University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Virginia Tech | The novelty of the process and lack of scale-up information created a high level of
uncertainty in designing the POC test circuitry. In light of this, a conservative scale-up approach
was used in determining the critical dimensions and relative sizes of the POC units. In theory,
this approach can be used by applying the principle of similarity for a same process step (i.e. in
batch and POC). This principle involves maintaining geometric ratios and dimensionless groups
(characterizing the phenomena of interest) constant at a larger scale. The phenomena involved in
the HHS process are complex; therefore, it is difficult to maintain constant values for all the
dimensionless groups associated with the process.
The conceptual flowsheet shown in Figure 4.1 provided crucial steady-state information,
such as slurry flow rates, solids mass rates, etc., necessary for sizing of the reactors and pumps.
Another prime parameter for sizing is the residence time for each process step. Although in
batch-scale testing, the residence times for unit operations were significantly low, the POC units
were designed conservatively with in excess capacity to provide much higher residence times.
The design of over-capacity reactors was a cautious step taken to give flexibility in the pilot
plant testing, particularly in terms of accommodating wide variations in operating parameters.
4.4 Development of Mixing Devices
Translation of mixing processes from a laboratory scale kitchen blender to the POC pilot
operation was challenging. Though data is available in the literature from several thousand
mixing applications, identifying an exact scale-up correlation was not possible. The oil
agglomeration process involves two stages mixing. The high-shear mixing step is needed to
break hydrophobic bridging oil into small droplets and facilitate the coal particle-oil contact. The
low shear mixing step is needed to promote agglomerate growth.
The two most common parameters generally used for mixing scale-up are constant P/V
(actual horse-power drawn by an impeller/ active volume in a tank) and constant T/V (actual
impeller torque/ active volume). Unfortunately, the kitchen blender used in laboratory testing
does not provide any credible information for actual power drawn or impeller torque. On the
contrary, the blender did provide the information that was used to attain dynamic similitude
among the four fluid forces in a mixing tank. One is the inertia force put in by the mixer and the
other three are opposing forces that resist mixing. These are the viscous force, gravitational
force, and the surface tension force. The ratio of the inertia force with these opposite forces
94 |
Virginia Tech | generate dimensionless quantities called Reynolds number (Equation 4.1), Froude number, and
Weber number (Equation 4.2) respectively, which can be used for designing POC pilot plant
mixing applications. It must be noted that Froude number is usually important only in cases
where gross vortexing exist and thus can be eliminated by installing baffles in a tank in turbulent
conditions (Amanullah, 2004). The Reynolds number for mixing applications is defined as:
[4.1]
where N is speed (rpm) of an impeller, ρ is the density of slurry (Kg/m3), d is the impeller
diameter (m) and µ is the slurry viscosity (N-sec/m2). Values of Re define the flow conditions.
For Re <100, the flow is laminar while for Re > 1000, the flow is defined as turbulent. Similarly,
Weber number for a mixing application can be defined as:
[4.2]
where ‘σ’ is the interfacial tension (N/m) of the hydrophobic liquid.
4.4.1 High-Shear Mixer
The 1.25 L glass jar used for high-shear mixing on the bench-scale system has a
cylindrical shape and tapered-flat bottom. The internal diameter (D) of the jar was 4.5 inches
with two baffles that were each 0.5 inch wide (b). A custom paddle-type two-flat-blade impeller
of 1.5 inch diameter (d) was installed at the bottom. The blender’s top speed (N) was up to
18,000 RPM and was adjusted using a variable-speed controller. For batch-scale testing, the
high-shear mixing was conducted at 11,000-12,000 RPM for 15-20 seconds, and later the speed
was lowered down for low shear mixing. Additionally, in each test 600 ml of coal slurry (6%
solid) was used to make the liquid level (Z) in the blender equal to 3.5 inches above the top of
the impeller. For sizing the high-shear prototype unit, the geometrical ratios such as impeller
diameter/ tank diameter (d/D); baffle width/ tank diameter (b/D); and impeller diameter/ liquid
level height (d/Z) were kept constant. Two baffles were also designed in the POC unit similar to
the laboratory scale blender.
Of the three aforementioned dimensionless quantities for scale-up, the Weber number is
theoretically the most appropriate group for high-shear mixing scale-up since the phenomenon
involves dispersing hydrophobic liquid into small droplets. Additionally, Reynolds number was
95 |
Virginia Tech | evaluated to keep the same order of turbulence in the mixing system. For a similar feed type and
hydrophobic oil and using Equation 4.2, one can establish the following relationship:
( ) [4.3]
For effective dispersion of oil droplets, a 4-inch diameter (d ) high-dispersion blade,
poc
which is very common in applications such as paint dispersion, was selected. These impellers
convert the majority of the energy of the motor to shear force. These impellers have much lower
power number, of the order of 0.45, but they are run at very high speed to produce desired
dispersion. From Equation 4.3, a scale-up correlation is established to identify the required speed
(rpm) for the POC prototype unit (illustrated in Figure 4.2). Additionally, Reynolds number at
these operating speeds, calculated from Equation 4.1, shows same order of turbulent flow in
laboratory scale tests (Re = 2.5 x 105 at 11000 RPM), as well as in the pilot-scale tests (Re =
lab poc
4.1 x 105 at 2500 RPM).
The process flowsheet shows the slurry feed rate to the mixing units is 4.3 gallons/min.
Based on the given flow rate and scale-up ratios, the high-shear mixing tank was fabricated using
the specifications shown in Figure 4.3. The constructed tank provides approximately 2 minutes
of residence time for high-shear mixing. Table 4.1 summarizes the scale-up ratios and calculated
design parameters used for designing POC high-shear mixing unit.
Figure 4.2 Scale-up correlation between lab-scale high-shear mixing and POC prototype.
96 |
Virginia Tech | 4.5 Selection of Sieve Panel
The overflow of high/low shear mixing is a solid-oil-water system. This three-phase
system is treated in a stationary curved sieve bend to separate the agglomerated coal from the
dirty aqueous phase. Typically these sieve bends are sized based on material flow rate over the
panel. A 1 feet x 1 feet square panel can dewater up to 100 gallons/minute (Lutttrell, 2013) and
removes a maximum of 45% water (Woodie, 2013).
Conn-Weld Industries donated two curved sieve screens for the POC plant, a 60 mesh
and a 45 mesh wedge wire designed perpendicular to flow. The screens were 2 feet wide x 1 feet
long and slightly curved with an angle of 37° (Figure 4.6a). For safety reasons, the whole system
was enclosed in a custom-made stainless steel screen-box (Figure 4.6b) manufactured and
donated by the Eriez Manufacturing Company. The screen-box was designed to hold up to 5
PSIG of pressure.
Figure 4.6 (a) Sieve panel from Conn Weld (b) Installed enclosed screen-box in POC pilot.
100 |
Virginia Tech | Figure 4.8 (a) Set-up of POC mechanical vibrating device (b) Constructed vibrating mesh
system for POC pilot (c) Pneumatic saw used to provide oscillations
4.6.2 Parametric Studies and Scale up Criteria
Two dimensionless quantities were investigated at batch-scale, vibration strength and the
dimensionless length. The two parameters are considered to be critical for the de-agglomeration
unit to achieve a low-moisture product. Vibration strength, which relates vibration energy as a
function of both frequency and amplitude, has been defined previously in Chapter 3 (Equation
3.7). The dimensionless length is defined as the ratio of the oscillation amplitude (A) to the
hydrophobic liquid column height (Z ) in the vibrating mixer. It can be described as the
p
characteristics length traveled by a completely dispersed particle reporting to the overflow port.
In light of this, a detailed response surface two factorial (2FI) model was developed using
Design Expert® software designed by Stat-Ease. The software is commonly used to locate ideal
process conditions utilizing the existing data. The designed model provides a combinatorial
relationship in terms of coded factors that can be used to make prediction about the response for
parameters within the data range. In the coding system, high levels are coded as +1 values and
low levels are coded as -1 values. The coded equation for the vibrating mixer calculated by
model was found to be:
102 |
Virginia Tech | ( ) ( ) ( ) [4.5]
Where “VS” is vibrational strength and “DL” is dimensionless length. Equation 4.5 shows that
the coded values are large positive coefficients, which indicates high impact of both the
parameters individually as well as in combination on final product moisture. The actual equation
derived using the batch-scale data is given by:
( ) ( ) ( ) [4.6]
Equation 4.6 with actual factors may be used (though it is not recommended due to the empirical
nature of the expression) to make predictions about the moisture response for each factor. The R-
squared value for the model is 0.717 and is in reasonable agreement with the adjusted R-squared
value, which is 0.813, showing the model can be used to navigate within the design space. The
response surface based on two-factorial combinatorial model is illustrated in Figure 4.9. It is
apparent that, higher vibration strength (energy) and high dimensionless length escalate the
product moisture.
Figure 4.9 Response surface curve showing the effect of vibration strength and
dimensionless length on product moisture
103 |
Virginia Tech | 4.6.3 Operating Conditions for POC unit
To identify the operating conditions of POC unit, the scale-up correlation was developed
to achieve same product moisture. Figure 4.10 shows the effect of variable dimensionless length
(A/Z ) on the HHS process product moisture obtained from the batch-scale unit. The
p
experimental data clearly identify the effective dimensionless length, which is less than 0.1 for
single-digit product moisture. Therefore, for fixed amplitude of 1.125 inches in the POC unit, the
hydrophobic liquid column height (Z ) was kept 11 inches. This height placed the bottom screen
p
in the oil phase, 1-inch above the oil-water interface.
For vibration strength, Equation 3.7 establishes following scale-up criteria:
( ) [4.7]
where, f is the operational frequency of the de-agglomerator at POC pilot unit, ζ is the
poc Lab
vibration strength at laboratory scale, A is the amplitude of oscillatory frequency in the pilot
poc
unit and g is the gravity (9.81 m/sec2). A series of experiments was conducted on the batch unit
to identify an effective range of vibration strength required for consistent product moisture.
Figure 4.11 shows a semi-log plot between the product moisture with respect to the vibration
strength. From the plot, it is evident that the low vibration strength (<2.5) always produced
consistent moisture reductions irrespective of dimensionless length. The moisture values were
more inconsistent at higher vibration strengths (>10). From a thermodynamic standpoint, lower
energy in the process is necessary to facilitate desirable water coalescence and to eliminate the
unwanted formation of stable micro-emulsions.
Figure 4.12 provides a graphical correlation that identifies the operating frequency for the
POC unit at constant vibration strength and given amplitude 1.125 inches. For low vibration
strength (0.5 to 2), the required operating frequency is only in the range of 2-4 Hz. The low
frequency is easy to manage and eliminated concerns over the mechanical stability of the
oscillating screen-shaft structure at higher operating frequencies.
104 |
Virginia Tech | Figure 4.12 Scale-up correlation between lab-scale vibratory mixer and its POC prototype.
4.7 Development of Process Thickeners
Solid-liquid separation systems are the key part of any process involving liquid for
separation. The separation can be achieved by several ways such as in settling tanks, thickeners,
centrifuges etc. For the HHS process POC pilot-plant, both thickener and solid-bowl centrifuge
units were initially considered for processing of the tailings and vibrating mixer overflow
(pentane-in-coal product), respectively. Thickeners typically produce 15-30% solids products
and have a longer residence time. Solid-bowl centrifuges can produce up to 60% solid products
and are more common for recovering fine particles. Although latter can be more suitable in
reducing the pentane evaporation load, the cost associated with a smaller solid-bowl centrifuge
(based on a given flow rate) was 20-fold higher compared to a stationary tank. Hence, a
thickener was designed to treat the process stream containing the coal-pentane product from the
POC vibrating mixer unit.
4.7.1 Development of Hydrophobic Liquid Thickener
The hydrophobic liquid thickener intercepts the clean dewatered coal in the pentane
liquid stream to thicken the solids and ultimately reduce the load in the evaporation step. The top
106 |
Virginia Tech | size of the particle to be treated in the pilot plant is 44 microns. A simple Stokes equation
calculation for the top size coal particle suggests 9 times higher terminal velocity in pentane as
compared to water (Figure 4.13). During the period of POC development and construction,
settling studies of particles in pentane were not studied and, therefore, the HL thickener for the
POC unit operation was designed based on an upper flow rate dictated by an approximate
residence time target. Later, a detailed study for particle settling was conducted using the Coe
and Clevenger method for developing a thickener model and is discussed later in this document.
The process flowsheet shows approximately 5 gallons/minute pulp flow rate to achieve 30%
solids from the thickener underflow at 15 minutes residence time.
An 80 gallon capacity tank was designed as shown in Figure 4.14 with a 360°
circumferential overflow launder and a large 45°conical bottom. Sizing for a conventional
thickener is typically based on a rise rate (i.e. gallons per minute feed per square feet of tank
surface area) of 0.5 gallons/minute/feet2 (Smith, 2010). For 5 gallons/minute feed rate, the
required tank area should be around 10 ft2, which provided a diameter of the POC unit
approximately 3 feet 6 inches.
Figure 4.13 Schematics showing differences in settling velocity of particle in water and
pentane using Stokes law
107 |
Virginia Tech | Figure 4.16 Installed pentane recovery systems in the POC pilot plant
The dryer has nominal length 4 feet with screw area 10 ft2. The total volume of the unit is
1.7 ft3 with a jacket outside to circulate heating medium (Figure 4.16). The unit was designed to
use thermal oil for heating at 500°C to dry sludge. However, pentane is a low volatile liquid with
very low boiling point (98°F) and low specific heat of vaporization (156 BTU/lbm at STP, 153.7
BTU/lbm at 98°F) that can be evaporated with only gentle heating. For this reason, water was
considered as a heating medium as it was safe and easy to handle.
The purchased dryer unit was cleaned, repaired, and refurbished as per the requirements.
The modifications include installation of an explosion-proof pneumatic motor with gear-box and
a custom discharge chute with a nitrogen gas chamber and two pneumatically controlled
butterfly gate valves. The nitrogen chamber was added to collect dry clean coal product and to
eliminate any atmospheric oxygen that could cause a safety risk. The residence time of solids in
the dryer was 8-12 minutes when operating at a screw speed of 30 RPM.
110 |
Virginia Tech | different modules specifically for each process described above. A P&ID flowsheet was
designed as described in Figure 4.17.
Module #1 is an oil agglomeration process, where fine coal slurry mixed with a small
dosage of pentane in a high-shear mixing tank (T-100). Due to high-shear mixing, the pentane
liquid breaks into small droplets, which adhere to fine coal particles and cause them to start
coagulating. These micro-agglomerates float on the top of the slurry and are transferred to a low-
shear mixing tank (T-200) by gravity flow. The low-shear mixing tank has a long retention time,
which helps tiny agglomerates to grow. Since only coal is hydrophobic in nature, it only takes
part in the agglomeration process, leaving behind water and mineral matter from the slurry.
The low-shear tank overflow, which carries coarser agglomerates, is passed to a
stationary curved sieve bend (F-300) to separate these agglomerates from water and clay. The
oversized coal agglomerates from the sieve bend are transferred to a phase separation process in
Module #2. The underflow of water and clay is transferred to a classifying thickener (T-800),
which is equipped with a traditional floating oil skimmer inside the tank. In the thickener (T-
800), any un-agglomerated fine coal particles and absorbed pentane reporting in the sieve
underflow will float on top of water due to its lower density. The skimmer is set to continuously
pump (P-800) the top layer from the classifying thickener back to the high-shear mixing tank.
In Module #2, the coal agglomerates are dispersed in the vibrating mixer (or de-
agglomerator) reactor (T-400). These agglomerates contain considerable water that is bridged
between the coagulated oil-coated fine coal particles. The vibrating mixer disperses these
agglomerates in pentane using mechanical vibration energy inside the reactor. The small water
drops bridged within the agglomerates are displaced from the coal surfaces and quickly coalesce
with other small water drops. The coalescence eventually leads to the formation of large water
droplets, which is a thermodynamically favorable process. These larger droplets grow in mass
and start settling. The droplets eventually cross the pentane-water interface and enter the
submerged aqueous phase inside the reactor. The water column inside the reactor is constantly
maintained by pumping (P-400) this extra amount of water to the classifying thickener in
Module#1. The dispersed low moisture clean coal in pentane is pumped (P-900) to the HL
thickener (T-500) for settling ultrafine particles. The settled coal from the HL thickener is
pumped (P-600) to Module #3 in the oil-recovery circuit. The overflow from T-500, which is
114 |
Virginia Tech | Nitrogen Generator
Nitrogen gas was needed for purging the POC units to eliminate any oxygen inside the
operational tanks. A pilot-scale PRO-N-8 model nitrogen generator from Onsite Gas was
procured for the pilot plant. The unit uses micro-sieves to remove oxygen from air and thus
provides nitrogen at three purity levels at 95%, 97%, and 99%. In addition, the generator is
mounted on an in-built 60 gallon receiver tank to maintain the pressure in the POC units. The
nitrogen gas was supplied in 1/2” air hose to a nitrogen reservoir tank located on Module #1,
which ultimately supplies nitrogen to the POC units at a regulated pressure.
Nitrogen Pressure Regulators
The whole POC system was pressurized with nitrogen at a slightly above atmospheric
pressure (2.5-7” of water column). For regulating this pressure, two ¾” Fischer type Y690AH
spring loaded pad-depad pressure regulators were installed on the hi-shear mixing tank and with
the nitrogen chamber below the product discharge chute at the dryer.
Pressure Vents
Enrado High Performance pressure relief valve, Model 953 and designed for 1”-7” of
water column pressure, were installed on each of the five operating tanks and at the condenser.
The vents on the tanks were set at 7” of water column pressure while the vent installed on the
condenser was set at 5” of water column making it as an emergency vent, where any released
nitrogen and pentane vapors were discharged outside the shed.
Pentane and Oxygen Sensors
Pentane gas sensors were required to detect the traces of pentane vapors, if there any,
below its LEL (lower explosion limit = 1.4 % by volume) around the moving shaft outside the
tanks, as well as in the near vicinity of the pilot plant area. An oxygen gas sensor was required to
detect oxygen percentage inside the POC system. Typical minimum oxygen concentration
(MOC) required initiating a fire with pentane vapors is 11.8% by volume (Mashuga and Crowl,
1998); therefore, during the POC operation, oxygen level was always kept below 10% by volume
as a precautionary measure, by managing nitrogen gas flow rate.
118 |
Virginia Tech | Hazard Material (HazMat) Storage Cabinet
A rental temperature controlled storage cabinet was procured to store pentane barrels
during inactive hours. The unit can accommodate up to 12 barrels and is self-equipped with an
air-conditioning unit and a dry chemical based fire-suppression system.
Fire Extinguishers and Fire Suppression System
Four standard fire extinguishers were installed at each corner of the POC plant at a
distance of 10 feet. In addition, each module was equipped with helon-gas based fire suppression
system (very commonly used in Nascar) connected with gas transfer aluminum lines, installed at
several points across the POC units, which can be more prone to a hazard.
Fail-safe and Emergency Shutoff Valves
Three fail-safe pneumatically controlled spring loaded ball valves were installed at the
pentane discharge point in the tanks installed in Module # 2. In addition, manually controlled 3-
way emergency ball valve was installed intercepting the air supply from the compressor to the
POC pilot plant.
Spill Containment System
Spill containment barriers were installed all across the shed as per recommendation from
EHS. In addition, the test site was equipped with a spill containment kit specifically designed for
hydrocarbon cleaning and disposal.
Grounding Wires
The whole POC system was grounded such that each operational unit and hose connected
to each other with ground wire and ultimately to the ground rods installed outside the shed, as
per recommendation from state Fire Marshal to eliminate hazards caused either due to static
generated from material flow or any nature activity.
Back-up Power Supply and Safety Signs
An additional gasoline based power generator was equipped to supply power (in-case of
power failure) to nitrogen generator, PLC control box and other ancillary unit installed for safety
120 |
Virginia Tech | In addition several other items were tested such as proper functioning of ancillary units — water
heater and chiller, testing nitrogen purging system and accurate detection from gas sensors,
pumps operations, and testing PLC controls and HMI interface for effective operation and
control of the POC pilot plant.
4.12.1 Material Procurement
For shakedown testing, a total of six barrels of screenbowl main effluent sample was
procured from Consol’s Buchanan preparation plant. The sample was high-grade metallurgical
coal with 3.8% solids and only 7% dry ash. The top particle size (d ) was found 34 micron with
90
98% finer than 38 micron. After each test, the feed slurry was reconstituted in the sample barrels
and perfectly mixed again before charging to the POC pilot plant for next shakedown testing.
Pentane was procured in bulk (3 barrels - 156 gallons; 99.5% pure) from South Hampton
Resources, a supplier based in Texas.
4.12.2 Shakedown Test #1
The first shakedown testing of the POC pilot plant, utilizing coal slurry and pentane
liquid, was conducted under the constant supervision of a Virginia Tech EH&S official. This
included proper start-up of pilot plant, constant monitoring of gas levels, proper initial charging
and handling of pentane liquid into the vibrating mixer tank and hydrophobic liquid thickener.
The coal slurry was first agitated thoroughly using an electric mixer located outside the shed.
The well-mixed slurry was pumped from the sample barrels at a pre-determined flow rate into
the high-shear mixing tank. The pentane feed was also pumped at the same time into the high-
shear tank.
Two major issues were encountered in the testing. First, during the period of the initial
charge of the pentane and the slurry transferred into the system, only a small amount of vapors
condensed in to the pentane-receiving column, which was unusual because the condenser was
flooded with pentane liquid. An additional by-pass vent line was installed from the condenser to
the pentane-receiving column, which eliminated the vapor lock in the receiver column and
allowed the liquid pentane to flow to the receiver column. The condenser appeared to be working
efficiently with the modified arrangement. Second, after few minutes of POC operation, just
when agglomerates begin to detect over the sieve bend, the laboratory air compressor (rated 50
122 |
Virginia Tech | CFM at 100 PSI) failed to maintain the air pressure needed to operate all the pneumatic drives.
This caused ultimately the shutdown of the pilot plant. No samples were collected during the first
shakedown testing. To eliminate the issue, a 185 CFM diesel powered portable air compressor
was rented and installed in the further testing.
4.12.3 Shakedown Test #2
Similar procedures were followed in the start-up of the pilot plant. The two major issues
were encountered during the operation. First, because of the small flow rate from the low-shear
mixing tank, the sieve screen was quickly blinded and agglomerates appeared to be stick on the
top of the screen. With only a small amount of agglomerates overflowing from the screen, the
plant was continued to run in a hope to collect a product sample. The second issue encountered
was all the air supply lines were flooded with condensed water. This flooding is believed to be
due to the high humidity in the atmospheric air utilized by the air compressor. To eliminate any
damage to the pneumatic motors, the plant was shut down before any sample could be collected.
To rectify the first issue, several major modifications were conducted in the screen box.
The feed port for the screen was re-located at the back of the box (initially there were two feed
ports on the side of the screen-box) just opposite to the low-shear tank overflow port to
streamline the fluid flow. Two wash-water lines were also installed inside the completely
enclosed screen box with additional pneumatic rapper from outside, below the box. In addition,
as the screen area was much larger for the given flow rate, almost 2/3 of the area was covered by
a thin sheet (1/4”) of Teflon™. Although this caused some extra amount of dirty water reported
in the overflow of the screen, the sheet facilitated smooth flow of agglomerates into the vibrating
mixer.
The second issue was resolved by installing a small water collecting tank (high pressure
rated) between the air compressor and air supply line. This did not completely eliminate the
problem but resolved flooding of condensed water in to the air-supply lines. Any residual
moisture in compressed air was trapped by water-filters installed before the pneumatic motors.
4.12.4 Shakedown Test #3
The third shakedown test also experienced problems. When all the units appeared to be
working well, the peristaltic pump tube of P-900 burst, which led to emergency shutdown.
Emergency spill procedures were followed to clean the pentane spill. The spilled material was
123 |
Virginia Tech | collected and disposed carefully. As no spare pump tube was available during the time of the
test, the plant was ultimately shut down without collecting any sample.
4.12.5 Shakedown Test #4 and 5
A common problem was encountered during shakedown test 4 and 5: the dryer was
observed to be flooded with water. The presence of unwanted water ultimately affected the final
product moisture. Two main causes were identified for the aforementioned problems. First,
during the plant operation, water vapors generated inside the processing units were carried by
nitrogen gas through inter-connected nitrogen lines to the dryer (green dotted lines). It was
visibly noticed that the water vapors fogged the sight-windows of the dryer, which ultimately
condensed into large water droplets during the long operating period. To eliminate this issue,
nitrogen lines connected to the dryer were re-routed (blue dotted lines) and connected directly to
the condenser unit so as to by-pass the dryer, as outlined in Figure 4.20(a).
The second reason for water reporting to the dryer was related to the control of the
pentane-water interface level inside the vibrating mixer unit. For an efficient separation, the
interface level must be below the bottom most oscillating screen. In POC system, this level is
Figure 4.20 (a) Schematic showing re-routed nitrogen path in the POC system (b) Image of
additional sight glass to monitor oil-water interface in vibrating mixer
124 |
Virginia Tech | Table 4.3 Shakedown test 4 and 5 assays obtained from POC pilot plant
Test Number Feed Product Tails
%Solids Ash% % Moisture Ash%
20.2 2.1
NOT
Shakedown 4
COLLECTED
3.8 7.0 21.4 2.2
Shakedown 5 5.4 2.0
monitored visually through the sight window installed in the tank and controlled manually by
adjusting the vibrating mixer underflow pump.
In shakedown test #4, the sight glass (made with 1” thick Plexiglas) was completely
smudged with coal-oil mixture, thus it was almost impossible for the operator to locate the
interface level. When the water level rose to more than the marked limit, it passed to the
overflow stream, which ultimately flooded the dryer. Two clean coal product samples were
collected before the plant shutdown, but without a tailings sample. The results from this test are
shown in Table 4.3.
To resolve the issue, prior to shakedown test #5, the sight window was equipped with a
float (painted fluorescent) made of oak wood. The oak wood has a specific gravity of 0.78,
which is in between pentane liquid (0.62 SG) and water (1.00 SG). The system worked for
several minutes, but the float was also quickly coated with coal-oil mixture, making it impossible
to monitor the interface level. Because of the high possibility of the dryer flooding again with
process water, the POC plant was shut down before steady-state could be achieved. Only one
clean coal product sample was collected in this test and again without a tailings sample. The
results are shown in Table 4.3.
Though there were several hiccups in shakedown tests #4 and #5, the preliminary results
obtained (as outlined in Table 4.3) were very promising. The high moisture obtained in test #4
was significantly reduced to a single digit in the test run #5, demonstrating that the HHS process
can drop moistures to levels only provided by thermal dryers. In addition, the product ash (~2%),
(similar values as achieved in the bench-scale system) showed the excellent cleaning capability
of the process on a large-scale. Unfortunately, the reject samples in both the tests could not be
collected because of untimed shutdown.
125 |
Virginia Tech | The setbacks in pilot-scale testing caused by the inability of the operator to monitor the
interface level was finally overcome by installing an additional sight glass level, as shown in
Figure 4.19(b). The glass level was positioned such that the top port was located in the middle of
pentane column and the bottom port was located in the middle of aqueous phase. The system was
tested in the next shakedown test, which showed the sight glass level worked very well for
monitoring the interface level, although the Plexiglas window smudged again.
4.12.6 Shakedown Test # 6
Shakedown test #6 was conducted with the same feed sample used in previous tests. The
pilot plant was run for few hours until all the reconstituted feed was consumed. A total of two
sets of sample were collected in a gap of 30 minutes of operation and the results are summarized
in Table 4.4. The 97% of carbon recovery in the product from the HHS process indicates almost
all the combustible was recovered from the feed.
Even though no major operational issues were encountered during shakedown test #6, a
couple of minor issues were observed. Therefore, some modifications were made in the pilot
plant to avoid these issues in the full POC pilot-scale testing. First, it was hard for an operator to
move feed from one barrel to another in every 15-20 minutes and simultaneously operate the
plant. This issue also introduced inconsistency into the feed quality to the pilot plant. To rectify
this problem, a 480 gallon feed sump was installed outside the shed with an electric mixer. The
feed sump can accommodate up to seven (7) barrels of feed slurry easily. Second, it was
observed that a substantial amount of pentane was condensing and pooling in the nitrogen
chamber installed below the dryer discharge chute, thus reporting in the final coal product, which
was a loss. To address this issue, a water heating jacket (by-pass copper tubing from hot water
lines) was installed all around the chamber. The whole system was insulated to maintain a hot
environment inside the chamber. This kept the pentane in a vapor form in the chamber allowed
the vapor to be carried by flowing nitrogen to the condenser.
Table 4.4 Shakedown test 6 assays obtained from HHS process POC pilot plant
Test
Feed Product Tails
Combustible % Ash
Number
Recovery % Rejection
%Solids Ash% % Moisture Ash% Ash%
4.2 2.1 64.0 97.0 71.8
Shakedown 6 3.8 7.0
3.8 2.1 64.8 97.1 71.9
126 |
Virginia Tech | 4.13 Conclusions
The proof-of-concept (POC) pilot plant for the novel HHS process was constructed in an
intrinsically safe manner under a constant monitoring of multiple regulatory agencies. The
primary motivation to develop the POC plant was to demonstrate the thermodynamic concept of
dewatering-by-displacement at a large scale. After the completion of construction, shakedown
testing was conducted on the newly developed POC system. Several necessary modifications
were made during the shakedown testing to streamline the pilot plant operation.
The preliminary data obtained from newly engineered HHS process POC pilot plant
tested with a screenbowl main effluent sample showed encouraging results and proved this
revolutionary concept at a large scale. The data shown in Table 4.4 elucidate the separation
capabilities of the process in terms of producing a high quality premium coal product from a
currently discarded stream. The consistency in producing low-moisture and low-ash product
from the shakedown testing signified an efficient operation of the pilot plant in achieving the
primary goal. The pilot-scale testing was moved to next stage where different coal feed stocks
were evaluated to obtain critical data needed to develop a full-scale demonstration unit.
References
1. Amanullah, A., Buckland, B.C., and Nienow, A.W. (2004), “Mixing in Fermentation and
Cell Culture Industries”, Handbook of Industrial Mixing – Science and Practice, Edited
by Paul, E.L, Atiemo-Obeng, V.A., and Kresta, S.M, Chapter -18, Page 1078, John Wiley
& Sons, Inc.
2. Luttrell, G.H. (2013), [Personnel Communications]
3. Mashuga, C.V. and Crowl, D. A. (1998), “Application of the Flammability Diagram for
Evaluation of Fire and Explosion Hazards of Flammable Vapors”, Chemical Safety
Progress, Fall Edition.
4. Osborne, D.G. (1988), Volume 1, Coal Preparation Technology. Graham & Trotman
Limited, London.
5. Primo, J. (2010), “Shell and Tube Heat Exchangers Basic Calculations”, PDH online
course M371, page 11, PDH Center, Fairfax VA, www.PDHcenter.com
6. Smith, J.H., III (2010), “Coal Refuse Thickeners”, The Coal Prep Primer, Edited by Kip
Alderman, Chapter 19, Page 19-4, The Coal Preparation Society of America.
7. Woodie, M. (2013), Conn Weld Industries, [Personnel Communications]
127 |
Virginia Tech | CHAPTER 5 – Pilot-Scale Testing and Evaluation of HHS Process
5.1 Introduction
The coal waste impoundments are considered a permanent disposal sites. According to a
2002 National Research Council report, 70-90 million tons/year of fine coal refuse discarded to
impoundments due to the lack of appropriate separation technologies (Orr, 2002). Unfortunately,
these so-called “waste” impoundments are unexploited energy resources that could not be
recovered by existing commercial technologies and are, therefore, discarded. Researchers from
Virginia Tech have developed a new process called “Hydrophobic-Hydrophilic Separation”
(HHS) based on a thermodynamic concept of dewatering-by-displacement. In the concept, the
surface properties of hydrophobic coal are exploited with hydrophobic liquid for the size fraction
(150 microns x 0) where differential gravity is ineffective for separation. Bench-scale system for
the proposed technology was developed and tested, which showed promise in removing both
mineral matter and water simultaneously from these fine coal discarded streams. Therefore, a
successful commercialization for the HHS process will help the mining industry in recovering
the amount of coal that is currently discarded in impoundments as well as generating a
substantial increase in revenue.
In light of this, Evan Energy LLC., an investment company based in Richmond-VA, has
sponsored to construct a Proof-of-Concept (POC) pilot plant in Virginia Tech, which can
demonstrate the capabilities of HHS process on a larger-scale. The construction of the pilot plant
of rated capacity 100 pounds/hour feed was completed in Spring 2013. Shakedown testing
commenced in June 2013 and was completed in August 2013 with the necessary modifications.
Preliminary test results from shakedown testing obtained from Buchanan’s screenbowl main
effluent sample tested on the newly engineered POC pilot plant were highly encouraging.
Though the real challenge is to evaluate the pilot plant with variety of feeds, particularly high ash
feed, on a large-scale. This chapter details the pilot scale test program, the test results and the
performance evaluation of the HHS process POC pilot plant.
5.2 Samples Procurement
A total of three different samples were procured for the evaluation of POC pilot-scale
plant from three Appalachian coalfields regions. The details of each sample and site descriptions
128 |
Virginia Tech | are discussed in the following sections. All these coal preparation plants are considered as the
potential sites for installation of the first HHS process next scale demonstration plant in future.
5.2.1 Samples from Lone Mountain
Lone Mountain processing facility is owned by Arch Coal and can be the first probable
site for the installation of the first HHS process demonstration plant. The facility is located in the
heart of Central Appalachia coalfields, Lee County, in South West Virginia. The preparation
plant produces premium quality steam coal product at a production rate of 1200 tons/hour and
equipped with typical size-size separators, dense-media gravity separators and fine coal flotation
circuit. The flotation feed is first deslimed by multiple banks of 6.5-inch diameter classifying
cyclone. The cyclone underflow (typically 150 x 44 microns) is treated in flotation circuit while
the cyclone overflow (below 44 microns) is discarded, because of several obvious reasons
mentioned earlier in the document. At present, the processing facility is suffering a loss of
approximately 74 tons/hour solids in this ultrafine waste stream (Lone Mountain, 2013), which
has ash in the range of 55-60%. A total of two batches of sample were procured during the pilot-
plant testing. The first batch of sample (Sample A) was used for first three pilot test runs (pilot
test #1-3). Whereas the second batch of sample (Sample B) was tested in pilot test #4, which was
specifically conducted for an external laboratory, based in Beckley, West Virginia, to collect and
analyze samples around each intermediate unit operation involved in the POC pilot plant for a
detailed evaluation.
Sample A
For the first three pilot-scale test runs, a total of five (5) barrels of the deslime cyclone
overflow raw coal sample was procured from the facility. The percent solid in the sample was
recorded 7.3% with an ash of 55.7%. The size distribution of the raw coal sample, as illustrated
in Figure 5.1, indicates particle top size (d ) 44 microns (size where 10% material is retained).
90
Sample B
For pilot test #4, a total of six (6) barrels of the deslime cyclone overflow raw coal
sample was procured from the preparation plant. The percent solid in the sample was 7.8% with
an ash of 60.5%. The size distribution of the raw coal sample is exhibited in Figure 5.2, which
indicates particle top size (d ) approximately 38 microns.
90
129 |
Virginia Tech | washing. The fine coal cleaning circuit implements reflux classifiers and column flotation for
recovering combustible material. Prior treating the fine coal from flotation circuit, the feed is de-
slimed using bank of 6-inch classifying cyclone, which cuts at 45 microns. The deslime cyclone
overflow (size below 45 microns) is discarded to the plant thickener because of high percentage
of clays (Chafin et al., 2012).
A total of seven (7) barrels of 6-inch de-slime cyclone overflow raw coal sample were
procured from Arch’s Leer preparation plant. A percent solid in the sample was recorded 3.8%
with a high ash of 60.5%. The size distribution of the raw coal sample, as shown in Figure 5.3,
indicates particle top size (d ) approximately 55 microns.
90
5.2.3 Sample from Sentinel Preparation Plant
The Sentinel processing plant facility is located in Barbour County, West Virginia and
also owned by Arch Coal. Mining operations extract coal from the Clarion coal seam. Coal from
the Sentinel mining complex is processed through 615 tons/hour preparation plant. The plant is
well equipped with heavy media vessel and cyclones for coarse coal processing while spiral
gravity concentrator and conventional flotation cells for fine coal processing (Coal Age, 2013).
The flotation product is dewatered with screen-bowl centrifugal dryer. The centrifuge main
effluent, which carries typically 50% ultrafine coal particles (-325 mesh) of the flotation
concentrate, is discarded into the plant thickener.
Figure 5.3 Size distribution curve for Leer plant feed sample
131 |
Virginia Tech | After pilot test #5, the vibrating mixer tank, the key unit operation in the process, was re-
designed and replaced with a very low-volume tank.
The POC pilot test results are discussed in the following sub-sections individually for each test.
5.3.1 Pilot Test #1
Pilot-scale testing was started utilizing the first batch (Sample A) of de-slime cyclone
overflow sample procured from Lone Mountain plant facility. The fresh sample feed (as-
received) from the preparation plant was used in the first test and recovered back for the later test
runs after analyzing samples. A total of three product and tails samples were collected at a fix
interval of 30 minutes. The direct data obtained from the pilot-test is outlined in Table 5.1. The
consistency in low ash clean coal product and high reject ash indicates excellent cleaning
capabilities of the HHS process even for high-feed ash tested on the POC unit. In addition,
single-digit product moisture values indicated the potential in this innovative dewatering method
involved in the HHS process.
From the analyzed assay values, performance parameters such as combustible recovery,
ash rejection and sulfur rejection were determined. The HHS Process POC pilot plant system
was able to recover up to 86% carbon and rejected impressively more than 98% ash (mineral
matter) from the discarded deslime cyclone overflow feed sample of Lone Mountain. On the
contrary, the sulfur rejection was only 9%. The low sulfur rejection is expected because most of
the sulfur is in organic form, which is also hydrophobic in nature and, therefore, recovered with
the clean coal product. The complete assessment is discussed later in the chapter. In later tests,
sulfur was not analyzed (except pilot test #4) as a parameter for performance evaluation.
Table 5.1 Pilot Test #1 samples assay from the HHS process POC plant
Feed Product Reject
Test
Dry Dry Dry Dry Dry Dry
Number % Pounds/ %
Ash Sulfur Ash Sulfur Ash Sulfur
Solids hour Moisture
% % % % % %
1.8 4.6 2.2 89.4
Pilot Test
7.3 55.7 0.4 2.6 5.3 2.5 0.63 89.8 0.06
#1
5.4 8.5 2.5 89.3
133 |
Virginia Tech | 5.3.2 Pilot Test #2 and #3
After collecting enough samples required for analysis from pilot test #1, the tails and
clean coal product of the Lone Mountain cyclone overflow feed sample was re-constituted and
mixed well to utilize in pilot test #2 and similarly later on for pilot test #3. A total of two product
and reject samples were collected in both the test and analyzed. In addition, as anticipated that
the feed characteristics will change, feed samples were collected in each test run. The complete
set of assays is illustrated in Table 5.2.
The feed to POC plant deteriorated after each test run. Even with a higher feed ash, the
HHS process pilot plant produced consistently low-ash and low-moisture product with
exceptionally high reject ash. Even though, the assay values for product and reject are very
similar to pilot test #1, the increase in feed ash has decreased combustible recovery to a range
from 69 to 73%. On the other hand, the ash rejection was still reported as high as 98.8%. The
complete evaluation is discussed later in the chapter.
5.3.3 Pilot Test #4
The second fresh batch of cyclone overflow sample (Sample B) from Lone Mountain was
procured and used in pilot test #4, which was conducted primarily for Precision Laboratory,
located in Beckley, West Virginia. The laboratory analyzed the samples collected from pre-
defined sample points located in each intermediate process streams around the POC pilot plant.
In addition, the pentane loss associated with clean coal product was also determined by the
laboratory. The complete data obtained from the test is instrumental in development of the next-
scale 1 metric ton/hour HHS process demonstration plant. The feed, product and reject assays are
outlined in Table 5.3. The detailed description with sample assays from the intermediate streams
is discussed in Chapter 6.
Table 5.2 Pilot Test #2 and #3 samples assays from the HHS process POC plant
Feed Product
Reject
Test Number
Ash %
%Solids Ash% % Moisture Ash%
5.9 3.3 86.0
Pilot Test #2 7.8 66.2
5.2 3.3 86.1
6.8 3.8 88.3
Pilot Test #3 8.6 67.5
9.6 3.6 86.9
134 |
Virginia Tech | Table 5.3 Pilot Test #4 samples assay from the HHS process POC plant
Feed Product Reject
Test
Dry Dry Dry Dry Dry Dry
Number Pounds/ Air-Vented
%Solids Ash Sulfur Ash Sulfur Ash Sulfur
hour % Moisture
% % % % % %
7.8 60.5 0.3 2.8 12.5 2.7 0.64 90.6 0.04
Pilot Test #4
4.5 77.3 0.13 5.7 9.8 2.7 0.66 88.7 0.04
NOTE: The samples were collected and analyzed by Precision Laboratory, West Virginia.
During the test run, two sets of samples were collected from each sample point. After
collecting the first set of samples, the feed to the pilot plant appeared to be low in the feed sump,
and therefore could not be nixed properly by sump agitator. This can be noticed in the second
feed sample characteristics outlined in Table 5.3. Nevertheless, the plant continued to run until
the feed sump was emptied and second set of sample was collected.
The low-product ash reported by Precision Laboratory from both the samples, as well as
from previous pilot test runs, clearly indicates the coal cleaning capabilities of the HHS process
are independent of the feed ash. In addition, the tail ash was reported as high as 90.6%. From the
assay analysis, performance parameters were evaluated, which showed the plant recovered
84.4% combustibles with ash rejection 98.5% from the first set of sample. The inconsistent feed
for second sample affected the performance, as the combustible recovery was only 56.8% with
ash rejection 99.5%. Due to inconsistency in the feed, only the first set of samples was used later
in development of process flowsheet (discussed in Chapter 6) for 1 metric ton/hour
demonstration plant.
Another purpose for conducting the pilot test #4 was to determine the pentane loss
associated with the clean coal product. The pounds/ton losses reported by the external laboratory
for both the samples were 26 and 148 respectively. Even for the first sample, the loss was very
high. In addition, the product moisture disclosed by the lab was also slightly higher as compared
to previous pilot test runs. After communicating with the lab, it was found that the samples were
vented for pentane vapors in open atmosphere overnight in humid conditions. The humidity is
expected to increase sample moisture and bring irregularities in the reported analysis. In later
tests, attempts were made to determine the exact amount of pentane and moisture in clean coal
product using thermo-gravimetric and gas chromatography analysis.
135 |
Virginia Tech | 5.3.4 Pilot Test #5 and #6
After successful testing with Lone Mountain samples, pilot test #5 was conducted with
the fresh feed procured from Leer’s deslime cyclone overflow stream, while pilot test #6 was
conducted with the re-constituted feed from the same sample. Prior to the pilot test #6, a major
modification was conducted, which was replacing the old vibrating mixer unit with a reduced-
volume vibrating mixer. The details of this investigation are discussed in later sections. Standard
operational protocols were followed for the pilot test #5. Only one set of samples was collected
in the pilot run and the direct data obtained from the tests are outlined in Table 5.4. The low-ash
results elucidate the high performance of HHS process with this type of coal feed. Furthermore,
the combustible recovery calculated was 79.7% with the ash rejection 97.5%.
For moisture studies, thermo-gravimetric analysis was conducted on clean coal product
sample at 50°C until all the liquids (pentane and water) are removed and combined weight loss is
reported in Table 5.4. This is because the initial method used for moisture determination in the
tests, prior to pilot test #4, involved preheating of the clean coal sample at 40°C for a 6-8 minutes
in the oven to remove the traces of pentane and then conduct the moisture analysis at 105°C. It is
observed that the method has several issues, which are discussed in the next section. To
eliminate any irregularities with analysis, combined weight loss is reported from pilot test #5.
5.3.5 Pilot Test #7
Pilot test #7 was conducted with the fresh screenbowl main effluent sample procured
from Sentinel preparation plant. Similar to the pilot test #6, the small vibrating mixer unit was
used in this pilot-test. Three sets of sample were collected in the pilot run and the direct data
obtained for the best set from the test are outlined in Table 5.5. The low-ash results further
confirm the high performance of HHS process with this type of coal feed. Moreover, the
combustible recovery calculated was 90.8% with the ash rejection 86.8%. In addition, an
investigation was conducted during the pilot test #7 to study the effect of high dosage of frother
concentration in the feed on the pilot plant performance. The details are discussed later in the
chapter.
136 |
Virginia Tech | Table 5.4 Pilot Test #5 and #6 samples assays from the HHS process POC plant
Test Number Feed Product
Reject
Combined Ash%
%Solid Ash% Ash%
Weight % Loss
Pilot Test #5 3.8 53.0 8.5 3.4 84.4
Pilot Test #6 3.5 50.9 14.0 3.3 85.7
Table 5.5 Pilot Test #7 sample assay from the HHS process POC plant
Test
Feed Product
Number Reject
Combined Ash%
%Solids Ash% Ash%
Weight % Loss
Pilot Test #7 4.5 28.5 16.6 5.5 78.9
5.4 Method for Moisture Determination
To determine the moisture in the final product obtained from the POC pilot plant, a
gravimetric method was established, which is based on several assumptions. It must be noted that
there is no standard method available to exactly determine volatile fractions (pentane and water)
in a coal-water-pentane mixture. In this method, the sample was first exposed at 40°C for a pre-
determined time. After the time period, initial sample weight was noted and the sample was
reheated in an oven at 105°C for 1 hour and the final sample weight was recorded. The difference
in the weight yielded the weight percent loss, which was considered as percent moisture in the
sample.
To identify the pre-determined time for pentane evaporation, an investigation was
conducted. Synthetic samples were prepared with known weights of pentane, water, and dry coal
powder. A moisture balance (manufactured by A&D, Model MF-50) was set at 50°C (which is
the lowest temperature that can be set on the equipment) and connected with a computer. Each
synthetic sample was heated isothermally at the set temperature inside the moisture balance
chamber and percent weight loss was plotted with time as exhibited in Figure 5.5. The graph
shows two evaporation rates with a sharp elbow at approximately a percent weight loss very
close to the known initial weight percent of pentane in the synthetic samples. All the experiments
conducted with different weight percent of pentane and water showed the time for the inflection
point at approximately 4 minutes.
137 |
Virginia Tech | Figure 5.5 Synthetic samples weight loss curves with time
Therefore, the evaporation time for pentane was identified at the inflection point, and the
time was generalized for POC product samples. The POC samples were first exposed at 40°C in
an oven for 6-8 minutes (time increased because of lower temperature) prior to conduct moisture
analysis. This method was not ideal because some moisture will also evaporate at this
temperature, and there is a possibility that some percentage of pentane will be left in the POC
test run samples. However, the method was expected to provide a fair estimation of moisture in
the clean coal product. Several assumptions were made to generalize this gravimetric method to
use for determination of moisture in POC test samples:
Evaporation of pentane was independent of surface area exposed at higher temperature.
All the pentane was evaporated in 4 minutes from the sample irrespective of its weight
percentage in the sample.
There was no absorption of pentane with coal.
Pentane was easily available for evaporation from the coal surface.
The method was abandoned after pilot test #4, as sample characteristics that obtained
from POC dryer unit and the synthetic sample was observed to be different. Because of the
helical motion of the Holoflite® screw dryer used in the pilot plant, the clean coal product
obtained was conglomerated into coal balls (illustrated in Figure 5.6) instead as a fine powder. In
contrast with synthetic samples results, the weight % loss versus time curve for the POC sample
did not showed a similar sharp elbow (illustrated in Figure 5.7), rather showed a blunt curve
when tested. The blunt curve indicates that some portion of pentane and water may be trapped in
the conglomerated coal sample. Similar curves were obtained for all the samples received from
138 |
Virginia Tech | the POC dryer unit. The initial method developed with synthetic samples was not suitable with
the POC test samples for moisture determination and, therefore, the method was abolished.
There may be multiple reasons but three possible reasons are discussed here. First,
generally evaporation of a liquid can be considered as the movement of molecules from the
surface into the vapor phase. Pentane like other hydrocarbons is (almost) insoluble in water. In
pentane-water system, pentane floats on the top of water surface. Typically, in low
concentrations, pentane makes a very thin layer, covering the entire water surface. Previous
studies (Smith, 2008) conducted with synthetic samples showed evaporation of pentane differs
with varying conditions. Smith (2008) further revealed that for water-pentane mixture only,
pentane molecules evaporate much quicker as pentane does not like water. On the contrary, in
pentane-coal system pentane likes coal and, therefore, tends to stay longer when heated. The
second reason could be that the coal is a bad conductor for heat transfer. Since the POC product
coal is conglomerated, the rate of evaporation is very slow and therefore it is hard to identify any
distinction between rate of evaporation of moisture and pentane. The third reason could be that
the POC samples may have a very minor percentage of pentane and all the liquid is moisture
only. The synthetic tests were conducted at high percentages of pentane (5-9%), which provide a
sharp elbow in the weight loss curve as described in Figure 5.5.
Figure 5.6 Conglomerated dry coal product obtained from POC dryer unit
139 |
Virginia Tech | Figure 5.7 Weight loss curve for sample collected in POC pilot test #5
Due to this complexity and in absence of a viable method, a combined weight % loss
(moisture and pentane) was reported from pilot test #5. Further attempts were made using gas
chromatography analysis to estimate the amount of pentane in these samples. Once pentane is
determined, moisture can be recalculated knowing the total weight percent of liquids in the clean
coal sample.
5.5 POC Performance Assessment
The direct data obtained from the HHS process pilot-scale plant has proved that
combined cleaning and dewatering for ultrafine coal can be achieved in a single process using
the concept of hydrophobic displacement. The consistent low-moisture and low-ash clean coal
products were obtained from three different type of coal samples tested on the POC pilot plant.
Crucial investigations were conducted during the pilot test runs, such as testing vibrating mixer
with small residence time impact of skimming, effect of feed size distribution, effect of excessive
frother in feed and pentane absorption analysis. These investigations were necessary to collect
information for designing the next-scale demonstration unit. The following section discusses
these investigations in detail in addition with a general performance assessment of the POC pilot
plant.
140 |
Virginia Tech | 5.5.1 Separation Efficiency
Separation efficiency, which is defined as the recovery of the desired material and
rejection of unwanted material in the clean coal product, is a performance indicator for any
process. The separation efficiency curves are commonly used to determine optimum conditions
of the process for certain type of feed. These curves normalize any variation in the feed to
provide optimum conditions. Table 5.6 outlines complete set of data for each pilot-test with their
separation efficiencies.
In shakedown test # 6, conducted with Buchanan screenbowl main effluent, almost all the
carbon (97% from feed) was recovered with substantial amount of ash also rejected (72% from
feed) using the novel process. The ash rejection was relatively high considering the feed ash was
only 7%.
The pilot test conducted with deslime cyclone overflow feed (Pilot test #1 to #6) also
produced enviable performance with the POC plant. The fresh feed from Lone Mountain (pilot
test #1 and #4) provided carbon recoveries as high as 86% with ash rejection unexceptionally
high (> 98%). Similar results were demonstrated with Leer deslime cyclone overflow feed. One
of the main reasons to discard minus 44 micron raw feed using 6-inch deslime cyclone in the
preparation plants is that it carries large amount of clay material. It is apparent that HHS process
can easily remove this clay material and produce premium quality sellable product from this
ultrafine raw coal stream.
Pilot test #7 test results, conducted with Sentinel screenbowl main effluent, are in
compliance with other feeds tested on POC plant. The high combustible recoveries with relative
high ash rejection further corroborate the outstanding performance of the HHS process with
ultrafine coal slurries.
A batch scale system typically depicts a plug flow system, which indicates optimum
performance level of any process that can be achieved. The POC plant was developed with
several engineering assumptions based on bench-scale testing surveys. The separation
efficiencies obtained from POC pilot test results are in good agreement with the HHS process
bench-scale system, exhibiting the excellent performance of HHS process pilot plant. Figure 5.8
illustrates the comparison between the batch and pilot-scale separation efficiencies.
141 |
Virginia Tech | 5.5.2 Particle Size Effect – Lone Mountain
In pilot test #4, conducted with Lone Mountain deslime cyclone overflow feed, the
collected samples (feed, product and tails) were screened at three size fractions. These fractions
were generated using sieves of aperture 60 mesh, 100 mesh, 325 mesh, and 500 mesh to
investigate the performance of the process with particle size. Figure 5.9 shows the recovery and
rejection plot with geometric mean size of the particles. The process performance is escalated for
particles below 75 microns, however, as the particle size increases, both combustible recovery
and ash rejection is decreased. Furthermore, the process is inefficient to separate organic sulfur
from the feed. This is because the organic sulfur is hydrophobic in nature.
5.5.3 Effect of Re-designed Vibrating Mixer
A new vibrating mixer unit, with significant volume reduction (1/6th of old unit), was
installed by replacing the old vibrating mixer in the POC pilot plant (Figure 5.10). As compared
to other unit operations in the pilot plant, the vibrating mixer is the most unknown unit in the
process. The sole purpose of this investigation was to evaluate the plant performance by reducing
the volume of pentane in the unit, but keeping the same design parameters (cylindrical reactor
with conical bottom) and flow rates. This helped to determine if the scale-up of next-stage unit
needed to be volume based or not. The complete comparison between old and new vibrating
mixer is outlined in Table 5.7.
Figure 5.10 New reduced volume vibrating mixer tank design and constructed unit
144 |
Virginia Tech | Table 5.7 Design changes between old and new vibrating mixer unit
Old-Vibrating Mixer New Vibrating Mixer
Total Tank Volume 120 gallons 20 gallons
Volume – Aqueous Phase 53 gallons 2.2 gallons
Volume – Pentane Column 35 gallons 10 gallons
Total Flow In 5.04 gallons 5.04 gallons
Residence Time ~7 minutes ~2 minutes
Sweco Screens Diameter 24” (165 mesh & 35 mesh) 18” (165 mesh & 35 mesh)
Operating
2.5 Hz, 1.125” 2.5 Hz, 1.125”
Frequency, Amplitude
At a given process flow rate of 5.04 gallons/minute to the unit, the residence time in the
process significantly reduced from 7 minutes to 2 minutes. An additional advantage with smaller
unit was handling a lower volume of pentane (10 gallons as compared to 35 gallons) in the
vibrating mixer. Pilot test #5 and #6 were conducted with old and new vibrating mixer
respectively with the same feed sample obtained from Leer de-slime cyclone overflow.
Results (outlined in the Table 5.4) obtained from pilot test #5 and #6 clearly indicates that
there is no significant difference in cleaning aspect of the process. The product ashes only differ
by 0.1% and reject ashes by 1.3%. In addition, the combustible recovery calculated for pilot test
#6 improved to 83.2% as compared to 79.7%, with an ash rejection 97.3%. This signifies that
Figure 5.11 Performance comparison of two vibrating mixer unit tested in POC pilot plant
14 5 |
Virginia Tech | the process scale-up might not be volume-based but on screen area-based. The moisture varies
by 5.5 percentage points, which may be due to the misplacement of water droplets in the product.
The performance comparison between the old and new design is illustrated in Figure 5.11.
5.5.4 Effect of Oil Skimmer
The pilot test #6 was also conducted to unravel the impact of oil-skimmer on the POC
plant performance. After collecting the first sample in usual operating conditions, the oil-
skimming pump (P-800) was deactivated, and the plant continued to run for another 30 minutes.
The second set of samples (product and reject) were collected and analyzed. Figure 5.12 exhibits
comparison of combustible recovery and ash rejection between the two sets of sample. With
skimmer in operation, there was no effect on ash rejection (97.3% as compared to 97.7%),
whereas, the combustible recovery improved significantly by 10.6% as outlined in Figure 5.12.
Clearly, the investigation showed implementation of oil-skimmer was necessary in the POC pilot
system to achieve better performance from the process. The direct data obtained from the
analysis is exhibited in Table 5.8.
Figure 5.12 Performance comparisons of POC plant with and without skimmer in
operation
146 |
Virginia Tech | Table 5.8 Pilot scale test data showing the effect of skimming on POC performance
Pilot Test #6 Feed Product Tails
Combustible % Ash
Ash
Ash Combined Ash Recovery% Rejection
%Solid %
% Weight % Loss %
With
14.0 3.3 85.7 83.2 97.3
Skimmer
3.5 50.9
Without
14.5 3.2 78.7 72.6 97.7
Skimmer
5.5.5 Effect of High Frother Dosage
Typical concentration of an alcohol-based frother used in the conventional flotation
process is 15 ppm (Kohmuench et al., 2010). After collecting first set of sample in the pilot test
#7, a high dosage of frother (30 PPM of MCHM type) was added into the feed sump and
conditioned for 20 minutes. The investigation was conducted to evaluate the effect of residual
frother present in the in-plant streams of screenbowl main effluent. The pilot plant was operated
for another hour and the second set of sample was collected and analyzed. The direct data
obtained from the test is illustrated in Table 5.9. It is apparent from Figure 5.13 that there is no
significant effect of high dosage of frother on the POC pilot-plant performance.
Figure 5.13 Performance comparisons of POC pilot-plant with and without frother
147 |
Virginia Tech | Table 5.9 Pilot-scale test data comparison between with and without high dosage of frother
Pilot Test #7 Feed Product
Tails Combustible % Ash
Ash% Recovery% Rejection
%Solid Ash% % Moisture Ash%
Without
4.5 28.5 16.6 5.5 78.9 90.8 86.8
Frother
With Frother
4.6 29.1 12.9 5.0 82.2 92.2 88.3
30 PPM
5.5.6 Estimation of Pentane Loss - Gas Chromatography Analysis
Gas chromatographic (GC) analysis was conducted with Leer product sample collected
from HHS process POC pilot plant. Several individual samples of weight 100 mg in special
sealed glass vials (25 ml) were prepared. Each vial was heated at 50°C for 30 minutes and it was
assumed that all the pentane would be evaporated in the headspace inside the vial. Then, a 1.5 ml
sample was collected from the headspace and injected into a “flame ionization detector” type GC
system. The reported analysis was used to calculate pentane lbs/ton loss in the coal samples
(illustrated in Figure 5.14). The data indicates that on an average at least 0.182% pentane is
absorbed in the POC coal samples, which is equivalent to 3.64 lbs/t of clean coal product.
Figure 5.14 Gas chromatography analyses for Leer clean coal sample obtained from HHS
process POC pilot plant
148 |
Virginia Tech | CHAPTER 6 - Modeling and Economic Analyses of HHS Process
6.1 Introduction
Modern coal processing plants incorporate as many as four separate cleaning circuits for
treating coarse (plus 2 inches), intermediate (2 inch x 1 mm), small (1 x 0.15 mm) and fine (0.15
x 0.044 mm) size fractions of feed coal. The ultrafine particles below 44 microns are often
discarded due to low recoveries, poor dewatering performance and high cost. To address this
problem, researchers at Virginia Tech have developed the Hydrophobic – Hydrophilic
Separation (HHS) process to more efficiently remove ash and water from these ultrafine streams
in a cost effective manner. Bench-scale tests conducted in the current work have successfully
demonstrated that the HHS process simultaneously clean and dewater ultrafine coal streams,
which are currently discarded in many existing coal preparation plants. The bench-scale tests
consistently showed moisture and ash levels in final product below 10% with variable types of
coal feedstocks. In light of this, a Proof-of-Concept (POC) plant was successfully designed,
constructed and tested in the current work. The consistent low-moisture low-ash coal products
obtained from POC plant further demonstrated that the technology could be commercialized at a
larger production rate.
Data obtained from bench and POC-scale testing of the HHS process provided critical
engineering information for the development of a next-scale demonstration unit. However, the
HHS process is an advanced separation technology that incorporates several complex sub-
processes such as oil-agglomeration, solid-liquid separation, oil-recovery process and the
innovative de-agglomeration process using vibrating mesh device. Therefore, successfully
simulating each unit operation using reliable models would be of great benefit in designing and
improving each processing unit.
The chapter presents the results of several bench-scale studies conducted to develop
mathematical models for each unit operation. The models were then incorporated into the
LIMN® flowsheet simulator, which is a powerful Microsoft Excel-based software package. The
resulting simulation package can be used to predict the performance of each unit operation in the
HHS process even before actually running the testing equipment. In the current work, this
process-engineering tool has been used to develop a 1 metric ton per hour (1 t/hr) process
flowsheet for the HHS process. The flowsheet for the 1 t/hr demonstration plant was based on
150 |
Virginia Tech | data obtained in the POC plant using coal samples from the Lone Mountain preparation plant.
The models were incorporated in the flowsheet simulator to provide preliminary design
specifications for the plant. Furthermore, economic evaluations of the HHS process were
conducted for Lone Mountain facility and for the Buchanan preparation plant. The economic
evaluations examined potential increases in productivity and direct revenue generation that may
be realized by implementing the HHS technology in these existing fine coal cleaning circuits.
6.2 Model for Mixing Operation
Oil-agglomeration, which is the first step of the HHS process, utilizes both high and low-
shear mixing. Mixing is itself a complex process involving several dimensionless quantities,
which are application specific. Until it is explored, it is unknown which quantity can be used for
scale up. On the bench-scale system, mixing was achieved using a standard kitchen blender,
which could not provide crucial data for scale-up. The POC system was designed overcautiously
with excess volume (residence time) to deal with uncertainties created by a lack of scale-up data.
Therefore, to identify the governing criteria and develop an accurate model for the agglomeration
step in the HHS process, a detailed investigation was conducted on a batch-scale.
In POC-scale testing, it was determined that the loss of pentane associated with coal is
substantially higher as compared to water in the tails. Any unrecovered coal particle in the HHS
process reject will also carry pentane coated on its surface, which ultimately is a loss. This loss is
a key economic driver in the process. Therefore, attempts were made to determine optimum
mixing conditions and scale-up factors that provide minimal carbon loss in the agglomeration
tails. To evaluate mixer performance, a parameter called ash recovery ratio (ARR) was used,
which is defined as the ratio of tails ash to feed ash. The ratio was studied with different mixing
parameters such as impeller RPM, impeller diameter, specific power input, and mixing time.
6.2.1 Experimentation and Results
Figure 6.1 illustrates the setup used to conduct the mixing studies. The apparatus includes
a mixing device with an in-built feature of measuring torque (in N-cm). The speed can be
controlled from 60 to 2000. The device is connected with a mixing shaft with impeller at the end.
Two types of impellers, each of different diameters (1, 2, and 3 inches) were evaluated.
151 |
Virginia Tech | Torque and
RPM indicator
Mixing
shaft
Reactor
Tailings Port
Figure 6.1 Bench-scale apparatus used for agglomeration studies
T he first was a laboratory propeller, which has low intensity and high pumping number. The
second was a high-dispersion blade, which has high intensity and a very low pumping number.
The internal diameter of the reactor at the location of impeller was held constant at 4 inches of
depth. Furthermore, a port was installed at the bottom of the reactor to withdraw high-ash tails
after agglomeration.
The mixing tests were carried out at variable time periods and RPMs for each impeller
using a high-rank bituminous coal sample (as-received 4.5% solids) procured from the Leer
facility. Pentane was used as the agglomerating agent. After each test, the agglomeration process
was terminated and the torque values were recorded. The reactor was then kept stationary for 30
seconds. During this time, the phase separation was visible where all the agglomerated coal
floated on the top of dirty aqueous phase. After an additional 30 seconds, all of the tails collected
152 |
Virginia Tech | from the tailings port were collected and analyzed. The agglomerates that remained in the reactor
were not collected or analyzed.
Figure 6.2 shows the direct data obtained during the investigations at different impeller
diameter (d)/reactor diameter (D) ratio, which is also considered as another critical parameter. A
higher ash-recovery ratio indicates that a higher tails ash was achieved in the test. At d/D ratio =
0.5, the tail ash recovery was reported maximum at 2,000 RPM and 4 minutes. As anticipated,
the ash-recovery ratio improves with increasing residence time except in case of d/D = 0.75 with
high RPM, where the ratio is highest at 15 seconds and then decreases with time. It was observed
that, at extremely high specific energy and increasing mixing time, agglomerates formed very
quickly but soon broke into smaller size aggregates. This may be due to excessive shearing effect
per unit volume inside the system.
Figures 6.3 and 6.4 illustrates several plots showing the effect of impeller tip speed
(feet/second), power/volume (KW/m3), d/D ratios and residence time on the tailings ash for both
types of impeller respectively. The trends with tip speed in both the cases (Figure 6.3a; Figure
6.4a) indicate an optimum condition for the given reactor geometry, which is approximately 17
feet/sec, to achieve the highest reject ash. The tip speed is a direct indication of shear intensity
provided to the system, showing it is one of the prime factors affecting the process. Therefore,
tip speed must be considered for the scale up of this unit operation in the HHS process.
On the contrary, in case of propeller power/volume curve (Figure 6.3c) does not produce
a significant correlation with ash ratio, though it appears to improve with increasing power. This
may be due to the fact that propeller type impellers have high pumping number (0.5), while high-
shear blades have low pumping number (0.26), where trend is more significant (Figure 6.4c).
Residence time defines the reactor volume needed in any process. Figure 6.3e depicts significant
information for designing the agglomeration unit for the HHS process. At high impeller speed
and d/D ratio, substantial reject ash is achieved at very short mixing time with propeller type
blade. However, the effect is different with high-shear blade (Figure 6.4e). It was observed that
the former type impeller created vortex and were excellent for mixing while the latter was only
good for dispersion only. From the trends, it is evident that propellers are good for agglomerates
growth while high-shear blade can be used prior to agglomeration for dispersion of pentane
liquid and coal-oil contact in the ultrafine coal slurry for a very short residence time.
153 |
Virginia Tech | Table 6.1 Response matrix for fitting rate constant (k) at variable RPM and d/D ratio
RPM
k 500 1000 2000
0.25 0.035 1.12 1.23
d/D
0.5 1.2 1.05 2.6
0.75 6.3 10.2 12.2
Table 6.2 Response matrix for fitting correction factor (α) at variable RPM and d/D ratio
RPM
α 500 1000 2000
0.25 0.33 0.81 0.69
d/D
0.5 0.919 0.25 0.21
0.75 0.63 0.454 0.415
The model equation for rate constant (k) was ca lculated empirically (R2 = 0.99) in terms of tip
speed (TS) in meter/second as described in Equation 6.2. Similarly, empirical equations for the
correction factor (α) were determined as a function of tip speed in meter/sec and specific power
(SP) in KW/m3 for individual d/D ratio (R2 = 0.82), as shown in Equation 6.3.
( )
[6.2]
( )
[6.3]
6.2.3 Discussion
The investigation on oil-agglomeration with ultrafine coal and pentane provided useful
information that can assist the designing units for the demonstration plant. The data shows that
tip speed is an important criterion in the HHS process agglomeration step and can be used for
scale-up process. The low-intensity and high-pumping propeller type impellers assist
agglomeration growth process. While the high-shear blade is good for dispersion of immiscible
157 |
Virginia Tech | Figure 6.6 Correlation between measured and predicted values of normalized ash ratio
liquid in coal slurry. For high-dispersion blade s, very low residence time will be preferred.
Additionally, a high reject ash is achievable at a high impeller speed and high d/D ratio at a very
short retention time.
The developed model is fairly accurate (R2 = 0.9) and found a discreet function for
specific d/D ratio, which is a best fit using the available experimental data. Figure 6.6 illustrates
the correlation between experimentally measured and predicted values from empirical model of
normalized ash ratio.
6.3 Model for Vibrating Mixer
The vibrating mixer unit is equipped with a novel oscillatory mesh device that facilitates
three important mechanisms necessary for achieving low-moistures and high recoveries. These
are de-agglomeration (breaking of agglomerates), homogenization (keep particles in suspension)
and water coalescence. The combination of the first two is dispersion mechanism. The kinetic
study was conducted on batch-scale vibrating mixer and was discussed in Chapter 3. In the
investigation, the rate of homogenization (K = 0.36 min-1) in the reactor was determined using
h
dry coal powder, while the rate of dispersion (K = 0.25 min-1) was identified using the same
d
procedure but with spherical coal agglomerates. Therefore, the difference in rates defined the
158 |
Virginia Tech | Figure 6.7 Semi-log concentration plots for mechanisms involved in vibrating mixer
rate of de-agglomeration or breakage (K
b
=K
d
- K
h
= 0.11 min-1). The investigation indicated that
the agglomerate dispersion mechanism follows a first order rate when operating under steady
st ate conditions. Figure 6.7 illustrates the semi-log concentration plot with time showing the
rates of dispersion and homogenization mechanism.
From the dispersion rate data, a simple kinetic rate model can be defined for the
concentration in the reactor as a function of time, as described in Equation 6.4. The
experimentally measured data and predicted values fit excellent correlation with a coefficient of
determination (R2) equals to 0.9976 and are outlined in Table 6.3.
( ) [6.4]
Furthermore, the relationship between the rate constant and time can be analyzed for
reactor modeling using the Levenspiel equation (described in Equation 6.5), which is appropriate
for first-order kinetic processes. The relationship provides recovery as a function of the
dimensionless Peclet number and the dimensionless product of mean residence time of particles
and the process rate constant (Levenspiel, 1999). Peclet number is a mixing intensity indicator
used to study transport phenomena of particles in fluids. The Peclet number is zero for perfectly
mixed reactors and infinite for plug-flow reactors.
159 |
Virginia Tech | Table 6.3 Measured and predicted values correlation for vibrating mixer
C/Co (in Reactor)
Time (minutes)
Measured Predicted
0 1.00 1.00
1 0.76 0.78
2 0.56 0.61
5 0.26 0.29
10 0.08 0.08
15 0.03 0.02
( ⁄ )
( ) ( ⁄ ) ( ) ( ⁄ ) [6.5]
√
From the experimental kinetics data, the mean residence time (τ) can be calculated using
Equation 6.6, which was found to be 4.45 minutes. Similarly, the Peclet number can be estimated
using its relationship with dimensionless variance (σ2) as described in Equation 6.7. The Peclet
number for the reactor was calculated approximately 0.93, which indicates the vibrating mixer is
close to a well-mixed reactor.
∑
[6.6]
∑
∑
( ) [6.7]
∑
6.4 Model for Regent Thickener
In the novel HHS process, the thickening of suspended ultrafine clean coal particles in
pentane from the vibrating mixer is a very important step, particularly in designing the heat
exchanger units. Higher thickener underflow solids will reduce the liquid pentane load that has to
be evaporated in the dryer. The force of gravity concentrates the suspended particles in the
thickener. Therefore, to properly apply gravitational sedimentation in identifying the right size
thickener, both basic and applied theory must be considered.
160 |
Virginia Tech | Figure 6.9 Pentane thickener dual model developed in LIMN® software
Considering the particle size, the settling rate of ultrafine coal solids in pentane was
observed to be very high. The unexpectedly high rate may be because particles do not carry any
electrostatic repulsive force when suspended in a non-polar liquid. Therefore, the remaining
attractive force naturally coagulates the ultrafine particles, which increases the settling rate.
6.5 Model for Reagent Dryer
A basic pentane evaporation model was developed in LIMN® based on the bench-scale
data reported in Chapter 2. The thermo-gravimetric experiments were conducted with known
amounts of ultrafine coal mixed with known amounts of liquid pentane in sealed vials. Each vial
was exposed to an isothermal environment at different temperatures, and the weight loss was
recorded with time. Figure 6.10 shows plots of concentration ratios (C/C ) with time at each
o
temperature. The evaporation of pentane for each temperature closely follows an exponential
relationship with time with R2 values ranging from 0.89 to 0.93, and can be best fit using the
following equation:
⁄
[6.14]
In Figure 6.10, the red circles indicate the batch-scale experimental data while the predicted
values from the empirical equation is shown with a black line.
163 |
Virginia Tech | It must be noted that the model is based on bench-scale experimental data and is expected
to vary depending on the heat transfer coefficient of the drying technology. Industrial dryers
have a high heat transfer coefficient as compared to the bench-scale system and therefore, an
appropriate dryer-condenser system is expected to reduce the loss of pentane as compared to
what is predicted with the proposed model.
6.6 Development of 1-Metric ton Plant Flowsheet
In pilot scale test #4, conducted with Lone Mountain deslime cyclone overflow feed,
various samples were collected from intermediate sample streams in addition with regular feed,
product and tail samples. Figure 6.12 shows the simplified flow diagram indicating the sample
points across the POC pilot plant. The data obtained during the test was utilized in the
development of a flowsheet for a 1 metric ton per hour demonstration plant (Table 6.4). The
LIMN® processing software was used to design the material and heat balance flowsheet, which
shows the steady-state flows necessary to specify and design each unit operation. Figure 6.13
shows the newly drafted flowsheet used as the first-step for the prototype demonstration plant.
Figure 6.12 Schematics showing sample points across the POC pilot plant
16 6 |
Virginia Tech | Table 6.4 Direct data obtained from POC pilot scale plant at various sample points
Sample 1 Sample 2
Ash% 60.5 77.3
Feed %Solids 7.8 4.5
%Sulfur 0.3 0.13
Hi-Shear O/F Ash% 41.4 68.8
Sieve Screen U/F Ash% 83.2 73.6
Vibrating Mixer U/F Ash% 90.7 83.3
Vibrating Mixer O/F Ash% 1.9 2.1
HL Thickener U/F Ash% 2.8 2.4
Skimmer Ash% 78.6 74.9
Ash% 2.7 2.7
Product %Moisture 12.5 9.8
%Sulfur 0.64 0.66
Ash% 90.6 88.7
Reject
%Sulfur 0.04 0.04
The major difference between the existing POC design and the proposed demonstration
plant is the replacement of sieve-bend with a stationary tank called as ‘phase separator’. The
phase separator allows the newly formed agglomerates to float on the top of dirty aqueous phase,
which will be skimmed from the top of the tank to the vibrating mixer. This modification is
proposed because of reasons listed below:
To eliminate maintenance associated with sieve bend.
The screen allows undersize micro-agglomerate to pass to the tailings tank, while the
phase separator will transfer all coal mass floating on the top to the vibrating mixer tank,
thus minimizing the carbon and pentane losses.
167 |
Virginia Tech | 6.6.1 Thermal Energy Requirements
The newly drafted 1 metric ton HHS process flowsheet indicates that a high amount of
energy will be required for evaporation and condensation of pentane. This high value is because
the hydrocarbon thickener underflow stream was only 15% solids by weight during the pilot test
#4. At this low value, the pentane load to the dryer would be very high. It is important to note
that the energy requirement can be reduced by increasing the solids concentration in thickener
underflow stream using an appropriate solid-liquid separation device, such as a solid-bowl
centrifuge. Fortunately, during pilot test #7, the thickener underflow concentration was observed
to be as high as 45.3% from the same static thickener. This higher value would substantially
reduce pentane losses. Due to high discrepancies in percent solids in the thickener underflow
stream, the process flowsheet was simulated at various solids concentration in the thickener
underflow stream to predict energy requirements (Kwh/ton) and associated cost ($/ton) for
pentane loads to the dryer (Figure 6.14). This analysis provides critical information for the
design the heat exchanging system for the 1 metric ton demonstration plant.
Note:- Cost estimation is based on natural gas price $5.50/MMBTU
Figure 6.14 Prediction of thermal energy requirement for dryer and associated cost relative
to thickener underflow solids
169 |
Virginia Tech | 6.7 Projected Economics of the HHS Process
Prior to economic analysis of any process for coal, it is necessary to address the market
value of this commodity. Typically the payments in the coal market are reported on a per
MMBTU basis, which is the heat content of the as-received product. The following assessment
is made with the current electric power generation fuel cost, which is $2.36 per MM BTU (EIA,
2014), and $5.84 per MM BTU or $135 per ton (EIA, 2012) for premium coking coal.
Appalachian mined coals are typically high-rank bituminous type with an average heat
value of 15,000 BTU/lb (dry ash-free basis). The existing preparation plants in this region
produce coal for either the coking and steam coal markets, depending on the seam
characteristics. However, the very fine coal, due to its high moisture content, is often discarded.
The newly developed HHS process can produce premium quality coal product from these
discarded ultrafine streams. This premium quality product can be blended for either of the
abovementioned markets, thus providing revenue generation for coal producers.
6.7.1 Revenue Assessment from POC Data
As an example, in the existing Lone Mountain facility, the discarded raw coal deslime
cyclone overflow is 74 tons solids/hour (Lone Mountain, 2013). With an ash content of 60.5%
(dry basis), as recorded in pilot test #4, 29.23 tons/hour (74 x 0.395 = 29.23) of combustible
material is lost in this stream. The HHS process POC pilot scale data demonstrated that 84.4%
carbon could be recovered with 2.7% ash and 12.5% moisture. The heat content of the as-
received product is 12,333 BTU/lb (15,000 BTU/lb x 0.975 x 0.875 = 12333 BTU/lb). The total
carbon produced per hour is 54,373 pounds (29.23 ton/hour x 0.844 x 2204 lb/ton = 54,373
lb/hr). Therefore, the total heat value recovered per hour is 671 MM BTU (12,333 BTU/lb x
55,606 lb/hour = 671 MM BTU/hour). If this high quality product is sold to the coking coal
market, the annual gross revenue generation for the coal producer will be $23.5 million (29.23
ton/hour x $135/ton x 6000 hour/year = $23.5 million/year). Similarly, if the HHS process
product blended with utility coal, the annual gross revenue generation will be $9.5 million at the
current market price (671 MM BTU/hour x $2.36/MMBTU x 6000 hour/year = $9.5
million/year).
A similar assessment can be conducted for the low-ash screenbowl main effluent stream.
POC pilot test conducted with Buchanan samples showed as high as 97.1% combustible recovery
170 |
Virginia Tech | with product ash 2.1% and moisture 3.8%. The facility is discarding this stream at a flow rate of
450 gallons/minute (Meenan, 2014). At 3.8% solids and 7% feed ash, total combustible loss is
8,110 lb/hour (450 gallon/min x 8.5 lb/gallon x 0.038 x 0.93 x 60 min/hour). The heat value of
the HHS process as-received product will be 14,123 BTU/lb (15000 BTU/lb x 0.979 x 0.962 =
14123). Therefore, total heat content generated per hour is 115 MM BTU (8,110 lb/hour x
14,123 BTU/lb) and the producer will gain $3.3 million/year (4.1 ton/hour x $135/ton x 6000
hour/year = $3.3 million/year) gross revenue, if the product is sold to the coking coal market.
Similarly, if the product is blended with utility coal, the annual gross revenue generation will be
$1.6 million (115 MM BTU/hour x $2.36/MMBTU x 6000 hour/year = $1.6 million/year).
6.7.2 Generalized Revenue Model
A generalized revenue model is developed to identify the minimum size of production
plant profitable to the coal producers. At present, a detailed economic analysis cannot be possible
due to lack of cost data; therefore, considerable assumptions are made into the model.
The annual gross return can be calculated as a function of clean coal tons produced. It is
assumed that the HHS product coal has specifications 10% moisture and 3% ash. Therefore, the
as-received heat value of the product will be 15000 x (1 – 0.1) x (1 – 0.03), which is 13095 MM
BTU/ lb. For 4800 operating hours in a year (300 days, 2 shifts of 8 hours each), the gross return
from an ‘X’ ton/hour plant will be:
( ) ( ) [6.17]
Where, V is the coal value in $/MM BTU, which varies with the type of market.
The investment and the front cost are divided in three parts: capital cost, operating and
maintenance cost, and personnel cost. These costs are discussed individually in the following
sub-sections.
Capital Cost
Capital cost may be sub-divided into two portions: fixed cost and ramp-up cost. The fixed
capital cost is required to construct the plant and procure its ancillary units, whereas the latter is
necessary to bring the plant into full production. Let’s assume, the fixed capital cost for 1 metric-
ton plant is ‘C’. Usually, the installation cost is approximately 10% of the fixed capital cost
i
171 |
Virginia Tech | (Osborne, 1988). Therefore, the total capital cost for a 1 metric-ton plant is ‘1.1 C’. To predict
i
capital cost for ‘X’ ton/hour plant, the sixth-tenths rule is applied (Osborne, 1988). Furthermore,
this capital cost will be distributed over the plant life (assumed here 10 years) resulting, in this
case, in a depreciation rate 10%. A straight-line depreciation is assumed with no salvage value.
Therefore, annual capital cost for ‘X’ ton/hour plant can be simply described as illustrated in
Equation 6.18.
( ) ( ) [6.18]
Power, Operating Supplies and Maintenance Cost
Three major operating and maintenance costs are considered: electrical costs, a regent
costs, and annual maintenance cost. Annual power costs are usually determined by (Osborne,
1988):
Cost ($/year) = Connected power (KW) x hours/year x Equipment-utilization factor x $/KW-hr
For 1 metric-ton plant, connected power is assumed to be 300KW and equipment-utilization
factor is 0.75. The current price for electricity in the United States is $0.12 per kilowatt-hour. For
‘X’ ton/hour plant, the sixth-tenths rule is applied to predict the electrical cost for large
production plants. Therefore, annual electrical consumption cost is:
( ) ( ) = ( ) [6.19]
Reagent cost is distributed in two portions: fixed cost and reagent-loss cost. Considering the
reagent is recovered and recycled, the fixed reagent cost is required at the start of production.
The annual fixed reagent cost is divided over the plant life, assuming straight-line depreciation.
Any reagent loss during the production will be needed to compensate by procuring new reagent,
which is the reagent-loss cost. The 1 metric-ton/hour flowsheet, illustrated in Figure 6.13, shows
approximately 500 gallon of reagent will be needed for this capacity. It is assumed that the
regent requirement will linearly escalate with increase in tons/hour capacity. The approximate
bulk cost for the reagent is $1.00 per pound. Furthermore, a conservative percentage of reagent
loss is assumed, which is 0.5% (11.02 pounds/metric-ton) of clean coal produced. Therefore, the
annual reagent cost for ‘X’ ton/hour plant operated for 4800 hours/year can be described from the
following:
172 |
Virginia Tech | ( ) [6.20]
With no data available for maintenance cost, it is assumed to be 15% of the total annual capital
cost for the production unit. Therefore, from the above analysis, the total annual O&M cost for
‘X’ ton/hour plant is:
( ) ( ) ( ) [6.21]
Personnel Cost (Labor Cost)
Typically in a coal production plant requires a manager, a plant supervisor and operators.
For small production units (<120 tons/hours), one manager and one supervisor would be enough
for monitoring the plant (or even multiple plants); however, number of operators may vary with
the production scale as well as amount of instrumentation implemented in the plant. It is a
discreet function with the production scale, because one person/shift can operate a 5 ton/hour
unit as well as 25 ton/hour unit. It is assumed that one operator/shift will be added after every 25
ton/hour module in a fully instrumented plant. This is described in detail in Table 6.5.
The average annual base salary for one person is assumed $80,000. With 50% benefit, the
average cost per person is $120,000. Therefore, the annual labor cost can simply be calculated:
( ) ( ) [6.22]
From the above economic analysis, the annual net revenue can be determined by subtracting
Equation 6.17 from Equation 6.18, Equation 6.21 and Equation 6.22, which will be shared by
involved parties.
Furthermore, to identify a profitable production unit in the current market situation, the
return on investment (ROI) was determined before taxes. The return on investment can be
defined as a measure of revenue generated from the process relative to the total amount of
investment required to produce that gross-revenue. Therefore, from the above economic analysis:
173 |
Virginia Tech | d/D (impeller diameter/tank diameter) ratio. Therefore, this reactor geometric ratio must be
considered in designing larger scale units.
A kinetic model developed for vibrating mixer shows that the dispersion mechanism
follows a first-order rate. The model can be used to determine solid concentrations with time for
the process streams leaving the unit. In addition, settling data for ultrafine coal particles in pure
pentane assisted in developing a dual model for the hydrophobic liquid thickener. The model can
predict the required unit area and thickener underflow solid concentrations needed to design the
thickener. A pentane loss model, which was based on bench-scale test studies, was also
developed to predict pentane consumption at a given residence time and temperature. However,
it is highly recommended that dryer manufacturers be consulted for pentane loss estimation.
Finally, a revenue model was developed based on current coal values and projected initial
costs. As anticipated, the model predicted substantial profits for large production plants. In
addition, due to increasing personnel cost per ton of coal produced, it is recommended that any
full-scale HHS production units be fully automated to minimize this cost.
References
1. Coe, H.S. and Clevenger, G.H. (1916), “Methods for Determining the Capacities of
Slime Thickening Tanks”, Transaction of AIME, Volume 55, Pages 356-384.
2. Dahlstrom, D.A. (2003), “Liquid-Solid Separation”, Principles of Mineral Processing,
edited by Fuerstenau, M. and Han, K., Page 338, Society for Mining, Metallurgy, and
Exploration.
3. EIA (2014), “Short-Term Energy Outlook”, February Monthly Report, Page 8, U.S.
Energy Information Administration.
4. EIA (2012), “State Energy Price and Expenditure Estimates 1970 through 2010”, U.S.
Energy Information Administration.
5. Kynch, G.J. (1952), “A Theory of Sedimentation”, Transactions of the Faraday Society,
Volume 48, Pages 166-176.
6. Levenspiel, O. (1999), Chemical Reaction Engineering, Third Edition, John Wiley and
Sons, Inc.
7. Lone Mountain (2013), Lone Mountain Preparation Plant Officials [Personnel
Communication]
8. Meenan, G. (2014), [Personnel Communication]
9. Osborne, D.G. (1988), “Economics of Coal Preparation”, Coal Preparation Technology –
Volume Two, Pages 869-899, Graham and Trotman.
175 |
Virginia Tech | CHAPTER 7 Conclusions and Recommendations
7.1 Summary of the Research
This research involves the development of an innovative technology called the
Hydrophobic-Hydrophilic Separation (HHS) process. The process can simultaneously recover
and dewater the ultrafine (44 microns x 0) coal fraction. Existing preparation plants currently
discard this size fraction, because there are no commercial technologies that can provide high
recovery and low moisture for ultrafine coal. Although the particles in ultrafine streams are well
liberated and can produce a low-ash product, they carry high surface moisture for the obvious
reason: the smaller the particle size, the higher the surface area.
The foundation of the innovative technology is the novel thermodynamic concept of
dewatering-by-displacement (DBD). The DBD concept is based on a naturally occurring
phenomenon, which makes the process thermodynamically favorable. Using this concept,
surface moisture on a hydrophobic particle can be reduced to a level that can only otherwise be
achieved by thermal dryers. In the current research work, the HHS process has been successfully
demonstrated on a bench-scale system as well as on a proof-of-concept (POC) pilot-scale plant.
In the initial phase of the research, several HHS-process bench-scale systems were
constructed to achieve drying and cleaning of ultrafine particles. The gist of the innovative
technology is the effective dispersion of agglomerates in liquid pentane. For this purpose, a low-
energy mechanical vibrating mesh device was developed. The device can be scaled-up as
compared to the initially used ultrasonic probe. Several batch-scale tests conducted with various
coal feedstocks using the innovative mechanical vibrating mesh device have shown that
consistent low-moisture (0.7 – 10%) and high combustibles recovery (>85%) can be achieved
from the HHS process. The successful testing with a bench-scale vibrating mesh device provided
the needed results to develop the technology on a pilot-scale. Alternative methods were also
tested for dispersion of agglomerates. These methods can also be engineered to produce a similar
quality clean coal product as achieved using the vibrating mesh device.
To understand the separation mechanism in the HHS process, several fundamental studies
were conducted. The thermodynamic studies exhibit that the novel dispersion step is a complex
process. It has been identified that the performance of the process is governed by three
mechanisms: breakage of coal-agglomerates, suspension of particles in pentane column, and
176 |
Virginia Tech | coalescence of released water droplets. The thermodynamic study also indicated that the process
is efficient in a specific range of external energy provided to the system for dispersion. At a very
low energy level the agglomerates will not disperse effectively in the pentane column. In
contrast, providing excessive energy in the system may hinder the water coalescence mechanism.
Additionally, it was observed during experimentation that an excessive energy sometimes led to
the formation of stable micro-emulsions in the reactor, which resulted in high product moisture
from HHS process. To determine breakage and dispersion rate, kinetic studies were conducted.
The batch-scale kinetic studies for the vibrating mixer revealed that the breaking of agglomerates
in pentane column with a vibrating mesh device is a slow process. Furthermore, the kinetics of
novel dispersion step can be described as a first-order rate process.
A proof-of-concept (POC) pilot-scale plant with a rated feed capacity of 100 pound/hour
of raw dry feed for the HHS process was constructed at the Virginia Tech Mining and Minerals
Research Laboratory. The primary intent was to demonstrate the capabilities of the HHS process
on a larger scale. The POC units were designed conservatively to provide flexibility in testing
various coal feedstock and to accommodate unknown factors that might hinder the testing
program. Several safety features were implemented in the newly constructed plant under
constant monitoring of multiple regulatory agencies. Necessary modifications were made during
the shakedown testing to rectify any operational issues prior to run the pilot-scale test program.
The POC pilot-scale test results successfully demonstrated that low-moisture and high
separation efficiency can be obtained consistently from the discarded ultrafine (44 microns x 0)
streams by using the HHS process. Testing with low ash feed samples, such as screen bowl main
effluent, moisture in the range of 4 – 16% and carbon recovery as high as 97% were achieved.
Similar results were obtained when high-ash feed samples, such as deslime cyclone overflow,
were tested in the POC system. Here, moisture was reported in the range of 5 – 14% with
combustible recovery as high as 86%. The performance measuring index, i.e. separation
efficiency (E), obtained from the HHS process POC plant was reported as high as 83.4%.
Furthermore, the separation efficiencies are in good agreement with the results obtained with
HHS process bench-scale system.
During the pilot-scale test program several parameters were evaluated. These studies
indicate that the HHS process is robust. The technology is able to process raw coal samples from
various sources with wide range of feed ash. Additionally, POC testing shows that the
177 |
Virginia Tech | technology can also process chemically treated ultrafine streams as-received from a preparation
plant. The findings from the investigations are listed below:
The POC pilot-scale test data indicate that the cleaning capability of the HHS process is
independent of feed ash. A 7% feed ash sample produced 2.1% ash product while a
77.3% feed ash sample produced 2.7% ash product.
The performance of the process depends on the particle size. This is because smaller
particles are well liberated and the surface forces are more prominent in ultrafine size
fraction; therefore, separation is higher for the fraction. The finer the particles the better
the separation.
Efforts were made to reduce the volume of pentane required in the system by replacing
the high volume vibrating mixer unit (120 gallons) with a very low volume unit (19
gallons). The comparative study indicates that there is no significant effect on the product
quality by using a small volume reactor.
To determine the effect of residual frother in the screen bowl main effluent streams, POC
performance was evaluated with a feed dosed with excessive frother (30 ppm). The
results from the investigation showed no detrimental effect on the performance of the
HHS process POC pilot-plant.
Finally, the critical economic factor, i.e. the loss of pentane associated with clean coal
product, was analyzed. The loss was determined using the gas chromatographic (GC)
technique. The analysis showed that at least 3.6 pounds of pentane per ton of clean coal
(0.18%) were lost with the HHS process POC product. This estimated loss is in support
with the results obtained from the batch-scale pentane absorption rate study (3.75 lb/ton
after 30 minutes), as well as with the theoretical analysis (3.97 lb/ton). Although the
detail pentane recovery study can only be conducted at a production scale, preliminary
results obtained from the three different methods predict a pentane loss of less than 4
pounds per ton of HHS product.
To identify scaling-up criteria for each unit operation involved in the HHS process,
several batch-scale investigations were conducted. Using the experimental data obtained from
the investigations, empirical models and simulators were developed. Following are the findings
that can be significant in designing the next-scale HHS process prototype plant:
178 |
Virginia Tech | The batch-scale agglomeration study revealed that for high/low shear mixer, the tip speed
is one of the primary scale-up criteria that must be considered. High reject ash is
achievable using high impeller speed with short residence time and high d/D (impeller
diameter/ tank diameter) ratio.
For the vibrating mixer, a rate model can very well define the dispersion mechanism.
Settling rate investigation of ultrafine coal in pentane exhibited a high settling velocity of
the particles before they enter into a compression zone. The study was conducted with
variable solid concentration in the feed. The results indicate that 28-45% solid
concentration in thickener underflow can be achieved in less than 4 minutes retention
time. Based on the data a dual simulator/model on LIMN® was developed. The model
can predict underflow concentration at a given thickener geometry and residence time.
The model can also identify the required unit area of the thickener to achieve desired
underflow concentration.
A 1 metric-ton per hour clean coal process flowsheet was drafted on LIMN® using the
data and information obtained in this research. The flowsheet was further simulated at
variable thickener underflow solid concentration. The simulations provide information
that can assist in identifying the heat-exchanging system for the next-stage HHS process
prototype plant.
Finally, an economic assessment of the HHS process was conducted and a revenue model
was developed. At the current coal market value the model showed that a 5 ton/hour clean coal
production unit will be profitable to process coking coal while at least a 15 ton/hour unit will be
required to process utility coal to gain revenue. Moreover, two conclusions can be made from the
economic assessment:
The process shows promise to be very profitable for the producers, if implemented in a
high tons/hour processing facility.
The economic assessment was conducted considering the plant will be fully
instrumented. The additional high labor cost may decrease the revenue generation
significantly.
The HHS process is a transformative technology. The novel process produces a premium
quality coal product from currently discarded ultrafine streams. The high quality coal will
provide an incremental value to the coal producers at no additional mining cost. A successful
179 |
Virginia Tech | commercialization of the HHS process will be instrumental for the industry, because with the
innovative technology all the ultrafine coal from run-of-mine feed can be recovered at low cost.
In addition to revenue generation, environmental issues, such as blackwater pollution, can be
reduced significantly.
7.2 Future Recommendations
In view of the results and conclusions of the experimental work included in this
dissertation, some recommendations regarding potential future investigation are provided below:
The newly engineered POC pilot plant has been successful in achieving the goal of low-
moisture product with high carbon recovery. The batch-scale investigation for
agglomeration showed that shear effect (impeller tip speed) is a primary parameter to
obtaining high reject ash. The shearing effect can also be achieved by using In-Line
mixers, which have very low footprint and no moving parts. Therefore, it is
recommended to study In-Line or dynamic mixers on POC pilot plant for agglomeration
of ultrafine coal slurry. A successful operation will reduce the footprint of production
units as well as reduce the operating and maintenance cost associated with this unit
operation.
Throughout the research, attempts were made to determine the exact amount of moisture
and pentane in the HHS product. Due to lack of any established method, the pentane loss
can only be estimated/ predicted on the basis of several assumptions. In light of this, it is
highly recommended to establish a method for an accurate measurement of pentane loss
associated with HHS clean coal product. This is a critical parameter in defining the
economics of the novel process.
Despite a large number of batch scale tests conducted on the newly developed vibrating
mesh device to identify the parameters, a detailed parametric study is suggested. To gain
further knowledge and improve the design for the novel dispersion device, a CFD study
is also recommended. These detailed investigations can be significant for the successful
development of the HHS process production plants in future.
The vibrating mesh design has been very successful for effective dispersion of
agglomerates, which is the requirement in the process. Due to the oscillating structure
and novel design, it may be expensive to fabricate such a device on a larger scale.
180 |
Virginia Tech | Therefore, it is recommended to test alternative existing methods that can provide similar
coal quality. Some examples of these alternatives, which are well established in
industries, are air-pulsated jigging mechanism, dispersion with Rushton-type impellers at
low speed, vibrating trays, etc.
The POC pilot plant described in the research has been tested with in-plant discarded
streams. In 2002, report published by National Research Council suggests that there is at
least 2 billion tons of fine coal lying on abandoned coal impoundments. Recovering coal
from these ponds will be beneficial for the environment as well as for the producers. It is
recommended to test various pond tailing samples on a POC plant to demonstrate that the
proposed method can be useful in recovering carbon from these abandoned resources.
The flotation feed samples tested on the bench-scale system showed very encouraging
results. To further corroborate this, it is recommended to conduct detailed investigations
on various flotation feed samples on the POC pilot plant, to demonstrate that the HHS
process can be beneficial for fine coal feed (150 x 44 microns).
The unique concept of hydrophobic displacement exploits the high affinity of the
hydrophobic particles in water with a hydrophobic liquid for separation. Coal is a low-
price commodity. The HHS process may be used to recover ultrafine particles in other
mineral industries that are currently discarded due to limited technology. For example,
the common practice in phosphate processing is to discard fine particles (below 100
microns). Similarly, in copper processing, typical discard size is 10-15 microns. The HHS
process can be beneficial to recover ultrafine value product from these discarded streams.
Hence, it is suggested to investigate the HHS process for other minerals.
181 |
Virginia Tech | Nikhil Gupta ©, 2014
Pre-Start Checks
Processing Plant Area
Ensure that the fire extinguishers are readily available around the area.
Ensure the back door to the Research Laboratory is open to provide access to the
process control.
Ensure the gate is open to provide access from the highway to the processing unit
and storage area.
Personnel operating the processing unit or entering the area must wear flame
retardant and/or 100% cotton clothing.
Cells phones a prohibited within 25 feet of the processing unit.
Processing Unit
All pipe hoses are connected properly and securely.
All nitrogen lines are connected properly and securely.
All air supply lines are securely connected.
All ground wires are securely connected –
Hose connections
Tanks
Modular frames
Earth ground
All guards are in place around the moving equipment.
Visually check all shaft seals for wear, damage, and other conditions that may
result in leaks.
Temperature sensors operating properly.
Flow meters operating properly.
All emergency vents are intact and set at 7” of water column pressure.
Drain valves closed on Tank-100 and Tank-200.
Discharge valve closed on Tank-800.
Agitator shafts secure on drives for Tank-100, Tank-200, and Tank-400.
Nitrogen Generator
Nitrogen generator operating properly - check for system faults, alarms.
Operating pressure at nitrogen generator receiver/supply tank at 80-100 psi.
Pressure at POC regulator set at 20 psi for the nitrogen purge regulators to purge
complete process.
183 |
Virginia Tech | Nikhil Gupta ©, 2014
PROCESS START UP PROCEDURE
1) Ancillary Equipment
Turn the power switch on for the processing area ventilation fan.
Turn circuit breakers ON for water heater and chiller.
Water chiller:
Turn power switch ON for chiller.
Check temperature control, set at 35 degrees F.
Check for proper water flow to condenser.
Water heater:
Turn power switch ON for water heater.
Check temperature control, set at 140 degrees F.
Turn power switch ON for circulating pump.
Check for proper water flow to dryer.
Compressed Air Supply
Air supply pressure properly set at 80 psi at the POC manifold.
Set agitators at speed (RPM) required for processing conditions.
Set pumps at flow rate (GPM) required for processing conditions.
Set dryer screw at speed (RPM) required for processing conditions.
Shut off the air supply to agitator drives and pump drives.
Monitor pentane and oxygen levels on the PLC control system.
2) Charging Processing Unit
Check pentane and oxygen levels to ensure nitrogen gas has purged the entire
system.
Charge Tank-400 and Tank-800 to required level with fresh water.
Move two barrels of pentane liquid from the storage cabinets to POC plant.
Connect ground wire and charging hose and fittings to the pentane barrel.
OPEN pentane charging valve, CLOSE Tank-400 discharge valves.
OPEN air valve to operate Pump-900.
Pump pentane to charge Tank-500 to the overflow weir.
Continue pumping until the overflow of T-500 charges pentane in Tank-400 to the
required level.
CLOSE air valve to Pump-900.
CLOSE the pentane charging valve.
Remove the charging hose and fittings from the barrel.
185 |
Virginia Tech | Nikhil Gupta ©, 2014
POC – OPERATING PROCEDURE
1) Set Process Feed and Reagent Rates, Start Agitators
OPEN air valve to start Pump-100 for mineral slurry feed (mineral/water mixture) to
Tank-100 and set at required GPM.
OPEN air valve to start Pump-200 for pentane reagent feed to Tank-100 and set at
required GPM.
OPEN air valve to start Agitator-100 (on T-100) and set at required speed.
When mineral slurry/pentane mixture begins overflowing to Tank-200, OPEN air
valve to start Agitator-200 (on T-200) and set at required speed.
Monitor Screen Box (F-300) through sight glass windows for mineral slurry/pentane
mixture flow from T-200.
Monitor the pentane vapors and oxygen levels on PLC control system.
2) Balance Flows through Process
As slurry levels approach proper operating levels in Tank-400 and Tank-800:
OPEN air valve to start Pump-800 and set at required speed.
OPEN air valve to start Agitator on Tank-400 and set at frequency.
OPEN air valve to start Pump-400 and set at required speed.
OPEN air valve to start Pump-800 and set at required speed.
OPEN air valve to start Pump-900 and set at required speed.
Monitor Tank-400 level and maintain level by controlling Pump-400 flow to Tank-
800.
Monitor level Tank-800 and maintain level by controlling elevation of tails
discharge hose.
Allow sufficient time for product to settle in Tank-500, then OPEN air valve to start
Pump-600 and set at required speed.
OPEN air valve to start Dryer Screws (D-700) and set at required speed.
Allow sufficient time for dried product to begin discharging from Dryer (D-700),
then activate timed cycling of the product discharge air-locking valves.
Collect the dried mineral product in the collecting pan at the Dryer (D-700)
discharge port.
Monitor the pentane vapors and oxygen levels on the PLC control system.
Monitor the liquid level in the condensed pentane barrel (HL Collecting Barrel).
187 |
Virginia Tech | Nikhil Gupta ©, 2014
POC – SHUTDOWN PROCEDURE
1) Clear Material from Processing Unit
CLOSE air valve to stop Pump-100 (mineral slurry feed pump).
CLOSE air valve to stop Pump-200 (pentane reagent feed pump).
CLOSE air valve to stop Pump-800.
Connect fresh water supply to Pump-100 inlet, OPEN air valve to start pump and
flush inlet, pump, and discharge tubing completely with water.
Remove the pentane reagent feed hose and fittings from the barrel.
Disconnect the ground wire from the pentane barrel.
Move the pentane barrel to the storage cabinet.
Connect fresh water supply to Pump-200 inlet, OPEN air valve to start pump and
flush inlet, pump, and discharge tubing completely with water.
OPEN the drain valve slightly on Tank-800 to start transferring residual tails to tails
collecting barrel.
Continue operating Pump-100 to completely flush all residual pentane and mineral
solids from Tank-100, Tank-200, Screen Box (F-300), and Tank-800.
CLOSE the air valves to stop Pump-100, Pump-200, Agitator-100, and Agitator-200.
OPEN the drain valves on Tank-100 and Tank-200 to drain the water into an empty
barrel.
Move two (empty) barrels of from the storage cabinets to POC plant.
Connect ground wire and transfer hose and fittings to the pentane barrel.
OPEN pentane charging valve.
OPEN air valve to start Pump-900.
OPEN Tank-400 discharge valves to completely remove any residual pentane.
CLOSE air valve to stop Pump-900.
CLOSE the pentane charging valve.
Remove the charging hose and fittings from the barrel.
CLOSE air valve to stop Pump-600.
Connect transfer hose and fittings from Pump-600 to the pentane barrel.
OPEN air valve to start Pump-600.
Completely remove any residual pentane from Tank-500.
CLOSE air valve to stop Pump-600.
Remove the charging hose and fittings from the barrel.
Disconnect the ground wire from the barrel.
188 |
Virginia Tech | Nikhil Gupta ©, 2014
Move the pentane barrels back to the storage cabinets.
Connect fresh water supply to Tank-400, completely fill tank with fresh water.
OPEN air valve to start Pump-900, fill Tank-500 to overflowing.
OPEN air valve to start Pump-400, transfer rinse water to Tank-800.
Continue to add fresh water to Tank-400, pumping to Tank-500, and transferring to
Tank-800 until mineral solids are cleared from tanks.
CLOSE fresh water valve to stop flow to Tank-400.
OPEN air valve to start Pump-600.
Pump residual rinse water from Tank-500 to a barrel, then CLOSE air valve to stop
Pump-600.
Operate Pump-400 to remove the water from Tank-400, then CLOSE air valve to
stop Pump-400.
Monitor the condensed liquid tubing at the condenser liquid discharge port to
determine when all the pentane from the system has been transferred to the pentane
barrel (HL Collecting Barrel).
Remove the hose and fittings from the barrel.
Disconnect the ground wire from the barrel.
Move the pentane barrel back to the storage cabinets.
Operate the air-lock valve system on the Dryer (D-700) until all the dried mineral
has been discharged from the Dryer.
CLOSE the air valve to stop the Dryer Screws.
Deactivate the automatic timed cycling of the Dryer discharge valves, then CLOSE
both valves.
OPEN the drain valve on Tank-800, drain all the water to a barrel to collect all the
tails from the test session.
Monitor the pentane vapors and oxygen levels on the PLC control system.
2) Ancillary Equipment
Water chiller:
Turn power switch OFF for chiller.
Water heater:
Turn power switch OFF for circulating pump.
Turn power switch OFF for water heater.
Compressed Air Supply
Shut off the air supply to POC processing unit.
Maintain the nitrogen blanket pressure on the whole POC system.
189 |
Virginia Tech | Nikhil Gupta ©, 2014
EMERGENCY SHUTDOWN PROCEDURE
Emergency shutdown procedure must be followed in the following events:
1) Any leaks in the hoses, pipe fittings, nitrogen lines, or air supply lines.
2) If the pentane sensor indicates pentane vapor levels approaching the lower explosive limit
(LEL).
3) If the oxygen sensor indicates sufficient levels to create a potential explosive atmosphere.
4) Faults or failures with any mechanical moving devices (Agitators, Dryer Screws, Pumps)
5) Faults or failure of the nitrogen generator unit.
6) Electrical power outage.
7) Spill of pentane at the POC test area.
8) Any personal injury.
POC – Emergency Shutdown Procedures:
CLOSE compressed air supply valve to stop all Pumps, Agitators, and Dryer Screw.
CLOSE any manual valve required to stop, prevent, and/or isolate a spill condition.
Institute the “Emergency Spill Handling Procedure”.
Monitor the pentane vapors and oxygen levels on the PLC control system.
Provide additional ventilation in the area of the spill if a spill occurs.
As soon as safely possible, remove the pentane from the processing unit and clear the
processing unit per the procedures under “POC – SHUTDOWN PROCEDURE” above.
Additional Requirements:
Case 5 – Nitrogen Generator Malfunction
Monitor the pressure in the nitrogen generator supply tank.
Connect a backup nitrogen bottle to the nitrogen supply valve at the nitrogen
generator.
Proceed with removal of the pentane from the processing unit and clear the
processing unit per the procedures under “POC – SHUTDOWN PROCEDURE”
above.
190 |
Virginia Tech | Nikhil Gupta ©, 2014
Emergency Spill Handling Procedure
In the event of a minor spill of pentane (less than 1 gallon) and mineral slurry:
Follow the Emergency Shutdown procedures to stop the POC processing.
Use spark-proof tools and intrinsically safe equipment when responding to any spill
involving pentane.
Follow the hazard information from the MSDS for the pentane:
Keep sources of ignition and hot metal surfaces isolated from the spill to a
distance of at least 30’
Use safety glasses, chemical goggles, or face shields
Use impermeable gloves to avoid pentane liquid from irritating the skin
Utilize the emergency spill kit to contain the liquids within the area of the POC plant and
recover as much of the pentane as possible. Pentane must be absorbed and containerized
as quickly as possible to minimize vapor generation.
Ensure continuous ventilation with fresh air in the area of the spill. Do not direct
ventilation air towards any potential source of ignition.
If possible, isolate the pentane for recovery –
Pump as much of the pentane as possible into a storage container
Place the storage container in the storage cabinet
Use absorbent material from the spill kit to absorb residual amounts of liquid
Pump the mineral slurry into storage containers with open tops.
Sweep and clean the floor with the fresh water.
Collect the dirty water in open top containers.
Test the slurry and dirty water containers for residual pentane vapors.
If pentane vapors are present at the top of the storage containers, place the containers in
an open well-ventilated area to dissipate the vapors.
Operate the ventilation fan located near the test site to remove any residual vapors from
the POC plant area.
Keep fire extinguishers readily available.
When the pentane vapors has dissipated from the slurry and dirty water containers, move
the containers to the research lab and follow the normal waste disposal procedures.
192 |
Virginia Tech | Nikhil Gupta ©, 2014
In the event of a major spill of pentane (more than 1 gallon) and mineral slurry:
Follow the Emergency Shutdown procedures to stop the POC processing unit.
As possible, spread absorbent pads across the spill to minimize vapor generation.
Immediately evacuate the POC test site area.
Call the Virginia Tech Environmental Health and Safety (after normal office hours, call
the Virginia Tech Police Department):
Specify large spill of pentane, estimate amount of spill
Specify exact location of spill (800 Plantation Road, storage shed behind
laboratory)
Follow all instructions from the Environmental Health and Safety or Police
Department
Prevent entry into the area of the spill until the emergency response team arrive
and take charge.
Designate one person to immediately go to the driveway entrance on Plantation
Road to inform the emergency response team of the emergency situation and
actions instituted.
Operate the ventilation fan located near the test site to remove as much vapor as possible
from the POC plant area.
Keep fire extinguishers readily available.
Assist the emergency response team as instructed.
When the emergency response team has cleared the area, proceed to clean up the mineral
slurry from the area.
Pump the mineral slurry into storage containers with open tops.
Sweep and clean the floor with the fresh water.
Collect the dirty water in open top containers.
Test the slurry and dirty water containers for residual pentane vapors.
If pentane vapors are present at the top of the storage containers, place the containers in
an open well-ventilated area to dissipate the vapors.
Operate the ventilation fan located near the test site to remove any residual vapors from
the POC plant area.
When the pentane vapors has dissipated from the slurry and dirty water containers, move
the containers to the research lab and follow the normal waste disposal procedures.
193 |
Virginia Tech | Nikhil Gupta ©, 2014
APPENDIX B. POC PILOT PLANT DESIGNS
B.1 Detailed POC Tanks Design
Table B.1 Data for development of POC mixing devices
σ = 0.0519 N/m σ = 0.0519 N/m
D (Batch)= 1.5 inches D (Batch)= 1.5 inches
0.0378 0.0378
D (POC)= 4 inches D (POC)= 5 inches
0.1008 0.126
ρ = 1026
Kg/m3
ρ = 1026
Kg/m3
Hi-Shear Mixing Low-Shear Mixing
Batch POC Unit Batch POC Unit
N (RPM) Weber No. N (RPM) Weber No. N (RPM) Reynold No. N (RPM) Reynold No.
2000 1186 100 56 2000 46268 200 51409
4000 4744 300 506 4000 92536 400 102818
6000 10675 600 2024 5000 115670 600 154227
8000 18977 1000 5623 6000 138804 800 205636
9000 24018 1300 9503 7000 161939 1000 257045
10000 29652 1500 12651 8000 185073 1200 308454
11000 35879 1800 18218 9000 208207 1400 359863
12000 42699 2000 22492 10000 231341 1600 411272
13000 50112 2300 29745 11000 254475 1800 462681
14000 58118 2500 35143 12000 277609 2000 514091
15000 66717 2700 40991 13000 300743 2200 565500
16000 75909 3000 50606 14000 323877 2400 616909
17000 85694 15000 347011 2600 668318
18000 96072 16000 370145 2800 719727
196 |
Virginia Tech | Nikhil Gupta ©, 2014
POC Spill Containment Evaluation
MRC PROCESS - SPILL IMPACT EVALUATION
EXISTING STRUCTURE: SPILL CONDITION: PROCESS VOLUMES:
Depth 30 ft Footprint Width 40 ft MRC Tanks 302 gal
Width 60 ft Barrier Height 2.05 inch Storage Tanks 220 gal
Offset 3 inch Spill Volume 70.042 ft3 Total Volume 522 gal
Angle 0.48 degrees Spill Volume 523.91 gal
Grade 0.83 % Spill Distance 20.5 ft
Volume 225 ft3 Spill Occur? No
Volume 1683 gal Safety Factor 1.0
Footprint Width (ft) 40 Barrier Height (Inch) 2.05
205
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 30
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 29
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 28 30
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 27
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 26
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 25
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 24
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 23 25
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 22
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE 21
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 20
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 19
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 18 20
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 17
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 16
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 15
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 14
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 13 15
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 12
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 11
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 10
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 9
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 8 10
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 7
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 6
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 5
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 4
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 3 5
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 2
FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE 1
60 55 50 45 40 35 30 25 20 15 10 5
Figure B.12 Spill containment analysis for POC pilot plant test site
209 |
Virginia Tech | Nikhil Gupta ©, 2014
D.4 Economic Analysis – Work Sheet
COAL QUALITY SPECIFICATIONS: OPERATING COST: LABOR COST: 1-20 tph 25-50 tph 55- 75 tph 80- 100 tph
Clean Coal Feed Coal Power used (KW) 300 Shifts 2 2 2 2
DAF As-Received Cost/KWH ($) 0.12 Hours/shifts 8 8 8 8
Ash (%): --- 3.00 60.00 Load Equip Factor 0.75 Manager 1 1 1 1
Heat Value (Btu/lb): 15000 13095 420 Reagent Cost ($/lb) 1.00 Supervisor 1 1 1 1
Moisture (%): --- 10.00 93.00 Reagent Lost (%) 0.50 Operator 1 2 3 4
COAL VALUE: Oprt Hours/year 4800 Avg $/Labor/year 120000 120000 120000 120000
Coking ($/MMBTU) 5.84 Plant Life (years) 10 Total ($/year) 600000 840000 1080000 1320000
Utility ($/MMBTU) 2.36 Fixed regent (gal/tph) 500
Ci (1 TPH) 2000000
HHS PROCESS ECONOMIC ASSESSMENT
Metric tons/hour Gross Revenue ($/Year) Capital Cost O & M Cost Labor Cost Total Cost NET Revenue ($/Year) Net Revenue Ratio
Clean Product Coking Utility $/Year $/Year $/Year $/year Coking Utility Coking Utility
1 809042 326942 220000 512761 600000 1332761 (523719) (1005819) (0.393) (0.755)
3 2427127 980825 425300 1047973 600000 2073274 353853 (1092448) 0.171 (0.527)
5 4045211 1634709 577836 1472957 600000 2650793 1394418 (1016085) 0.526 (0.383)
10 8090422 3269417 875836 2361311 600000 3837146 4253276 (567729) 1.108 (0.148)
15 12135633 4904126 1117062 3131060 600000 4848122 7287511 56003 1.503 0.012
20 16180844 6538834 1327519 3836527 600000 5764046 10416798 774788 1.807 0.134
25 20226055 8173543 1517703 4499644 840000 6857346 13368709 1316196 1.950 0.192
30 24271266 9808251 1693150 5131974 840000 7665124 16606142 2143127 2.166 0.280
35 28316477 11442960 1857221 5740538 840000 8437758 19878719 3005201 2.356 0.356
40 32361688 13077669 2012142 6329988 840000 9182130 23179558 3895538 2.524 0.424
45 36406899 14712377 2159485 6903605 840000 9903090 26503809 4809287 2.676 0.486
50 40452110 16347086 2300407 7463809 840000 10604216 29847894 5742869 2.815 0.542
55 44497321 17981794 2435793 8012447 1080000 11528240 32969082 6453555 2.860 0.560
60 48542532 19616503 2566335 8550968 1080000 12197304 36345229 7419199 2.980 0.608
65 52587743 21251211 2692593 9080536 1080000 12853129 39734614 8398082 3.091 0.653
70 56632954 22885920 2815020 9602103 1080000 13497123 43135831 9388797 3.196 0.696
75 60678165 24520628 2933995 10116458 1080000 14130454 46547712 10390175 3.294 0.735
80 64723376 26155337 3049837 10624267 1320000 14994105 49729272 11161232 3.317 0.744
85 68768587 27790046 3162817 11126097 1320000 15608914 53159674 12181132 3.406 0.780
90 72813798 29424754 3273167 11622434 1320000 16215601 56598197 13209153 3.490 0.815
95 76859009 31059463 3381091 12113701 1320000 16814792 60044217 14244671 3.571 0.847
100 80904220 32694171 3486765 12600269 1320000 17407034 63497186 15287137 3.648 0.878
228 |
Virginia Tech | Permissions
T & F Reference Number: P041812-01
4/18/2012
Michael Kiser
Virginia Tech
We are in receipt of your request to republish Figure 4 on p. 42, Figure 5 on p. 44, and Figure 1 on p. 55
Figure 3 on p 59 and Figure 4 on p. 60, respectively from
“Coal Flotation Washability: An Evaluation of the Traditional Procedures” by M. K. Mohanty et al,
International Journal of Coal Preparation and Utilization Volume 19 Issue 1-2 (1998), pp. 33-49
“Coal Flotation Washability: Development of an Advanced Procedure” by M. K. Mohanty et al,
International Journal of Coal Preparation and Utilization Volume 19 Issue 1-2 (1998), pp. 51-67
in your Master’s thesis for Virginia Tech.
We will be pleased to grant you permission free of charge on the condition that:
This permission is limited to non-exclusive English world print and electronic rights for this usage only.
This permission does not cover any third party copyrighted work which may appear in the material
requested.
Full acknowledgement must be included showing article title, author, and full journal title, copyright,
reprinted by permission of (Taylor & Francis, http://www.tandfonline.com).
Thank you for your interest in our journal.
Sincerely,
Brittany Alderfer
Permissions Coordinator
Telephone: 215.625-8900
E-mail: [email protected]
325 Chestnut Street, Suite 800, Philadelphia, PA 19106 Phone: 215-625-8900 Fax: 215-625-2940
Web: www.informaworld.com |
Virginia Tech | D -D T A
OUBLE IFFERENCE OMOGRAPHY PPLIED TO
M G C
ONITORING OF EOLOGIC ARBON
S A O F , U
EQUESTRATION IN THE NETH IL IELD TAH
Brent Slaker
A
BSTRACT
Double-difference seismic tomography is performed on a carbon sequestration operation in the
Aneth Oil Field in southeast Utah as part of a Department of Energy initiative on monitoring,
verification, and accounting of sequestered CO . A total of 1,211 seismic events were recorded
2
from a borehole array of 22 geophones. Aneth Unit data were divided into four time periods for
time-lapse analysis. A low velocity zone spanning the lateral extents of the observable region,
likely representing a CO plume, is detected when considering voxels containing the highest ray
2
path coverage. A series of synthetic tomography tests simulating different CO plume sizes and
2
locations was performed to assist in characterizing velocity changes associated with Aneth Unit
data. Inferences about the existence of a CO plume should be made by comparing actual data to
2
synthetic data resulting from simulations performed under similar conditions. Considering
synthetic simulation similarities and a derivative weight sum analysis, a CO plume can be
2
imaged within the Desert Creek reservoir, but the resolution of the CO2 plume is too low for
proper monitoring, verification, and accounting of injected CO . Recommendations, for
2
improving CO plume resolution through double difference seismic tomography, are made to
2
increase the ray path distribution throughout the Aneth Unit by varying geophone locations. |
Virginia Tech | Chapter 1: Introduction
Climate change is a significant international concern, often attributed to increasing
concentrations of greenhouse gasses in Earth’s atmosphere, namely Carbon Dioxide (CO ).
2
Fossil fuels are responsible for the majority of anthropogenic CO emissions, but also sustain the
2
global economy. This conflict has created one of the greatest geo-political environmental issues
in recent history. One method for reducing the amount of CO released into the atmosphere is
2
geologic carbon sequestration. Geologic carbon sequestration is the capture of CO before it
2
reaches the atmosphere, and its storage under deep geologic formations, such as depleted oil
reserves, unmineable coal seams, or deep saline formations.
One of the first challenges encountered by the storage of CO is the selection of a reservoir with
2
sufficient storage capacity, appropriate sealing mechanism, and an absence of faults or fractures
through which fluid migration may occur. Post-injection, the engineering challenge shifts to
monitoring, verification, and accounting (MVA) of injected CO . There are three primary
2
objectives of MVA:
- Monitoring - monitor both the location and impact of sequestered CO .
2
- Verifying - verify movement of CO and ensure that sequestered CO is not
2 2
permeating the sealing mechanism, or migrating to an unsealed area.
- Accounting - account for the amount of injected CO by comparing it to the amount
2
of CO estimated to be in place, through the chosen monitoring method.
2
The United States Department of Energy has established seven regional carbon sequestration
partnerships to test the feasibility of geologic carbon sequestration as a means of storing
CO and preventing its release into the atmosphere. The goal of each of these
2
partnerships is to assist in the development of technology, infrastructure, and regulations
necessary to implement large-scale CO sequestration projects in different regions and
2
geologic formations [1].
1 |
Virginia Tech | Seismic tomography is a useful tool for characterizing geologic structures, and is anticipated to
providing accurate method of monitoring, verification, and accounting of injected CO . Similar
2
in principles to the more widely recognized and practiced medical tomography, seismic
tomography utilizes seismic energy generated either passively or actively in rock masses to make
inferences about the state of the medium through which the seismic energy passes. Expanding
further on traditional tomography, double-difference seismic tomography is a method of
improving precision of seismic source locations and accuracy of velocity model reconstruction
by reducing travel path residuals among closely spaced seismic events.
A practical application of double-difference tomography to the monitoring, verification, and
accounting of injected CO is conducted on a carbon sequestration operation in the Aneth oil
2
field. Data for this experiment are provided by Resolute Natural Resources and the National
Energy Technology Laboratory as part of the Southwest Regional Partnership on Carbon
Sequestration. Analysis of event, receiver, and travel time data provided for the Aneth Unit of
the Aneth oil field is performed on a time-lapse basis to determine CO plume extents, CO
2 2
concentration changes, and any leakage of CO from the reservoir. Appropriate damping and
2
smoothing values are selected to optimize the performance of the velocity model reconstruction.
Simulations with synthetic travel times for a variety of different CO plume sizes and locations
2
are also conducted to assist in the verification of results obtained from the Aneth Unit data. This
thesis will demonstrate the degree to which sequestered CO can be monitored, verified, and
2
accounted for in the Aneth Unit through time-lapse analyses of velocity changes and
comparisons of synthetic plume simulations to Aneth Unit data.
2 |
Virginia Tech | Chapter 2: Literature Review
Predictions of global energy use throughout the next century suggest increasing levels of carbon
emissions and rising concentrations of carbon dioxide (CO ) in the atmosphere [1].
2
Concentrations of greenhouse gases have increased since the Industrial Revolution, to the level
that measurable climate change has been attributed by many to human activity. Greenhouse
gasses trap energy from the sun within Earth’s atmosphere, warming the planet. The extent to
which humans increase the concentration of greenhouse gasses in the atmosphere, and the global
warming that occurs as a result, is referred to as the enhanced greenhouse effect. The gas
considered most responsible for the enhanced greenhouse effect is CO , largely as a consequence
2
of fossil fuel combustion for energy production. Currently, about 33 gigatons of CO are
2
produced annually from anthropogenic sources [2]. Fossil fuels provide roughly 85% [3] of the
world’s energy, and are likely to remain a large component of the world’s energy portfolio for
many decades due to their low cost, wide availability, ease of transport, and large reserves.
2.1 Geologic Carbon Sequestration
Geological sequestration of CO is the most likely method to provide the first large scale
2
opportunity for concentrated sequestration of CO because the technology exists and has been
2
used extensively in enhanced recovery projects [4]. Geologic carbon sequestration is a means of
reducing carbon emissions by capturing carbon dioxide from the exhaust of fossil fuel power
plants and other sources and sequestering them within geologic formations. CO can be trapped
2
in these formations as either a gas or a supercritical fluid. It is generally injected in a
supercritical phase at pressures above 69 bars to reduce the injected volume. The costs of
separation and capture of CO ,including compression to sequestration pressures, are estimated to
2
account for approximately 75% of the cost of geologic sequestration [1]. Geologic formations
selected for carbon sequestration should be deeper than 762 meters to ensure that the injected
CO will remain in a supercritical state [43]. While the density of supercritical CO is greater
2 2
than that of gaseous CO , it is still less than the density of other in-situ fluids. As a result, the
2
3 |
Virginia Tech | migration of CO away from the injection well is dependent on gravity and an unfavorable
2
mobility ratio.
The permanence of sequestered CO depends on the effectiveness of the trapping mechanism.
2
To ensure that the injected CO does not buoyantly rise to the surface, an impermeable barrier,
2
known as a caprock, is required to trap the CO within the geologic formation. This is similar to
2
the natural geologic trapping of oil, where oil is unable to permeate an overlying caprock.
Depending on the formation in which the CO is injected, it may eventually dissolve into the
2
ground water. CO may also become trapped in the form of carbonate minerals formed by
2
chemical reactions with the surrounding rock [6].
The three most widely suggested storage sites for large scale geologic carbon sequestration are:
oil and gas reservoirs, saline formations, and unmineable coal seams. Each storage location has
its own advantages and disadvantages. Oil and gas reservoirs have a known caprock, a potential
for fuel recovery to offset the cost of sequestration, and a well characterized volume, due to
extensive exploration. Oil and gas reservoirs also have good access to in-place CO pipelines,
2
however the scarcity of these pipelines is a disadvantage to geologic carbon sequestration
overall. There are several disadvantages to sequestering CO in oil and gas reservoirs. One
2
disadvantage is both the limited number of oil and gas reservoirs, and the storage capacity of
these reservoirs. The U.S. Department of Energy estimates that the total storage capacity of oil
and gas reservoirs in the United States is limited to 140 gigatons of CO [6]. Another
2
disadvantage is the large number of boreholes used for oil and gas development may need to be
remediated in order to prevent possible leakage pathways.
Advantages to deep saline formations include their large storage capacities and wide distribution.
The U.S. Department of Energy estimates the storage potential of deep saline aquifers to be
around 12,600 gigatons of CO in the United States. Due to a lack of economic incentive to
2
explore deep saline formations, they have been very poorly characterized and have the most
geologic uncertainty concerning seal effectiveness and storage capacity. Storage capacity in
saline formations is difficult to determine because multiple methods of storing the CO are
2
4 |
Virginia Tech | available: trapping underneath an impermeable cap rock and dissolution and mineralization of
the CO [6].
2
Storage in unmineable coal seams has the advantage of nearby power plants, offering the
potential for methane recovery to offset the cost of sequestration. The unmineable coal seams,
however, suffer from the same storage capacity issues as oil and gas reservoirs, with capacities
numbering in the tens to hundreds of gigatons of CO . In addition to these problems, unmineable
2
coal is hard to define, as coal seams that are unmineable today may be capable of being mined
years later [6]. Initial sequestration projects are more likely to occur in depleted oil and gas
fields or unmineable coal seams due to the quality of subsurface data and the ability to mitigate
sequestration costs through enhanced oil and gas recovery. However, due to the large storage
potential available, most long-term geologic carbon sequestration will likely occur in deep saline
formations [7].
2.2 Induced Seismicity
Geologic carbon sequestration has the potential for inducing seismic events through fluid
injection. Mining activity, reservoirs, long-term fluid withdrawal wells, and long-term fluid
injection wells have all created events on the microseismic level, but also have the potential to
cause events with large enough magnitudes to damage surface facilities. Seismic activity was
first linked to deep well injection at the Rocky Mountain Arsenal in Denver, Colorado [5]. The
seismic activity was a result of waste water disposal in a deep injection well, drilled in 1961 by
the Army Corps of Engineers. Between 1962 and 1967, over 1,500 seismic events were detected
in the area with magnitudes as high as 5.5 [8]. Induced seismicity is recognized to be a hazard in
nearly any engineering activity that alters the stress or pore-pressure within the earth’s crust.
The earth’s crust is generally supporting very high stress levels and is often close to failure [9].
There are several potential mechanisms of induced seismicity, such as hydraulic fracturing,
slipping resulting from redistribution of elastic stresses, slipping resulting from pore pressure
relaxation, or a combination of these mechanisms. It is hypothesized that the dominant
5 |
Virginia Tech | mechanism of triggering induced seismicity is the diffusive process of pore-pressure relaxation
in porous, saturated rocks. The tectonic stress in the earth’s crust is, at some locations, close to a
critical state that would cause brittle failure of rocks. If fluid pressure in a reservoir is increased,
pore-pressure in the critical locations is also increased. This increase in the pore-pressure causes
the effective normal stress to decrease, allowing for slipping along pre-existing cracks [10].
When injected fluid enters pre-existing microfractures within the rock, part of the normal stress
is supported. Fluid has no shear strength and, as a result, the frictional resistance to sliding is
lowered by an amount equivalent to the pressure of the fluid [11]. Fault stability assessments
and estimates for the maximum sustainable fluid pressure are typically based on the Mohr-
Coulomb failure criterion.
Several conditions, which must be satisfied to induce seismic events by injection of fluids, are
suggested by McClain. First, the most important condition to be satisfied is the presence of
regional tectonic stresses near the breaking strength of the rocks, prior to injection. Second, the
reservoir formation must be porous enough to accept the fluids and simultaneously have a
permeability low enough for pore pressure to increase. Last, the fluid must be injected into the
formation at a high enough rate to significantly increase pressures over a wide area [12].
In determining displacement along faults due to induced slipping, the fault-plane solutions from
the first motions detected by the recorded seismograms should be calculated. This requires a
well calibrated array of seismic monitoring stations [13]. As a means to map reservoir dynamics,
the deployment of geophones, or other acoustic sensors, is promoted to compliment standard
engineering gauges. These monitoring devices can measure microseismic events related to
induced seismic movements along existing fractures or the creation of new fractures.
Microseismic data can be used to determine the location of the fracturing or to infer
geomechanical details of the formation.
Monitoring can take place from either surface arrays or borehole arrays. Borehole deployments
generally allow for closer placement of instrumentation to the seismic events, maximizing
sensitivity. Surface arrays, while easier to deploy, provide a limited event monitoring resolution
and are unable to detect seismic events with travel paths that do not reach the surface. A study at
6 |
Virginia Tech | the Ekofisk field in the North Sea had a maximum travel path recorded for seismic events of 2
kilometers, while the reservoir was 3 kilometers in depth. Without borehole monitoring, the
seismic activity would have gone undetected [14].
2.2.1 Intensity and Frequency of Occurrence
The intensity and frequency of induced seismic events is controlled by a change in fluid injection
rates, the net amount of fluid injected into a formation, the orientation of the stress field relative
to the pore-pressure increase, and the size and extent of the local fault system [15]. Suckale
suggests several reasons for the absence of a simple relationship between the quantity of fluid
injection and seismicity. An absence of a correlation could be explained by: inaccuracies in the
location of hypocenters, a lack of data regarding the precise injection pattern over time, a lack of
high-quality instrumentation during the onset of seismic activity, or the clustering of seismic
activity could obscure the fact that selected events are closely related to injection patterns [11].
The most well documented seismic events induced by fluid injection are associated with water
flood operations for enhanced recovery. These operations often include large arrays of wells
injecting fluids at high pressures into confined reservoirs with low permeability. Waste disposal
wells tend to inject fluids at lower pressures into large, porous aquifers with high permeability,
away from fault structures. As a result, only three waste disposal sites have been conclusively
shown to have induced significant seismic activity: Ashtabula, Ohio, El Dorado, Arkansas, and
Denver, Colorado [8].
Certain regions are more inclined to produce a higher seismic moments and more seismic
activity than others. This inclination depends on the existing stress conditions within the upper
crust. In some regions, elevating formation fluid pressures by only tens of bars can trigger
shallow seismic events. The Great Lakes region of the Appalachian Plateau is a prime example
of a region especially vulnerable to increased fluid pressures. Injection pressures ranging from
60 to 100 bars have triggered earthquakes in northeastern Ohio, western New York, and
southwestern Ontario [8].
7 |
Virginia Tech | Many forms of deep well injection have resulted in significant numbers of induced seismic
events, but CO injection for enhanced oil recovery has a proven history of seismic stability. 30
2
megatons of CO is currently being injected worldwide, annually, for enhanced oil recovery. A
2
cumulative amount of 500 megatons of CO has been injected to date, yet there have been no
2
high magnitude seismic events attributed to enhanced oil recovery using CO injection [16].
2
2.2.2 Seismicity Mitigation
There are several steps that can be taken to reduce the chance of experiencing induced seismic
events or to reduce the intensity of induced seismic activity. The most important step is
choosing sequestration sites that avoid formations or locations prone to fracturing or earthquakes
[17]. Basins located in tectonically active areas, such as locations around the Pacific Ocean and
northern Mediterranean Sea, are more likely to experience seismic events as a result of CO
2
injection. Basins formed in mid-continent regions or on the edges of stable continental plates, are
stable long-term storage locations. Exceptions to this rule include the Los Angeles Basin and the
Sacramento Basin, both of which have demonstrated stable storage capacity, yet lie in
tectonically active regions. General guidelines can be established for selecting an ideal reservoir
location, but care must still be taken to analyze the reservoir on an individual basis, considering
site-specific conditions [16].
Another precaution that can be taken to reduce seismic activity is to set injection rates that keep
pressures below the fracture strength of the surrounding rocks [17]. As discovered in Rangely,
Colorado, changing injection rate correlates far better to seismic activity than net injected
volume. With proper seismic monitoring of the region surrounding the injection well, the
injection rates can be normalized or reduced if seismic activity becomes too extreme. Carefully
selecting an injection rate that will not exceed the critical pressure of the injection site will help
reduce seismic activity and intensity resulting from fluid injection. The scope of a fluid injection
project could also play a significant role in the amount of seismicity experienced. Long-term
fluid injection generally carries a higher risk of inducing high-magnitude seismic events than
8 |
Virginia Tech | short-term injections, since the total volume of rock where the stress field has been altered is
expected to be larger [9].
2.3 Tomography
Tomography is derived from the Greek word tomos, meaning the study of slices. The process of
recreating the internal structure of a body by compiling a series of projections, or slices, is
generally referred to as tomography [18]. Generating an image, or tomogram, of an object by
examining its reaction to the passive, probing energy from an external source forms the
foundation of tomography [19]. The most common application of tomography is in the form of
medical imaging, known as CAT (Computer Assisted Tomography) scans, which create a two-
dimensional image from a series of one-dimensional line projections. Not limited to the most
familiar imaging signal, X-rays, used in CAT scans, the signals used in tomography are
incredibly varied in both type and scale. The most common types of tomographic signals
include: seismic waves, ultrasound, magnetic resonance (MRI), X-rays, gamma rays, neutron
beams, and electron microscopy. These signals can vary in scale from nanometers, during
electron microscope reconstructions of viruses and atomic lattices, up to kilometers for seismic
tomography applications [18]. Seismic tomography may be new by name, but it is
fundamentally very similar to techniques that have been in use for many years, not only in the
medical community but geophysical community as well. One example of a geophysical
application of tomography is the wide range of engineering problems encountered in the oil
industry. Many of these problems, from exploration and development to production can make
use of seismic tomography. Applying tomography to previously existing or newly collected
seismic data can allow for the generation of subsurface velocity models, providing a non-
invasive procedure for assisting in these phases of operation [19]. Global seismology is another
field which has been making use of tomography for many years. Global seismology has,
however, lagged behind seismic exploration in development of accurate tomographic imaging for
several reasons: earthquakes used in global seismology are both uncontrolled in their magnitude
and location, coverage of the Earth in seismometers is limited, and instrument responses have
been inconsistent [20].
9 |
Virginia Tech | 2.3.1 Seismometers
The means by which signals are detected, in seismic tomography, are through seismometers.
Modern seismometers first came into use around the end of the 19th century. Since their
implementation, seismic waves have been used to locate remote objects and earthquake
epicenters. Efforts were even made during World War I to locate heavy artillery batteries by
examining seismic wave propagation [20]. Seismic sensors can be delineated into two main
categories: inertial seismometers and extensometers. Inertial seismometers measure the motion
of the ground relative to an inertial reference. Extensometers measure the movement of a single
point on the ground relative to another point. Extensometers are well suited for low frequency
measurements, but the large relative moment associated with earthquakes makes inertial
seismometers well suited to seismic tomography [21]. Inertial seismometers make use of a
suspended mass to convert ground motion into an electrical signal. The signal generated
depends on both the amplitude of ground motion and how suddenly the event occurs.
2.3.2 General Principles
While the methods may differ, the principles behind the different forms of tomography remain
the same. X-ray absorption tomography, for example, constructs images by analyzing the
reduction in radiation as X-rays pass through a sample along a straight line [18]. Seismic
tomography accounts for wave propagation along curved paths, but rather than measuring the
intensity of a wave, the speed at which the wave traveled may be used to infer properties of the
medium through which it passed.
The curved paths of seismic waves are governed by Snell’s Law, which is derived from Fermat’s
Principle of Least Time. Fermat’s Principle states that light will take the shortest possible travel
path from point A to point B. Snell’s law is derived from Fermat’s Principle with dimensional
variables presented in Figure 2.1.
10 |
Virginia Tech | 2.3.3 Seismic Tomography Principles
Medical tomography and seismic tomography also differ in their ray path distribution. Ray path
distribution is determined by the event location, receiver location, and the medium through
which the ray path travels. Ray bending, as a result of heterogeneous media, was explained
through Snell’s Law. Event and receiver location, however, are determined by the researcher or
experimental conditions. In the case of medical tomography, the receiver and event locations are
both known and chosen by the technician, so accurate ray paths can be determined. Event and
receiver placement are more difficult when performing seismic tomography. With insufficient
ray path distribution, velocity features may be smeared when performing seismic tomography
[23]. Receiver locations are precisely known, but event locations are often unknown, unless
active sources are used. Active source locations, artificially created and with a known
hypocenter, have been used in the past to image individual pillars or tunnel stress distribution,
but are labor-intensive, and therefore impractical for long-term monitoring programs [23].
Passive seismic events are seismic events already occurring at the location of interest. Passive
seismic monitoring has applications in a wide range of engineering projects that make use of
induced seismic events, such as oil and gas production [24], carbon sequestration [25], or mining
[23]. For mine-stress analysis, using mining-induced microseismic events as ray path sources
can be advantageous due to their frequency of occurrence and location, most likely occurring
near the area of active mining [26]. Passive microseismic events are ideal for noninvasive,
remote, time-lapse monitoring [26], because they allow for continuous observation of stress
levels, fluid migration, or other time-dependent analyses.
Passive seismic imaging requires that both the seismic event locations and velocity structure be
calculated simultaneously. A homogeneous velocity model can be assumed for initial source
location, but more accurate source locations can be determined if velocity heterogeneities are
accounted for prior to locating sources. The accuracy of source locations is one of the most
important factors in any acoustic emission experiment [27]. The accuracy of reconstructed three-
dimensional structures is sensitive to the initial velocity model, increasing the importance of
obtaining an accurate or reliable starting model [28].
12 |
Virginia Tech | In addition to characterizing geologic structures, tomography can be used to determine the
distribution of stress magnitude or monitor fluids within a rock mass. Higher stressed areas tend
to produce higher seismic wave velocities due to the closing of microfractures and reduction of
pore space within a rock mass. This trend does not continue as the rock begins to fail, because
new fractures are being formed, creating additional void space that reduces the seismic wave
velocity [26]. Many different factors influence the velocity of seismic wave propagation through
rock masses, including: joints or faults, porosity, rock strength, density, depth, stress magnitude,
stress anisotropy, degree of saturation, and the type of saturation fluid [29].
Fluid displacement within a rock mass is also capable of being imaged through tomography.
Rocks in the subsurface may be saturated with brine, which has a specific associated p-wave
velocity. If the brine is displaced by a fluid with different elastic properties, the p-wave velocity
of waves traveling through that medium will change. This principle has applications to carbon
sequestration, as CO has different elastic properties than brine. The increased compressibility
2
of CO will result in subtle velocity changes in the host rock, which can then be imaged through
2
the use of tomography [30]. Reservoirs with favorable injection and storage characteristics, such
as high porosity and permeability, tend to have seismic properties suitable for CO monitoring.
2
Reservoirs with low permeability and low porosity are likely to have poor seismic conditions for
imaging [31].
Cross-well seismic tomography is one common method of applying these tomographic
techniques. This form of tomography involves generating events in one borehole and detecting
them in another borehole with an array of geophones. Shown in Figure 2.2, cross-hole seismic
tomography can be useful for determining the location and extent of low velocity zones between
two boreholes. This method of velocity model reconstruction has revolutionized subsurface
seismic measurements for a variety of engineering applications, such as tunnel development,
deep foundation construction, and petroleum reservoir monitoring [29].
13 |
Virginia Tech | Figure 2.2: Cross-Well Seismic Tomography
Regardless of method or application, the success of arrival-time tomography is dependent on the
quality of data, the travel-time calculation method, the constraints placed on the model, the
weighting values selected, and the algorithms chosen to invert the system of linear equations
[28].
2.3.4 Double-Difference Tomography
Double-difference tomography began as the double-difference earthquake location algorithm,
developed by Waldhauser and Ellsworth [32]. This method attempts to minimize the residuals
between observed and theoretical event-to-receiver travel times. If the hypocentral separation
between two seismic events is small when compared to the absolute distance between the event
and receiver, then it can be assumed that the ray paths are similar along the entire ray path,
except for the small region near the spatial offset between the two events. The difference in
these two travel times is a result of the spatial offset between the events rather than any velocity
differences encountered between the event and station, due to their absolute errors being shared
along the rest of the ray path [32]. Figure 2.3 shows an example event arrangement for a station
X.
14 |
Virginia Tech | 2.4 Rock Mechanics
Waves are the phenomena through which tomography is possible. A wave can be described as
“a disturbance that transfers energy and momentum progressively from one particle to another
particle in a medium [34].” These waves are separated into two broad categories:
electromagnetic waves and mechanical waves. Mechanical waves require a medium, such as
water, air, or rocks to transfer mechanical energy between points, whereas electromagnetic
waves can transmit energy in a vacuum [34]. Quantities frequently used to describe wave
motion include frequency (f), amplitude (A), wavelength (λ), and period (T) [35]. Frequency is
number of cycles per unit time and period is the time taken to complete one cycle. Wavelength
and amplitude are presented in Figure 2.4.
Figure 2.4: Wave Properties
Waves can further be separated into transverse waves and longitudinal waves. Particles in
transverse waves oscillate perpendicularly to the direction of propagation of the wave, whereas
particles in longitudinal waves oscillate parallel to the direction of propagation of the wave [34].
Seismic body waves are composed of both longitudinal (compression) and transverse (shear)
waves, called P and S waves respectively. P waves are capable of traveling through liquids and
propagate at a velocity related to their frequency (f) and wavelength (λ), given by the equation:
Equation 2.5
16 |
Virginia Tech | S waves travel at slightly more than half the velocity of P waves and create shear forces in the
medium they pass through, allowing for division of S waves into S waves, causing horizontal
H
motion, and S waves, causing vertical motion. Due to liquids not having a shear strength, S
V
waves are incapable of propagating through them [36].
Considering that P and S waves are mechanical waves, they require that an elastic medium
transmit them. The elasticity of that medium determines how well the seismic waves are
transmitted, represented by several elastic moduli. These varying elastic moduli are [37]:
- The Young’s Modulus: given by the ratio of extensional stress to the resulting
extensional strain.
- The bulk modulus: a measure of uniform incompressibility of the material.
- the shear modulus: a measure of a material’s resistance to shearing, or changing shape
while maintaining a constant volume
- Poisson’s ratio: compares a material’s transverse strain to its axial strain.
The approximate bulk modulus, shear modulus, Poisson’s ratio, and density of several common
materials, and their corresponding P and S wave velocities are given in Table 2.1.
Table 2.1: Typical Material Properties (Averages adapted from [37])
Bulk Modulus Shear Modulus Density Poisson's
Medium Vp (km/s) Vs (km/s)
(GPa) (GPa) (kg/m3) Ratio
Air 0.0001 0 1 0.5 0.32 0
Water 2.2 0 1000 0.5 1.5 0
Sandstone 24 17 2500 0.21 4.3 2.6
Limestone 38 22 2700 0.19 4.7 2.9
Granite 56 34 2610 0.25 6.2 3.6
Basalt 71 38 2940 0.28 6.4 3.6
These moduli are dependent on stress and strain, whose relationship is commonly reported as a
stress-strain curve, shown in Figure 2.5. The gradient of the linear portion of this curve is known
as the Young’s Modulus.
17 |
Virginia Tech | Figure 2.5: Stress-Strain Curve
Strain, by its simplest definition is the length of deformation of a body divided by its original
length [38]. Stress, as it pertains to engineering, is a measure of the internal forces acting upon a
deformable body, presented in force per unit area. Rock masses exist in a stressed state
naturally, regardless of human activities. Understanding those natural stresses can assist in the
mitigation of harmful effects the stresses might have on human activities, such as mining. Many
civil or mining engineering projects require extensive knowledge of in-situ and induced stresses,
such as: the stability of underground excavations, pillar design, drilling and blasting, and slope
stabilities [39]. While there is no widely accepted terminology for describing the states of stress
in a rock mass, it is described by Hudson in ten different ways [40]:
- Tectonic Stress – Stress resulting from the release of energy caused by tectonic plate
movement. Tectonic stress is important to consider for safety factor calculations in
underground construction projects, such as tunnels or caverns for gas and waste storage
[41]. It can be delineated further into Global Tectonic Stress, concerning the shifting of
plates, and Local Tectonic Stress, concerning plate bending and isostacy [42].
- Gravitational Stress – Stress from Earth’s gravitational field, also called vertical stress.
This stress encompasses the weight of the overburden, which increases with depth, as
well as accounting for stress resulting from surface topography [42].
- Natural Stress – Also called in-situ stress or virgin stress, this is the stress that exists
naturally in the rock mass before any man-made disturbance [42].
- Regional Stress – The stress state in a relatively large geological domain [40].
18 |
Virginia Tech | - Local Stress – The stress state in a small domain [40].
- Near-field stress – The stress state in the region of an engineering perturbation [40].
- Induced Stress – The stress state as influenced by man-made disturbances. Examples of
induced stresses are those that result from excavating, explosions, drilling, or pumping
[42].
- Residual Stress – Stresses that remain in the rock mass after the original triggering
mechanism has occurred. These stresses exist in equilibrium or at a near-equilibrium
state within the rock mass [42].
- Thermal Stress – The stress state caused by temperature change [40].
- Palaeostress – A form of residual stress occurring as a result of palaeo-tectonic events
[42].
Stress is neither a scalar nor vector quantity, but a tensor quantity. It contains a magnitude,
direction, and plane under consideration [38]. Nine components of stress exist at any given point
in a rock mass, as shown in Figure 2.6. A normal stress acts on the x, y, and z plane and two
shear stresses act within each plane. In the case of excavation, unsupported surfaces become
principal stress planes because no shear stress can act on them. This is important to consider
when orienting underground openings.
Figure 2.6: 3D Stress State
19 |
Virginia Tech | 2.4.1 Failure Criteria
There are many criteria for explaining the failure of rocks due to stresses, known as failure
criteria. Perhaps the most well-known of these criteria are the Mohr-Coulomb Failure Criterion,
the Hoek-Brown Failure Criterion, and the Griffith Failure Criterion. The Mohr-Coulomb
Failure Criterion, presented in 1900, suggests that materials fail due to a critical combination of
normal stresses and shear stresses, rather than from one individual component. The Mohr-
Coulomb Failure Criterion is written as:
Equation 2.6
where
τ = shear stress on the failure plane
f
C = cohesion
σ = normal stress on the failure plane
ϕ = angle of internal friction
The Mohr-Coulomb failure criterion can be visually represented by a Mohr’s circle, shown in
Figure 2.7. The line lying tangent to the circle is the failure criterion. Shear and normal stresses
plotted below the failure criterion will not result in shear failure, whereas those plotted on the
line will cause a shear failure along that plane. The angle of shear failure is given by ϴ and the
diameter of the Mohr’s Circle is determined by the minor principle stress, σ , and the major
3
principle stress, σ [44].
1
20 |
Virginia Tech | Figure 2.7: Mohr's Circle
In 1980, Evert Hoek and E. T. Brown introduced the Hoek-Brown Failure Criterion to provide an
empirical relation for characterizing stress conditions that lead to failure in rock masses. The
Hoek-Brown method can be used when either many joints are present in the rock mass or no
joints are present. It cannot be used in anisotropic rocks. The Hoek Brown failure criterion is
defined by the equation [45]:
Equation 2.7
where
= major principal stress
= minor principal stress
= uniaxial compressive strength of the intact rock
m and s are material constants, where s = 1 for intact rock.
The Griffith theory of fracture was developed to explain why brittle materials, such as glass,
have a much smaller mechanical tensile strength than the theoretical molecular bond strength of
that material [46]. Griffith uses a conceptual model of a two-dimensional rock containing
randomly oriented thin cracks to derive a failure criterion for rock under both tensile and
compressive loads. This leads to a nonlinear failure surface that is more realistic in several
respects than the linear Coulomb Law. Griffith’s Failure Criterion is given as:
21 |
Virginia Tech | Chapter 3: Double-Difference Tomography Applied to
Monitoring of Geologic Carbon Sequestration in the
Aneth Oil Field, Utah
3.1 Abstract
Double-difference seismic tomography is performed on a carbon sequestration operation in the
Aneth Oil Field in southeast Utah as part of a Department of Energy initiative on monitoring,
verification, and accounting of sequestered CO . A total of 1,211 seismic events were recorded
2
from a borehole array of 22 geophones. Aneth Unit data were divided into four time periods for
time-lapse analysis. A low velocity zone spanning the lateral extents of the observable region,
likely representing a CO plume, is detected when considering voxels containing the highest ray
2
path coverage. A series of synthetic tomography tests simulating different CO plume sizes and
2
locations was performed to assist in characterizing velocity changes associated with Aneth Unit
data. Inferences about the existence of a CO plume should be made by comparing actual data to
2
synthetic data resulting from simulations performed under similar conditions. Considering
synthetic simulation similarities and a derivative weight sum analysis, a CO plume can be
2
imaged within the Desert Creek reservoir, but the resolution of the CO2 plume is too low for
proper monitoring, verification, and accounting of injected CO . Recommendations for
2
improving CO plume resolution, through double difference seismic tomography, are made to
2
increase the ray path distribution throughout the Aneth Unit by varying geophone locations.
3.2 Introduction
Climate change is a significant international concern, often attributed to increasing
concentrations of greenhouse gasses, namely Carbon Dioxide (CO ), in Earth’s atmosphere.
2
Fossil fuels are responsible for the majority of anthropogenic CO emissions, but also sustain the
2
global economy. This conflict has created one of the greatest geo-political environmental issues
26 |
Virginia Tech | in recent history. One method for reducing the amount of CO released into the atmosphere is
2
geologic carbon sequestration. Geologic carbon sequestration is the capture of CO before it
2
reaches the atmosphere, and its storage under deep geologic formations, such as depleted oil
reserves, unmineable coal seams, or deep saline formations. It is important to monitor, verify,
and account for CO during injection and post-injection. One tool that can be used to perform
2
this task is seismic tomography.
Generating an image, or tomogram, of an object by examining its reaction to the passive, probing
energy from an external source is the foundation of tomography [19]. The external source, in the
case of seismic tomography, is a seismic event. Seismic events can either be passive or active by
nature. Active seismic event locations, artificially created and with a known epicenter, have
been used in the past to image individual pillars or tunnel stress distribution, but are labor-
intensive, and therefore impractical for long-term monitoring programs [23]. Passive seismic
events are seismic events already occurring near the location of interest, possibly induced by
humans through mining, fluid injection, or other engineering projects. Passive microseismic
events are ideal for noninvasive, remote, time-lapse monitoring [26], as they allow for
continuous observation of stress levels, fluid migration, or other time-dependent analyses.
Fluid displacement within a rock mass is capable of being imaged through tomography. Rocks
in the subsoil may be saturated with brine, which has a specific associated p-wave velocity. If
brine is displaced by a fluid with different elastic properties, the p-wave velocity of waves
traveling through that medium will change. This is useful for carbon sequestration, as CO
2
different elastic properties than brine. The increased compressibility of CO will result in subtle
2
velocity changes in the host rock, which can then be imaged through the use of tomography [30].
This thesis will demonstrate the degree to which sequestered CO can be monitored, verified, and
2
accounted for in the Aneth Unit through time-lapse analyses of velocity changes and
comparisons of synthetic plume simulations to Aneth Unit data. Analysis of event, receiver, and
travel time data provided for the Aneth Unit of the Aneth oil field is performed on a time-lapse
basis to determine CO plume extents, CO concentration changes, and any leakage of CO from
2 2 2
the reservoir. Appropriate damping and smoothing values are selected to optimize the
27 |
Virginia Tech | performance of the velocity model reconstruction. Simulations with synthetic travel times for a
variety of different CO plume sizes and locations are also conducted to assist in the verification
2
of results obtained from the Aneth Unit data.
3.2.1 Double-Difference Tomography
Double-difference seismic tomography is based on the assumption that two seismic waves
propagating along the same path should generate similar travel times. Studies have shown that
substantial improvements in both locating seismic events and increasing geologic velocity model
resolution can be obtained through the use of simultaneous multiple event ray path analysis. For
double-difference tomography, the absolute and differential travel times are used to reconstruct a
three-dimensional velocity model as well as relocate seismic events. The double-difference is
given by the equation
Equation 3.1
where is the arrival time of the seismic wave, emanating from location i, arriving at station k
and is the arrival time of the seismic wave, emanating from location j, arriving at station k.
The difference between the calculated and observed arrival time offsets is or the double-
difference [49].
The double-difference tomography code, tomoDD, is being used to perform pseudo-bending ray
tracing to calculate the travel times between seismic events and receivers. This code is based on
hypoDD, a double-difference event location algorithm developed by Waldhauser and Ellsworth
[50].
28 |
Virginia Tech | 3.2.2 Site Description
The Aneth oil field, located in southeast Utah, shown in Figure 3.1, is part of ongoing carbon
sequestration experiments conducted under the Southwest Regional Partnership on Carbon
Sequestration (SWP), which is represented by state and federal agencies, universities, electric
utilities, non-governmental organizations, coal, oil and gas companies, and the Navajo Nation
[51]. The Aneth Oil Field is Utah’s largest producer of oil, having produced over 440 million
barrels since its discovery in 1956. The Aneth Unit, the northwestern part of the Aneth Oil Field,
was selected by the SWP to demonstrate the monitoring, verification, and accounting of injected
CO . The Aneth Unit was acquired from ChevronTexaco Exploration and Production Company
2
in 2004 by Resolute Natural Resources Company and is jointly owned by both Resolute and the
Navajo Nation Oil and Gas Company. The Aneth Unit was selected for this project for several
reasons: the practice of CO injection in neighboring parts of the oil field, the presence of a CO
2 2
source and pipeline nearby [52], the amount of remaining oil for enhanced oil recovery, and it
represents the archetypal oil field of the western United States [53].
Figure 3.1: Location of the Greater Aneth Oil Field [48]
The Aneth Unit has produced roughly 149 million barrels of oil over an area that covers
approximately 16,800 acres. Secondary recovery through water injection was first implemented
29 |
Virginia Tech | in the field in 1962, and CO injection began in the Aneth oil field in 1985. CO injection had
2 2
not occurred in the Aneth Unit until 2007 [54]. The injection fluids being used at the site consist
of salt water and CO . The salt water is produced through oil and gas production. The injection
2
CO is produced from the McElmo Dome Field located about 35 miles northeast of the Aneth
2
Unit. The typical contents of the sequestered fluid are 72.6% CO , 15.4% methane, and 1.58%
2
nitrogen, with the remainder being other hydrocarbon gases [55].
3.2.3 Geology
The Aneth Oil Field is located in the Pennsylvanian Paradox Basin of southeastern Utah. The
basin is a stratigraphic trap, containing fractures and minor faults. The primary reservoir for the
Aneth Unit is the Desert Creek carbonate, deposited on the Chimney Rock shale and overlain by
the low-permeability Gothic shale, which acts as a caprock for the reservoir. Wells reveal a
geology consisting of limestone, both oolitic and algal, and finely crystalline dolomite. Fractures
in core samples are common, indicating minor faults that may affect fluid flow.
In addition to the Gothic Shale, the Desert Creek reservoir is overlain by approximately 800 feet
of tight carbonates, evaporites, and fine-grained siliciclastics. Above that is up to 3,200 feet of
shales, siltstones, sandstones, and evaporites, providing multiple layers of confinement against
CO migration [55].
2
The average vertical extent for the Desert Creek reservoir in the Aneth Unit is 14.9 meters (49
feet). The porosity of the reservoir is 10.3% with a permeability ranging from 6 to 27
millidarcies and water saturation of 23.3 percent [56]. Previous exploration has revealed the
reservoir extents and geometry. The depth of the top of the Desert Creek reservoir is shown in
Figure 3.2. The buoyancy of CO should result in an upward migration of CO until it is stopped
2 2
by an impenetrable layer, the Gothic shale. If CO migration is buoyancy driven, accumulation
2
would most likely occur in the 1,708 to 1,702 meters in depth zone.
30 |
Virginia Tech | Figure 3.2: Depth of the Top of the Desert Creek Reservoir [48]
3.2.4 Microseismic Monitoring
There has been a history of monitoring at the Aneth Unit and Greater Aneth oil field. Since
2006, there have been several monitoring programs in place in the Aneth Unit for the purpose of
monitoring CO . Establishing a background was performed through 3-D seismic reflection
2
imaging surveys, passive seismic monitoring, vertical seismic profiles (VSP), crosswell seismic
imaging, active doublet imaging, groundwater chemical analysis, and remote sensing. Time-
lapse vertical seismic profile surveys, conducted from 2007 to 2009, found that injection of CO
2
registered an interpretable signal in VSP-CDP (vertical seismic profile to common depth point)
images [57]. The monitoring well, C313SE, was used in many of these methods as well as the
seismic monitoring associated with this tomographic study of the Desert Creek reservoir.
The geophone arrangement, in place at the C313SE monitoring well, is a 22-level vertical
geophone array. The monitor well contains a 60-level geophone cable that extends to 1,704
meters in depth, with levels 15.2 meters apart. 16 geophones are evenly spaced between 1,445
31 |
Virginia Tech | Several outputs were compared for optimizing the damping and smoothing values. The
maximum velocity, shown in Figure 3.9, and the minimum velocity, shown in Figure 3.10,
should be kept within reasonable limits. One method for quantifying if the data have acceptable
velocity change is reported in Figure 3.11, which reports the percentage of data within an
expected 10% variation in velocity due to carbon sequestration. A 10% variation allows for
velocities in the range of 4.70 km/s to 5.74 km/s. The minimum and maximum velocity change
for each smoothing and damping value should be kept reasonably close to the background
velocity, as neither the resolution of the velocity model nor known reservoir geology suggests
that extreme deviations in velocity should be seen. A damping of 80 and smoothing of 100 was
chosen to avoid underfitting or overfitting the data. The 1,000 and 500 damping values were
found to be identical in absolute mean and variance, maximum recorded velocity, minimum
recorded velocity, and percentage of velocities within expected limits, likely meaning the data
are being overfitted. The 100 smoothing line shows the first separation from the 1,000
smoothing and 500 smoothing lines. The 100 smoothing line shows the first separation from this
overfitting, and also represents the most linear relationship. A smoothing of 100 allows for
separation from the 500 and 1,000 smoothing lines, likely overfitting the data, while also keeping
the minimum values, maximum values, and overall velocity change within acceptable limits. A
damping of 80 is chosen because it represents approximately the last point before the data begin
to wander in Figure 3.8, representing underfitting of the data.
38 |
Virginia Tech | 3.3.3 Tomography Calibration
Calibration of both the MATLAB travel time calculator and tomoDD are performed. The ability
to test the accuracy of tomoDD with this data set is limited, however one test is performed to
ensure that tomoDD would not change the velocity model when no changes are necessary. The
same event locations, receiver locations, and velocity model from the Aneth Unit are used.
Using the squared distance between each event-receiver pair, and a desired average velocity of 5
km/s, each travel time is artificially set. The background velocity model for tomoDD was also
set to a velocity of 5 km/s. Given that each travel time reflects a travel path of a 5 km/s average
velocity and the velocity model contains a constant background velocity of 5 km/s, tomoDD
should make no changes to the initial velocity model. TomoDD accurately made no changes to
the initial velocity model, with a constant velocity of 5 km/s being reported for 100% of nodes.
For the synthetic travel time calculation calibration, two tests are performed to assess the
potential for flaws in the calculation method that would prevent the accurate determination of
travel times. The first test involves the creation of a 1000x1000x1000 meter cubic area
containing 4,000 seismic events. Two receivers are placed within this region as well as a
200x200x200m low velocity zone, all shown in Figure 3.13. The background velocity for travel
time calculation is 5 km/s and the low velocity zone is 3 km/s. Using a 20x20x20 node grid size,
the travel-time calculator was able to output a travel-time input for tomoDD. Due to data
constraints within tomoDD, the number of events was randomly reduced to 1,980, and then
further reduced within the program to 1,091 events after relocation. The background velocity for
tomoDD analysis is set to 4.56 km/s, to reflect the average velocity of the velocity model after
travel times were calculated. A cross-sectional tomogram of this calibration is shown in Figure
3.14. The resulting low-velocity zone ranges from approximately 4.4 km/s to 4.9 km/s. The low
velocity zone is not a cube, and also extends beyond the extents of the low-velocity cube. The
symmetric nature of the velocity changes, and the existence of a low-velocity zone of the
approximate dimensions centered in the correct location suggest that the travel time calculator
may be able to provide an approximate representation of low-velocity zones.
42 |
Virginia Tech | velocity anomalies. As with the tomoDD calibration, the event-receiver combination from Aneth
is used, with the travel-time calculator velocity model having the same layered velocity model
used with the actual Aneth data. This will be the control simulation to which synthetic tests,
considering the Aneth event-receiver arrangement, can be compared. Some reduction in velocity
is expected as a result of node-to-node travel paths rather than a continuous curve. Ideally, this
velocity reduction would be very small and not reflected in the tomoDD output.
3.3.4 Synthetic Experimental Plumes
Synthetic plumes, designed to represent potential conditions, or assist in the understanding of
velocity change in the Aneth Unit, are all created and processed in an identical manner. Event
and station locations used in processing the original Aneth Unit data are used to assess the
accuracy with which the current event-receiver arrangement can image the CO accumulation in
2
the oil field. The velocity model for the synthetic tests is the same layered velocity model
provided for the Aneth Unit, with the exception of an artificially created low velocity zone at
varying locations. These low velocity zones are assigned a 2 km/s velocity reduction from the
background. A traditional plume shape is not assumed, nor is velocity dependent on position
within the plume. The use of multiple injection wells and a changing caprock elevation are
likely to create an unpredictable plume shape and CO concentration distribution. A simple
2
rectangular prism of low velocity is created instead to represent an area likely to be influenced by
CO .
2
For all synthetic data post-processing, voxels with derivative weight sums (DWS) of zero will be
excluded from the final results. The DWS is an indicator of the number of ray paths traversing a
node, and can thus be considered a form of confidence in the value of that voxel. Voxels with a
DWS of zero had no ray paths traverse them and remained unchanged, because no information
can be inferred from the ray path behavior.
44 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.