sizes of the steel bars were all selected in a way to produce splitting bond failure. Transverse hoops consisting of 10 mm diameter Grade 60 steel and spaced 100 mm apart were provided in the shear spans of all the beams to prevent shear failure.The original test variables included size of the spliced bars (20 mm, 25 mm), ratio of concrete cover to bar diameter (c/db of 1.4 and 2.0), volume fraction of steel fibers, and concrete compressive strength (NSC, HSC).The spliced bars consisted of Grade 60 steel having actual yield strength of 508 MPa and 548 MPa for the 20 mm and 25 mm diameter bars, respectively.The applied load was composed of a sequence of displacement-controlled cycles as shown in . The imposed deflections at different cycles varied between a minimum of 0.5% and a maximum of about 3% of the shear span of the beam. Two cycles were conducted at each deflection level. The loading system was designed to produce a constant moment within the splice region as shown in All specimens experienced splitting mode of bond failure. The load resistance of the specimens dropped significantly as the number of displacement cycles increased beyond splitting. Typical response of applied cyclic load versus midspan deflection for the NSC specimens reinforced with a volume fraction of fibers Vf of 1.0% in comparison with the plain control specimens (Vf

= 0.0%) is shown in . More details of the experimental program of the original specimens and test results are reported elsewhere Nine of the specimens described above were repaired for restoring the bond strength of the spliced bars and upgrading the flexural capacity of the damaged zone. The repair procedure consisted of removing totally the damaged plain or steel fiber concrete within the splice zone, adding internal confinement using steel ties or wire mesh reinforcement and then casting new concrete (see ). No special measures were taken to bond the old and new concrete at the cold joints. For the FRP strengthened specimens, plain concrete was first cast within the splice region and then wrapped with FRP sheets along the splice length after the lapse of a minimum of 15 days. Note that while wire mesh reinforcement may theoretically be classified similar to steel ties, because the wires in a wire mesh are closely spaced and oriented in two directions, it is believed that for the same value of Atr/s (Eqs. ), confinement using wire mesh reinforcement may be more effective in improving the bond strength in comparison with conventional steel ties. Forms and sizes of commercially available wire mesh reinforcement are reported by the by ACI Committee 549 Typical views of the actual specimens during or after strengthening are presented in . A summary of the confinement method and materials used for bond strengthening is presented along with the specimens designations in . The specimens were divided into three test groups depending on the diameter of the spliced bars and the strength of concrete used (NSC, HSC).The transverse steel reinforcement used for confinement consisted of 10 mm diameter Grade 60 deformed bars with actual yield strength of 616 MPa. Three closed ties were added within the splice zone at a spacing s equal to one-third the splice length. The corresponding area and spacing of the ties produce values of Atr/s (see Eqs. ) equal to 1.18 mm2/mm and 0.94 mm2/mm for the 20 mm and 25 mm splices, respectively.The wire mesh reinforcement consisted of a square galvanized welded mesh having 2 mm diameter plain wires and wire spacing of 25 mm. The yield strength and modulus of elasticity of the wires, determined by testing wire samples, was estimated at 406 MPa, and 200 GPa, respectively. The mesh was added in either one or two layers across the length of splice, producing values ofAtr/s of 0.25 mm2/mm and 0.5 mm2/mm, respectively, and wrapped along the full circumference of the section with 50 mm overlap.The FRP reinforcement consisted of carbon polymer flexible sheets (SikaWrap Hex-230 C) having the following design properties reported by the manufacturer: thickness per one layer tf

= 0.13 mm; modulus of elasticity Ef

= 230,000 MPa; tensile strength = 3500 MPa; and strain at tensile fracture of the fibers = 1.5%. The concrete substrate was first prepared for FRP application in accordance with the manufacturer specifications and with the corners rounded to about 15 mm in accordance with the recommendation of ACI Committee 440 The mix of the newly added concrete within the splice zone was prepared using Portland cement type I, crushed limestone aggregate and beach sand. The mix proportions of aggregate:sand:cement by weight were 1:0.73:0.35 with water–cement ratio of 0.45 for the NSC, and 1:0.7:0.55 with water–cement ratio of 0.30 for the HSC specimens. The strength of the added concrete was determined using three standard 150 mm by 300 mm cylinders. Values of fc′ for the old concrete and the newly added concrete in the current test program are provided in All repaired specimens were subjected to the same load geometry and load history to which the original specimens were subjected (). Testing of the specimens was conducted at least 28 days after repair.The deflections of the beams were measured using linear variable differential transformers (LVDT) placed at midspan. The stresses in the spliced bars were measured using electric strain gages. Two strain gages were attached to the top bars and two strain gages were attached to the bottom bars at the lead ends of the splice. All test data was monitored automatically using a computerized data acquisition system.Throughout the discussion that follows, the control unconfined specimens in the three test groups (N20, N25, and H25) which were tested earlier before repair are referred to as the “original” specimens, while the specimens of the current experimental program are referred to as the ‘repaired” specimens. Also, the cycles generated by downward (positive) deflection, producing tension stresses in the bottom spliced bars, are referred to as compression cycles, while the cycles with upward or negative deflection, producing tension stresses in the top splices, are referred to as tension cycles.Relevant test data pertaining to the maximum load attained during the cyclic response in both the compression and tension cycles, the deflection at which the maximum load or splitting failure occurred, as well as the bond strength of the spliced bars for the various specimens are provided in . The bond stress U was estimated from strain/stress measurements at the splice lead end using equilibrium between the bar force Abfs and the bond force mobilized at bond failure,UπdbLs, where fs is the steel stress, db is the bar diameter, and Ls is the splice length (20db). For the specimens in which the strain gage readings were deemed unreliable, the steel stresses needed for evaluating U were estimated from the applied bending moment assuming a lever arm between the tension and compression forces across the depth of the section of 0.9d, where d is the depth of the top or bottom splices. Note that because the concrete strength of the specimens in each test group varied relative to the original control specimens, the peak load and bond strength were normalized to the concrete compressive strength using the 1/4 power of fc′ as suggested by Zuo and Darwin . The normalized values (bond strength and peak loads) are obtained by multiplying the raw values by [fc′(control)/fc′]1/4 where fc′ is the compressive strength of the specimen under consideration. The bond ratio in represents the ratio of the normalized bond strength of the repaired specimen to that of the control specimen in each test group.During the early stage of cyclic loading, all repaired specimens developed flexural cracks at the interface (cold joint) between the old concrete and the newly added concrete within the constant moment region. With increasing level of applied load, all specimens developed splitting failure. Splitting failure occurred in the bottom splice first, followed with splitting failure of the top splice during the same cycle. Further increase in downward or upward deflection beyond splitting increased quickly the slip of the spliced reinforcement and resulted in substantial widening of the cracks formed at the cold joints or lead ends of the spliced bars. The splitting mode of bond failure in the specimens was generally evident in the formation of horizontal side and bottom cracks which started from the splice lead ends and propagated quickly along the full splice length. Typical mode of failure is shown in . For the FRP repaired specimen, because the splice region was shielded by the FRP sheets, it was only possible to detect bond failure during the test from the sudden change in behavior of the load–deflection response. For the FRP confined specimens in series II (N25FRP), bond failure was accompanied by fracturing of about 50 mm wide strip of the FRP sheets at the lower corner of the section (see show the cyclic load–deflection response of the repaired specimens against the control unconfined specimens in the three test groups. For the original unconfined specimens, splitting failure was followed by sudden and dramatic descent of the load resistance and almost total loss of flexural stiffness. On the other hand, the presence of confinement in the repaired specimens restricted the growth and propagation of the splitting cracks, leading to a more ductile bond failure when compared to the control specimens. One of the most evident observation in the current test is that the repaired specimens were not only able to regain the strength that was totally lost due to bond failure under earlier cyclic loading but also mobilized slightly larger loads, higher energy absorption and dissipation capacities, and experienced less damage in comparison with the original unconfined specimens.Before splitting, the stiffness of the load–deflection response of the repaired specimens in all test groups were consistently equal to that of the unconfined specimens (except N25S), indicating no significant influence of the cold joint between the new concrete and old concrete on the specimen performance. For the repaired specimens in group I, using steel ties for strengthening (specimen N20S) increased the normalized peak load (average between the compression and tension cycles) by about 25% (see ) and resulted in considerably lower stiffness degradation in comparison with specimen N20. On the other hand, because of its relatively low value of Atr/s (see ), the use of one layer of wire mesh reinforcement such as in specimen N20WM1 did not increase the normalized peak load in comparison with N20 but improved slightly the ductility of the response. Doubling the area of wire mesh reinforcement from one to two layers (or Atr/s from 0.25 mm to 0.5 mm) improved the normalized peak load by about 14% and also enhanced the post-splitting response in comparison with one layer. The corresponding improvement was better for the tension cycles as compared to the compression cycles, possibly due to inferior bond resistance of the top splice as a result of the difficulty encountered in consolidating the concrete at the top of the specimen when two layers of wire mesh reinforcement were used. Comparing the performance of specimen N20FRP with the remaining specimens, it can be seen that the use of only one layer of CFRP laminates increased the average normalized peak load of the compression and tension cycles by about 4% and reduced the stiffness degradation similar to specimens N20WM2 and N20S.Similar observations can be made regarding the cyclic behavior of the repaired specimens in group II. Note that the response of specimen N25S shown in is not representative of the actual behavior since this specimen was subjected by error to a large upward force which caused splitting of the top splice before starting the actual test and recording data. Similar to the effect of wire mesh reinforcement in test group I, the use of two layers of wire mesh in specimen N25WM2 did not lead to improvement of the normalized peak load of the specimen when compared with the original specimen (N25), but it reduced considerably the stiffness and load degradation with number of cycles. The use of external FRP confinement (specimen N25FRP) resulted in a substantial increase in the normalized peak load of about 33% and also enhanced significantly the ductility of the cyclic response.Comparing the behavior of the specimens in group I with those in group II which have different ratios of concrete cover to bar diameter c/db, it can be seen that except for the fact that the normalized peak load of the specimens in group II are all higher than those in group I due to larger area of the spliced bars, no characteristic difference can be observed in the overall response to warrant clear conclusion of the effect of c/db on the cyclic response of the repaired specimens.For the HSC specimens in group III, the use of wire mesh reinforcement resulted in a less violent bond failure that was observed in the original unconfined HSC specimen (H25). It also increased the normalized maximum load capacity by a sizable 15% for specimen N25WM1 and 10% for N20WM2, and reduced the stiffness degradation with number of loading cycles. While the use of two layers of wire mesh did not improve the peak load when compared to one layer, it did improve substantially the energy absorption capacity in the post-splitting range.Note that the cyclic response of the specimens, including the original ones (see ), experienced pinching-in effect. Pinching is attributed to the concentration of deformation at the cracks formed at the splice lead ends (or cold joints in the repaired specimens) as a result of the significant slip of the spliced bars following splitting failure (see ). Because all specimens, including the original specimens, mobilized splitting failure and consequently localized cracks at the lead ends, all specimens, irrespective of the type of repair mechanism used experienced pinching in their hysteresis response. The level of pinching increased with increase in cyclic deflections due to increase in splice slip and crack width.It should be mentioned that because the original specimens were not subjected to axial load, for the purpose of comparisons with the original specimens, no axial load was applied to the repaired specimens that would simulate loading conditions of columns or bridge piers. In the presence of axial load, the pinching-in effect of the cyclic load–deflection response of the original and repaired specimens are expected to increase. shows the envelope or backbone load–deflection response and shows the energy absorbed and dissipated per cycle for all specimens. The energy per one cycle is equal to the area enclosed within a full loop of the load–deflection response. Note that the odd cycle numbers correspond to the first cycle at a given imposed deflection level, while the even numbers correspond to the second cycle at that same deflection level. generally show an increase in the peak load, improvement in the ductility of the response, and considerably larger energy absorption and dissipation capacities of the repaired specimens in the post-splitting stage when compared to the original unconfined specimens. As an indication of reduced structural damage or improved performance of the repaired specimens, it can be observed from ) that while at a deflection level of about 10 mm (which corresponds to twice the deflection at which the control unconfined specimens developed splitting failure) the control specimens suffered almost total loss of load capacity, the repaired specimens were still able to sustain sizable load. The magnitude of the corresponding load (taken as average between the tension and compression cycles) varied in percentage of the load resistance of the original unconfined specimens between a minimum and a maximum of 58% (Specimen N20WM1) and 130% (N20S) in test group I, 80% (N25WM2) and 100% (N25FRP) in group II, and 57% and 88% in group III, respectively. At three times the deflection at which the original specimen failed by splitting, the repaired specimens were still able to sustain, on average, a load equal to 53%, 67%, and 43% of that of the original control specimens in test groups I, II, and III, respectively.Typical variations of splice strain with applied cyclic normalized load are presented in . Because the spliced bars in all specimens, except N20S, did not yield, the variation of splice strain in tension with applied load for the repaired specimens displayed practically a linear trend. Unlike the response of the original specimens in which the linear trend was preceded by a stiff response until flexural cracks developed, because the cold joint did not offer resistance to tension stresses, the linear trend for the repaired specimens started as soon as the load is applied. Also, since the compression force across the depth of the beam section is resisted jointly by the spliced bars and concrete following closure of the flexural cracks at the cold joints, the compression steel strains developed in each of the bottom and top splices were relatively small compared to the tension strains. Note that while the variation of tension strain with applied load for the bottom and top splices were intrinsically identical, larger compression strains were mobilized for the top splices when compared to the bottom splices. This is most likely attributed to partial closure of the flexural cracks (developed at the cold joints) at the level of the top splice when the cycle is reversed as compared to full crack closure at the level of the bottom splice, a trend which is influenced essentially by the history of applied load and the progress of damage in the specimens.The average bond stress u of the various specimens at any load level during the response were calculated using equilibrium between the bar force Abfs and the bond force uπdbLs, as follows:in which Ab is the area (equal to πdb2/4) and Ls is the splice length (20db) of the bar, and fs is the steel stress obtained from strain gage measurements.The average bond stresses u for the bottom or top splices, calculated using Eq. , were normalized to the concrete compressive strength of the original control specimen in the same group by multiplying the bond stress by [fc′(control)/fc′]1/4 and plotted versus specimen deflection as shown in . The responses were traced up to the deflection beyond which the strain gage measurements were deemed unreliable due to strain gage damage. that the bond stress versus deflection responses when the spliced bars are in tension are practically similar to the load–deflection responses. This similarity in the behavior is attributed to the direct relation between the flexural resistance of the specimens and the bond strength of the spliced bars. It can be seen in that, in as far as bond strength and bond performance of the spliced reinforcement are concerned, the repaired specimens were able to recover the full bond strength or even higher strength, and to mobilize larger ductility of bond failure when compared to the original unconfined specimens. Depending on the type of confinement used, the bond ratio (given in the last column of ) attained maximum values of 1.23 (specimen N20S), 1.29 (N25FRP) and 1.13 (N25WM2) in test group I, II, and III, respectively.Comparing the normalized peak bond strength or Bond Ratios of the specimens in the various test groups (), it is interesting to observe that internal confinement by wire mesh tended to be more effective in improving the bond strength for the HSC specimens (H25MW1 and H25MW2) in comparison with the NSC specimens (N20WM1, N20WM2, and N25WM2).The average peak bond strength results of the tension and compression cycles of the specimens of the current test program were compared with the predictions of the various design equations presented earlier (Eqs. were used to predict the bond strength for the wire mesh repaired specimens, because the surface condition and distribution of wires in wire mesh reinforcement are different from conventional steel ties, more experimental data is needed to support the applicability of Eqs. It can be seen from the comparisons and statistical data (average and standard deviation) provided in ) leads to more reasonable predictions of the average bond strength when compared to the other approaches. Also, except for the slightly higher standard deviation, the predictions of the approach proposed by Zuo and Darwin by Zuo and Darwin and normalization of bond strength to the 1/4 power of fc′ have already been proposed by ACI Committee 408 as a base for modifying the ACI Building code approach for calculating the development and splice length of steel bars in tension While the predictions for the FRP repaired specimens appear to be reasonable, the number of specimens is unfortunately small to verify the validity of Eq. . It should be indicated however that the validity of Eq. has already been supported by analysis and verified with earlier available test data can be combined linearly with either Eq. to produce a powerful design expression that can be used for estimating the minimum area of FRP jacket needed to achieve “adequate bond strengthening” of the steel bars at the location of the structure where spliced reinforcement is used for plain unconfined concrete (since Eqs. are both normalized to the 1/2 power of fc′) leads to the following:nftf=1000nsdbEf(Ld/db)κfyfc′-16.6-Lddbcmdb+0.4 is limited to 0.4fc′ and in the absence of experimental data to justify a larger upper limit, there exists a theoretical limit beyond which the steel stress can not be increased any further irrespective of the area of FRP confinement used. The corresponding steel stress fs can be estimated from Eq. by replacing κfy with fs and the value of nftf on the left hand side of the equation by its upper limit of nftf=1600nsdbEf calculated from Eq. Concrete beam specimens which encountered considerable damage and loss in load resistance due to splitting bond failure of their spliced reinforcing bars under cyclic load reversals were repaired for upgrading the bond strength and flexural capacity. The repair procedure consisted of removing the loose and damaged concrete within the splice region, adding confinement reinforcement, and casting new concrete. Three different types of confinement were evaluated, namely, internal confinement by steel ties, internal confinement using wire mesh reinforcement, and external confinement by carbon FRP laminates. Following repair, the specimens were subjected to inelastic cyclic load reversals to assess their strength and behavior under seismic loading.Based on the results of this investigation, the following conclusions or observations can be drawn:In the absence of confinement, concrete members in which the reinforcement is spliced at the critical hinging regions are expected to experience splitting bond failure when subjected to cyclic load. Splitting failure leads to considerable concrete damage within the spliced zone, loss in flexural strength, and total stiffness degradation of the cyclic load–deformation response. The loss in load resistance and stiffness degradation are expected to be more pronounced for HSC in comparison with NSC.Repairing and strengthening the damaged zone of the concrete members using internal confinement by steel ties or wire mesh, or external confinement by FRP laminates allowed the damaged members to regain strength up to or exceeding the strength capacity of the original unconfined members, and improved substantially their energy absorption and dissipation capacity under cyclic loading.Depending on the amount of confinement used and supported with analytical verification, the maximum increase in peak bond strength of the repaired specimens in comparison with the original plain concrete specimens in the various test groups attained 25% for the steel confined specimen in test group I, 33% for the FRP confined specimen in group II, and 13% for the HSC specimen confined with one layer of wire mesh reinforcement in group III.While the original unconfined specimens lost most of their load resistance in the early stage of cyclic loading immediately after splitting, at a deflection level of three times the deflection at which the original unconfined specimens mobilized bond failure, the repaired specimens were still able to sustain, on average, a load equal to 53%, 67%, and 43% of that of the original unconfined specimens in test groups I, II, and III, respectively.Internal confinement of the spliced zone by wire mesh reinforcement was more effective in increasing the bond strength for the HSC specimens in comparison with NSC specimens.Available expressions for predicting the bond strength of developed/spliced bars in plain unconfined concrete or confined concrete showed generally good agreement with the test results. However, it was evident that normalization of average bond strength to fc′1/4 as proposed by Zuo and Darwin Localization control of carbon nanotubes in immiscible polymer blends through dynamic vulcanizationFiller localization behaviors determine the morphology and final properties of immiscible polymer blends composites. In this study, we report a simple and efficient approach to tailor localization behaviors of carbon nanotubes (CNTs) in immiscible polylactide/crosslinking polyurethane (PLA/COP) blends through dynamic vulcanization. CNTs dispersed in the dispersed COP phase if they were added before dynamic vulcanization while localized in the continuous PLA phase and at the interface when they were added after dynamic vulcanization. The particle size of dispersed phase increased by dispersion of CNTs in the COP phase and decreased through dispersion of CNTs in the continuous PLA phase. The composites with different CNTs localization behaviors showed totally different electrical conductivities, rheological and mechanical properties. This study provides a facile way to tailor filler localization behaviors in immiscible polymer blends especially for those are fabricated by dynamic vulcanization, such as thermoplastic vulcanizates.Immiscible polymer blends composites represent a special class of hybrid materials, as there are at least two phases in which the particulate fillers can locate and the specific fillers localizations determine the final properties of the composites []. Therefore, localization control of fillers is the key to design and fabricate immiscible polymer blends composites with desired properties. The localization behaviors are usually governed by thermodynamic factors based on polar and dispersion interactions between the fillers and the polymer components []. In general, the fillers tend to disperse in the phase with higher affinity during melt processing. With decreasing difference in intrinsic properties of the polymer components, the fillers prefer to locate at the interface to reduce interfacial energy. Therefore, for a specific immiscible polymer blends composite, the localization of the filler is hard to control due to the fixed intrinsic interactions of the components. Surface modification of fillers represents an efficient way to control their interactions with polymer pairs thus to change their localization behaviors []. However, the surface modification of some fillers would partially sacrifice their excellent intrinsic properties such as conductivity of carbon nanotubes, due to disruption of their translational symmetry or surface wrapping of insulating modifiers [Except for the thermodynamic aspects, kinetic effects, such as the rate of mixing process and viscosity ratio of polymer pairs, are also important for localization control of particulate fillers []. In fact, kinetic aspects should also be considered during discussion of thermodynamic factors, because they may restrict the attainment of thermodynamic equilibrium during melt processing. Kinetic effects provide an efficient way to control particle localizations through sequential mixing, where the fillers are premixed with the less interacting component and then mixed with the second polymer component in the second mixing step []. In this case, particles may migrate toward more interacting phase driving by thermodynamic effects. Therefore, the particle localizations could be theoretically regulated to disperse either in less interacting phase or at the interface and or in more interacting phase by sequential mixing. However, the migration mechanisms are very complicated and affected by many factors, such as the interactions between fillers and the polymer pairs, viscosity ratio of the polymer pairs, and mixing parameters []. Therefore, very careful control of mixing process is required to get the specific localizations through kinetic aspects. Furthermore, the specific localizations may be unstable with respect to subsequent melt processing for products manufacturing, due to their possibly thermodynamic nonequilibrium state. Therefore, it remains a challenge to fabricate stable immiscible polymer blends composites with desired filler localizations from specific fillers and polymer pairs.Dynamic vulcanization, involving a process of selective crosslinking of a rubber during melt blending with a thermoplastic polymer, is usually employed to fabricate thermoplastic elastomers (also known as thermoplastic vulcanizates, TPVs), which are phase-separated materials with a cross-linked rubber dispersed in a thermoplastic matrix []. Dynamic vulcanization is also demonstrated as an efficient way to toughen brittle polymers such as the most promising biobased and brittle polylactide (PLA) []. Based on the characteristics of dynamic vulcanization, we suppose that this technique may provide a possible way to fabricate immiscible polymer blends composites with desired and stable fillers localization by simply controlling the feeding sequence especially for the polymer pairs, in which the crosslinking component is more interacting phase for the fillers. In this case, the fillers should disperse and be fixed in the crosslinking component when they are added at the same time with the polymer pairs before dynamic vulcanization. Otherwise, they should locate at the interface or disperse in the non-crosslinking phase when they are added after dynamic vulcanization, due to the prohibited dispersion of the solid fillers in the solid-response crosslinking component with rather high viscosity and low mobility.To confirm the aforementioned point, we herein investigate the effect of feeding sequence on the localization behaviors of carbon nanotubes (CNTs) in immiscible polylactide/crosslinking castor oil-based polyurethane (PLA/COP) blends prepared by dynamic vulcanization []. The effect of CNTs localizations on the conductivity, thermal behavior, and rheological and mechanical properties of the ternary PLA/COP/CNT composites were studied in detail.Polylactide (PLA, 4032D) with weight average molecular weight of 1.76 × 105 g mol−1 was purchased from NatureWorks and was dried at 80 °C for 8 h prior to processing. Castor oil (AR grade) with 2.7 mol hydroxyl groups per molecule was procured from Kelong Chemical Reagent Factory (Chengdu, China) and was used directly. 4,4′-Diphenylmethane diisocyanate (MDI, 98%) was purchased from Micxy Chemical Co., Ltd (Chengdu, China) and was used without any purification. Multi-walled carbon nanotubes (CNTs) (NC 7000) with an average diameter and length of 9.5 nm and 1.5 μm were obtained from Nanocyl Corporation (Belgium) and were used directly.PLA/COP/CNT composites were prepared by dynamic vulcanization in torque rheometer (Shanghai Kechuang, China) at 180 °C with a roller rotation rate of 80 rpm. Two feeding sequences were used. For the first sequence, PLA, castor oil and CNT were first mixed in the chamber of the torque rheometer for ∼6 min, then predetermined amount of MDI was added into the chamber to perform dynamic vulcanization for another ∼6 min to obtain the composites. For the second sequence, PLA and castor oil were first mixed for ∼3 min, then predetermined amount of MDI was added to perform dynamic vulcanization. CNT was added after the dynamic vulcanization (∼6 min) to mix with the PLA/COP blends for another ∼3 min to generate the composites. The weight ratio of PLA to COP was fixed at 80:20 and the -OH/-NCO equivalent ratio was kept at 1:1. For convenience, the composites prepared by the first and second feeding sequences were designated as PLA/COP/1CNT-x and PLA/COP/2CNT-x, respectively, where x represents the loading of CNT in wt%. PLA/COP binary blends were also prepared with the same procedure for property comparison. Standard tensile bars (ISO 527) were prepared by injection molding (Xinshuo MiniJet, Shanghai, China) with a nozzle and mold temperature of 180 and 40 °C, respectively.The morphologies for the cryo-fractured surface of the composites were observed with a field emission scanning electron microscope (FE-SEM, JSM-7800F) at an accelerating voltage of 2 kV. The surface was gold-sputtered before observation.Electrical conductivity was measured on a digital high resistance test fixture PC68 (Shanghai Precision Instrument Manufacture, China). The samples with dimension of 100 × 100 × 1 mm3 were prepared via compression molding for the experiment. Average results from three measurements were reported for all samples.Rheological behavior was measured on a TA DHR-1 rotational rheometer in dynamic frequency sweep mode from 0.01 to 100 Hz at 180 °C with an oscillation strain of 1.0%.Tensile properties were measured on a MTS E44 Universal Testing Machine with a crosshead speed of 10 mm/min at room temperature. The gauge length between the two pneumatic clamps was 25 mm. Averaged results from five measurements were reported for all the samples. shows the variation of melt torque with processing time for preparation of PLA/COP binary blends, PLA/COP/1CNT and PLA/COP/2CNT composites with CNT loading of 0.5 wt% as typical example. The total time for the mixing of all samples was fixed at 12 min to avoid processing time induced different properties. All samples showed abrupt increase in melt torque after addition of MDI due to the occurrence of dynamic vulcanization through crosslinking of castor oil with MDI. To get a good dispersion of CNT, the time for premixing of PLA, castor oil and CNT was set at 6 min before addition of MDI for preparation of PLA/COP/1CNT composites. It is noted that the final torque of PLA/COP/1CNT-0.5 was almost the same with that of PLA/COP binary blends but was much lower than that of PLA/COP/2CNT-0.5, of which a further increase in melt torque occurred by addition of CNTs. From such a special phenomenon, we can expect that PLA/COP/1CNT and PLA/COP/2CNT would have different CNTs dispersions. For elastomer toughened plastic blends, the stiffness, modulus and melt torque are usually dominated by the major continuous plastic matrix and affected by the minor dispersed rubber phase. Therefore, the composite with rigid fillers dispersed in continuous plastic phase should have higher melt torque than the counterpart with fillers located in the dispersed elastomer phase. So, we can speculate that CNTs should mainly disperse in COP phase for PLA/COP/1CNT and in PLA matrix for PLA/COP/2CNT.To confirm abovementioned speculation, the CNTs localization behaviors in the composites were studied by SEM. As a shows, the PLA/COP binary blends showed a phase-separated morphology with COP particles (blue circle) dispersed in the PLA matrix. For the PLA/COP/1CNT composites, CNTs could not be observed in PLA matrix but was only found in the COP phase, regardless of the CNTs loading. It is noted that aggregation of CNTs in the dispersed COP phase was very prominent as highlighted by red circles in b–e. The aggregation of CNTs could be clearly observed in the enlarged inset with PLA/COP/1CNT-0.1 (b) as a typical example. In contrast, for the PLA/COP/2CNT composites, CNTs could not be found in the COP phase (blue circle in f–i), they dispersed in the continuous PLA matrix uniformly without obvious aggregation, as shown in f–i. Except for dispersion in the PLA matrix, CNTs were also found to localize at the interface between COP and PLA phases, which can be observed obviously for the composites with higher CNTs loadings, as shown by the high magnification SEM images in , in which the CNTs dispersed in PLA matrix were designated with blue arrows and those located at the interface were designated with red arrows.The results perfectly proved our idea of controlling CNTs localization behaviors through feeding sequences during dynamic vulcanization. The schematic illustration for CNTs location control is shown in . When CNTs were added before dynamic vulcanization, they prefer to disperse in castor oil due to its rather low viscosity compared to high molecular weight PLA; with addition of MDI, CNTs were fixed in the dispersed phase through rapid crosslinking of castor oil during dynamic vulcanization. In the cases of PLA/COP/2CNT composites, CNTs were added after crosslinking of castor oil. The diffusion of CNTs to the crosslinking COP phase was prohibited, therefore, the dispersion of CNTs in COP phase was avoided. Thus, CNTs either dispersed in PLA matrix or located at the interface for the PLA/COP/2CNT composites.It is worth noting that the size of dispersed COP phase in PLA/COP/1CNT composites was greater than that in PLA/COP binary blends and in PLA/COP/2CNT composites. Such a phenomenon should be attributed to the different viscosity ratios between dispersed COP and PLA phases with different CNT dispersion behaviors. The dispersion of CNT in COP would increase the viscosity of the dispersed phase and thus reduce its dispersion level.PLA/COP/CNT composites with different CNTs localizations would exhibit different electrical conductivities. As a shows, PLA/COP/1CNT composites were non-conductive regardless of CNTs loadings with the electrical conductivity (σ) at the order of magnitude of 10−17 S cm−1, which was similar to that of the PLA/COP binary blends. Such low electrical conductivities for PLA/COP/1CNT composites were anticipated and attributed to absence of CNTs conductive networks, due to the localization of CNTs in the dispersed COP phase (). In contrast, PLA/COP/2CNT composites showed better electrical conductivity, which was attributed to the dispersion of CNTs in the continuous PLA phase, making it possible to form CNTs conductive networks to improve electrical conductivity (). A drastic improvement in electrical conductivity occurred with σ value increased by 10 orders of magnitude when the content of CNT was ≥0.5 wt%, indicative of the occurrence of electrical percolation.The electrical percolation threshold can be calculated from the relationship between filler content and electrical conductivity as described by the classic percolation theory [where σ is the measured electrical conductivity, σ0 is a scaling factor, ϕ is the filler content, ϕc is the percolation threshold, and t is a percolation exponent to predict the mechanism of network formation. The predicted t values for two- and three-dimensional conductive networks were ∼1.3 and ∼2, respectively []. However, experimental data usually diviated from the prdicted values, as reported in various literature []. Although the reason for the diviation is still unknown, it is widely accepted that the value of t > 2 indicates three-dimensional conductive networks []. The ϕc value was obtained as 0.39 wt% by fitting the experimental data to Equation (1) in a logarithmic plot, as shown in b. The percolation threshold is much lower than CNTs filled PLA composites or other immiscible PLA blends composites fabricated by melt mixing, usually showing percolation threshold of higher than 1.0 wt% []. The t value was obtained as 2.74, indicating the formation of three-dimensional conductive networks in the PLA/COP/2CNT composites with CNTs loading higher than 0.39 wt%.The microstructures of polymer composites are well known to have significant effect on their rheological parameters such as storage modulus and complex viscosity, since these parameters are highly related to the fillers dispersions and their interactions with polymer components [a and b shows, PLA/COP/1CNT composites showed similar rheological behavior with PLA/COP binary blends, although both storage modulus (G′) and complex viscosity (|η*|) of the PLA/COP/1CNT composites were slightly higher than those of PLA/COP binary blends at given frequency. In addition, both G′ and |η*| of PLA/COP/1CNT increased slightly with increasing CNTs loading. As displayed in c and d, PLA/COP/2CNT composites with low CNTs loading showed similar rheological behavior to PLA/COP binary blends. However, the high CNTs loading composites showed much different rheological behaviors with the G’ and |η*| much higher than those of PLA/COP at low frequency range.Such different rheological behaviors between PLA/COP/1CNT and PLA/COP/2CNT were again related to the different CNTs localizations for the two types of composites. For phase-separated polymer blends with elastomer dispersed in plastic matrix, the storage modulus and complex viscosity are dominated by the major continuous phase and affected by the minor dispersed phase. For PLA/COP/1CNT composites, CNTs located in the dispersed COP phase, thus cannot prominently influence melt properties of the PLA matrix. Therefore, PLA/COP/1CNT showed similar rheological behavior to PLA/COP binary blends. In the cases of PLA/COP/2CNT composites, CNTs mainly dispersed in the PLA matrix thus could modify the melt properties of the PLA matrix apparently to impart the composites with different rheological behaviors. It is well known that G′ is related to the elasticity of polymer melt and the increase in G’ indicates enhanced elastic response of the melt []. Obviously, the elastic response of the composites should be enhanced when the solid fillers CNTs dispersed in the dominated PLA matrix. Meanwhile, the complex viscosity of the composites should increase due to the restricted chain mobility of PLA matrix by the presence of solid fillers.It is worth noting that PLA/COP binary blends and PLA/COP/2CNT composites with low CNTs loadings (0.1 and 0.3 wt%) exhibited liquid-like response with low storage modulus and obvious Newtonian plateau at low frequency range. While, with CNTs loading increased to 0.5 and 1.0 wt%, the rheological behavior of the composites changed to solid-like response as evidenced by the presence of the plateau in the low frequency range of G′-f plots as well as the significantly enhanced shear-thinning behavior in the low frequency range of |η*|-f plots. The transition from liquid-like behavior to solid-like response was corresponding to the occurrence of rheological percolation through formation of CNTs networks with CNTs loading increasing from 0.3 to 0.5 wt%. The presence of CNT networks would restrict the long-range molecular chain motion of the continuous PLA matrix to show solid-like behavior []. It is interesting that the rheological percolation occurred with CNTs loading between 0.3 and 0.5 wt%, which is in accordance with the electrical percolation with threshold of 0.39 wt%.Different dispersions of fillers in immiscible polymer blends are known to have significant effects on the mechanical properties of the composites []. Specifically, the mechanical properties can be improved significantly when the fillers are located at the interface, because such filler localization, like a compatibilizer, is helpful to refine the phase morphology and enhance interfacial adhesion [ shows the stress-strain curves of PLA/COP/1CNT and PLA/COP/2CNT composites and summarizes the mechanical property parameters. The two series of composites showed different variations of mechanical properties versus CNTs loading. PLA/COP binary blends showed a yield strength, a Young's modulus and an elongation at break of 37.2 MPa, 1.43 GPa and 335%, respectively. With CNTs incorporated before dynamic vulcanization, the PLA/COP/1CNT composites showed poorer mechanical properties compared to PLA/COP binary blends and the mechanical properties deteriorated with increasing CNTs loading. The PLA/COP binary blends showed obvious strain hardening prior to break. The strain hardening became less prominent or even disappeared with increasing CNTs loading for PLA/COP/1CNT composites. In contrast, PLA/COP/2CNT composites with CNTs incorporated after dynamic vulcanization showed much better mechanical properties than the PLA/COP binary blends and all the mechanical property parameters increased first and then decreased with increasing CNTs loading. The PLA/COP/2CNT-0.3 showed the best mechanical properties with the yield strength, Young's modulus and elongation at break of 41.3 MPa, 1.52 GPa and 398%, respectively.Such different mechanical properties between PLA/COP/1CNT and PLA/COP/2CNT were once again attributed to their different CNTs localizations. For PLA/COP/1CNT composites, CNTs only dispersed in the COP phase, thus could not reinforce the continuous PLA matrix, which dominates the yield strength and Young's modulus of the composites. In addition, the dispersion of CNTs in COP phase enlarged the size of the dispersed phase (), which is harmful for the overall mechanical properties including yield strength, Young's modulus and elongation at break of the composites. In the case of PLA/COP/2CNT composites, CNTs dispersed in PLA matrix and also localized at the interface. The both localization behaviors are positive to reinforce the mechanical properties of the composites. The stiffness of the composites should be reinforced to exhibit enhanced yield strength and Young's modulus with CNTs dispersed in the dominated continuous PLA matrix. In addition, localization of CNTs at the interface is very helpful for stress transfer from PLA matrix to dispersed COP phase, which is very important for immiscible polymer blends to exhibit excellent mechanical properties. Besides, the size of dispersed COP phase did not increase or even decrease in PLA/COP/2CNT composites compared to PLA/COP binary blends, as shown in , which is also helpful for the mechanical properties of the composites. The interfacial located CNTs could work as a compatibilizer to tune the phase morphology of the immiscible PLA/COP blends. The slightly reduced mechanical properties of PLA/COP/2CNT composites with CNT loading higher than 0.5 wt% should be attributed to formation of CNT networks, which may restrict the chain motion of PLA matrix significantly to reduce elongation at break of the composites.In conclusion, dynamic vulcanization was demonstrated as an efficient technique to control localization behaviors of CNTs in immiscible PLA/COP blends. CNTs localized in the dispersed COP phase when they were added before dynamic vulcanization, while they located in PLA matrix and at the interface when they were added after dynamic vulcanization. The PLA/COP/CNT composites with different CNTs localizations showed much different morphologies and physical properties. Dispersion of CNTs in dispersed COP phase increased the size of the dispersed phase, whereas localization of CNTs in PLA matrix and at the interface reduced size of COP particles and refined the microstructure of the composites. The composites with CNTs dispersed in COP phase were nonconductive due to the absence of CNTs conductive path. The conductivities of the composites with CNTs dispersed in the continuous PLA matrix increased with increasing CNTs loading and electrical percolation occurred at threshold of 0.39 wt%. The rheological behavior of PLA/COP blends was not changed apparently by dispersion of CNTs in the dispersed COP phase but was changed obviously with localization of CNTs in the PLA matrix and at the interface. The transition from liquid-like to solid-like response occurred with CNTs loading increasing from 0.3 to 0.5 wt% for the composites with CNTs dispersed in PLA matrix due to the formation of rheological percolation. The composites with CNTs localized in PLA matrix and at the interface showed better mechanical properties than the PLA/COP binary blends, which showed better mechanical properties than the composites with CNTs dispersed in the COP phase. This study provides a simple and efficient technique to tailor filler localizations and thus final properties of immiscible polymer blends composites.Microstructure damage related to deformation properties of grain compositesMathematical model of micro-heterogeneous medium with voids and shrinkage porosity is considered. Green’s tensor is used to obtain the formulas for macro modules of elasticity. Discrete and continuous probability distributions are used to describe random properties of microstructure elements. Convergence of the method is discussed. Sample calculations are provided.Methods for prediction of deformational properties for many types of composites are well developed Grain composites represent a broad class of materials, which include metals with their polycrystalline structure, ceramics, powder metals, certain plastics, and concrete. The common feature of all these materials is the small size of their microstructure elements and random relative disposition of elements within the composite. Mechanical properties of individual elements are not fully determined. This has to do with manufacturing process, variation of chemical composition, or impurities. To study such materials we propose a model of micro-heterogeneous medium, containing elements of the first and second orders infinitesimal Elements of the first order infinitesimal Δ1V have deterministic macroscopic mechanical properties. The medium is macroscopically homogeneous and isotropic. The size Δ1V is approximately equal δL, where L is a characteristic dimension of the considered part, and δ

≪ 1. It is assumed that for the volume V the macroscopic boundary value problem is solved, and that the composite’s macroscopic deformation tensor e is known. Deformations are the smooth functions of their coordinates, so they can be considered constant inside a small volume Δ1V.Elements of the second-order infinitesimal Δ2V possess random microscopic mechanical properties. The tensor of stochastic elastic modules Q(X) is given for interior points X of a small neighborhood Δ2V. The strength properties of the microstructure are also random. Random microstructure stresses and deformations are found through the stochastic boundary value problems. Dimensions Δ2V are much smaller than Δ1V and are about δ3L

−

δ2L. The unit dimension can be chosen as the size of a grain or a fraction of that size. Grain size is determined based on the region of statistical dependency of microstructure properties. For metals is usually from 5 μm to 100 μm The distribution parameters of the microstructure’s random properties are obtained through experiments The microstructure damage of a composite part may be created during manufacturing process or develops during lifecycle of the part. Let us consider the two types of damaged grains. The grain of the first type has failed completely creating a void. The modules of elasticity in such grains are equal to zero. The grain of the second type has lost its continuity, but has not completely failed. Such grains can still carry some load. The variation in their deformational properties can be characterized by specially selected random functions. Shrinkage porosity is best described by the second type of grain damage; therefore the term shrinkage porosity will be used hereafter for this particular type of damage regardless of the way it was created.Grain boundaries of metals after service has developed not voids or wedge-like cracks but compact shrinkage porosities During the early stages of metal failure, the fracture is scattered or dispersed. Properties of microstructure can be considered statistically independent. The relative number of fractured microstructure elements can be found through the microstructure strength criteria. These criteria depend on the type of loading on the composite and include distribution parameters of its random microstructure stresses Microstructure analysis of the properties of the material makes it possible to find characteristics of random elastic modules at interior locations X

= (x1,

x2,