University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Colorado School of Mines
|
Table 9.10: Uncertainty validation for current “risk-averse” mine production plan
As a final comment, it is noted that although the plan obtained in Figure 9.10 is optimal,
and meets all the specified problem requirements, for practical reasons, it cannot be easily
implemented on the field, as the scheduler “smears” the production throughout the orebody.
Nonetheless, in important regards, these results are still informative and valuable because: (i) they
provide a provably optimal solution establishing an upper bound on the value of the optimal integer
solution; (ii) they can still offer guidance on the boundary of potential pushbacks or phases. For
instance, in presenting mine plans, the numbers inside each block represent the period by which
the respective block has been fully mined and this is very useful guidance in terms of the limit
times by which certain areas of the pit need to be fully depleted.
9.2 Three-Dimensional Synthetic Case Study – KD85
The solution methodology developed in this research is utilized to solve a three-
dimensional problem corresponding to a notional copper deposit referred to as KD85, so as to
demonstrate the advantage of the methods proposed. The results obtained are compared to those
originated from a risk-free approach and further validate the analytical framework proposed. KD85
contains a total of 5,400 blocks, each assigned its individual copper grade value and having the
same size and tonnage. In addition, all blocks are classified into one of the three resource
classification categories, so that the schedules including grade risk constraints can be contrasted
to risk-free schedules. Using the economic assumptions presented in Table 9.11 and assuming pit
wall slope constraints require 45-degree slope angles, it was possible to determine ultimate pit
164
|
Colorado School of Mines
|
limits for the deposit. These were subsequently loaded into Mintec’s MineSight software and
visualized as appropriate (see Figure 9.11 below).
Figure 9.11: Three-dimensional view of the KD85 ultimate pit limits.
It is customary to intersect the topographic contours of the surface terrain with the three-
dimensional model of the deposit, as the said contours represent an important consideration in
practice. However, for the purpose of the discussions in this methodology this aspect is considered
immaterial.
Table 9.11: Economic assumptions used in determining ultimate pit limits for KD85
Table 9.12 provides a general summary of the characteristics of the material contained
within the ultimate pit limits. It can be seen that for KD85 only a small proportion of blocks is
considered ore under the adopted economic assumptions, however, the average grade of the ore
blocks (3.64% Cu) is such that KD85 is considered a high-grade deposit.
Table 9.12: Characteristics of the ore material inside the ultimate pit limits
RESOURCE CLASSIFICATION
STATISTIC VALUE UNITS VALUE UNITS
CATEGORY
TOTAL BLOCKS 2016 ** INFERRED 20 %
ORE BLOCKS 64 ** INDICATED 31 %
WASTE BLOCKS 1952 ** MEASURED 48 %
AVG. GRADE 3.64 (% Cu)
165
|
Colorado School of Mines
|
If the set of conditional simulations is representative of the full spectrum of uncertainty, then
this procedure allows one to determine the robustness of a mining plan relative to grade
uncertainty.
The current solution to is tested against the ten available orebody realizations
(cid:3013)(cid:3017)(cid:3045)
to verify the number times the realized average grade is within a ±15% accuracy about the [min,
(cid:4666)(cid:1841)(cid:1842)(cid:1839)(cid:1842)(cid:1845)(cid:1842)(cid:4667)
max] grade interval. If, for any time period, the realized average grade is outside the bounds of the
predefined interval more than once, then the plan is considered to have failed the desired risk
criteria, or equivalently, not to have met the minimum required risk threshold. Obviously, this is
akin to the construction of a confidence interval about the target average copper grade. The results
of confronting the current to the ten conditional simulations are presented in Table
(cid:3013)(cid:3017)(cid:3045)
9.20.
(cid:4666)(cid:1841)(cid:1842)(cid:1839)(cid:1842)(cid:1845)(cid:1842)(cid:4667)
Table 9.20: Uncertainty validation for the current solution to (OPMPSP)LPr. The current solution
fails to meet the desired minimum risk threshold in period 2.
Recall that the current realized production plan results from a problem definition whose risk
requirements allowed for the proportion of Inferred material, the riskiest resource classification
category, to reach up to 15% of the total mill feed (see Table 9.16).
Given that the current plan fails the desired risk criteria (in period 2), one possible course of
action by the decision maker might be to constrain more strictly the proportion of Inferred material
allowed into the composition of the mill feed. Such a decision should, in principle, help the newly
generated schedules more easily meet the required risk threshold. Hence, the risk requirements are
modified such that the maximum allowed proportion of Inferred resources does not exceed 5% of
the total composition of the mill feed (see Table 9.21).
174
|
Colorado School of Mines
|
Figure 9.24: East-West two-dimensional cross-sectional view of the KD85 phase design solution.
Table 9.25 summarizes the results for phases obtained using said approach:
Table 9.25: Realized mine production plan for phase design.
PHASE ORE MINED (blk) WASTE MINED (blk) AVG GRADE (%Cu) INFERRED (%) INDICATED (%) MEASURED (%)
1 7 21 3.56 0 0 100.0
2 31 292 3.73 22.6 45.2 32.2
3 22 254 3.50 13.6 22.7 63.6
NPV ($M) 119.952
As expected, given that no mechanism is enforced that prevents violations of either risk or
blending constraints these might not be met. In effect, this aspect is referenced in other published
work involving rounding heuristics (see Espinoza et al., 2012; Brickey, 2015). However, it can be
shown graphically that the phases obtained provide a very good guide for the true integer optimal
solution. In order to further emphasize this proposition, the sections corresponding to the integer
production schedules are plotted side-by-side with those corresponding to the tentative phases,
both for plan and cross-sectional views.
178
|
Colorado School of Mines
|
CHAPTER 10.
CONCLUSIONS
In a multitude of dimensions, social, political, economic or technical, the mining industry is
faced with important challenges it needs to meet. It is possible - indeed common – to attempt to
address these challenges under a deterministic framework. However, one important topic of
concern is the effective management of grade uncertainty, specifically, the important ways in
which its effects percolate the mining system and impact project economics. In this context, the
practical success of geostatistical conditional simulations (GCS) have consolidated its importance
as the method of choice for grade uncertainty characterization. Unfortunately, from an
optimization perspective, there exists a crucial obstacle to the direct, explicit incorporation of GCS
into mine production schedules in the form of stochastic linear programs: the problems become
either intractable or very challenging to solve (by exact methods). This tendency for stochastic
models grow to dimensions beyond the reach of exact solution methods has led to a large
reorientation of research towards heuristic (or metaheuristic) techniques which - provided some
conditions are verified - can obtain a solution to said problems, although one whose optimality
cannot be verified.
In this dissertation a novel approach to mine production scheduling is presented which takes
into account geostatistical conditional simulations, but models grade uncertainty differently than
the current state-of-the-art practice of combining stochastic optimization models with heuristic
solution algorithms. We show how the so-called “curse of dimensionality” for stochastic models
can be circumvented by means of an integrated solution methodology which, rather than applying
the grade realizations of GCS as input parameters in a stochastic model, uses these as elements to
be incorporated in two distinct levels of analysis:
(i) as a complement to traditional geostatistical methods for mineral resource classification
(ii) as the means by which a specific, (user-defined) risk criterion can be established
Regarding the first level of analysis, GCS can be central to adequate classification of mineral
resources into the Inferred, Indicated and Measured categories which constitutes an integral
component of our methodology. We adopt said categories both as a valid proxy for grade
182
|
Colorado School of Mines
|
uncertainty and as a tool for quantifying risk in the production schedules generated. Specifically,
we show how the precise composition of the mill feed can be defined a priori in terms of the
proportions of material belonging to each of the resource categories, and how this impacts the
schedules obtained. To our knowledge, this is an approach not yet published in the technical
literature, but which carries important implications. On the one hand, these constitute the industry
standard for communicating confidence in estimated grades, and this is important because it
increases transparency and reduces resistance to adoption of the methods by industry. On the other
hand, the inclusion of resource classification categories into the mine plans helps to address the
disconnect between demands for robust schedules (on the part of management) and traditional
deterministic production schedules which ignore risk.
All optimization models formulated are solved to proven optimality by exploring well-
established exact solution algorithms for the mine production scheduling problem. In particular,
the Bienstock-Zuckerberg and the PseudoFlow algorithms (both state-of-the-art algorithms) are
combined to produce solutions to optimization problems including ore risk
constraints . This is in contrast to heuristic (or metaheuristic) techniques which have
(cid:3045)
the advantage of solving large-scale problems, but also the inconvenience that the solutions
(cid:4666)(cid:1841)(cid:1842)(cid:1839)(cid:1842)(cid:1845)(cid:1842)(cid:4667)
obtained cannot be proven optimal, and may rely significantly on initial starting conditions.
Regarding the second level of analysis, our method allows for the inclusion for a risk
preference measure, expressed in the form of a risk threshold which conditions the acceptance
(adoption) of any schedule obtained from the solver. For example, the risk criterion used in the
research consisted of finding the number of times (out of ten) that average grades obtained from a
production schedule stayed within the limits of some predefined grade interval.
Finally, by allowing direct specification of a critical threshold for risk on the part of the
modeler, the method enables decision makers to affect (“shape”) the outcome of the desired
production schedules transparently and explicitly by imparting risk preferences. In essence, the
method developed operates similar to a “price discovery” mechanism in which the variable of
interest is the proper level of risk in the mine production schedules, which can be determined by
consideration of a range of factors and expectations available to the decision maker, and which
may go beyond the simple technical inputs to an optimization solver.
183
|
Colorado School of Mines
|
10.1 Suggested Future Work
The solution framework presented in this dissertation meets the goal of practicality and
transparency, and it is expected that such characteristics should make for an easier incorporation
by industry. There remain, however, important directions for future research which should be
further explored. First, although uncertainty with regards to grades is considered in our models,
the true (full) spectrum of uncertainty is much wider and reflects itself in multiple dimensions,
including, most notably, the economical and operational.
The stochastic nature of metal prices, in particular, can have an important impact on the
production schedules generated (Sabour et al., 2009; Mokhtarian & Sattarvand, 2015) - and/or on
overall project economics (Davis & Samis, 2006, 2014) - and to the extent to which it affects
mining cutoff grades, it is also interdependent with grade uncertainty. Therefore, it is anticipated
that the integration of market uncertainty within the proposed framework – provided its
implementation be possible while avoiding the “curse of dimensionality” – would prove
advantageous.
Similar to (Froyland et al., 2007), our methodology can be used to derive an upper bound
on the economic value of additional infill drilling. However, the definitive answer to the question
of determining the optimal number of drillholes and their specific location in space encloses an
endogenous nature which our models cannot grasp. Indeed, additional drilling should be
formulated as an internal decision variable of the optimization model, but one which reduces
uncertainty itself. Such problems belong to a broader class of stochastic problems with decision-
dependent uncertainty (Goel & Grossman, 2005), and it is not clear how such an integration could
be accomplished without making the models intractable. On the other hand, recent years have
witnessed dramatic advances in the size of problems incorporating uncertainty which are now
solvable in reasonable amount of time, i.e., possibly not exceeding one or two hours per scenario
evaluated (Lamghari et al., 2012). The methods used frequently avoid the “curse of
dimensionality” by employing solution techniques belonging to the class of heuristic (or
metaheuristic0 algorithms. This newfound possibility has prompted a large body of research work
directed at holistically modelling increasingly large mine planning problems representing entire
mining complexes. In our methods all the solution algorithms used are exact, which implies that
the limitations resulting from the large size of mining problems must be overcome differently. In
184
|
Colorado School of Mines
|
this regard, one specific direction of research seems particularly promising, and involves
determining the minimum possible number of realizations (conditional simulations) required for
an adequate characterization of the true uncertainty spectrum (Romisch, 2008; Armstrong & Galli,
2010; Bayraksan & Morton, 2011). If scenario reduction techniques result in significantly smaller
sets of “high-quality” realizations, it is envisioned that our methods can be extended to solving
explicitly stochastic models of the production scheduling problem with risk constraints included
.
(cid:3045)
(cid:4666)(cid:1841)(cid:1842)(cid:1839)(cid:1842)(cid:1845)(cid:1842)(cid:4667)
Another important direction of future research relates to the improvement and scaling of
the implementation of our solution algorithms so that they are made more efficient and able to
tackle increasingly larger problems. Specifically, more computational testing is required to
establish (or refute) dependence of algorithmic performance on specific problem instances, as well
as comparisons with heuristic solution times. In this regard, the inclusion to the solution procedure
of “hot starts” – defining an initial feasible solution expected to be relatively close to optimal –
can accelerate convergence and should therefore be explored. Similarly, our methods provide
tentative indication to the generation of “Mining Phases” that differ from the ones obtained by
price parameterization, and seem to provide a very good guide to optimal production scheduling.
However, in line with some of the research published in the mining literature (Chicoisne et al.,
2012), much can be done regarding the study of approximation (or exact) techniques that make
possible the conversion of initial fractional solutions into integer ones.
185
|
Colorado School of Mines
|
Wackernagel, H. 2013. Multivariate geostatistics: an introduction with applications. Springer
Science & Business Media.
Webster, R., & Oliver, M. 2007. Geostatistics for environmental scientists. John Wiley & Sons.
Whittle J, 1999. “Four-x strategic planning software for open pit mines.” Reference manual.
Whittle, J., 2014. “Not for the faint-hearted.” Proceedings of Conference on Orebody Modeling
and Strategic Mine Planning, pp. 3 – 5, Dimitrakopoulos, R., (eds.) Perth Australia.
Zhang, J., & Dimitrakopoulos, R. 2015. “A dynamic ore-price-based method for optimizing a
mineral supply chain with uncertainty.” Application of Computers and Operations Research in the
Mineral Industry - Proceedings of the 37th International Symposium, APCOM, pp.1–9.
Zhang, M., & Middleton, R. 2007. “A study of the impact of price uncertainty to mining project
valuation.” Knowledge Creation Diffusion Utilization, p.1–5.
Zhang, Ye. 2009. Introduction to Geostatistics - Course Notes. Dept. of Geology & Geophysics,
University of Wyoming.
Zhao, Y. 1992. “Algorithms for optimum design and planning of open pit mines.” Ph.D. Thesis,
The University of Arizona, Phoenix, AZ.
195
|
Colorado School of Mines
|
ABSTRACT
With increasing population, industrial development, and global temperatures, water scarcity is the
biggest challenge the world is facing today. As a result, the importance of the research, development, and
implementation of advanced water treatment processes to treat waste streams previously considered too
difficult and expensive for beneficial reuse has never been higher, with an emphasis on waste streams
affecting the water-food-energy nexus. One such waste stream is the wastewater generated from
unconventional oil and gas operations, particularly waste from hydraulic fracturing (HF). This water,
termed produced water (PW), is currently being managed mostly through disposal by deep-well injection
as this is the least expensive way to deal with PW. Therefore, it is critical to develop treatment processes
that are effective in removing contaminants while reducing treatment costs closer to the costs of disposal.
Biological wastewater treatment of conventional waste streams (i.e., municipal wastewater) is an
effective pretreatment strategy for the removal of organic compounds before membrane desalination (e.g.,
reverse-osmosis (RO)) to mitigate biofouling. Unfortunately, PW can limit biological pretreatment and
desalination options due to the high concentrations of total dissolved solids (TDS) which can devastate a
microbial community and eliminate the effectiveness of RO when TDS concentrations exceed 50 g/L-
TDS (a common occurrence with PW with TDS ranging from 10 g/L to 250 g/L). A more thorough
understanding of how these processes perform when treating PW, through the use of pilot-scale
experiments and data collection, is therefore necessary to fill this knowledge gap.
Thus, the objective of this dissertation was to evaluate the efficacy and environmental impact of a
cost-effective treatment trains for high salinity PW. This was accomplished through long-term pilot-scale
testing on the effectiveness and resiliency of a biological pretreatment process on high salinity PW, a
membrane bioreactor (MBR). Additionally, this dissertation describes a comprehensive environmental
water quality analysis of PW chemistry, including toxic effects on human cells, throughout a complete
treatment train that includes MBR pretreatment and membrane distillation for desalination. Finally, this
dissertation evaluated the performance and practicality of a novel desalination process, LSRRO, to
desalinate high salinity PW after several pretreatment steps, including MBR.
iii
|
Colorado School of Mines
|
ACKNOWLEDGMENTS
As someone who never saw myself capable of getting through a PhD program, I know I never
could of have accomplished this on my own, and because of this, the acknowledgments section of this
dissertation is going to be quite long and I apologize to anyone that I did not mention. First, I would like
to thank the Zoma Foundation, the National Alliance for Water Innovation, the Department of Energy,
Exxon Mobile Corporation, Produced Water Society, the Edna Baily Sussman Foundation, and the
Virginia Hill Foundation for their continued funding and support throughout my doctoral research. I
would also like to thank Bayswater E&P, particularly Tyler Greenly, who supported my research by
generously donating thousands of gallons of authentic PW from the Denver-Julesburg Basin, giving me
the ability to study how the treatment processes performed on actual PW matrices. Thank you to the many
people from ExxonMobile Biomedical Sciences, Inc. for their hard work and assistance throughout my
studies, particularly Ted Healey and his patience as I learned how to properly ship water samples. I would
also like to give special thanks to the entities and individuals I was so fortunate to collaborate with:
Shwetha M. Acharya who was so instrumental in all of the microbial community analyses data from the
MBR, as well as Romy Chakraborty and Susannah G. Tringe from Lawrence Berkeley National
Laboratory and the Joint Genome Institute; all of the collaborators from the United States Environmental
Protection Agency, including Jie Liu, Mark J. Strynar, Christopher Corton, Hannah Liberatore and
Stephen Jackson for taking the time to help me analyze samples during my time in North Carolina, and a
special thanks to Treasure Bailley and Tricia R. Pfeiffer for taking the lead in assisting me coordinating
all aspects of our joint research project; Fluid Technology Solutions, Inc. and particularly Keith Lampi,
Ed Beaudry, Mark Schutter, and Jack Herron for providing the LSRRO membranes and for their direct
technical support during my research.
I could not have operated any of my experiments for this dissertation without the overwhelming
support of Tani Cath, who programmed and installed every VI that my systems were controlled by. I also
want to acknowledge the analytical assistance I received throughout my research from Estefani Bustos,
Kate Spangler, Nate Rothe, and Amy Boczon. I want to thank Kate Fiore, my one and only undergraduate
assistant, who helped immensely with data collection and experimental operation at a time when I needed
it most. And my biggest thanks possible to Mike Veres, without whom none of my experiments could
have been built. His understanding of practical engineering, his patience with students, and his relentless
work ethic inspired me to be better and work harder and I now consider him a friend after our time
together.
The completion of this dissertation was only possible with the tremendous help and support
throughout my doctoral research from the members of my PhD committee. I would first like to thank my
thesis advisor, Dr. Tzahi Cath, who challenged me to be a better engineer, let me vent all of my
xii
|
Colorado School of Mines
|
frustrations, and gave me the opportunity to become a PhD when no one, including myself, thought it
would be possible for me to complete. A tireless worker, Dr. Cath always made time to help me with any
issues I was facing, both academically and personally. His mentorship has been invaluable, and I know
the type of engineer I want to be only because of the tremendous example he gave me. I only hope that I
can continue to look to Dr. Cath after graduation for advice and support as I start my career. I want to
acknowledge and thank my thesis co-advisor, Dr. James Rosenblum, for challenging me to be a better
researcher. His help and insight during my research, including securing PW, spending valuable personal
time to help me understand the analytical machines so that I could know what was in my water,
forwarding scholarship and employment opportunities, and establishing collaboration partners so that my
research could move forward with minimal effort, was above and beyond what was expected. We may not
have always seen eye to eye, but he was always there with the support I needed. Next, I would like to give
a massive thanks to Dr. Junko Munakata-Marr, who helped me first as an undergraduate when I was in
desperate need of academic advising. Dr. Munakata-Marr not only helped me with advising, but she also
gave me my first opportunity as an undergraduate to conduct research, igniting a passion I did not know I
had. Working on several research projects at Mines Park that Dr. Munakata-Marr’s students were
responsible for, including the Turf Grass Project and Dr. Andy Pfluger’s anaerobic digester, allowed me
to show promise as a researcher and eventually get invited to join Dr. Cath’s research group. I literally
wouldn’t be here if not for her guidance and support. I would also like to thank Dr. Chris Bellona who
seemed to be the only other professor who came to the Denver laboratory. Dr. Bellona always asked me
for research updates and offered suggestions to any obstacles I was facing whenever he was in Denver.
He also gave me my first introduction to teaching by inviting me to be a part of the department’s Field
Session in 2019, letting me create the experiment and literature for the module I was asked to run. His
dog, Gizmo, was the most well-behaved and friendly dog and I am jealous that I cannot get my own dogs
to behave like that. The next committee member is the chair of my committee, Dr. Hussein Amery. I have
had to take several courses while at Mines, with many professors, most of whom were either cold and
distant, or unremarkable in any way. However, when I was able to take a humanities course, Water
Politics and Policy, I was introduced to Dr. Amery who quickly became one of, if not the, favorite
professors on campus. I am sure that most engineering students are not fond of taking humanities courses,
as they take time away from studying of engineering material. But, as a much older student, I was very
interested in this course as it was about my two favorite subjects, water, and politics. Not only was I able
to learn about these topics from the assigned readings, I found myself, quite frequently in fact, staying
after class to continue a conversation with Dr. Amery that had begun during the normal class hour. These
discussions were helpful, insightful, engaging, and thought provoking in ways that no other professor,
regardless of the course being taught, was able to induce. I cherished these after-class discussions so
xiii
|
Colorado School of Mines
|
much, that when asked to name a professor for my committee that was not in the CEEN department, I had
only one name to offer, Dr. Amery’s. Thank you for being on my committee and thank you for teaching
beyond the textbook on a subject I am very interested in. My final committee thanks goes to Dr. John
Spear, arguably the coolest person on the planet. Dr. Spear’s engaging talks, whether about
microorganisms or local beers, showed me it was possible to be the smartest person in the room while
also being the most down-to-earth person in that same room. He does not come across as a professor, but
rather an old friend that you would want to catch up with over a couple of those local beers. He is also the
most dedicated environmentalist I know on campus, literally removing trash from the recyclables bin,
while also never buying anything until his previous one was too worn to operate anymore. I have not
genuinely enjoyed being in the company of anyone I have ever met more than Dr. Spear and I thank him
for being himself.
Finally, I would like to thank my friends and family who encouraged me during my time at
Mines. Brian Wiley, my Colorado brother, gave me a place to stay while starting school and showed me
what a true friend is. Andrew Wipple, my co-manager at Patient’s Choice, who showed me that it was
indeed possible to go back to school after such a long delay. I literally would not be here without the
example he set and help he gave me during my first years at Mines. John Velicsanyi, who has put up with
me and my nomadic life all while helping me in any way he could. John Mauro, my best friend in Las
Vegas, showed me what true work ethic looked like. Chris Poe and his wife, Brandi, whom I looked to for
inspiration throughout my life, as they showed me what was possible to achieve in life, even after some
bad times. They are my true role models. My younger cousin, David Doyle, whom I have considered a
brother my whole life. My aunt Kitty Hunt, who has been my mother in Colorado and a second mom my
whole life, her love and support has never wavered, and I would not even know that Mines existed
without her. And to the two most important people in my life, my mother, Sherri McKenn, and my
forever person, Jennie Callahan. My mom has been and continues to be, the most supportive person in my
life. No matter if I was an unemployed bartender or an engineering PhD candidate, she always showed
nothing but love and support in anything I was doing. I literally could not have done this without her
always being there for me. To Jennie, thank you so much for being there every step of the way on this
journey. I never dreamed when I went back to school that I would meet someone I had so much in
common with, from literally being in all the same classes, to your kindness towards animals, to your
religious and child-rearing beliefs, I could not have found a better match and I am so happy and proud to
say you are my person. I also want to take a couple of sentences and mention some of the people who
could not be here to see this, but who nevertheless were extremely instrumental in my getting to this
point. My friend, Tim White, my grandparents, King and Charlotte Greenwood and Jack and Petie
McKenn, my father and example on what it takes to be a man, Kirk McKenn, and finally, my brother,
xiv
|
Colorado School of Mines
|
CHAPTER 1
INTRODUCTION
1.1 Problem Statement and Significance
Water scarcity and increasing energy demands, particularly in the arid western United States,
have highlighted the importance of improving the efficiencies of industrial production and waste
management practices at the heart of the national water-energy nexus as well as identify any potential
new sources of freshwater. According to the U.S. Energy Information Administration, 40% of the energy
consumed by the United States in 2020 came from one type of unconventional oil and gas (O&G)
production called hydraulic fracturing (HF).1 HF uses large volumes of freshwater while also generating
large volumes of a highly contaminated wastewater, called produced water (PW). Due to its high
concentrations of several contaminants, including total dissolved solids (TDS), heavy metals, disinfection
byproduct (DBP) precursors, and organic material, PW poses a hazard to the environment if not managed
correctly. Treating PW so that it can be reused beneficially can reduce freshwater usage from HF as well
as increase available water resources due to the large volumes of PW potentially treated. However,
because of the chemical complexity of PW, it is a major challenge to treat PW to a level acceptable for
reuse. As such, most PW is currently injected underground by deep-well disposal or left in evaporation
ponds. But, these PW management practices can increase the risk of environmental contamination due to
accidental spillage or leakage. Additionally, these methods do nothing to reduce the amount of freshwater
used during O&G exploration and production. However, the costs currently associated with PW treatment
make it the last option for energy companies looking to manage their PW volumes. It is therefore vital to
find PW treatment processes that do not add costs and can successfully treat PW to reusable levels. To
keep costs down, these processes must be simple to operate and maintain while also increasing overall
treatment efficiencies.
A membrane bioreactor (MBR) is one possible treatment process that can be used for PW
reclamation. MBR removes contaminants from PW in two ways, biologically and physically. The
microorganisms in an MBR can reduce the concentrations of organic constituents in PW, removing the
potential for biofouling during membrane-based desalination. Additionally, by incorporating an ultra-
filtration (UF) membrane, an MBR also removes suspended solids, increasing treatment efficiency with
multiple contaminants removed within one process. However, using biological processes such as MBR in
the treatment of PW presents additional challenges. For example, PW TDS concentrations can range from
less than 10 g/L to more than 200 g/L.2 This complicates treatment due to the biological degradation of
contaminants being reduced at TDS levels higher than 10 g/L.3 Additionally, biological treatment systems
require a stable environment to perform optimally, an impossibility with PW as its chemistry can vary
1
|
Colorado School of Mines
|
widely depending on the location and time in operation of a HF well. When parameters such as
temperature, pH, salinity, or nutrient concentrations are out of the optimal range, the biological system
can be negatively impacted, causing deactivation of the biological community with slow or no potential
future recovery.
Suspended solids and organic compound removal are just two of the constituents that need to be
removed from PW before it can be recycled or reused. A complete treatment train that contains several
processes for contaminant removal needs to be implemented for PW to be reused. Additionally, treated
PW, regardless of constituent concentrations, must be evaluated for potential environmental hazards if it
is to be reintroduced. While several studies have looked at environmental hazards of raw PW, there are no
studies assessing toxicity of treated PW as it moves through an entire treatment train. This type of
evaluation would give a deeper understanding of the water chemistry of PW and how it changes during
treatment, allowing researchers to better recognize which treatment processes are removing toxicity and
which ones are not.
The final process of any effective PW treatment train must be desalination. However, due to its
high salinity, PW desalination options are currently limited to thermal processes, such as membrane
distillation (MD), thermal distillation, or evaporation ponds because conventional reverse-osmosis (RO)
becomes increasingly inefficient once TDS concentrations elevate above 50 g/L. These processes have
downsides though, with evaporation ponds requiring large footprints to hold the volumes of PW
generated and MD not yet being commercially available. As such, several novel high salinity desalination
processes that use RO membranes have recently been gaining interest amongst researchers. Osmotically
assisted reverse-osmosis (OARO), ultra-high-pressure reverse-osmosis (UHPRO), and low-salt rejection
reverse-osmosis (LSRRO) are some examples of processes currently being researched.
This doctoral research focuses on the application of novel and well-established processes for the
sustainable treatment of PW for beneficial reuse. The treatment train consists of coagulation/flocculation
for the removal of suspended solids, followed by an MBR for additional solids removal as well as organic
compound reduction. The next process is adsorption columns containing granular activated carbon for
heavy metals removal and additional organic chemical reduction. The final process is desalination, either
by MD or LSRRO. It is expected that the construction, operation, and demonstration of this treatment
train, including never before used processes, will continue leading to additional research in PW treatment,
and also show the feasibility of actual implementation in the field.
1.2 Objectives and Scope of Work
The main goals of this doctoral research were to demonstrate the effectiveness and practicality of
introducing a MBR into a PW treatment train for beneficial reuse and to evaluate the performance of a
small pilot-scale LSRRO system to desalinate PW. To accomplish this, three objectives were established:
2
|
Colorado School of Mines
|
(a) determine the effectiveness and resiliency of an MBR as a pretreatment process for high salinity PW
through the long-term evaluation of dissolved organic carbon (DOC) and suspended solids removal from
real PW with varying TDS concentrations (from 27 g/L to 100 g/L), (b) evaluate the performance and
determine optimization of a complete PW treatment train, from raw PW to desalination, using MBR as
one process before desalination with MD, and (c) to validate previously published models on the
effectiveness of LSRRO to desalinate PW. The objectives of this research are met through the publication
of three distinct studies that investigated the performance of a MBR individually, toxicity removal from
PW after each process in a treatment train, and water recovery during the testing of five unique high-
salinity water streams with LSRRO.
1.3 Structure of Dissertation
This dissertation builds, expands, and improves upon previous PW research conducted by Dr.
Cath and his research team at the Colorado School of Mines, Golden, Colorado. This dissertation is a
collection of three journal articles that were written over the course of the doctoral research. Chapter 1 is
this introduction. Chapter 2, summarized in section 1.3.1, is an article that was published in ES&T Water
and is reprinted (not requiring copyright permission). Chapter 3 is a manuscript that was under review at
the time of this dissertation for publication in ES&T Water and is detailed in section 1.3.2. Chapter 4,
summarized in section 1.3.3, is a manuscript that was under review at the time of this dissertation for
publication in Desalination. Multiple author release agreements signed by all co-authors are included in
the appendix of this dissertation.
1.3.1 MBR for the Pretreatment of High Salinity PW
A review article on biological treatment of PW that surveyed 59 published studies found that on
average, 73% of chemical oxygen demand (COD) was removed from PW with TDS concentrations less
than 50 g/L. One study using biologically active filtration (BAF) in combination with ultrafiltration
membranes (UF) as pre-treatment before desalination with nanofiltration (NF) was able to remove over
75% of organic contaminants while reducing the fouling and increasing the efficacy of NF membranes.4
Another BAF study was able to achieve 95% removal of organic matter.5 These studies not only showed
the effectiveness of biological systems in treating moderate salinity PW, but also highlighted that
combining biological and physical treatment technologies can substantially reduce the number of
processes needed.
In Chapter 2 of this dissertation, the evaluation of the ability of a small pilot-scale MBR
(bioreactor volume of 70 L) to remove various constituents (e.g., TSS, DOC, nutrients, metals) from PW
during 9 months of continuous operation, using real PW was conducted. Performance was evaluated using
select water quality indicators and targeted organic constituents (e.g., surfactants). Additionally, the
3
|
Colorado School of Mines
|
microbial community and its changes were also analyzed over the course of the study using 16S rRNA
gene amplicon analysis, revealing that the core microbiome in the MBR is made up of a few key
microbial groups that can adapt to varying salinities.
1.3.2 Comprehensive Water Quality and Toxicity Analysis on Complete Treatment Train with
MD
In Chapter 2, the MBR was shown to be a successful pretreatment process for high salinity PW.
However, this is just one part of the treatment process. Treatment steps before and after the MBR in a
complete treatment train must also be evaluated, including final desalination. Water quality analyses,
including toxicity analysis, will best determine the reusability of treated PW.
Chapter 3 of this dissertation involves the evaluation of a complete PW treatment train, from raw
PW to final desalination, with four-unit processes, including an MBR and an MD. Performance of this
treatment train was evaluated by comprehensive environmental water quality analyses that included
testing for biotoxicity, naturally occurring radioactive material (NORM), contaminants of emerging
concern (CEC), heavy metals, organic compounds, and inorganic contaminants. Biotoxicity was reduced
to field blank levels in the MD distillate, while NORM and inorganic constituents, including TDS
concentration, were reduced over 99%. Organic compound concentration analysis conducted by
measuring total organic carbon (TOC) was reduced by 93%, indicating that this small treatment train can
be effective at treating high salinity PW to levels suitable for reuse.
1.3.3 Mini-Pilot System Demonstrating the Effectiveness of a Novel Desalination Process, LSRRO
In Chapter 3, MD was shown to be effective at desalinating high salinity PW to drinking water
standards. However, as mentioned, MD is not currently an option for large-scale commercial applications.
For this reason, a different desalination process, one that can be implemented quickly, should be
considered. Research into viable RO options for desalination of high salinity waters has focused either
reducing the required hydraulic pressure needed to overcome the large osmotic pressure to drive water
through an RO membrane, or on developing an RO module that is capable of operating at elevated
hydraulic pressures, typically three times higher pressure (3,000 psi) than conventional RO (1,200 psi).
In Chapter 4 of this dissertation, a novel RO process, LSRRO, that has been successfully modeled
with theoretical data, is extensively tested with five unique and increasingly complex water streams.1
Recovery as a percent of the initial feed volume, as well as rejection of salts in the final permeate, were
the parameters used to assess this process. A conventional one-pass RO system can typically achieve 50%
recovery while rejecting over 99% of the salts in the feed.6 The results from the LSRRO pilot study
compared favorably to those standards, reaching 69%, 64%, 66%, 53%, and 32% recovery for the five
tested waters. The 32% recovery was during the PW test, which showed, even with several pretreatment
4
|
Colorado School of Mines
|
effective at removing low-density TSS and colloids through physical and chemical processes, but not as
effective at removing dissolved organic carbon (DOC).2, 3 With each additional process added, treatment
becomes more complex and expensive, reducing the ability of oil and gas companies to choose between
water treatment for reuse and disposal. If the treatment of PW is to be adopted for reuse outside the oil
field, the complexity and costs of treatment must be reduced.
One way to achieve cost reduction is by using treatment processes that can remove several
contaminants in one step, thereby shortening the treatment train.4 Successful treatment was achieved in
one study by combining forward and reverse osmosis systems.5 This study also highlighted through life
cycle analysis that pretreatment to remove organic foulants could substantially increase efficiency and
reduce operating costs of treatment. Several processes can remove organic contaminants from water,
including adsorption, chemical oxidation, and biodegradation. Biological processes are usually preferred
because they do not require the use and storage of chemicals or require the disposal of the organics that
have adsorbed to media or precipitated out. However, using biological processes in the treatment of PW
can be challenging. PW total dissolved solids (TDS) concentrations can range from less than 10 g/L to
more than 300 g/L.6 This complicates treatment because biological degradation of contaminants has been
shown to be substantially reduced at TDS levels higher than 10 g/L.7-9 Additionally, biological treatment
systems require a stable environment to perform optimally. When parameters such as temperature, pH,
salinity, or nutrient concentrations are out of the optimal range, the biological system can be negatively
impacted, causing deactivation of the biological community with slow or no potential future recovery.10
The challenges associated with biologically treating PW are substantial; yet, there have been
several studies that have demonstrated success at the bench-scale. A review article on biological treatment
of PW that surveyed 59 published studies found that on average, 73% of chemical oxygen demand (COD)
was removed from PW with TDS concentrations less than 50 g/L.11 One study using biologically active
filtration (BAF) in combination with ultra-filtration membranes (UF) as pre-treatment before desalination
with nanofiltration (NF) was able to remove over 75% of organic contaminants while reducing the fouling
and increasing the efficacy of NF membranes.12 Another BAF study was able to achieve 95% removal of
organic matter.13 These studies not only showed the effectiveness of biological systems in treating
moderate salinity PW, but also highlighted that combining biological and physical treatment technologies
can substantially reduce the number of processes needed.
One successful process that combines biological and physical processes into a single system is a
membrane bioreactor (MBR). In addition to combining multiple processes, an MBR offers other
advantages over other biological treatment systems such as reduced footprint, easy and independent
control of hydraulic retention time (HRT) and solid retention time (SRT), and simple control systems that
allow for automated operation and maintenance (O&M).14 Also, because MBRs have been used to
7
|
Colorado School of Mines
|
successfully treat municipal wastewater for many years, it is a widely accepted commercial process that
can be rapidly implemented.15 These advantages have led to several examinations into treating PW with
different TDS concentrations. Frank et al. (2017) explored the treatment of a combined residential
wastewater and PW stream using a hybrid sequencing batch reactor-membrane bioreactor process (SBR-
MBR) and was able to achieve over 90% soluble COD (sCOD) reduction.16 Two different studies using
the same laboratory-scale membrane sequencing batch reactor (MSBR) system showed removal of total
organic carbon (TOC) at 92% and 91% at TDS concentrations of 16 g/L and 35 g/L, respectively, using
synthetic and real PW.17, 18 In another set of experiments 83% and 95% removal of COD was obtained
from synthetic PW with a TDS of 64.4 g/L and 144 g/L, respectively, using a laboratory-scale MBR.19, 20
These studies were able to show that an MBR at a laboratory-scale can be effective in the
pretreatment of PW; however, they were limited in their overall scope. For example, in three of the four
studies reviewed, synthetic PW was used instead of real PW. Synthetic PW may not be able to accurately
represent the chemistry of real PW as not only does the composition change depending on the location of
the well and how long it has been in production, but many of the reagents that energy companies use
during hydraulic fracturing are proprietary, and therefore not available for use in a synthetic PW. Real
PW also contains native microbes which may be well adapted to the salinity and hydrocarbon content of
PW. Another limitation seen in these studies is the TDS concentrations of the PW treated. TDS
concentrations of real PW can range from 10 g/L to over 300 g/L; yet, only one of the previously
referenced studies tests PW with a TDS concentration over 100 g/L, and that was synthetic PW.19 A third
limitation to these studies is their length and/or scale. Most of the studies found in the literature were done
for a limited timeframe (weeks to a few months) or performed on a bench-scale setup (approximately 5
liters) or both.21-23 And the last limitation these previous studies have is their use of a single PW source,
which may not accurately represent the conditions seen during actual well production, particularly the
organic chemicals, where the characteristics of PW can vary substantially over the lifetime of the well.3
Therefore, the main objective of our study was to evaluate the ability of a small pilot-scale MBR
(bioreactor volume of 70 L) to remove various constituents (e.g., TSS, DOC, nutrients, metals) from PW
during 9 months of continuous operation, using real PW from the Denver Julesburg (DJ) basin that had its
TDS gradually increased to 100 g/L, and culminating in using real PW from the Permian basin with a
natural TDS concentration of 110 g/L. Performance was also evaluated using select water quality
indicators and targeted organic constituents (e.g., surfactants). Additionally, the microbial community and
its changes were also analyzed over the course of the study using 16S rRNA gene amplicon analysis,
revealing that the core microbiome in the MBR is made up of a few key microbial groups that can adapt
to varying salinities. As such, this work presents a unique long-term pilot study of the effectiveness of
biological treatment for moderate and high salinity PW, illustrating an efficient pretreatment process prior
8
|
Colorado School of Mines
|
to desalination (e.g., reverse osmosis (RO) or membrane distillation (MD)).
2.3 Methods and Materials
2.3.1 MBR Feed Water
The PW used in this study was obtained from multiple well sites in the DJ basin located in the
northeastern section of Colorado. The PW was stored in 950 L (250 gal.) water totes at ambient
temperature (~ 20 °C) until fed into the MBR. A 950 L batch of PW from the Permian basin was brought
to the laboratory for experiments with naturally occurring high salinity PW. The water quality of the PW
received throughout this study is summarized in Table 2.1. No pretreatment was performed on the PW
before use.
Table 2.1. Water quality of PW and the dates collected from the DJ-Basin. DJ-Basin PW was
collected from composites obtained from numerous well sites. Permian basin PW was obtained in March
2021 and introduced after DJ-Basin experiments. Most constituents were observed to remain consistent,
with a few outliers seen at each collection date (i.e., DOC concentration fluctuating from 83 mg/L to 207
mg/L).
Permian
Analytes (mg/L) Feb. 8th July 26th Aug. 14th Sep. 27th Dec. 24th
Basin PW
DOC 78 83 207 68 189 71
TN 63 18 114 22 33 455
NH 55 14 103 20 28 405
3
B 27.3 15.1 19.9 20.1 20.3 48
Ba 30.5 5.32 3.59 12.2 10.9 2.6
Ca 990 97.3 150 239 303 4052
Fe 77.4 BDL 0.22 0.865 1.77 14
K 63.1 20.6 22.3 31.9 42.7 1020
Li 7.52 2.37 3.46 4.03 5.07 36
Mg 126.5 18.9 23.0 40.3 42.8 752
Na 10,288 3,486 4,384 5,823 7,012 45,841
P 3.92 1.4 BDL BDL 1.36 0
S 7.95 23.8 14.5 13.0 8.30 27
Si 43.3 73.3 46 104 55.3 13
Sr 263.0 75.7 264 1148 67.3 716
Cl 15,000 6,648 8,300 10,702 13,000 69,659
PO BDL BDL BDL BDL BDL BDL
4
NO BDL BDL BDL 0.8 BDL BDL
3
SO 0.09 60.2 32 31.3 16.5 602
4
Br 26.7 83 46.2 10 128 592
Source TDS 26,800 10,900 14,000 21,420 21,800 111,500
NaCl added 0 16,000 26,000 58,580 78,200 0
Adjusted TDS 26,800 27,000 40,000 80,000 100,000 111,500
2.3.2 MBR System
A schematic drawing of the MBR system is shown in Figure 2.1 (p. 11). Raw PW was held in a
200 L (55 gal) drum and was refilled weekly with fresh PW. The volume of the bioreactor, including
9
|
Colorado School of Mines
|
displacement for the submerged ultra-filtration membrane module and stirring paddle, was 70 L. The
peristaltic pump feeding PW into the continuously stirred MBR was operating at a constant flowrate of 24
mL/min. Aeration of the bioreactor and air scouring of the membrane was accomplished by pumping air
into the membrane aeration port at a rate of 16 L/min using a 5 min on and 2 minutes off cycling of the air
pump (AL-15A, Alita Industries, Inc., Arcadia, CA). Permeate was removed from the bioreactor by
peristaltic pump at a flowrate of 24 mL/min which, when coupled with the 70 L reactor, produced an
HRT of 48 hours. A constant water flux of 2.9 L per m2 per hour (LMH) through the membrane was
maintained during the entire study. Backwashing of the membrane was performed for 20 seconds every
10 minutes at a rate of 300 mL/min. To sustain the slow-growing microorganisms in the MBR, no solids
were removed from the reactor throughout the entire study, resulting in a theoretically infinite SRT. HRT,
water flux, and backwash cycle were chosen based on previous PW biological studies for more
straightforward comparisons.12, 13, 24, 25
The ultrafiltration (UF) membrane used in this system was a submersible Puron® 0.04 µm pore-
size, hollow fiber module with a surface area of 0.5 m2 (Koch Separation Solutions, Wilmington, MA).
Because flux was kept at a constant 2.9 LMH, transmembrane pressure (TMP) was monitored for signs of
membrane fouling. Cleaning was performed on the UF membrane whenever the TMP approached 50% of
the membrane’s maximum filtration TMP of 9 psi, which occurred approximately four times over the
course of the study. The cleaning procedure consisted of acid/base wash cycles, including one hour
backwashing with HCl solution (pH 2), one hour backwashing with NaOH solution (pH 10), another hour
of backwashing with HCl solution, and concluding with a 30 minute backwashing rinse with deionized
water. It is worth noting that in two separate incidents a single membrane fiber physically detached from
the membrane module. In both instances the entire membrane module was replaced.
The reactor was seeded with activated sludge from a municipal wastewater treatment facility.
Activated sludge was acclimated to the high salinity PW by diluting raw PW with dechlorinated tap water
at a starting ratio of 20:1. The fraction of PW in the feed was increased every 48 hours to correspond with
an increase of TDS concentration by 2 g/L until 100% of the feed water was raw PW. To further increase
TDS levels from the average 40 g/L of the raw PW, sodium chloride (Culinox®, Morton Salt, Chicago,
IL) was added to each feed batch at the same acclimation rate as described above to reach the desired
salinity. TDS levels were maintained at 40 g/L, 60 g/L, 80 g/L, and 100 g/L for extended time to evaluate
MBR performance at each of these concentrations.
10
|
Colorado School of Mines
|
Figure 2.1. A flow diagram of the MBR system used in this study. PW and air are pumped into the
bioreactor where they are mixed with the activated sludge. PW is continuously fed into the reactor, which
maintains an average HRT of 48-hours. Treated water is pulled through the UF membrane, keeping all
suspended solids in the bioreactor, and producing a treated permeate stream.
2.3.3 Sampling and Bulk Analytical Procedures
All feed water samples were collected at the point just before the feed water enters the bioreactor.
All permeate water samples were collected after the peristaltic pump that draws the permeate through the
UF membrane. Conductivity and pH were determined using a handheld digital meter with appropriate
probe (HQ40d, PHC10101, CDC40101, Hach Co., Loveland, CO) and conducted once a week. Alkalinity
and ammonia were measured using Hach test vials (TNT 870, TNT 832, Hach Co., Loveland, CO) and
diluted below levels of interference. Analysis for dissolved organic carbon (DOC) and total nitrogen (TN)
(TOC-L, Shimadzu, Columbia, MD) was also performed weekly. Samples for DOC and TN analysis were
filtered through a 0.45 µm polytetrafluoroethylene (PTFE) filter (VWR International, LLC., Radnor, PA),
acidified with concentrated HCl to pH 2, and stored at 3 °C until analysis was performed. Ion
chromatography (IC, ICS-900, Dionex, Sunnyvale, CA) for negatively charged ions and inductively
coupled plasma-atomic emission spectroscopy (ICP-AES, Optima 5300, Perkin-Elmer, Fremont CA) for
positively charged ions were performed monthly. Samples for IC analysis were filtered through a 0.45 µm
PTFE filter and stored at -4 °C until analysis was performed. Samples for ICP-AES analysis were filtered
through a 0.45 µm PTFE filter, acidified with concentrated HNO , and stored at 3 °C until analysis was
3
performed. TDS, MLSS, and MLVSS quantifications were performed according to standard methods
(EPA 160.1, 1684).
2.3.4 Surfactant Analysis
Solid-phase extraction (SPE) was performed on all samples for liquid chromatography and time
of flight mass spectrometry (LC-qTOF) analysis for the semi-quantitative abundance and identification of
11
|
Colorado School of Mines
|
polyethylene glycols (PEGs), polypropylene glycols (PPGs), PEG dicarboxylates (PEG-diCs), and PEG
carboxylates (PEG-Cs). SPE cartridges (Oasis HLB 6cc-500mg 60 µm, Waters Corp., Milford, MA) were
pre-conditioned with methanol and HPLC water.26 First, 5 mL of methanol was vacuumed through the
cartridge at a rate of 5 mL/min followed by 5 mL of HPLC water at 5 mL/min. Next, 10 mL of sample
was pulled through at 5 mL/min. This was followed by 10 mL of HPLC water to flush out any salts that
adhered to the cartridge, at a rate of 5 mL/min. Elution of the samples was performed using 10 mL
methanol at a rate of 1 mL/min. The eluted samples were then concentrated down to 1 mL by a gentle
stream of N gas (XcelVap, Biotage, Uppsala, Sweden). Samples were then pipetted into 2 mL amber
2
vials and stored at -4 °C until analyzed.
Analysis was non-targeted and conducted on a SCIEX X500R QTOF (Framingham, MA) using
high resolution liquid chromatography. The operating parameters of the LC-MS were obtained from
published methods for PEG, PPG, PEG-diCs, and PEG-Cs identification in PW.27 All organic solvents
used throughout this analysis were of HPLC grade or higher (Sigma-Aldrich Corp., St. Louis, MO). For
quantification of these compounds, their hydrogen, ammonium, and sodium adducts were extracted from
samples and analyzed on the SCIEX OS Analyst Software (Framingham, MA). The peak areas of each of
these adducts were summed to give a semi-quantitative concentration of that surfactant.26, 28 Removal
percentage was then determined using the following equation:
(2.1)
𝐶𝐶0−𝐶𝐶(𝑡𝑡)
𝑅𝑅% = ∗100
where is the removal percentage,𝐶𝐶 0 is the relative abundance (defined at the counts per
second intensit 𝑅𝑅y %(cps)) of a specific PEG in the
𝐶𝐶
0feed water, and is the relative abundance of the same
PEG in the permeate. same PEG. 𝐶𝐶(𝑡𝑡)
2.3.5 Microbial Community Analysis: DNA Extraction and 16S rRNA Gene Amplicon
Sequencing
Feed water and sludge samples were collected for analysis at regular intervals during the study.
Sludge samples were collected in 50 mL sterile tubes. Feed water samples (50-100 mL) were filtered
through a Sterivex™ filter (0.22 um, PES filters, Millipore-Sigma, MA). Feed and MBR permeate were
shipped on ice to LBNL for further analysis. Sludge and feed solids collected on the filters were shipped
on dry ice to LBNL, where the samples were stored at -80 °C until DNA extraction. Triplicate 1.5 mL
aliquots of MBR sludge collected at each time point were centrifuged at 10,000 x g for 5 mins, followed
by DNA extraction of the pellet with DNeasy PowerLyzer PowerSoil kit (Qiagen, Hilden, Germany) as
per manufacturer’s instructions. Sterivex filters containing feed microbes were extracted using DNeasy
PowerWater Sterivex kit (Qiagen, Hilden, Germany).
12
|
Colorado School of Mines
|
16S rRNA gene amplicons were amplified from DNA extracts using 515F and 806R primers
targeting V4 hypervariable region, followed by PCR-free short library preparation and sequencing on
Novaseq 6000 (Illumina, PE250) at Novogene Corporation Inc. Sequencing reads were processed in
QIIME2 v. 2020.8.29, 30 Specifically, reads were demultiplexed, quality filtered and denoised with
DADA2.31 Taxonomy was assigned to the amplicon sequence variants (ASVs) using a naive Bayes
taxonomy classifier trained on SILVA 138 99% OTUs from 515F/806R region of 16S rRNA gene
sequences.32, 33 The ASV table generated was then manipulated in R to remove singletons, perform
statistical analyses and generate plots using phyloseq package34, 35 The raw sequencing reads for MBR
sludge and feed samples are deposited at NCBI SRA under the BioProject PRJNA768964.
2.4 Results and Discussion
2.4.1 Removal of Organic and Inorganic Contaminants
DOC and TDS concentrations in the feed and permeate streams of the MBR for the entire study
are shown in Figure 2.2 (p. 14). DOC concentration in the permeate stream remained relatively constant
at ~12 mg/L over the almost 10-month testing period, even as TDS concentrations in the feed increased
from 27 g/L at the start to 100 g/L by the end; all while the feed DOC concentration fluctuated between
30 and 170 mg/L. These results are in contrast to studies that observed a marked decline in biodegradation
of organic matter when the salinity of the feed water was increased.36, 37 The MBR average effluent DOC
concentrations were 12.3 mg/L, 6.2 mg/L, 10.5 mg/L, and 10.3 mg/L at bioreactor salinities of 40 g/L, 60
g/L, 80 g/L, and 100 g/L TDS, respectively, which translates to 90%, 82%, 87%, and 89% DOC removal
over these time periods, respectively. These results compare favorably with several other studies
involving biological treatment of PW. Freedman et al. (2017) demonstrated DOC removal of 95% using
biologically active filtration (BAF) treating PW with TDS concentrations of ~20 g/L TDS, while Riley et
al. (2016) were able to achieve over 75% DOC removal using a similar BAF system.12, 13 Pendashteh et al.
(2012) observed 91% total organic carbon (TOC) removal with the use of a laboratory-scale MSBR
treating PW with TDS concentrations of 35 g/L, and Frank et al. (2017), using a pilot-scale hybrid
sequencing batch reactor-membrane bioreactor, were able to remove over 90% sCOD from residential
wastewater that was dosed with 6% PW.16, 18
While the MBR permeate DOC concentration was relatively constant, the feed DOC
concentration declined over time for each of the water batches acquired for the study. This in turn affected
the calculated percent removal of DOC over time, with lower feed concentration resulting in lower
percent removal of DOC. The changes in influent DOC can be attributed to the slow degradation of
organic matter in the PW storage tanks—the longer the PW was kept in the totes after collection from the
O&G wells, the more DOC concentrations declined, including in high salinity raw PW.
13
|
Colorado School of Mines
|
240 120
Feed Permeate TDS
200
160 80
120
80 40
40
0 0
56 84 112 140 168 196 224 252 280 308 336
Day
Figure 2.2. DOC concentration of the feed and permeate of the MBR beginning after the acclimation
period (day 56 of operation). DOC concentration is given by the primary Y-axis, with feed represented by
blue triangles and permeate represented by orange circles. The solid red line corresponds with the
secondary Y-axis to show the gradual increase in TDS concentration from 27 g/L to 100 g/L.
As expected, the MBR’s DOC removal is heavily dependent on the DOC concentrations in the
PW used in this study. Throughout this study, when influent DOC levels exceeded 100 mg/L, percent
removal averaged 91% at all TDS concentrations. Considering the relatively long HRT of 48 hours in the
bioreactor, most of the labile organic compounds were likely degraded, leaving behind biologically
recalcitrant organic compounds, which remained in the permeate and consisted of an average of 10.4
mg/L DOC throughout the study. It is possible that with the addition of supplemental nutrients, the
microorganisms would perform better and reduce the permeate DOC concentration even further. Nicholas
et al.,38 using a similar sequencing batch reactor (SBR), were able to further reduce sCOD concentrations
in PW by an additional 20% with the addition of phosphorus at a level of 7.5 mg-P/L. Since the
completion of this study, additional testing with supplemental phosphorus has been performed in the
MBR, with preliminary results showing no additional DOC removal. It is likely that despite the addition
of phosphorus, the lack of additional DOC removal is due to the very high PW salinity in this study.
A comparison of the concentrations of inorganic constituents between the PW feed and MBR
permeate observed for bioreactor salinities ranging from 40 g/L to 100 g/L is summarized in Table 2.2 (p.
15). Except for iron, little to no reduction was observed in any of these constituents. Riley et al. reported
similar results using a BAF and NF treatment processes on similar PW.4 The reduced iron concentrations
are most likely due to oxidation from aeration in the reactor, causing the iron to precipitate out. While no
removal was observed, Table 2.2 does show the substantial variability in the concentration of constituents
typically seen in PW over time. This characteristic of PW has been well documented in previous studies
and again highlights the challenges associated with PW treatment.3
14
)L/gm(
COD
)L/g(
SDT
|
Colorado School of Mines
|
Table 2.2. Average feed and permeate inorganic concentrations throughout the study. Little to no
reduction in inorganics was observed during this study. However, the accumulation of these ions in the
bioreactor over the 10-month study did not hinder organic biodegradation. Phosphate and nitrate levels
were below detection limits.
Analyte (mg/L) MBR Feed MBR Permeate
TN 50±23 48±26
NH 45±21 40±18
3
B 21±1.5 20±2.7
Ba 11±4.3 9±3.3
Ca 222±51 232±74
Fe 2±1.1 0.1±0.14
K 47±18 46±27
Li 5±1.3 4±1
Mg 34±7.2 33±6.6
Na 24,198±11,152 25,025±11,651
P 1±0.6 1±1.2
S 15±10 27±11
Si 41±6.5 49±19
Sr 50±17 44±5.2
F 4±2.5 2±1
Cl 38,893±14,178 38,066±13,882
Br 123±29 117±30
SO 19±11 24±10
4
2.4.2 Removal of Targeted Organic Compounds
The MBR’s ability to remove targeted organic chemicals was also evaluated. Several surfactants
were targeted that are commonly found in PW, which included PEGs, PPGs, PEG-diCs, and PEG-Cs.26, 27
PEGs and PPGs are used as surfactants to enhance recovery of O&G, and they can remain in PW after it
is brought to the surface for over a year.39 Several PEGs and PPGs were identified in the PW with the use
of LC-qTOF analysis and evaluated over the course of the study. Identification of PEGs is determined by
the number of ethylene oxide (EO) units, where PEG-EO6 have 6 ethylene oxide units, PEG-EO7 have 7,
and so on. The average mass difference between these different PEGs is 44.0262 mass units, which
corresponds to the addition or subtraction of an ethylene oxide unit [-CH -CH -O-]. PPGs have a
2 2
difference of mass of 58.0419 mass units, which corresponds to the addition or subtraction of a propylene
oxide unit [--CH -CH(CH )-O-]. For convenience, EO6 refers to all surfactant compounds with 6
2 3
additional units, i.e., PPG-PO6 will be referred to as “EO6”. Quantification of volatile and even semi-
volatile chemicals through the MBR is very difficult due to the inability to accurately identify if removal
from the PW was due to air stripping or microbial degradation, as demonstrated by Sitterley et al.40 As
such, we opted to evaluate the non-volatile chemicals and target the surfactants present in the raw vs.
MBR treated PW.
Contrary to DOC removal, it appears that TDS concentration did have a negative impact on the
15
|
Colorado School of Mines
|
MBR’s ability to degrade PEG, PPG, PEG-diCs, and PEG-Cs surfactants. Kawai showed that PEGs are
aerobically metabolized first through the oxidation of a PEG compound to a carboxylated PEG.
This is done through the microorganism’s use of alcohol and aldehyde dehydrogenases enzymes.
In a second step, the terminal ether bond is cleaved, reducing the PEG by one glycol unit.41 Figure
2.3 (p. 17) illustrates the negative impact TDS concentration had on biodegradation of PEGs by using the
relative abundances obtained from LC-qTOF analysis to show the removal percentage from feed to
permeate for PEGs, PPGs, PEG-Cs, and PEG- diCs of size E06-E09 in PW with TDS concentration of 40
g/L and 100 g/L. The percent removal is semi-quantitative because relative abundance was used. This is
due to individual PEG standards not being easily obtained or readily available. Additionally, this semi-
quantitative method was used for this study because, as Rosenblum et al. reported, “..a quantitative
measure would be challenging due to matrix-induced ionization effects and specific response factors for
these types of compounds, relative abundance was used as a way to compare these compound levels over
time”.39, 42 As shown in Figure 2.3 (p. 17), PEG removal averaged 87% when TDS concentrations were 40
g/L but only 58% when TDS concentrations were at 100 g/L. This trend was also observed with PPG
removal, which averaged 81% at 40 g/L but averaged only 2.3% removal at 100 g/L; and also with PEG-
Cs, where removal averaged 96% at 40 g/L but dropped down to 67% at 100 g/L. These results are
possibly due to two previously described phenomena. The first is that biodegradation of PEGs is
compromised in salty environments. Bernhard et al. showed that short-chain PEGs, while remaining
completely biodegradable, require a much longer time of treatment in saline environments compared to a
freshwater environment.43 In artificial seawater (TDS of 35 g/L), short-chain PEGs did not fully
biodegrade until after 37 days of treatment. The second phenomena that might explain these results is the
biodegradation of other ethoxylated additives present in the PW that have a mass that falls outside of the
mass range analyzed in this study. Sitterley et al. demonstrated the presence of these compounds in PW
(same source that was used in the current study) and that their aerobic biodegradation can lead to the
formation of straight-chain PEGs.26 McAdams et al. showed how alkyl ethoxylates used in fracturing
fluid transform to PEGs through cleavage of the alkyl group from the polyethoxylated chain as a result of
aerobic biodegradation.44
The inability to remove surfactants from PW with high TDS levels is a concern and additional
treatment steps may be necessary if treated PW is to be reused in applications that require these residual
surfactants to be removed to a greater degree. For every TDS level, PEG-diCs concentration increased
after MBR treatment. This is due to PEGs biodegrading first to singly and then to doubly carboxylated
metabolites during treatment as previously described by Sitterley et al.40
16
|
Colorado School of Mines
|
Figure 2.3. Heatmap of removal percentage of selected PEGs, PPGs, PEG-diCs, and PEG-Cs
remaining in the MBR permeate at TDS concentrations of 40 g/L and 100 g/L. Lighter colored boxes
represent higher removal percentage, with increasingly darker boxes representing lower removal
percentages.
Overall, these results illustrate the variable nature of constituent removal, and shed some light
into the 10.4 mg/L of remaining DOC present in the PW studied. Unfortunately, without the ability to
obtain individual PEG standards (e.g., an analytical HPLC standard of PEG-E06), only semi-quantitative
analysis was performed.45 However, these compounds are only expected to make up a small percentage of
the remaining DOC, as shown by Thurman et al.46 Regardless, the persistence of these chemicals in the
permeate suggests that additional treatment processes may be needed for complete removal. For example,
research on PW has shown the removal of surfactants through other processes, like activated carbon,
which could be utilized in a water reuse treatment train.47
2.4.3 Transition to Treatment of Permian Basin
PW characteristics can vary dramatically depending on the source basin. To test the MBR’s
ability to treat naturally occurring high salinity PW, once all experiments had been completed using DJ-
Basin PW, Permian basin PW was fed into the MBR for one month. The characteristics of this new feed
stream are summarized in Table 2.1 (p. 9). Because the microorganisms in the MBR had already been
acclimated to a TDS concentration of 100 g/L, no acclimation period was given for the Permian PW that
had 110 g/L TDS. The performance of the MBR, during days 338-368 of operation, in removing DOC
from Permian basin PW is shown in Figure 2.4 (p. 18). Like treatment of DJ basin PW, consistent DOC
concentrations in the MBR permeate were observed over the entire one-month study period (average of
17.5 mg/L vs. 10.4 mg/L when operating with DJ basin PW).
17
|
Colorado School of Mines
|
80
Permian PW MBR Permeate
70
60
50
40
30
20
10
0
338 343 348 353 358 363 368
Day
Figure 2.4. DOC concentrations of feed and permeate, during days 338-368 of operation, using
Permian PW. No acclimation period was used with the Permian PW being fed into the MBR immediately
after the DJ basin study. The TDS concentration of the Permian PW was naturally 110 g/L. Additionally,
this PW contains much higher concentrations of several other constituents than the DJ basin PW, such as
ammonia, calcium, and potassium.
While consistent, DOC removal from the Permian basin water was lower compared to DJ basin
water (66% vs. 86%). This was due to the combination of the lower DOC concentration found in Permian
PW (average of 54 mg/L vs. 76.2 mg/L in the DJ water) and the lower average DOC concentrations of the
treated DJ basin PW. The higher DOC concentrations observed in the permeate during the 31-day period
of testing with Permian basin PW feed may be due to one or a combination of several factors. As shown
in Table 2.1 (p. 9), the PW from the Permian basin contained substantially higher concentrations of many
inorganic constituents compared to DJ basin PW, including ammonia, bromide, calcium, iron, potassium,
magnesium, lithium, and sulfate. These substantial increases, without the benefit of an acclimation period,
might have shocked the microorganisms in the bioreactor, creating a less than ideal environment that
hindered their ability to biodegrade the DOC in the feed. Additionally, the amount of recalcitrant organics
in the Permian PW may be naturally higher than that found in the DJ basin PW, leading to overall higher
concentrations remaining in the permeate.
Upon conclusion of the experiment with Permian basin PW feed stream, operation of the MBR
was resumed with DJ basin PW feed having an artificially raised TDS concentration of 100 g/L. This
TDS concentration was maintained for the next 6 weeks. At that point, in an attempt to further test the
robustness of the MBR to maintain performance during a sudden and substantial change in salinity, the
TDS concentration was reduced to 75 g/L without any acclimation period. This procedure was repeated 6
weeks later when the TDS concentration was reduced to 50 g/L without an acclimation phase. The final
phase of this experiment involved the raising of the TDS concentration from 50 g/L to 100 g/L with no
acclimation period. The DOC removal percentage results observed were 77%, 85%, and 84%.
18
)L/gm(
noitartnecnoC
COD
|
Colorado School of Mines
|
2.4.4 Microbial Community Analysis
The diversity of the microbial community in MBR sludge declined dramatically during the
acclimation phase, as indicated by the decrease in Shannon index and increase in abundance of a few key
taxa (Figure 2.5, p. 20). This is likely attributable to the inability of bacteria originating from the
inoculum (activated sludge) to adapt to the increased salinity. The core community composition of the
MBR was relatively stable beyond the acclimation phase, with a few persistent taxa constituting greater
than 50% of the reads. Specifically, members of the genera Roseovarius and Iodidimonas first became
established in the reactor when it was fed with 100% PW, followed by other genera like Rehaibacterium,
Methylophaga, and unclassified Rhodobacteraceae, which together constituted close to 80% of the reads
when the reactor was fed with 100% PW in the salinity range of 40-70 g/L. Both Roseovarius and
Iodidimonas have been reported/isolated from PWs/hydrocarbon contaminated environments from around
the world; however, experimental evidence for hydrocarbon degradation by Roseovarius and Iodidimonas
isolates is not available.48-51 Isolates belonging to these genera have salinity tolerance in the range of 5-
100 g/L.52, 53 Roseovarius and Iodidimonas have been identified as Iodide Oxidizing Bacteria (IOB)
because they can produce iodine from iodide in the presence of oxygen.48, 54 A wide range of salt
tolerance, coupled with the reported bactericidal properties of iodine, may allow these genera to colonize
the reactor first by inhibiting susceptible bacteria.9 Rehaibacterium is a newly described genus with one
reported isolate so far, but a recent study correlated this genus with DOC removal in a BAF treating PW
from Sichuan Basin, China.55, 56 Members of genus Methylophaga are primarily known for metabolism of
C1 compounds; however, one isolate has been reported to degrade alkanes (n-hexadecane).57, 58
As the salinity of the feed was further increased beyond TDS of 80 g/L, there was an increase in
diversity, which coincided with an increase in abundance of other genera, including Marinobacter,
Malaciobacter, and Marinobacterium. Marinobacter and Malaciobacter (formerly classified as
Arcobacter) were predominant groups in early flowback period natural gas brines from Utica and
Marcellus Shale, and isolates belonging to these genera demonstrated salinity tolerance up to 150 g/L.59
Marinobacterium has been detected as a dominant genus in several PWs, and studied isolates have a wide
salinity growth range (5-180 g/L).49, 50, 60, 61. Because the microbial community data is in the form of
relative abundance, it is unclear if increase in diversity results from an absolute increase in these
populations better adapted to higher salinities or a decrease in the dominant reactor populations. We
therefore hypothesize that this increased diversity could be related to out competition of the dominant
populations by produced water-derived populations better adapted to higher salinities coupled with the
reactor design that results in retention of all microbial cells, even those that are not proliferating in the
reactor.
19
|
Colorado School of Mines
|
Figure 2.5. Microbial community diversity in MBR sludge. (a) Shannon diversity index and (b)
relative abundance of dominant genera based on 16S rRNA gene amplicon sequencing reads. Sludge
samples collected at each timepoint were processed as three technical replicates. Genera with relative
abundance greater than 2% at any time point are shown and the remaining minor taxa are grouped
together as “Others.”
Additionally, Roseovarius, Iodidimonas, Rehaibacterium, and Marinobacterium have been
reported as dominant genera in the effluent of an aerated BAF treating DJ-Basin PW, indicating that these
microbes can consistently grow in DJ-Basin PW under aerobic conditions.62
In contrast to the sludge community, the microbial community of the feed had higher diversity
and included several nitrate and sulfate-reducing bacteria like Denitrovibrio, Desulfotignum,
Desulfomicrobium, Desulfovibrio, and Desulfuromonas. The dominant and prevalent genera in feed
samples were Marinobacterium and Malaciobacter. The dominant genera found in the MBR sludge
(Roseovarius, Iodidimonas, Rehaibacterium, Methylophaga, unclassified Rhodobacteraceae and
Marinobacter) are also present in the feed at relatively low read abundances, indicating these organisms
likely derive from the PW rather than the sludge inoculum.
In contrast to the feed, aerobes dominate in the aerated sludge. Thus, a combination of aeration,
salinity, and addition of PW shapes the evolution of the microbial community in the MBR from the
conventional sludge inoculum to a community adapted to PW treatment (Figure 2.6, p. 21). The microbial
community in the MBR sludge falls into two major clusters in a principal coordinates analysis (PCoA)
plot—one at TDS of 29-80 g/L that shows little similarity to inoculum or feed, and another at higher
salinities (80-100 g/L TDS), which is closer to the community of the feed.
While the reason for this is not entirely clear, relative abundance at phylum level of sludge and
20
|
Colorado School of Mines
|
water (PW); and there is a growing interest in reusing PW outside the oilfield, particularly in water scarce
regions in the U.S. such as Texas and New Mexico.1 An estimated 490,000 acre-feet (1.98B m3) of PW
were generated in the Permian Basin in 2019, and these volumes will continue to rise in the future as
exploration and production activities in the region are increasing.2 This volume of water could help
alleviate some of the region’s water scarcity issues. However, before this water resource could be utilized,
adequate treatment and testing are required given that PW can have high concentrations of oil and grease,
suspended solids, microorganisms, particulate and dissolved organic contaminants (e.g., natural organic
matter (NOM), biocides, corrosion inhibitors, friction reducers), heavy metals, total dissolved solids
(TDS), and naturally occurring radioactive materials (NORM), all of which are challenging from both
water treatment and environmental health perspectives.3
Many studies have looked at singular treatment processes targeting specific classes of chemicals
in PW; however, only a few studies to date have investigated multi-barrier treatment trains to achieve a
substantial reduction in organic and inorganic contaminants present in PW. Riley et al., using biologically
active filters (BAF) and closed-circuit desalination with nanofiltration (NF) and reverse osmosis (RO)
membranes, were able to achieve 99.6% TDS and 89% dissolved organic carbon (DOC) reduction from
raw PW.4 Maltos et al. reported over 99% rejection of all measured ions and over 95% rejection of
hydrocarbons with a pilot-scale hybrid forward osmosis-RO system.5 While these studies showed
promising results, they were conducted with PW that had relatively low salinity (~30 g/L). Further
research is needed on multi-barrier treatment trains that can be scaled up to handle the large volumes of
high-salinity PW generated while being robust enough to maintain effective treatment of variable PW by
each unit process throughout the treatment train.
The evaluation of PW and the treatment processes can be difficult and expensive.6 While
conventional analytical methods such as pH, alkalinity, TDS, dissolved organic carbon analysis (DOC),
cation and anion analysis, and traditional water and wastewater chemical indicators (e.g., volatile and
semi-volatiles) are used for general water quality determination, PW and its numerous unknown
constituents cannot be evaluated by these methods alone.7 In a recent literature review, almost 1200
different chemicals were identified and prioritized illustrating the vast number of chemicals potentially
present in PW.8 Detection and quantification of these chemicals require more sophisticated analytical
instruments such as gas chromatography (GC), liquid chromatography (LC), and mass spectrometry
(MS); and necessitate the development of unique analytical methods due to the many constituents found
in PW at variable concentrations. For example, high salinity PW can damage delicate analytical
equipment, requiring substantial dilutions of samples. This, in turn, raises the detection limits of these
methods, which can lead to chemicals being not detected even though they are likely present in the
undiluted sample.
28
|
Colorado School of Mines
|
Identification and quantification of chemicals in PW can be challenging as detailed above, which
is compounded by PW chemicals often lacking analytical standards, preventing quantification. As such,
alternative approaches to identifying the presence of potentially harmful chemicals are required. One such
approach is using toxicity testing to determine whether the mixture may cause a harmful effect.6 To date,
several studies have evaluated untreated PW toxicity using in-vitro bioassays and aquatic organisms. Hull
et al. observed significant cytotoxicity in PW from the Denver-Julesburg basin in Colorado via
BioLuminescence over a 220-day period.9 McLaughlin et al. studied the acute toxicity on Daphnia magna
from a minimally treated PW discharged into a stream used for agriculture and livestock and found
several chemicals above their MCL thresholds at the discharge site, including known carcinogens.10, 11
Other studies have used Daphnia magna, such as Boyd et al. whom investigated the toxicity of flowback
water from the Montney Formation in Canada.12 Azetsu-Scott et al., using Microtox® tests, found that
toxicity increased over time following discharge of PW from offshore oil production installations due to
the precipitation of heavy metals present in PW.13 Overall, there is a lack of research on toxicity of treated
PW.
This paucity of literature to date impacts our ability to perform risk assessments to evaluate
mixture toxicity and know what chemicals or toxicological end-points to look for through a PW treatment
train. Danforth et al. were unable to find safety evaluations or toxicological studies on 56% of the 1198
chemicals detected in PW, with 85% of these chemicals lacking enough data to perform a risk
assessment.8 Additionally, even less is known about how PW chemicals interact among themselves and
how such interactions affect overall PW toxicity as it moves throughout a multi-step treatment train. It is
not just the presence of toxic chemicals in PW that is of concern, but also what could form during
treatment. An example of this was demonstrated by Almaraz et al. whom found that BAFs used as
pretreatment of PW before NF membranes led to the formation of several halogenated-disinfection
byproducts.14 Additionally, select studies have investigated PW following some level of treatment, with
Mehler et al. observing reduced toxicity to Lumbriculus variegatus with PW that had been treated with
activated carbon to remove organic compounds.15 However, this was done to determine any possible
correlation to organics concentrations in PW and inorganic toxicity to worms, not to test the individual
treatment processes. Yang et al. studied the effects of using treated PW for the germination and irrigation
of wheat.16 However, the PW tested was the final product water from a six-stage wastewater treatment
facility; water from the individual processes was not studied.
To further understand the challenges associated with high salinity PW treatment, including the
ability to reduce various chemicals (inorganic and organic) and specific toxicological end-points (e.g.,
cytotoxicity and aryl hydrocarbon receptor (AhR) activation), a treatment train consisting of four separate
unit processes was selected. In this study, comprehensive water quality analysis at each treatment stage
29
|
Colorado School of Mines
|
was performed, from raw PW through desalination. The change in specific toxicological endpoints in PW
after each treatment process was also analyzed. As such, this work presents a unique framework for a
more thorough understanding of PW, its risks to environmental health, and effective treatment processes
to mitigate those risks and promote water reuse.
3.3 Methods and Materials
3.3.1 Produced Water Characteristics
Produced water from the Permian Basin was collected in a 1000 L tote and transported to our
laboratory in a non-climate-controlled trailer. The general quality of the raw PW was analyzed in our
laboratory and the results are summarized in Table 3.1.
Table 3.1. Initial water quality of produced water. Water was obtained from the Permian Basin and
stored for one month before it was used in this study. BDL=Below detection limit.
Analyte (mg/L) Analyte (mg/L)
Dissolved organic carbon (DOC) 79.7 Sulfur (S) 349
Total nitrogen (TN) 396 Silica (Si) 19.8
Ammonia (NH +) 389 Strontium (Sr2+) 1,009
4
Boron (B+) 56.2 Bicarbonate (HCO–) 248
Barium (Ba2+) 2.04 Chloride (Cl–) 69,672
Calcium (Ca2+) 4,959 Phosphate (PO 3–) BDL
4
Iron (Fe3+) 12.7 Nitrate (NO –) BDL
3
Potassium (K+) 651 Sulfate (SO 2–) 431
4
Lithium (Li+) 16.9 Bromide (Br–) 389
Magnesium (Mg2+) 687 Iodide (I–) 52.0
Phosphorous (P) 1.0 pH 6.5
Sodium (Na+) 39,994 Total dissolved solids (TDS) 119,000
3.3.2 Treatment Train
A treatment train consisting of four-unit processes was chosen for this study. Chemical
coagulation was the first process. This was followed by a membrane bioreactor (MBR) and adsorption
columns consisting of granular activated carbon (GAC) and ion exchange resin (IX). The final process
was desalination through membrane distillation (MD).
3.3.3 Coagulation/Flocculation
Coagulation/flocculation was used for the removal of particles and suspended solids. Aluminum
chloride hexahydrate (AlCl •6H O, Sigma-Aldrich, St. Louis, MO) was used as the coagulant. Jar testing
3 2
was conducted to determine the optimal concentration of coagulant, which was found to be 10 mg/L.
Coagulant was applied, at 10 mg/L to the 1000 L tote of raw PW. The water was then pumped into the
first stage of a 3-stage tapered flocculation system operating at a hydraulic retention time (HRT) of 10
30
|
Colorado School of Mines
|
min/stage (velocity gradient, G=80, 40, and 20 s−1), followed by 30 min settling time.50
3.3.4 MBR
After coagulation, the PW was fed into an MBR to further remove suspended solids and reduce
organic content. The operating parameters of the MBR have been described in our previous publication.18
Briefly, prior to the beginning of this study, the MBR was acclimated to the high TDS concentration of
the Permian Basin PW. The 70 L, lab-scale MBR system was constantly stirred and intermittently aerated
and was operated for several months under these conditions. Water was fed into the bioreactor at a
constant flowrate of 24 mL/min (~35 L/day). A hollow fiber ultrafiltration (UF) membrane with a 0.04
µm nominal pore-size and 0.5 m2 surface area (Puron® Koch Separation Solutions, Wilmington, MA) is
submersed in the bioreactor. An air pump is connected to the UF membrane module for air scouring of
the membrane fibers and aeration of the MBR. The air flow rate was 16 L/min and was turned on for 5
minutes and then off for 2 minutes, for continuous 7 minute on/off cycle. Water flux through the
membrane was kept constant throughout this study at 2.9 L per m2 per hour (LMH). A second peristaltic
pump pulled permeate through the membrane at the same 24 mL/min flowrate, resulting in an HRT of 48
hours. After every 10 minutes of permeation, membrane backwashing was performed for 20 seconds with
a flowrate of 300 mL/min. No solids were removed from the MBR during this study, resulting in an
infinite sludge retention time (SRT).
3.3.5 GAC/IX Adsorption Columns
MBR permeate was pumped into a column containing GAC (CS-HAC A8009-14, Clack Corp.,
Windsor, WI) followed by a column containing IX (DOWEX MAC-3, Dow Chemical Co., Midland, MI)
connected in series. The purpose of this process was to remove residual organic compounds and scaling
agents by adsorption before the final desalination process. The columns were 1-inch diameter PVC pipes
and filled to a height of 19 inches with media for a total bed volume of 14.9 in3 (244.5 cm3). Flowrate into
the columns was kept the same as the flowrate of the MBR (24 mL/min). This resulted in an empty bed
contact time (EBCT) of 10 mins, which is over three times the manufacturer’s minimum recommended
EBCT of 3 minutes.
3.3.6 Membrane Distillation
The final treatment process implemented was direct-contact membrane distillation (DCMD).
Feed water was heated to 50 °C while the distillate stream was held at 20 °C, and the two streams were
flowing concurrently through the DCMD membrane cell on the opposite side of the membrane. An
elongated polytetrafluoroethylene (ePTFE) hydrophobic membrane (CLARCOR Industrial Air, Overland
Park, KS) with a thickness range of 127-203 µm, a porosity range of 70-85%, and a nominal pore size of
0.45 µm was used for this study. This membrane was selected due to its previous published literature for a
31
|
Colorado School of Mines
|
similar MD application.19 MD testing was performed using an acrylic flow cell. The cell can hold a
membrane with an area of 193 cm2 and flow channel dimensions of 254 mm length, 76.2 mm width, and
2.5 mm depth. The flowrates of both the feed and distillate streams were kept at 2 L/min. Feed and
distillate conductivities were measured to determine water recovery and membrane integrity during
desalination.
3.4 Water Quality Analysis
Comprehensive water quality analysis was performed on the raw and treated PW. Additionally,
toxicological tests were conducted in the human breast cancer cell line MCF-7 using an MTS tetrazolium
bioassay for cytotoxicity and assessment of the expression of HMOX1, CYP1A1 and CYP1B1 genes as
indicators of AhR activation.
3.4.1 Organic and Inorganic Chemical Analysis
The treatment train, except for the MD, was operated for a two-day period, and samples were
collected from each of the treatment processes, including raw PW. The MD, due to its distillate
production capacity and the volumes required for analysis, was operated for 4.5 days. Additionally, one
field blank and one field duplicate were collected for each analysis. The analyses are summarized in
Table 3.2. All samples were stored at 4 °C and shipped to meet their respective holding times on ice to the
individual EPA laboratory performing the analysis. Gamma-ray spectrometry and Radium-226 analysis
were performed by the USEPA’s National Analytical Radiation Environmental Laboratory (NAREL).
Table 3.2. EPA testing methods used for each analysis.
Analysis EPA Method
Dissolved organic carbon (DOC) 415.2
Total organic carbon (TOC) 415.1/SM 4
Dissolved metals 200.8/6020
Total metals 200.8/6020
Volatile organic carbon (VOC) 8260B
Semi-volatile organic carbon (SVOC) 8270D
Diesel range organics (DRO) 8015
Gasoline range organics (GRO) 8021/8015
Anions 300.0
Naturally occurring radioactive material (NORM) 901.1/9320
MCF-7 cytotoxicity MTS tetrazolium assay
AhR activation qPCR analysis of CYP1A1, CYP1B1
3.4.2 Evaluation of Cytotoxicity
For cytotoxicity determination, an MTS tetrazolium assay was chosen due to its simple operation,
fast reaction times, better accuracy, and good repeatability.20 This assay is based on tetrazolium salts
being reduced by the mitochondrial activity of viable cells into a dark colored, soluble formazan
32
|
Colorado School of Mines
|
product.21 The color of the formazan directly correlates with the number of viable cells present.22 After 24
hrs of exposure of cells to the water, 10 μL of CellTiter AQueous One MTS solution (Promega, Madison,
WI) was added, and after 3 hrs the optical density values were measured at 490 nm.
3.4.3 Concentrating Samples
Samples were extracted and concentrated following the methods of Sitterley et al., 2020.23 In
brief, 100 mL water samples were concentrated on a 6cc/250 mg Waters Oasis HLB solid phase
extraction cartridge (Milford, MA) using a vacuum manifold. SPE cartridges were precleaned (5 mL
methanol) and conditioned (5 mL DI water) before use. Desalting of SPE cartridges (20 mL DI water)
was followed by centrifugation (5 minutes at 2,700g) to remove residual water on the cartridge. SPE
cartridges were then eluted with methanol (5 mL). To obtain sufficient extraction volume, this method
was repeated twice (200 mL water total) and the extracts combined (10 mL methanol). An aliquot (200
μL) was sampled for non-targeted analysis (NTA) via high resolution mass spectrometry (HRMS) and the
remainder evaporated to dryness under a gentle nitrogen gas stream at 40 °C using a Zymark TurboVap
LV (Biotage, Upsala, Sweden). The condensate was then resuspended in DMSO (400 μL) with vortex
mixing (2 minutes) and sonication until fully dissolved.
3.4.4 Cell Culture and Treatment
MCF-7 wild-type and AhR-null cells (Synthego Inc, Redwood City, CA) were cultured in
DMEM media (ThermoFisher, Durham, NC) supplemented with 10% FBS, 1 mM glutamine, 1 mM
sodium pyruvate (Sigma, St. Louis, MO), and 1x Antibiotic-antimycotic (ThermoFisher,Waltham, MA).
Cells were plated at 8 x104 cells per well in 24-well plates or 2 x104 in 96-well plates. After 22 h culture
when cells reach 80% confluency, media was replaced with dosing solutions containing aliquots from
500x concentrates to make final concentration of 1x, 2x, and 5x. DMSO (0.01%) was used as control.
Raw water treatment medium was prepared from DMEM powder (50 mL raw water containing 675 mg
DMEM powder, 185 mg NaHCO , 500 µl of 100 mM sodium pyruvate, and 500 µl of Antibiotic-
3
antimycotic as 100%), and then diluted with culture medium to 3.13, 6.25, 12.5, 25, 50, and 75%. For
cytotoxic determination, cells were treated for 24 h followed by the MTS assay. For gene expression
studies, cells were treated for 6 h.
3.4.5 Evaluation of Gene Expression by RT-qPCR
After 6 hrs of exposure, media was removed, and cells were lysed in 0.3 mL Trizol, followed by
RNA extraction and quantified at 260/280 nm in a Nanodrop 2000 spectrophotometer (Thermo Fisher,
Waltham, MA) with 260/280 nm ratio > 1.8. Approximately 500 ng RNA was reverse transcribed with
the SensiFAST cDNA Synthesis Kit per manufacturer instructions (Bioline, UK). The resulting cDNA
was then amplified in 384-well PrimePCR assay plates with Universal SYBR Green Supermix on a PCR
33
|
Colorado School of Mines
|
machine (Bio-Rad, Hercules, CA).
3.5 Results and Discussion
3.5.1 General Water Quality
3.5.1.1 Organic Matter
Removal of organic constituents was measured through analysis of TOC, VOC, and SVOC. TOC
was reduced after each treatment process, resulting in an overall removal of 93%, as shown in Figure 3.1.
The highest removal of TOC was seen after MBR treatment with the MBR feed having a TOC
concentration of 21.4 mg/L (following coagulation) and the MBR Permeate measuring 7.1 mg/L for a
67% removal. Van Houghton et al., using this same MBR, was unable to achieve these low TOC
concentrations in the MBR permeate when using Permian PW as the feed stream, indicating that the
coagulation treatment process prior to the MBR successfully removed a portion of the non-biodegradable
organic compounds from the raw PW, that likely persisted in the absence of coagulation.18
45
40
35
30
25
20
15
10
5
0
Raw PW MBR Feed MBR MD Feed MD
Permeate Distillate
Figure 3.1. TOC concentration in raw and treated PW after each process. Results shown are averages
obtained from duplicate samples with standard error bars.
Targeted and non-targeted analysis showed that none of the analyzed VOCs and SVOCs were
detected after MBR treatment, with one exception – the diesel range organics (DRO), shown in Figure 3.2
(p. 35), which remained in the PW throughout the entire treatment train. Interestingly, the concentration
of DRO almost doubled (177 µg/L to 323 µg/L) in the MD distillate sample. As MD is a vapor pressure
driven process, when volatile components of the water are heated up in the feed stream, these compounds
can diffuse through the membrane along with (and even faster than) clean water vapors, and concentrate
in the distillate stream.24, 25 As DRO were the only volatiles shown to be present in the feed water, this
would explain DRO concentration and accumulation in the distillate stream, as seen here. Due to these
34
)L/gm(
noitartnecnoC
COT
|
Colorado School of Mines
|
results, before this water is suitable for reuse, an additional treatment process must be employed either
before or after MD to remove these compounds. One solution would be to use batch MD as a two-step
process. In the first step, the DRO will be highly concentrated in a small volume of distillate because the
DRO preferentially diffuses through the membrane faster than water. In the second step, the distillate
(now without DRO) is directed to a new tank, and the concentrated DRO in the distillate from the first
step is further treated. An alternative solution could be further treatment with activated carbon. Further
research is needed to determine the optimal solution.
1000000
100000
10000
1000
100
10
1
Raw PW MBR Feed MBR MD Feed MD
Permeate Distillate
Figure 3.2. DRO concentration, in µg/L, after each treatment process. Results shown are average
values from duplicate samples from each process. The y-axis is a logarithmic scale to highlight the large
reduction observed after each process.
3.5.1.2 Inorganic Compounds
With few exceptions, all inorganic constituents in the raw PW were reduced by over 99% in the
final MD distillate, as shown in Table 3.3 (p. 36). The constituents with lower removal were still reduced
by substantial amounts: boron (97%, possibly due to the low volatility of boric acid and borates), barium
(97%), and manganese (94%). These results were expected as the pretreatment processes before the MD
is intended to remove suspended particles and organic contaminants, not inorganic constituents. Of note is
the increase in barium after MBR treatment. This is most likely due to barium concentrating in the reactor
over several years of continuous operation with previous feeds, and possible dissolution from the
suspended solids in the bioreactor when new streams are introduced (i.e., Permian PW in this case). The
concentration of barium; however, did not appear to affect overall MBR performance.
35
)L/gµ(
noitartnecnoC
ORD
|
Colorado School of Mines
|
Table 3.3. Inorganic constituent concentrations after each treatment process. MBR Feed was
coagulated with AlCl . MD Feed was run through GAC and ion exchange resin
3
Analyte (mg/L) Raw PW MBR Feed MBR Permeate MD Feed MD Distillate
Boron (B+) 54.9 55.9 49.6 49.1 1.9
Barium (Ba2+) 1.9 1.9 3.4 3.5 0.06
Calcium (Ca2+) 5,050 4,945 4,370 4,335 0.03
Potassium (K+) 701 682 609 627 0.48
Magnesium (Mg2+) 773 758 672 676 BDL
Manganese (Mn3+) 0.98 0.97 0.85 0.89 0.06
Sodium (Na+) 33,500 30,900 29,750 31,150 0.11
Strontium (Sr2+) 788 771 690 686 0.28
Bicarbonate (HCO–) 248 239 201 40.5 73.5
Bromide (Br–) 519 548 506 500 BDL
Chloride (Cl–) 67,750 71,600 67,000 63,700 13.3
Sulfate (SO 2–) 525 526 453 470 2.58
4
20 300
Water Flux
18
Feed Conductivity
250
16
14
200
12
10 150
8
100
6
4
50
2
0 0
0 24 48 72 96 120
System Run Time (hrs)
Figure 3.3. MD water flux and feed conductivity over the run time of the system during this study.
Feed conductivity is used as a representation of the inorganic concentration in the feed. Water flux was
determined using a digital pressure transducer over time, leading to the fluctuations observed here.
3.5.1.3 Naturally Occurring Radioactive Material
The average activity of eight radionuclides found in the two samples taken at each unit process
are shown in Figure 3.4 (p. 37). All eight radionuclides tested for in this study were reduced substantially
after MD treatment. Of note are bismuth 212 (Bi-212), potassium 40 (K-40), and lead 212 (Pb-212),
which were removed to a level below the detection limits (25 pCi/L, 21 pCi/L, and 4.5 pCi/L,
respectively). Specifically, K-40, which had an initial concentration in the raw PW of 634 pCi/L, was
reduced to non-detect limits.
Of the eight radionuclides detected, the EPA currently has a maximum contaminant level (MCL)
for only the bismuth isotopes (Bi-212 and Bi-214) and the radium isotopes (Ra-226 and Ra-228). The
MCL for bismuth isotopes is 15 pCi/L (combined activity for both isotopes <15 pCi/L) and the combined
36
)HML(
xulF
retaW
)mc/Sm(
ytivitcudnoC
deeF
|
Colorado School of Mines
|
MCL for radium (Ra-226 plus Ra-228) is 5 pCi/L. Overall removal percentage of Bi-212, Bi-214, Ra-
226, and Ra-228, from Raw PW to MD distillate, was 99.9%, 98.8%, 98.9%, and 99.5%, respectively,
with combined bismuth activity reduced to just 1.5 pCi/L (below the MCL) and combined radium activity
reduced to 6.2 pCi/L (slightly above the MCL).
1000
100
Raw PW
Coagulated PW
MBR Permeate
GAC/IX
10
MD Distillate
1
Bi212 Bi214 K40 Pb212 Pb214 Ra226 Ra228 Tl208
Figure 3.4. Average activity of radionuclides found in PW from the Permian basin after each step of
the treatment train. Minimum detectable concentrations for Bi212, K40, Pb212, and Ti208 in the MD
Distillate samples were 25 pCi/L, 21 pCi/L, 4.5 pCi/L, and 2.2 pCi/L, respectively.
The relatively higher removal rates seen after the MBR treatment is likely due to one or several
microbial species being present in the MBR that could interact with radionuclides. Previous microbial
community analysis performed on the MBR did show the presence of some genera that have been
reported at radioactive sites or used in the bioremediation of such sites, including Mycobacterium,
Shewanella, Roseovarius, and Methylophaga.18, 29-31 The mechanisms of these interactions may be
biotransformation, biosorption, bioaccumulation, bioprecipitation, and/or biosolubilization.32 These
mechanisms have been studied extensively as a result of research into bioremediation of radioactive
materials. Mathur and Dwivedy showed that bacteria belonging to the genus Arthrobacter were able to
remove 95% of Ra-226 from uranium mill effluent through biotransformation.33 Kokke et al. was able to
isolate several bacterial strains capable of removing radioactive matter through biosorption.34
Similarly, the high removal percentage observed in this study after MD agrees with previous
studies. Liu and Wang demonstrated removal to below detectable limits on selected radionuclides in low
level radioactive wastewater using DCMD.35 Likewise, Khayet et al., using DCMD, were able to achieve
both desalination and removal of radionuclides from real radioactive liquid waste from a nuclear
37
)L/iCp(
ytivitcA
|
Colorado School of Mines
|
laboratory.36 Overall NORM removal for gamma-ray testing showed a reduction percentage of 99.3%. It
should also be noted that while these two processes can effectively remove radionuclides from PW,
special care must be taken when disposal of solids from the MBR and retentate from the MD as both will
contain a concentrated level of activity.
3.6 In-vitro Bioassay Testing
3.6.1 Evaluation of Cytotoxicity
MCF-7 cells were exposed to the undiluted (100%) or dilutions of the PW sample (3.13%. 6.25%,
12.5%, 25%, 50%) for 24 hrs to identify dilutions that would not be overtly toxic for subsequent gene
expression analysis. The MTS assay used to assess cytotoxicity of diluted samples showed no cytotoxicity
when cells were exposed to any of the samples at 3.13% or 6.25% as illustrated in Figure 3.5.
Cytotoxicity increased concentrations for all samples except the MD distillate and the field blank. We
therefore used the 6.25% concentration to evaluate gene expression. Additionally, determination of
cytotoxicity of the concentrated water samples found that 1X, 3X, and 5X concentrations did not exhibit
cytotoxicity in MCF-7 cells (data not shown), and thus 2X concentrates were subsequently used to
evaluate gene expression.
140
120
100
80
60
Field Blank
40 Raw PW
MBR Feed
MBR Peameate
20
MD Feed
MD Distallate
0
3.13 6.25 12.5 25 50 100
Sample Water Concentration (%)
Figure 3.5. Cytotoxicity of raw water samples. MCF-7 cells were exposed to media containing raw
water at concentrations of 3.13% to 100% for 24 h, and 10 µl of MTS solution was added for 3 h to detect
cell viability at 490 nm. The data were expressed as % of medium blank.
3.6.2 AhR Activation
Evaluation of raw and treated waters using in-vitro bioassays have been used for various
municipal and industrial water treatment processes and trains.37-39 Of the bioassays that have been used,
38
)DO
knalb
fo
%(
etaR
lavivruS
lleC
|
Colorado School of Mines
|
aryl hydrocarbon receptor (AhR) is highly relevant for PW because polycyclic aromatic hydrocarbons
(PAHs) and dioxins have routinely been detected in various PW, with PAHs being detected in the water
studied.40-42 As such, we measured the expression of two genes from the cytochrome P450 (CYP) 1
family, CYP1A1 and CYP1B1 that are induced when AhR is activated by PAHs or dioxins.43 In addition,
we examined the expression of HMOX1, regulated by the oxidant-induced transcription factor NFE2L2
(NRF2). We analyzed two forms of the water samples. In the first set, we examined the raw water or
dilutions that were used to make the media. In the second set, we examined nonvolatile organics purified
through solid-phase extraction columns.
3.6.3 Effects of Diluted Samples on AhR Activation
Based on the cytotoxicity assay, 6.25% of raw water did not produce cytotoxicity in MCF-7 cells,
and this concentration was used to determine HMOX1, CYP1A1, and CYP1B1 expression changes. As
shown in Figure 3.6, the 6.25% Raw PW water activated HMOX1 (8.7-fold), CYP1A1 (17-fold), and
CYP1B1 (7-fold). Activation of the genes generally decreased with increased number of treatment steps.
However, the activation of both genes was greater in the MBR Permeate compared to the MBR Feed. The
final activation of the genes in the MD Feed was greatly decreased to 5-fold for CYP1A1 only. MD
distillate was ineffective in induction of all three genes. Consistent with the cytotoxicity assay, HMOX1
induction was only seen with raw water dilutions but not water concentrates.
25
HMOX1
6.25% Dilution
CYP1A1
CYP1B1
*
20
*
15
10 * * *
* *
*
5
* *
0
Field Raw PW MBR Feed MBR MD Feed MD
Blank Permeate Distillate
Figure 3.6. Effect of diluted water samples on HMOX1, CYP1A1, and CYP1B1 expression in MCF-7
cells. Cells at 80% confluency were treated with 6.25% of Raw water samples for 6 hours. Total RNA
was extracted and subjected to RT-qPCR analysis. Data are mean ±SE of 3 biosets. *Significantly
different from Field Blank at P < 0.05.
39
)knalb
dleif
fo
dlof(
noisserpxE
evitaleR
|
Colorado School of Mines
|
3.6.4 Effects of Solid Phase Extracted Samples on AhR Activation
As shown in Figure 3.7, the 2x concentrate of the raw PW extract activated CYP1A1 (~11-fold)
and CYP1B1 (~3-fold). Activation of the genes generally decreased with increased number of treatment
steps. However, the activation of both genes was greater in the MBR permeate compared to the MBR
feed. The final activation of the genes in the MD distillate was ~2.5-fold and ~1.4-fold for CYP1A1 and
CYP1B1, respectively, with only the CYP1A1 activation being significant.
14
2X Concentrates CYP1A1
* CYP1B1
12
10 *
8
6
*
4 * * *
*
* *
2
0
Field Blank Raw PW MBR Feed MBR MD Feed MD
Permeate Distillate
Figure 3.7. Effect of water concentrates on CYP1A1 and CYP1B1 expression in MCF-7 cells. Cells at
80% confluency were treated with 2X water concentrates for 6 hours. Total RNA was extracted and
subjected to RT-qPCR analysis. Data are mean ±SE of 3 biosets. *Significantly different from Field
Blank at P < 0.05.
3.6.5 Activation of CYP Expression is Dependent on AhR
To confirm the role of AhR in CYP1A1 and CYP1B1 activation, wild-type MCF-7 (WT) and
AhR-null MCF-7 cells were exposed to 3x water concentrates for 6 hours. As shown in Figure 3.8 (p. 41),
Raw PW extract activated CYP1A1 (8-fold) and CYP1B1 (3.3-fold) in wild-type MCF-7 cells, while
activation of both genes was decreased with treatment steps. The MD Distillate did not significantly
induce both genes. Importantly, activation of CYP1A1 and CYP1B1 expression was abolished in AhR-null
cells, indicating the critical role of AhR in activating these genes in the presence of a chemical activator.
40
)knalb
dleif
fo
dlof(
noisserpxE
evitaleR
|
Colorado School of Mines
|
10 5
CYP1A1 WT AhR-null CYP1B1 WT AhR-null
9
8 4
7
6 3
5
4 2
3
2 1
1
0 0
Figure 3.8. Effect of water concentrates on CYP1A1 and CYP1B1 expression in MCF-7 (WT) and
AhR-null cells. Cells at 80% confluency were treated with 3X water concentrates for 6 hours. Total RNA
was extracted and subjected to RT-qPCR analysis. Data are mean ±SE of 3 biosets. *Significantly
different from Field Blank at P < 0.05.
Here we found clear evidence that the PW examined in this study contains chemicals that activate
AhR, and the methods used to treat the PW eliminated the chemicals that activate AhR. We examined
AhR activity in a human cell line known to be responsive to AhR activators because of several studies
that showed that PW from various sources contains chemicals that activate AhR. In the earliest studies,
Hurst et al. examined the AhR activity using a trans-activation assay carried out in the rat H4IIE
hepatoma cell line of PW from offshore oil and gas drilling in the United Kingdom Continental Shelf.44
The authors found that most of the 53 oil platforms tested discharged PW that contained AhR activity. In
follow up work, the same group used an effects-directed approach to identify the components in the PW
that were responsible for the AhR activity.45 Using several fractionation methods, the authors were able to
identify chemicals in PW that are known to be AhR activators including persistent organic contaminants
hexachlorobenzene, decachlorobiphenyl, and octachlorodibenzofuran. PW from an oil platform in the
Norwegian sector of the North Sea was used to determine effects in the three-spined stickleback
Gasterosteus aculeatus.46 The authors found that the PW activated a number of responses, indicating that
the fish were under increased stress; genes that are under control of AhR were activated in the livers of
the exposed fish. Li et al. found that PW from an oilfield in northeast China contained AhR activity as
measured by increases in the activity of CYP1A1 protein (Ethoxyresorufin-O-deethylase (EROD)).47
Remarkably, in the same study AhR activity was also detected in water from a wastewater treatment
plant. He et al. examined samples of hydraulic fracturing flowback and produced water (HF-FPW)
collected from 2 wells in Alberta, Canada.48 Using the H4IIE assay system, the authors found that most of
41
)knalb
dleif
fo
dlof(
noisserpxE
evitaleR
)knalb
dleif
fo
dlof(
noisserpxE
evitaleR
|
Colorado School of Mines
|
the samples collected contained AhR activity. The authors did not determine if the activity was due to the
PW or the hydraulic fracturing fluid. The same group examined the effects of HF-FPW on the zebrafish
embryo assay and found increases in EROD activity and CYP1A1 and CYP1B1 expression indicative of
AhR activation.49 In summary, our observation of AhR activity in PW in the present study is consistent
with the presence of chemicals that activate AhR regardless of the location of the source or the method
used to assess AhR activity. Nevertheless, we have shown that the processes used to treat the PW in this
study eliminated the chemicals that activate AhR.
3.7 Conclusion
In this study we have investigated the toxicity and contaminant removal of individual treatment
processes of a high salinity PW treatment train utilizing four-treatment processes to reclaim PW from the
Permian basin to a theoretical discharge water. Removal performance of this treatment system was shown
to be effective at reducing constituent concentrations to levels at or below EPA effluent discharge limits.
Additionally, for those inorganic contaminants for which established drinking water standards exist,
concentrations were reduced to below the EPA primary and secondary drinking water standards,
including TDS < 500 mg/L. Toxicity testing performed after each treatment process showed a substantial
reduction to non-detect levels from the diluted raw PW to the MD distillate in both cytotoxicity and
HMOX1 gene expression. HMOX1 activation is an indicator of oxidative stress to the cells, consistent
with the cytotoxicity of the Raw water dilutions. However, after concentrating water samples by solid
phase extraction, cytotoxicity and HMOX1 induction were not observed (data not shown), suggesting
differences between raw water dilution and water concentrates in producing oxidative damage. In MD
Distillates, most contaminants were removed at over 99% from the raw PW, including NORM, select
organics, metals, and anions, no cytotoxicity and oxidative stress were observed.
Water concentrates, however, were able to induce CYP1A1 and CYP1B1 gene expression,
although to lesser extents compared to the raw water (e.g., 17-fold induction of CYP1A1 by 6.25% Raw
PW vs 11-fold by 2x concentrates), indicating that some AhR-like chemicals were still present in 2x
concentrates of raw PW, which could not be removed by solid phase concentration. However, in MD
Distillates, after most contaminants were removed at over 99% from the raw PW, no induction of
CYP1A1 and CYP1B1 was observed.
To further verify AhR-like chemicals existed in water concentrates, AhR-KO MCF-7 cells were
used. Induction of AhR biomarker gene CYP1A1 and CYP1B1 was evident with 3x concentrates in WT
MCF-7 cells, but no such induction was observed in AhR-KO cells, confirming AhR-like chemicals
existed in raw water concentrates and raw water dilutions (data not shown). Additional research is
required to identify these compounds, evaluate polishing steps for this treatment train, and better
understand why MBR treatment resulted in a higher AhR response as compared to the coagulated water.
42
|
Colorado School of Mines
|
where this was once considered impractical such as highly polluted industrial wastewater and brines. The
increasing costs of treating these waste streams to a level suitable for reuse has greatly hindered the
implementation of treatment as the preferred wastewater management method. Therefore, accelerating
research into novel treatment processes that are more effective and relatively easy to implement for cost-
conscience industrial managers is therefore necessary to promptly address water scarcity as quickly as
possible.
With 96% of the world’s water located in salty oceans, desalination by reverse osmosis (RO) has
been the most widely implemented solution to combat water scarcity.3, 4 RO desalination is popular
because it is a fully mature and commercial technology that has been in use for several decades and
reached energy efficiencies close to the thermodynamic limit of salt separation from water, especially due
to advances in energy recovery devices.5-7 RO is a separation process that pressurizes saline water against
a semi-permeable membrane, concentrating the salts in a brine stream while producing a clean permeate
stream. The hydraulic pressure required for operation of the process must be higher than the osmotic
pressure of the RO brine. When desalinating saline streams, the salinity of the feed water becomes higher,
the osmotic pressure increases, and the hydraulic pressure must be further increased to maintain
commercially acceptable water flux. Current RO membranes can typically operate at a maximum pressure
of ~1,200 psi (8.2 MPa) for seawater desalination; however, at feed concentrations higher than seawater
(~35 g/L TDS), the hydraulic pressure required to overcome the osmotic pressure of the brine and achieve
practical water production (flux) is much higher than 1,200 psi, and might cause irreparable damage to
current RO membranes.8 At that point, it becomes impractical to use RO for highly saline water, and other
desalination technologies such as distillation must be implemented.
The solutions currently available for desalinating high salinity waters are limited to technologies
that require a substantial footprint and much more energy such as in multi-stage flash distillation (MSF),
multi effect distillation (MED), and vapor compression distillation (VCD). Therefore, there has been
increased research into novel processes capable of desalinating high salinity water sources that are
comparable in size and energy efficiency to conventional RO. For example, many studies examined
membrane distillation (MD) for hypersaline water, and have found it to be a cost-effective solution.9-12
MD is a vapor-pressure driven process where the feed water is heated and clean water vapor diffuses
through a porous membrane. When using low grade heat or renewable energy sources such as solar or
geothermal, the operating costs can be substantially reduced.13 However, some of the major drawbacks
preventing MD from becoming commercial are lack of membrane availability, lack of membrane
robustness, and complex heat management. A different option for desalination of high salinity streams is
ultra-high-pressure reverse-osmosis (UHPRO). This process allows for much higher operating pressures
than traditional RO and is therefore able to overcome the higher osmotic pressures while simultaneously
48
|
Colorado School of Mines
|
reducing energy consumption.14-18 Unlike MD, the system design of UHPRO is similar to conventional
RO, giving it the potential to be more widely implemented in a shorter time, with multiple commercially
available high-pressure RO membranes recently coming to market.19, 20 However, several recent studies
have revealed the high potential of UHPRO membranes to exhibit substantial performance decrease due
to membrane compaction. Wu et. al., observed a 50% reduction in the porosity of a high-pressure RO
membrane (0.63 to 0.32) when exposed to pressures of ~3,000 psi (21 MPa) for an hour, with potentially
more compaction possible.21 Also, several recent molecular dynamics simulation studies from He et. al.
examined the effects of varying membrane properties (i.e., membrane thickness, degree of cross-linking,
manufacturing process) on membrane performance and integrity, observing rapid membrane compaction
at ultra-high pressures, regardless of membrane characteristics.16, 22, 23 These studies also highlight the
limitations of current UHPRO research, which is mostly theoretical modeling based, as experimental data
is difficult to attain due to limitations on the availability of components capable of operating under ultra-
high pressures (e.g., membranes that avoid significant compaction, pressure vessels rated for ultra-high
pressures).24
Osmotically-assisted reverse osmosis (OARO), another emerging desalination process, avoids the
difficulties associated with UHPRO by reducing the osmotic pressure differential across a series of
membranes by increasing the salinity of the permeate streams.25-28 The OARO process utilizes a multi-
stage design that recirculates the reject of each stage back into the permeate of the previous stage,
increasing the salt concentration and lowering the required hydraulic pressure needed on the feed side to
achieve meaningful water flux. The mixed permeate is then pumped into the feed inlet of the next stage
with an additional high-pressure pump, maintaining overall system pressure and flowrate. This novel
approach to hypersaline water treatment has shown promising results in desalination performance and
energy demands compared to thermally driven process.29, 30 Peters et. al. were able to demonstrate water
recovery rates of 74% and 44% when treating water with initial feed TDS concentrations of 35 g/L and 70
g/L, respectively.29 Comparing the difference in costs to generate 1 m3 of product water using traditional
RO, MVC, and OARO, Bartholomew et. al. developed a cost-optimization model to determine the
economic feasibility of achieving 30-70% recovery rates when treating water with an initial TDS
concentrations between 50 g/L and 125 g/L.30 At these system requirements, OARO was shown to be
equal or less than $6/m3, using 30-60% less energy than MVC. However, similar to UHPRO, unique
membrane elements are needed for productive operation; specifically, OARO requires bilateral RO
membrane elements to allow for the cross-flow of feed and permeate streams. These non-traditional RO
elements use loose spacers on both sides of the membrane, increasing the potential for membrane
deformation at higher operating pressures, and limiting the maximum hydraulic pressure to 800 psi (5.5
MPa). As such, OARO elements that both allow cross-flow of streams and maintain membrane integrity
49
|
Colorado School of Mines
|
are not yet widely available, limiting research and reducing OARO viability for commercial
implementation.
Another type of OARO process that has shown promising results during theoretical process
modeling is low-salt rejection reverse-osmosis (LSRRO), a process flow diagram of which is illustrated in
Figure 4.1.31-34 LSRRO, as the name suggests, uses RO membranes that have low salt rejections, creating
permeate streams with increased TDS concentrations, lowering the hydraulic pressure required to produce
acceptable water flux and salt rejection through the membrane. The saline permeate from each stage of
LSRRO is then used to dilute the TDS of the feed to the previous stage (i.e., permeate from the second
stage is routed to the feed of the first stage), while the concentrated reject is fed into the following stage.
The membranes utilized in LSRRO are manufactured from readily available traditional polyamide RO
membranes with the intention of assigning each membrane a distinct salt permeability coefficient (B-
value).
Figure 4.1. Process flow diagram of LSRRO. The 1st stage is operated as a traditional RO process,
generating the final product permeate of the system, while each stage after that uses low-salt rejection
membranes that recycle their permeate back into the feed of the previous stage. Figure adopted from Wang
et. al.31
The membranes in LSRRO are installed in series, with each subsequent membrane possessing
lower salt rejection capabilities than the membrane before it, with the first stage only generating the final
permeate and each succeeding stage having progressively higher permeate TDS concentrations, which is
recycled upstream. The design of the LSRRO results in the same benefit of OARO (i.e., desalination of
highly saline streams with hydraulic pressures comparable to traditional RO) while providing distinct
advantages over OARO; namely, using commercial membranes and reducing the number of high-pressure
pumps needed for each stage. These distinctions suggest that LSRRO can be implemented in shorter
times and with less costs than MD, UHPRO, or OARO while capable of generating clean water from
hypersaline brines. For example, during the development of their LSRRO process simulation, Wang et. al.
calculated that LSRRO was able to achieve hypersaline brines with TDS concentrations of 234 g/L
50
|
Colorado School of Mines
|
without exceeding traditional RO pressures.34 In another simulation, the specific energy consumptions
(SEC) of OARO and LSRRO were compared and it was determined that these two processes have similar
energy requirements, with both being much more efficient than current thermal desalination processes.31
The positive results obtained from process modeling studies on LSRRO have facilitated the need
to conduct laboratory experimentation on the proposed technology. Experiments should include testing on
authentic hypersaline sources (e.g., industrial wastewater) to corroborate the modeled performance of
LSRRO in real-world conditions. For example, wastewater (i.e., produced water (PW)) from
unconventional oil and gas (O&G) operations is a hypersaline water source that is difficult to desalinate
due to having many contaminants in high concentrations, including heavy metals, organic compounds,
and naturally occurring radioactive material (NORM).35 These constituents found in raw PW can
substantially reduce performance of membrane-based treatment processes, particularly membrane-based
desalination processes, such as OARO and LSRRO, due to the increased fouling and scaling potential that
PW manifests.36 Consequently, testing with feed solutions consisting only of NaCl cannot replicate the
complexity of PW.37, 38 Additionally, testing with real PW, as opposed to a prepared synthetic PW, will
provide more understanding of the capabilities and limitations of LSRRO (e.g., recovery percentages,
scaling potential, ion rejections).39
Therefore, the objective of this study was to evaluate the ability of a small pilot-scale LSRRO
system to desalinate several synthetic and authentic high salinity water streams, including PW from the
Permian Basin (133 g/L TDS). Five differing test waters, consisting of progressively more complex water
matrices, were used to extensively evaluate the LSRRO process. Performance of the LSRRO was
evaluated by overall TDS rejection (initial feed versus final permeate), inorganic constituent removal, and
percent recovery of the final product water. Additionally, water permeability testing was performed
throughout the study to monitor potential decreases in performance of the LSRRO membranes due to
fouling or scaling. These tests established baseline performance and the effectiveness of chemical
cleaning that were performed between each test. As such, this work presents a unique and comprehensive
pilot study of a novel RO process and assesses its potential as a solution in future minimal/zero liquid
discharge (MLD/ZLD) applications.
4.3 Methods and Materials
4.3.1 LSRRO Pilot System
The LSRRO pilot system was designed and fabricated in a 3-stage configuration, as illustrated in
Figure 4.2 (p. 52). Each stage consisted of three single LSRRO membranes (stage 1 M-1, M-2, M-3; stage
2 M-4, M-5, M-6; and stage 3 M-7, M-8, M-9), for a total of nine unique LSRRO membranes. Similar to
the OARO, an additional seawater RO membrane (DuPont SW-30) was used as the final, polishing step.
The nine LSRRO membranes were arranged in series such that each subsequent membrane has a lower
51
|
Colorado School of Mines
|
salt rejection than the previous membrane. Thus, the first membrane (M-1) was manufactured to have the
highest salt rejection while the last membrane (M-9) had the lowest rejection.
Figure 4.2. System flow diagram of the LSRRO pilot. During step 1, feed water from T-1 enters the
mixing tank, T-4 (1), from T-4, the mixed feed enters the LSRRO (5), as the water moves through each
membrane, reject is sent to the following membrane. Stage 1 permeate (7) is sent to the SWRO membrane
for final polishing until the TDS concentration becomes too high, at which point it is sent to T-3 to be
collected and will become the starting feed stream for the next cycle. Stage 2 permeate is sent back to T-4
to dilute (12); Stage 3 permeate (10) is sent to T-5, where a high-pressure pump moves it back to the feed
of M-4 to dilute (11). System reject (6) is collected in T-2 (3) to become the next step’s feed stream.
During SWRO operation, permeate (8) is recovered product, while reject (8) is sent back to T-4 for
additional dilution feed into stage 1 (5). During step 2, feed enters from T-2 (2) and system reject (6) is
routed to T-1 (4), and the process is continued.
Because it is difficult to scale down the LSRRO process, a system was designed and fabricated to
enable operation in a continuous-batch mode, step-wise concentrating the batch. For each experiment, a
batch of feed water is introduced into feed tank T-1, and a second feed tank T-2 is initially empty and
ready to accept the concentrate from membrane M-9 (6). When the process starts, water from T-1 is
transferred into mixing tank T-4 and pumped into the train of LSRRO membranes (1), with each
membrane’s concentrate stream feeding the next membrane in the line (i.e., M-1 through M-9). When T-1
becomes empty (end of a step), T-2 becomes the feed tank (valve 2 opens and valve 3 closes) and T-1
becomes concentrate tank (valve 4 opens and valve 1 closes), collecting LSRRO concentrate generated
during Step 2.
52
|
Colorado School of Mines
|
After several steps, when the feed batch becomes more concentrated than the LSRRO membranes
could effectively treat (water flux below ~3-5 LMH), feed to the system would be switched to tank T-3,
which had been collecting permeate from stage 1 after the first two steps. The collection of steps that ends
before feeding the system from tank T-3 is defined here as a “cycle”. We conducted three concentration
cycles in four of the experiments and two cycles in a fifth due to limited initial feed volumes, resulting in
46 steps, 30 steps, 21 steps, and 34 steps during the NaCl, BW, SW, and PW experiments with three
cycles and 26 steps during the MW test with two cycles.
As detailed in Figure 4.2 (p. 52), the permeates from the three membranes of each stage were
combined and introduced into the feed streams of upstream stages. Initially, while the feed concentration
is low, stage 1 permeate (7) is directed to the feed tank for the RO membrane (T-6), and when the feed
concentration becomes too high (usually after 2 steps in each cycle), stage 1 permeate (7) is directed to,
and collected in tank T-3 for later use as a feed to the system in the next cycle. Stage 2 permeate (12) is
directed into the mixing tank (T-4) to dilute the feed from T-1 or T-2, and stage 3 permeate (10) is
collected in T-5, and then boosted into the feed to Stage 2 while diluting it. These dilutions allowed the
LSRRO to continue to operate at brine concentrations much higher than traditional RO.
Stage 1 permeate, after the first two steps of each cycle, was removed from feeding the SWRO
membrane, and was directed to tank T-3 for later use until the end of the cycle. This allowed the SWRO
membrane to operate within acceptable TDS concentrations at the beginning of each cycle; but as the feed
continued to concentrate, stage 1 permeate TDS also increase, eventually past the point of efficient
SWRO membrane operation. This would repeat at the beginning of each cycle until the final cycle of each
experiment, when the stage 1 permeate would continue to feed the SWRO membrane until product
permeate generation had ceased as a result of the TDS concentration of stage 1 permeate reaching levels
beyond the SWRO membrane capabilities. While the SWRO membrane was in operation its permeate (8)
was collected and removed from the system while the reject (9) was directed to the mixing tank (T-4) to
combine with the feed from T-1/T-2 and the permeate from stage 2, thus increasing the dilution of the
concentrating feed entering stage 1.
Two high-pressure pumps (VFD automated) were used in the main system (Hydra-Cell, Wanner
Engineering, Inc., Minneapolis, MN), a primary pump that was fed by T-4 and a secondary pump that was
fed by T-5 (stage 3 permeate boosted into the feed of stage 2). A third high-pressure pump (VFD not
automated) was use during operation of the SWRO membrane. Feed pressures generated by the two
automated pumps was controlled using a digital pressure transmitter (WIKA Instrument, LP,
Lawrenceville, GA) through a SCADA system. Flowrates for the feed and reject were monitored using
digital flow meters (Titan Enterprises LTD, Sherborne, Dorset, United Kingdom) and flowrates for each
permeate (M-1 through M-9) were measured with analog flow meters (Blue-White, Huntington Beach,
53
|
Colorado School of Mines
|
CA). Conductivity was continuously monitored and converted to TDS concentration by Eq. 4.1
(4.1)
where K is the conversion factor and EC is the recorded conductivity reading (µS/cm for product
𝑇𝑇𝑇𝑇𝑇𝑇 = 𝐾𝐾∗𝐸𝐸𝐶𝐶
permeate, mS/cm for everything else) from the multiple probes located at various points in the pilot
system. Two different conversion factors were used to convert conductivity to TDS; K=0.72 for mS/cm
readings and K=0.55 for µS/cm readings.40, 41 Toroidal conductivity probes (3700 series, HACH Co.,
Loveland, CO) were placed at the permeates for each stage, tanks T-1 and T-2, after T-4 for the mixed
water entering the system, the effluent of T-5, and the final reject leaving the system after M-9. The
toroidal probes were connected to a master controller unit (SC1000, HACH Co., Loveland, CO), which
relayed the continuous conductivity data in real time to the controller’s digital display. An additional
conductivity probe (OAKTON Instruments, Vernon Hills, IL) was installed in the SWRO permeate line
for continuous monitoring of product permeate quality. Feed flow into the mixing tank (T-4) from T-1
and T-2 was controlled by a float valve attached to T-4 for the purposes of maintaining constant volume
in the tank (5 L). T-5 volume was set to an 8 L limit such that the secondary pump would increase or
decrease RPM to maintain the 8 L set point. This was controlled by the LSRRO pilot’s SCADA system
that received signals from a digital pressure transducer (Omega Engineering, Inc., Norwalk, CT) installed
at the bottom of T-5. Feed temperature was monitored with a temperature transmitter and maintained at
25 °C with a water chiller. In addition to the digital pressure gauge connected to the SCADA system, feed
pressure was monitored with a 0-2,000 psi pressure gauge (Ashcroft, Stratford, CT) while individual stage
reject pressures were each connected to a 0-1,500 psi pressure gauge (McDaniels Controls, Inc., Boutte,
LA). The SWRO membrane feed pressure was monitored with a 0-1,500 psi pressure gauge (Swagelok,
Solon, OH). All digital meters and probes, excluding the toroidals, were connected to pilot’s SCADA
system that monitored and controlled most of the LSRRO system components.
4.3.2 Membranes
The membranes used for the LSRRO were manufactured using a proprietary process to ensure the
proper individual rejection rates required from each membrane (Fluid Technology Solutions, Inc.,
Albany, OR). A list of each membrane’s NaCl salt permeability coefficients, as provided by the
manufacturer, are given in Table 4.1 (p. 55). The membranes were size 2540 and produced from
conventional, commercially available polyamide RO membranes. The membrane used in the final
polishing step was a conventional seawater RO (SW-30 2540, DuPont, Wilmington, DE).
54
|
Colorado School of Mines
|
Table 4.1. Operational order and salt permeability coefficients for each of the nine LSRRO
membranes. Membranes were manufactured to be in a specific order with cascading B-values. B-values
provided by manufacturer.
B values
Membrane Name Location in Series Configuration
NaCl (LMH)*
M-1 Stage 1 feed 5-20
M-2 Stage 1 middle 5-20
M-3 Stage 1 concentrate 5-20
M-4 Stage 2 feed 15-30
M-5 Stage 2 middle 15-30
M-6 Stage 2 concentrate 15-30
M-7 Stage 3 feed 15-30
M-8 Stage 3 middle 30-60
M-9 Stage 3 concentrate 30-60
4.3.3 Water Chemistry
For this study, five different feed waters of varying compositions and sources were used to test
the LSRRO pilot system. For the first experiment, a solution of NaCl and DI water (NaCl) was prepared
at a TDS concentration of 70 g/L using food-grade quality 99.9% pure (Culinox 999, Morton Salt,
Chicago, IL). The second experiment was carried out with a synthetic brackish groundwater (BW) that
was prepared using laboratory grade compounds (Sigma-Aldrich Corp., St. Louis, MO) of calcium
chloride, magnesium chloride, sodium sulfate, sodium bicarbonate, sodium nitrate, and sodium chloride
to reach the desired concentrations for each constituent. Concentrations were based on values in brine
from brackish water treated by reverse osmosis and electrodialysis.42, 43 The third experiment was carried
out with a synthetic seawater (SW) solution. The salts used for this experiment (Instant Ocean, Spectrum
Brands, Blacksburg, VA) were added at a concentration intended to represent concentrated seawater, with
a TDS concentration ~40% higher than common seawater, 56 g/L compared to 35 g/L. The fourth
experiment was carried out with a naturally occurring, lithium-rich mining wastewater (MW), and the last
experiment, and the most complex water matrix of the study, was carried out with hypersaline PW from
the Permian Basin. Pretreatment was minimal for the first four experiments; none performed for the NaCl
experiment, the pH was lowered to 7.2 with HCl for the BW experiment, the pH was lowered to 5.9 with
HCl and 2.5 mL of an antiscalent (Avista 7000, Avista Membrane Treatment Solutions, Inc., San Marcos,
CA) was added to the SW experiment, and the MW was pretreated with granular activated carbon (GAC)
and a 5-micron cartridge filter. Due to the high concentrations of scaling compounds, suspended solids,
and organic chemicals, the PW was pretreated with several processes, including coagulation (aluminum
55
|
Colorado School of Mines
|
chloride, Sigma Aldrich Corp., St. Louis, MO) at a concentration of 10 mg/L, a membrane bioreactor
(MBR), water softening based on jar testing results using soda ash and lime, GAC, and a 5-micron
cartridge filter. The concentrations of major constituents in the five streams are summarized in Table 4.2.
Analysis for cations was performed by inductively coupled plasma-atomic emission spectroscopy (ICP-
AES, Optima 5300, PerkinElmer, Fremont CA) while anion analysis was conducted by ion
chromatography (IC, ICS-900, Dionex, Sunnyvale, CA).
Table 4.2. General water quality of each of the test waters used in this study. The concentrations
listed represent the water as it was introduced to the LSRRO system after all pretreatment processes had
been conducted. A solution of NaCl at 70 g/L was also tested.
Analyte (mg/L) NaCl BW SW MW PW
B 13.7 9.0 576 74
Ca 148 676 389 1,232
K 187 639 226 701
Li 5.7 0.61 2,710 22.4
Mg 620 2,212 794 617
Na 27,500 11,925 17,420 2,591 51,684
HCO - 2,700 400 86 828
3
SO 2- 391 5,254 475 262
4
Cl- 42,500 20,041 29,283 26,747 77,093
TDS 34,129 56,000 34,600 134,000
pH 7 7.2 5.9 7.3 6.8
4.3.4 LSRRO System Operating Conditions
Initial feed volume in each experiment was 473 L (125 gal), except for the MW experiment in
which the batch volume was 284 L (75 gal) and therefore had only two concentration cycles, while the
other experiments had three concentration cycles. The main objective was to achieve maximum total
water recovery in each experiment while maintaining, as much as possible, constant flux through each of
the LSRRO membranes. This was done by setting an initial feed pressure for the primary LSRRO pump
to 600 psi (41.4 bar) at the start of each concentration cycle, which would be supplemented by the
hydraulic pressure from the secondary pump that dilutes the feed entering the second stage. As the
osmotic pressure of the feed began to increase and permeate flux began to decrease, we slowly increased
the feed pressure in 100 psi (6.9 bar) increments until a maximum set pressure of 1,100 or 1,200 psi (75.8
or 82.7 bar). At this point, cycles were run until steps reached the same length of time, usually <5 min
56
|
Colorado School of Mines
|
(limited by availability of water in the feed tank). The maximum RPM of the pump was limited to ~85%
of max RPM to prevent pump failures. Feed flowrates were maintained at an average of 6 L/min in all
experiments. All operating parameters were controlled by a SCADA system with inputs from the digital
probes (thermocouples, pressure transducers, flow meters, and conductivity probes) in the system, except
for the conductivity toroidal probes that have their dedicated data recording system.
The SWRO membrane was operated in batch-mode with each batch consisting of stage 1
permeate until the TDS concentration was too high, at which point the SWRO subsystem was shut down.
Similar to the LSRRO primary pump, the SWRO pressure was adjusted during operation to maintain a
constant flux of ~20 LMH. At the end of the last concentration cycle in each experiment, the SWRO
membrane was kept in operation until permeate flux was reduced to 3 LMH with the pump operating at
up to 85% of maximum RPM and up to 900 psi (62 bar) pressure. This was done to minimize scaling and
pressure damage to the membrane. SWRO pump speed was controlled by hand using a VFD controller.
At the conclusion of each experiment, an acid cleaning was performed followed by a pure water
permeability test to verify that the membranes were maintaining their integrity throughout the study. The
acid cleaning consisted of HCl solution adjusted to pH of 2 followed by 2-3 hrs of slow recirculation
through the LSRRO membranes and subsequently DI water recycling through the system to flush out any
remaining acid. After membrane cleaning, a new batch of DI water was introduced to the system and
integrity test was conducted at two different hydraulic pressures, 100 psi and 125 psi (6.9 bar and 8.6 bar),
while individual membrane permeate flowrates were recorded. The data from these integrity tests were
used to calculate the water permeability of each membrane (flux/pressure) and compared to previous
integrity tests to evaluate possible damage or scaling that may have occurred in the previous experiment.
4.4 Results and Discussion
4.4.1 LSRRO Membrane Performance
4.4.1.1 NaCl Testing
The TDS concentrations of the feed, reject, and the permeates from all three stages are shown in
Figure 4.3 (p. 58). Initial feed TDS concentrations were 70 g/L, 58 g/L, and 50 g/L for cycles 1, 2, and 3,
respectively. The initial volumes for each cycle were 473 L for cycle 1, 208 L for cycle 2, and 110 L for
cycle 3. Total runtime for this was ~10 hrs, with the first cycle being 5.5 hrs, the second cycle running for
2.7 hrs, and the final cycle taking 1.8 hrs to complete. During this experiment the LSRRO system was
able to produce a total of 327 L permeate, resulting in ~68% water recovery. It is important to note that it
is almost impossible to conduct a mass balance and calculate water recovery based on salt mass in a
system such as the one we used in this study. This is because salts at high concentrations are occupying
every space in the system (i.e., feed and concentrate channels, permeate channels, tanks, pipeline, etc.)
and it is impossible to measure their volume and concentrations in space and over time, and especially at
57
|
Colorado School of Mines
|
the end of each cycle.
Feed Reject Feed Pressure Stage 1 Perm Stage 2 Perm Stage 3 Perm
200 1200 200
(a) (b)
175 1050 175
150 900 150
125 750 125
100 600 100
75 450 75
50 300 50
25 150 25
0 0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Runtime (hrs) Runtime (hrs)
Figure 4.3. LSRRO membrane performance during NaCl experiment with, (a) showing overall
system feed and reject TDS concentrations along with feed pressures and, (b) showing the individual
Stage concentrations. There were three concentration cycles in this experiment shown when the maximum
pressure of 1,100 psi was reached.
Stage 1 and stage 2 permeate TDS concentrations were similar, with stage 1 on average 30%
lower. These concentrations begin to converge as the concentration cycle nearing the end and feed
concentration reaching its maximum. Unsurprisingly, stage 3 permeate TDS concentration was markedly
higher than in the permeate of the other two stages because the feed to the stage is not diluted like the
other two stages, making it the highest feed concentration in the train, and because stage 3 LSRRO
membranes have the lowest rejection properties. Interestingly, this resulted in stage 3 permeate TDS
being approximately the same as the feed TDS from T-4 during the entire runtime of this experiment.
Maximum feed and reject TDS concentrations were reached at the end of cycle 1, resulting in the highest
permeate TDS concentrations of 127 g/L, 130 g/L, and 149 g/L for stages 1, 2, and 3, respectively.
Water flux through the membranes during this experiment are shown in Figure 4.4a (p. 59) and
the average fluxes are shown in Figure 4.4b (p. 59). This experiment resulted in the highest average flux
per membrane as well as the highest water recovery of 68%. The average water flux through the first
seven membranes was 16.3±2.1 LMH throughout the NaCl experiment, while the last two membranes,
M-8 and M-9, had an average flux of 26.8±0.74 LMH, approximately 39% higher. These results are
consistent with the salt permeability of the membranes, with M-8 and M-9 having a B-value
approximately twice as high as M-7. The results of this experiment provide a baseline from which to
compare to the following experiments with more complex feed streams.
58
)L/g(
SDT
)isp(
erusserP
)L/g(
SDT
|
Colorado School of Mines
|
M-1 M-2 M-3
M-4 M-5 M-6 35
M-7 M-8 M-9
35 30
30
25
25
20
20
15
15
10
10
5
5
0 0
1 2 3 4 5 6 7 8 9
0 1 2 3 4 5 6 7 8 9 10 11
Runtime (hrs) Membrane
Figure 4.4. (a) water flux through the LSRRO membranes during the NaCl experiment. The two dips
seen at 5.5 hrs and 8.25 hrs are when one cycle ends, at maximum feed concentration, and another cycle
starts, at a low TDS concentration from tank T-3. (b) Average membrane flux through each LSRRO
membrane throughout the NaCl experiment.
4.4.2 Brackish Water Testing
The next experiment was conducted with BW, having more complex composition with the
addition of several salts along with NaCl. This added complexity resulted in slightly reduced performance
compared to the NaCl experiment, increasing the runtime by 0.5 hrs and reaching overall water recovery
of 64%. Yet, a maximum brine TDS concentration of 148 g/L was achieved at the end of cycle 1, as
illustrated in Figure 4.5a (p. 60). In the previous experiment, a maximum feed pressure of 1,100 psi (75.8
bar) was reached during all three concentration cycles. During the BW experiment, 1,100 psi was reached
only at the end of cycle 1, with cycles 2 and 3 able to reach maximum brine salinity at a lower hydraulic
pressure of 950 psi (65.5 bar). This was likely due to the lower batch TDS concentrations used during this
experiment compared to the NaCl experiment, reducing the osmotic pressures of cycles 2 and 3.
Permeate TDS concentrations of the three stages during the BW experiment are shown in Figure
4.5b (p. 60) and follow the patterns seen during the NaCl experiment (i.e., stages 1 and 2 having similar
concentrations and stage 3 displaying substantially higher values). As with the previous experiment,
maximum stage permeate concentrations during the entire runtime peaked at the end of the first cycle
with 91 g/L, 97 g/L, 117 g/L for stages 1, 2, and 3, respectively.
59
)HML(
xullF
retaW
)HML(
xulF
egarevA
|
Colorado School of Mines
|
Feed Reject Feed Pressure Stage 1 Stage 2 Stage 3
160 1200 160
(a) (b)
1000
120 120
800
80 600 80
400
40 40
200
0 0 0
0 1 2 3 4 5 6 7 8 9 10 11 0 1 2 3 4 5 6 7 8 9 10 11
Runtime (hrs)
Figure 4.5. LSRRO membrane performance during the BW experiment with, (a) showing overall
system feed and reject TDS concentrations and feed pressures and, (b) showing the individual stage
permeate concentrations.
The permeate flux recorded for the LSRRO membranes during this test experiment are shown in
Figure 4.6a (p. 61), while permeate flux averages are shown in Figure 4.6b (p. 61). The LSRRO primary
feed pump pressure was adjusted manually to maintain (as much as possible) constant flux through each
of the LSRRO membranes. As the feed progressively concentrated over time, there became a point when
maintaining constant flux was not possible, regardless of the level of hydraulic pressure. During the first
cycle, this point was reached at ~4 hrs runtime, reducing the flux for membranes M-1 to M-7 by an
average of 35% until the start of the next cycle. The last membranes in the series, M-8 and M-9, were able
to maintain flux for an additional 1.2 hrs (5.2 hrs runtime) before beginning to decline at the same 35%
rate. At the start of the second cycle fluxes were restored, achieving higher water flux than were observed
during cycle 1. This was a result of the lower initial TDS at the start of the cycle (28 g/L vs. 35 g/L) and
likely the composition of the feed water during the first stop of the second cycle – permeate from stage 1
of the system during the later steps of cycle 1, which has lower concentration of divalent ions. This was
repeated for cycle 3, as flux was restored to a higher level (+1.3 LMH) than cycle 2. Average water flux
for stage 1 and stage 2 membranes followed expected trends, with the first membrane of each stage
having a higher flux than the middle membrane, which had a higher flux than the last membrane of the
stage. The membranes in the third stage, though, did not follow the trend, with M-7 having the lowest
average flux of the stage and M-8 having the highest. While not initially expected, upon further analysis
these results can be attributed to two characteristics of stage 3 that are not found in stages 1 or 2; namely,
the undiluted feed entering M-7 resulting in much higher feed concentrations, and the significantly higher
salt permeability that M-8 and M-9 have over M-7. Stage 2 feed had on average 17% lower TDS
concentration than stage 3 feed, and that is without taking into account the additional dilution stage 2 feed
was receiving as a result of stage 3 permeate pumped into the stage 2 feed line. This added salinity was
first received by M-7, with salt permeability characteristics similar to M-5 and M-6, resulting in a
60
)L/g(
SDT
)isp(
erusserP
deeF
)L/g(
SDT
|
Colorado School of Mines
|
substantial drop in performance, while the next membrane (M-8) has double the permeability of M-7.
Thus, while receiving a higher concentration feed than M-8, the considerable increase in permeability
allowed M-8 to maintain higher water flux, averaging 22.6 LMH compared to M-7 with 16.4 LMH.
M1 M2 M3
M4 M5 M6
M7 M8 M9
35 35
(a) (b)
30 30
25 25
20 20
15 15
10 10
5 5
0 0
0 2 4 6 8 10 12 1 2 3 4 5 6 7 8 9
Runtime (hrs) Membrane
Figure 4.6. (a) water flux through the LSRRO membranes during the BW experiment. The two dips
seen at 6 hrs and 8.75 hrs represent when one cycle ends, at maximum feed concentration, and another
cycle starts, at a low concentration. (b) Average water flux through the LSRRO membranes over the BW
experiment runtime.
4.4.3 Seawater Testing
The feed and reject TDS concentrations of the LSRRO train of membranes alongside the feed
pressure during the SW experiment are illustrated in Figure 4.7a (p. 62). The SW experiment required an
additional 1.5 hrs of runtime compared to the BW experiment. During this experiment the LSRRO system
was able to produce a total of 312 L clean permeate, resulting in ~66% water recovery. Given that the
average permeate concentration during this experiment was below 50 mg/L, it is safe to assume that the
final brine concentration was close to 150 g/L. Based on a simple mass balance, this will be the result of
close to 77% water recovery from seawater having 35 g/L salinity.
The combined permeate TDS concentrations of the three stages are shown in Figure 4.7b (p. 62)
and exhibiting the familiar trend of stages 1 and 2 permeates being similar and stage 3 noticeably higher.
Stage 1 and 2 averaged a 1.7 g/L difference for the final 2.3 hrs of the first cycle, highlighting the
effectiveness of diluting the feed into stage 2 with the permeate from stage 3 to both dilute stage 2 feed
and restoring any pressure losses that occurred from stage 1 with a high-pressure input from the
secondary pump. Water flux through the LSRRO membranes is shown in Figure 4.8a (p. 63) with the
average flux for each membrane shown in Figure 4.8b (p. 63). Unlike the previous two experiments, M-1
and M-2 had the highest average water fluxes of 20.2 LMH and 17.4 LMH, respectively. Interestingly,
M-9, which displayed the second highest flux in the NaCl and BW tests, averaged lower flux than four
61
)HML(
xulF
retaW
)HML(
xulF
egarevA
|
Colorado School of Mines
|
other membranes, having an average flux of 15 LMH compared to M-8 at 16.1 LMH and M-4 at 16.1
LMH. Flux results also indicate a performance reduction from the BW experiment, with each membrane
having a lower average flux for this experiment than the BW experiment by an average of 18%, mainly
due to the increased TDS concentrations and higher concentrations of divalent ions combined with
elevated concentration polarization.
Feed Reject Feed Pressure Stage 1 Perm Stage 2 Perm
150 1200 120
(a) (b)
125 1000 100
100 800 80
75 600 60
50 400 40
25 200 20
0 0
0
0 2 4 6 8 10 12
0 2 4 6 8 10 12
Runtime (hrs) Runtime (hrs)
Figure 4.7. LSRRO membrane performance during the SW experiment with, (a) showing overall
system feed and reject TDS concentrations feed pressures and, (b) showing the individual stage permeate
concentrations.
The lowest water flux through all the LSRRO membranes was at the end of the first cycle, with
an average drop of 9.3 LMH from the initial flux, with M-1 having the largest decline in performance
from an initial flux of 25.4 LMH down to 2.8 LMH. As the feed concentrated up, the fluxes through the
higher rejection membranes in stage 1 declined by greater amounts than the membranes in stages 2 and 3,
with 16.2 LMH lost in stage 1 and 5.8 LMH being the average loss for stages 2 and 3. However,
performance returned at the start of the second cycle, with the average initial cycle flux for all membranes
increasing from 15.2 LMH in the first cycle, to 18.2 LMH to begin cycle 2. Flux was again restored at the
start of cycle 3, performing at a higher flux than cycle 1 with an initial value of 16.6 LMH, although this
was lower than cycle 2. While cycle 3 had a lower initial flux than cycle 2, over the runtime of each cycle,
cycle 3 had the highest overall average flux of 16.4 LMH, with cycles 1 and 2 averaging 13.1 LMH and
15.9 LMH, respectively. The cycle average results can be explained by the lower feed concentrations for
cycles 2 and 3 compared to cycle 1, particularly in the case of divalent ion concentrations, which were
substantially removed during cycle 1.
62
)L/g(
SDT
)isp(
erusserP
deeF
)L/g(
SDT
|
Colorado School of Mines
|
45 45
40 M1 40
35 M2 35
30 M3 30
25 M4 25
20 20
M5
15 15
M6
10 10
M7
5 5
M8
0 0
0 2 4 6 8 10 12 M9 1 2 3 4 5 6 7 8 9
Runtime (hrs) Membrane
Figure 4.8. (a) Water flux through each membrane during the SW experiment. (b) Average flux
through each membrane during the SW experiment.
4.4.4 Mining Water Experiment
The MW experiment began with cycle 1 having an initial TDS concentration similar to the BW,
at ~35 g/L. However, this was the first natural water tested, coming from a lithium-rich mining operation,
increasing the complexity of the water chemistry over the previously, laboratory prepared test waters. In
addition to the complex water matrix, another change from the previous three experiments was the initial
feed volume of 284 L, being 40% lower than the previous experiments. These differences resulted in a
large performance reduction compared to the three previous experiments, as illustrated by the TDS
concentrations shown in Figures 4.9a and 4.9b (p. 64). Though the initial feed TDS in cycle 1 was similar
to the BW experiment, the total brine TDS achieved during the MW experiment was ~69 g/L and
reaching a maximum of 112 g/L at the end of cycle 1, which was 24% lower than in the BW experiment.
The reduced initial volume had a high impact on these results, with 189 L less water to complete cycle 1,
and completely eliminating cycle 3 from the experiment. This also resulted in the total runtime of the MW
experiment being approximately 50% shorter than the previous SW experiment and the second lowest
recovery percent of the study, at 53%. The lower volume also had the effect of rapidly concentrating the
feed, requiring maximum hydraulic pressure at 2.5 hrs runtime compared to 4.5 hrs runtime for the SW
experiment.
The lower volume had a noticeably negative impact on water flux through the LSRRO
membranes during the MW experiment, as shown in Figure 4.10a (p. 64). Flux declined rapidly once peak
hydraulic pressure was reached, dropping 54% from an average flux of 9.7 LMH at 2.5 hrs runtime to a
4.4 LMH average at the end of cycle 1 (4.6 hrs runtime). The total decline in flux during cycle 1 was
substantial, losing 81% from the initial flux of 23.4 LMH. The 19 LMH decline was more than double the
decline during the first cycle of the SW experiment, which declined by 9.3 LMH. Further illustrating the
sizeable performance loss during the MW experiment, a considerably smaller fraction of flux was
recovered at the start of cycle 2 compared to the previous experiments, with average flux of 13.8 LMH
63
)HML(
xulF
retaW
)HML(
xulf
egarevA
|
Colorado School of Mines
|
initially during cycle 2 compared to 23.4 LMH at the beginning of cycle 1, a 41% performance loss. This
is substantially lower than the BW and SW experiments flux differences between cycles 1 and 2, with
BW having a 2% drop and SW realizing a gain in average flux of 17 LMH. Similar to the SW experiment,
the tighter membranes in stages 1 and 2 exhibited a higher fouling potential from the MW than the more
permeable membranes in stage 3, with 92% reduced flux for stage 1, 78% for stage 2, and 53% for stage
3. However, contrasting these results, the average fluxes for the individual membranes were the highest in
stage 1, totaling 46.7 LMH, 35.2 LMH, and 29 LMH, for stage 1, stage 2, and stage 3, respectively, as
shown in Figure 4.10b.
Feed Reject Feed Pressure Stage 1 Perm Stage 2 Perm Stage 3 Perm
120 1200 120
(a) (b)
100 1000 100
80 800 80
60 600 60
40 400 40
20 200 20
0 0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Runtime (hrs) Runtime (hrs)
Figure 4.9. LSRRO membrane performance during the MW test with, (a) showing overall system
feed and reject TDS concentrations and feed pressures and, (b) showing the individual stage permeate
concentrations.
M1 M2 M3
M4 M5 M6
M7 M8 M9
45 1200 45
40 40
1000 35 35
30 800 30
25 25
600
20 20
15 400 15
10 10
200
5 5
0 0 0
0 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9
Runtime (hrs) Membrane
Figure 4.10. (a) Water flux through the LSRRO membranes and operating feed pressure during the
MW experiment. (b) Average flux through each membrane during the MW test.
It should be noted that while a large portion of these performance reductions likely would have
been mitigated with an increased initial feed volume, the unique composition of this water (i.e., high
64
)L/g(
SDT
)HML(
xulF
retaW
Feed
Pressure
(psi)
)isp(
erusserP
deeF
)L/g(
SDT
)HML(
xulF
egarevA
|
Colorado School of Mines
|
concentration of lithium) would still result in decreased membrane performance. The osmotic pressure
exerted by each ion is in direct relation to its molar concentration in the water. As summarized in Table
4.2 (p. 56), lithium concentration in the source water was 2,710 mg/L (0.39 moles), compared to 0 moles,
8x10-4 moles, and 9x10-5 moles in the source waters for the NaCl, BW, and SW experiments, respectively.
This is approximately 500 and 4,000 times more osmotic pressure due to lithium than the BW and SW
experiments, which is noticeable at 2 hrs runtime, when water flux through the membranes declined by
approximately 39% in the next 45 minutes. Thus, similar feed TDS can result in much different osmotic
pressure characteristics. Additionally, it has been shown that the ionic radius of a ion can affect the rate at
which it is rejected by the membrane.44 Lithium has a 50% larger ionic radius than sodium (6 Å vs. 4 Å),
thus decreasing its permeability and further exacerbating concentration polarization effects.45
4.4.5 Permian Basin PW Experiment
The last water tested was PW from the Permian Basin. This PW is characterized by its high
concentrations of many different constituents, including ammonia, hardness, and TDS.46 This water
profile makes selecting appropriate treatment processes challenging because the high levels of
contaminants can hinder the ability of conventional treatment systems to operate effectively (e.g., high
TDS can devastate biological treatment process, inorganic scaling and organic biofouling can shorten the
lifespan of expensive membrane processes). For this reason, before testing PW in the LSRRO, several
pretreatment processes were utilized to mitigate the potential for the membranes to become damaged due
to fouling and scaling. The TDS concentrations achieved throughout the PW experiment are shown in
Figures 4.11a and 4.11b (p. 66). The PW had 133 g/L TDS, the highest initial feed TDS concentration of
all the waters tested. The considerably higher TDS resulted in the maximum hydraulic pressure being
reached upon the initial start of the experiment, lowering water flux through the membranes, and resulting
in cycle 1 taking 11 hrs to complete. The lower flux resulted in the lowest recovery in the study, 32%. At
the end of the experiment, the calculated brine salinity reached 195 g/L, close to the maximum
concentration achieved in this study during the NaCl experiment (~210 g/L).
The permeate TDS concentrations for the three stages are shown in Figure 4.11b (p.66). Each
stage, expectedly, continued the trend from the previous tests, with the exception of the last 2 hrs of
runtime during cycle 1, when stage 2 permeate salinity rose sharply, getting much closer to the TDS
concentration of stage 3, demonstrating the high initial feed TDS affecting each stage in the LSRRO
process. Abrupt swings in TDS were observed, particularly during cycle 3 when reject flowrate from the
SWRO was diluting the feed from T-4 and when it was not during intermitted SWRO operation.
The reduced performance during this test is also signaled by the noticeably diminished water flux
through all the membranes, as illustrated in Figure 4.12a (p. 67). Overall flux average during the PW
experiment was the lowest in the study (11 LMH), compared to the 18 LMH, 15 LMH, and 12 LMH from
65
|
Colorado School of Mines
|
the BW, SW, and MW experiments, respectively. While the large flux reduction that occurred during the
first cycle was similar to the previous experiments, with the increased runtime of the PW experiment, the
LSRRO operated at this lower flux for a much longer time, explaining the low TDS removal and the
longer runtime. Flux was restored at the start of cycles 2 and 3, similar to the previous experiments with
the peak fluxes averaging 18.3 LMH, 20.4 LMH, and 18.5 LMH for cycles 1, 2, and 3, respectively,
reinforcing the difficulty this water poses to treatment processes. These results were not matched by the
results found with the individual averages for each membrane, as can be seen in Figure 4.12b (p. 67).
Several membranes, unexpectedly, demonstrated higher averages during the PW than the averages
attained in the MW test. For example, M-6, M-8, and M-9 had averages that were 0.3 LMH, 2.3 LMH,
and 2.5 LMH higher than their fluxes during the MW test. Specifically with M-8 and M-9, the LSRRO
performed 20% better with a more difficult water stream. While any future testing should look further
into these flux results, the higher water fluxes through M-8 and M-9 during the PW experiment were most
likely the result of the additional volume of the PW experiment, which allowed fluxes to rebound at a
higher value than the was observed during the two-cycle MW experiment. The flux results reported here
should be able to allow manufacturers to further optimize membrane permeabilities for future studies.
One last notable point that must be mentioned regarding these results not shown in Figure 4.12; the low
flux increased the runtime required for stage 3 permeate to reach the volume setpoint for T-5. This meant
that the secondary pump feeding stage 2 was not operating as frequently while also running at reduced
power compared to the previous tests. Without the hydraulic pressure supplementation provided by the
secondary pump, the demands on the primary pump to maintain system pressure were amplified,
increasing the potential for damage to system components. Every metric used to evaluate the LSRRO
pilot’s performance reaffirms the challenges associated with PW treatment.
Feed Reject Feed Pressure Stage 1 Perm Stage 2 Perm Stage 3 Perm
250 1200 160
(b)
1100 140
200
1000 120
150 900 100
800 80
100 700 60
600 40
50
500 20
0 400 0
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
Runtime (hrs) Runtime (hrs)
Figure 4.11. LSRRO membrane performance during the PW experiment with, (a) showing overall
system feed and reject TDS concentrations and feed pressures and, (b) showing the individual stage
permeate concentrations.
66
)L/g(
SDT
)isp(
erusserP
deeF
)L/g(
SDT
|
Colorado School of Mines
|
M1 M2 M3
M4 M5 M6
M7 M8 M9
60 1200 30
(b)
1100
25
45 1000
900 20
30 800 15
700
10
15 600
500 5
0 400 0
0 2 4 6 8 10 12 14 16 18 1 2 3 4 5 6 7 8 9
Runtime (hrs) Membrane
Figure 4.12. (a) Water flux through the LSRRO membranes and feed pressure during the PW
experiment. (b) Average flux through each membrane during the PW experiment.
4.4.6 SWRO Polishing Performance
The SWRO was the final step in the LSRRO process, desalinating a portion of the permeate of
stage 1. The results for the SW-30 membrane are summarized in Table 4.3. Understanding how the
LSRRO membranes are able to reduce the osmotic pressure of hypersaline brines is crucial to advancing
research on this new technology; however, ultimately, final system recovery of high-quality permeate will
be the parameter used when comparing LSRRO to other RO configurations.
Table 4.3. Final permeate recovery percentages and TDS concentrations for each of the tested water
streams. The first three waters were prepared in lab, the last two are from natural sources.
Water Average permeate TDS
Feed Vol. (L) Permeate Vol. (L) Recovery %
Stream (mg/L)
NaCl 473 327 69 183
BW 473 304 64 41.2
SW 473 312 66 49.0
MW 284 151 53 30.9
PW 473 153 32 333
The individual permeate flux and TDS concentrations for each experiment, representing the
recovery percent and quality of product water, are shown in Figure 4.13 (p. 69). As was the case with the
LSRRO membranes, the SW-30 was operated to maintain a constant flux of ~20 LMH. Gaps in the data
points represent operation from one cycle to the next in which the SWRO membrane was not operating.
The operational runtime for the SW-30, dependent on how much volume stage 1 permeate was
generating, decreased in each successive experiment conducted, matching the observed decline in
performance discussed previously. Despite the differences in permeate volumes, a similar trend was
67
)HML(
xulF
retaW
Feed
Pressure
(psi)
)HML(
xulF
egarvA
|
Colorado School of Mines
|
observed at the end of each experiment, when feed concentrations were at their highest, permeate quality
decreased, increasing TDS concentrations. This was particularly true for the MW experiment, as TDS
concentration went from 12 mg/L to 160 mg/L during the final 2 hrs of the experiment. Interestingly, the
first test, NaCl, recorded the highest TDS among the first four experiments, despite consisting of the
simplest water matrix. Additionally, the stage 1 permeate TDS feeding the SW-30 for the NaCl test was
the highest in the study, at 21.1 g/L, while stage 1 permeates for BW, SW, MW, and PW were 6.2 g/L,
9.4 g/L, 6.0 g/L, and 13.9 g/L, respectively, indicating a strong, though not definitive, correlation between
SW-30 feed TDS concentration and final permeate quality. Despite the reduced permeate quality, the
NaCl test produced the highest recovery percentage of all experiments, at 68%. Recoveries were similar
from the BW and SW tests, achieving 64% and 66%, respectively. These performance results, while
generating brines of 109 g/L, 79 mg/L, and 110 g/L, respectively, compare favorably to traditional RO
that typically sees 50% recoveries and 70,000 mg/L-TDS brines. Recovery during the subsequent MW
experiment was notably reduced, achieving 53% recovery, on par with traditional RO. It should be noted,
though, as was the case with the previously discussed LSRRO membranes, the 40% reduction in initial
feed volume was the determining factor with the recovery results. When analyzing and comparing the
results from each experiment, specifically the permeate volumes generated during each cycle, percent
recovery during the MW experiment was similar to the three previous experiments. For example, during
their final cycles, BW and SW recovered approximately 67%, whereas during MW final cycle (cycle2),
with an initial volume comparable to BW and SW (117 L for MW, 114 L for SW, and 95 L for BW), 60%
was recovered, suggesting similar total recovery would have been achieved, albeit still slightly lower, if
the initial volume was the same. Nevertheless, actual recovery realized during MW experiment compares
favorably with traditional RO.
The PW experiment resulted in substantially reduced recovery, achieving 32%, a result of the
initial feed composition negatively affecting the flux from stage 1 membranes and feeding the SW-30
much less volume than the previous experiments. For example, the average water flux for stage 1
permeate during the NaCl, BW, SW, MW, and PW were 60 LMH, 53 LMH, 50 LMH, 47 LMH, and 38
LMH, respectively. Additionally, the high feed TDS continued to have an impact on the LSRRO’s
performance, generating the lowest quality final permeate of the study, with a TDS concentration of 333
mg/L. While the quality of permeate during the PW experiment was the poorest in this study, this TDS
concentration is within the range of the EPA’s National Secondary Drinking Water Regulations.47
68
|
Colorado School of Mines
|
indicating that the membranes had undergone compaction after the NaCl test operating at a hydraulic
pressure of 1,100 psi. Once deformation of the membranes had stabilized, the permeabilities did as well.
The tests were conducted at 100 psi and 125 psi, one-tenth the pressure during experimentation, and thus
not demonstrating any compaction during the initial testing. The consistency in permeability did not
continue after the MW experiment, with a distinct 17% decline from the previous experiments across all
membranes from the post-SW testing.
4
Initial Test NaCl BW SW MW PW 24 hr Acid Wash
3.5
3
2.5
2
1.5
1
0.5
0
1 2 3 4 5 6 7 8 9
Membrane Element
Figure 4.14. The water permeability of each membrane after the conclusion of each experiment are
presented. Initial testing was conducted on new membranes, before undergoing compaction. 24 hr acid
wash was not performed after an experiment, it was performed several weeks after the PW test.
As a result of the rapid decline in performance, the acid cleaning protocols were modified by
adding an additional base (pH=10) cleaning to remove any potential organic buildup on the membranes
and increasing in duration of the cleaning by several hours post the PW experiment. This resulted in a
notable improvement after the PW experiment, raising the permeability of each membrane by an average
of 10% over the post-MW test. The final integrity test was conducted 14 days after the PW following a 24
hr acid washing. Interestingly, while permeability was improved by ~7% on average for stages 2 and 3
membranes, declines in permeability were observed in stage 1 membranes by a similar average of 7%.
This was, perhaps, an indication of residual scaling sloughing off of the last six membranes and those
particles being recycled through the LSRRO and reintroduced to the stage 1 membranes. Regretfully,
membrane autopsies were not conducted before completion of this paper but should be included in future
studies as membrane availability becomes more accessible. Nevertheless, after accounting for compaction
after the NaCl experiment, the final integrity test showed a decline of 0.14, or 6%, after the completion of
70
)rab/HML(
ytilibaemreP
retaW
|
Colorado School of Mines
|
four high salinity experiments with a total runtime of ~57 hrs. The results of these integrity tests suggest
that LSRRO membranes, when properly maintained, are capable of consistent performance during
continuous operation desalinating hypersaline brines.
4.6 Water Quality
4.6.1 Brackish Water, Seawater, and Mining Water Quality
The concentrations of major ions and TDS concentrations during the last cycle of each
experiment (except NaCl experiment) in the are summarized in Table 4.4 (p. 72). Calcium, potassium,
lithium, magnesium, sodium, chloride, and sulfate are the ions found in highest abundance in the initial
feed. Results shown are from samples collected during the last cycle of each test, in the middle of a step
such that the LSRRO was operating as close to steady-state as possible. A final summary of ion
concentration for each experiment summarizing the initial and final concentration is detailed in Table 4.5
(p. 73).
Distinct, consistent trends were observed across all experiments, that being the concentrations of
all ions were lower at the beginning of a stage, increasing in each successive membrane (i.e., M-1 had
lower concentrations in the permeate than M-2, which had lower than M-3). There was one notable
exception to this trend as sulfate concentrations appeared to be higher in the leading membrane than in the
following one for each stage across all experiments, but only minimally. Noticeably, though, the LSRRO
displayed exceptional performance in the removal of divalent ions. For example, in the SW experiment,
calcium, magnesium, and sulfate concentrations were reduced by 94%, 99.3%, and 97%, respectively,
from the feed to M-1 permeate. Overall removal of all ions, from initial feed to final permeate, were
rejected at over 99% (Table 4.5, p. 73), producing high-quality product water with TDS concentrations of
183 mg/L, 41.2 mg/L, 49 mg/L, 30.9 mg/L, and 333 mg/L for the NaCl, BW, SW, MW, and PW
experiments, respectively, meeting or exceeding the performances of traditional RO systems.
71
|
Colorado School of Mines
|
Table 4.5. Initial and final TDS concentrations of select ions for each test.
Ca TDS
Test Sample K Li Mg Na Cl- SO -
(mg/L) 4 (mg/L)
Initial Feed 27,500 42,500 70,000
NaCl NA NA NA NA NA
Final Permeate 72 111 183
Initial Feed 149 187 5.7 619 11,925 20,041 391 34,100
BW
Final Permeate 0.14 0.5 <0.004 0.05 7.5 12.7 0.013 41.2
Initial Feed 676 639 0.6 2,212 17,420 29,283 5,254 55,500
SW
Final Permeate 0.07 0.05 <0.004 <0.001 16.0 26.8 <0.01 49.0
Initial Feed 389 226 2,710 794 2,591 26,748 121 34,500
MW
Final Permeate <0.004 0.15 3.5 <0.001 2.8 15.2 0.12 30.9
Initial Feed 1,232 701 22.4 618 51,684 77,093 262 133,000
PW
Final Permeate <0.004 0.4 0.1 <0.001 52 26 <0.1 333
4.7 Conclusion
In this study, a novel membrane-based desalination process, low-salt rejection reverse-osmosis
(LSRRO), was tested at pilot scale. LSRRO has been previously modeled and was shown to have a
distinct advantage over traditional RO – the ability to desalinate high salinity waters while operating at
typical RO hydraulic pressures. Our pilot system was configured in a 3-stage design, with each stage
containing three membrane elements arranged in series. The membranes were specially manufactured by
FTS-H2O, with each membrane having a progressively higher salt permeability. Five different water
compositions were used to evaluate LSRRO, from a simple NaCl/DI water solution to pretreated
produced water from oil and gas operations in the Permian Basin.
The LSRRO pilot was effective in meeting or exceeding current traditional SWRO system
standards in percent water recovery or generated brine salinity across all water compositions tested.
Additionally, the permeate streams produced during all experiments achieved TDS concentrations below
the EPA secondary standard for drinking water. Membrane integrity testing verified that the LSRRO
membranes did not experience permanent damage or incur irreversible scaling during the ~57 hrs of total
runtime in this study.
Initial feed volumes appeared to have a major influence on system operation and performance, a
variable unable to be tested during this study due to the size and configuration of the pilot system. Also,
the pilot system was not able to operate in full continuous mode, allowing only small windows of steady-
state operating conditions and water compositions. Testing the LSRRO process on a continuous, larger
scale, with a constant feed composition, will allow better characterization and optimization of the LSRRO
process, but will require a very large system with more than 36 standard membrane elements connected in
series. In a larger scale pilot system, teed concentrations will reach steady-state and would likely result in
improved performance over the observed in this study. Future research should also focus on membrane
73
|
Colorado School of Mines
|
permeability optimization and improving LSRRO water recovery without compromising membrane
integrity. Additionally, while simulated studies have investigated energy consumption of LSRRO
compared to other processes, evaluating the realized energy requirements of a large-scale pilot system
will improve the ability to accurately compare LSRRO to traditional RO. Nevertheless, the results of this
study indicate the ability of LSRRO to recover high-quality permeate from hypersaline brines, including
pretreated PW and is a promising desalination process for high salinity streams and should continue to be
investigated by researchers.
4.8 Acknowledgments
The authors would like to acknowledge the financial and technical support for this study provided
by the National Alliance for Water Innovation (NAWI) in connection with the U.S. Department of Energy
(DOE). Additional financial support was also provided by the ZOMA Foundation, Bayswater
Exploration, LLC, The Sussman Foundation, and the Produced Water Society. The authors are grateful
for the technical support provided by Mr. Michael Veres (design, construction, and operational support of
the LSRRO pilot system), Mr. Tani Cath (computer process controls), Nathan Rothe and Amy Ashford
Boczon (chemical analysis).
4.9 References
1. Nations, U., The United Nations World Water Development Report 2022: Groundwater: Making
the Invisible Visible. In UNESCO Paris, France: 2022.
2. Vargas Zeppetello, L. R.; Raftery, A. E.; Battisti, D. S., Probabilistic projections of increased heat
stress driven by climate change. Communications Earth & Environment 2022, 3, (1), 183.
3. Kalogirou, S. A., Seawater desalination using renewable energy sources. Progress in energy and
combustion science 2005, 31, (3), 242-281.
4. Misdan, N.; Lau, W.; Ismail, A., Seawater Reverse Osmosis (SWRO) desalination by thin-film
composite membrane—Current development, challenges and future prospects. Desalination 2012, 287,
228-237.
5. Glater, J., The early history of reverse osmosis membrane development. Desalination 1998, 117,
(1-3), 297-309.
6. Feinberg, B. J.; Ramon, G. Z.; Hoek, E. M., Thermodynamic analysis of osmotic energy recovery
at a reverse osmosis desalination plant. Environmental science & technology 2013, 47, (6), 2982-2989.
7. Hancock, N. T.; Black, N. D.; Cath, T. Y., A comparative life cycle assessment of hybrid osmotic
dilution desalination and established seawater desalination and wastewater reclamation processes. Water
research 2012, 46, (4), 1145-1154.
8. Fritzmann, C.; Löwenberg, J.; Wintgens, T.; Melin, T., State-of-the-art of reverse osmosis
desalination. Desalination 2007, 216, (1-3), 1-76.
9. Hickenbottom, K. L.; Cath, T. Y., Sustainable operation of membrane distillation for
enhancement of mineral recovery from hypersaline solutions. Journal of Membrane Science 2014, 454,
426-435.
74
|
Colorado School of Mines
|
CHAPTER 5
CONCLUSIONS
5.1 Research Synopsis
This dissertation is a collection of three full-length research articles detailing the comprehensive
assessment of advanced treatment processes for high salinity PW. These investigations include: (a) an
evaluation of a membrane bioreactor (MBR) to effectively remove organic compounds from PW with
different compositions while characterizing the changes in the microbial community of the MBR, (b) an
assessment of an MBR to be included in a short treatment train capable of treating high salinity PW from
raw PW through to final desalination including the ability to remove toxic substances from PW, and (c) a
performance evaluation of a novel reverse osmosis (RO) process, low-salt rejection reverse osmosis
(LSRRO), on the ability to desalinate high salinity water streams while operating under traditional
seawater reverse osmosis (SWRO) hydraulic pressures on a small pilot-scale system.
5.1.1 Summary of MBR Evaluation
This study investigated the ability of a pilot-scale MBR to effectively pre-treat high salinity PW
as the singular process before eventual membrane-based desalination treatment. Despite changes to TDS
concentrations (from 27 g/L to over 100 g/L), the MBR’s performance in removing suspended solids and
DOC remained consistent, averaging 86% with PW from DJ-Basin and 66% from PW from Permian
Basin. No acclimation periods were required following the initial acclimation at the beginning of the
study. After an initial acclimation period, the microbial community was largely stable up to ~80 g/L TDS;
at higher salinities, a more diverse community with a higher prevalence of PW-derived organisms was
observed but was not associated with any change in performance. Additionally, little change to the
dominant microbial genera was observed during Permian Basin PW treatment.
5.1.2 Summary of Complete Treatment Train for High Salinity PW
In this study we have investigated the toxicity and contaminant removal of individual treatment
processes of a high salinity PW treatment train utilizing four-treatment processes to reclaim PW from the
Permian basin to a theoretical discharge water. Removal performance of this treatment system was shown
to be effective at reducing constituent concentrations to levels at or below EPA effluent discharge limits.
Additionally, for those inorganic contaminants for which established drinking water standards exist,
concentrations were reduced to below the EPA primary and secondary drinking water standards,
including TDS < 500 mg/L. In MD Distillates, most contaminants were removed at over 99% from the
raw PW, including NORM, select organics, metals, and anions.
Toxicity testing performed after each treatment process showed a substantial reduction to non-
78
|
Colorado School of Mines
|
detect levels from the diluted raw PW to the MD Distillate in both cytotoxicity and HMOX1 gene
expression, with no cytotoxicity and oxidative stress observed in the MD Distillates. Water concentrates
induced CYP1A1 and CYP1B1 gene expression, although to lesser extents compared to the raw water,
indicating that some AhR-like chemicals were still present in 2x concentrates of raw PW, which could not
be removed by solid phase concentration. To further verify AhR-like chemicals existed in water
concentrates, AhR-KO MCF-7 cells were used. Induction of AhR biomarker gene CYP1A1 and CYP1B1
was evident with 3x concentrates in WT MCF-7 cells, but no such induction was observed in AhR-KO
cells, confirming AhR-like chemicals existed in raw water concentrates and raw water dilutions (data not
shown).
5.1.3 Summary of LSRRO Performance to Desalinate Hypersaline Brines
The LSRRO pilot was effective in meeting or exceeding current traditional SWRO system
standards (i.e., 50% recovery while generating brines with 70 g/L-TDS) in percent water recovery or
generated brine salinity across all water compositions tested. Additionally, the permeate streams produced
during all experiments achieved TDS concentrations below the EPA secondary standard for drinking
water. Membrane integrity testing verified that the LSRRO membranes did not experience permanent
damage or incur irreversible scaling during the ~57 hrs of total runtime in this study.
Initial feed volumes appeared to have a major influence on system operation and performance, a
variable unable to be tested during this study due to the size and configuration of the pilot system. Also,
the pilot system was not able to operate in full continuous mode, allowing only small windows of steady-
state operating conditions and water compositions. Testing the LSRRO process on a continuous, larger
scale, with a constant feed composition, will allow better characterization and optimization of the LSRRO
process, but will require a very large system with more than 36 standard membrane elements connected in
series. In a larger scale pilot system, feed concentrations will reach steady-state and would likely result in
improved performance over the observed in this study. Future research should also focus on membrane
permeability optimization and improving LSRRO water recovery without compromising membrane
integrity. Additionally, while simulated studies have investigated energy consumption of LSRRO
compared to other processes, evaluating the realized energy requirements of a large-scale pilot system
will improve the ability to accurately compare LSRRO to traditional RO.
5.2 Future Work
While this dissertation was able to report promising results for the use of an MBR as part of a
complete treatment train for high salinity PW as well as the capabilities of a LSRRO system to desalinate
hypersaline brines, additional research needs to be completed before these processes can become practical
to implement in the field. The largest focus for researchers moving forward should be on increasing the
79
|
Colorado School of Mines
|
scale of the experimental systems. The studies presented in this dissertation are an improvement in scale
over previous studies, but at the small pilot-scale level, these experiments are limited in their ability to
represent a true commercial system's capabilities. The MBR, for example, was much larger than other
studied systems at 70 L; however, for a more accurate representation, an MBR capable of treating many
thousands of liters per day will need to be constructed and evaluated. This is also true of the complete
treatment train investigation included in this dissertation in Chapter 3, where scaling up these processes
would give a more complete understanding of these processes in the field. Additionally, studying the
possible benefits of nutrient supplementation for the microorganisms in the MBR to optimize DOC
removal should be assessed, along with the economical tradeoff of increasing costs with the addition of
nutrients compared to any increased performance. Future work on LSRRO should focus on membrane
optimization to increase the percentage of water recovery as well as the maximum salt rejection
properties. This is in addition to scaling up the experimental system to be able to operate continuously,
with a constant feed TDS concentration, as this should reduce concentration polarization effects on the
membranes, allowing them to operate at their peak, while also generating increased volumes of high-
quality permeate. Finally, the research presented in this dissertation did not investigate any economic
variables of full-scale implementation of these processes, such as energy consumption, initial capital
costs, or tecno-economic assessment. These will need to be performed to provide a comprehensive
evaluation and increase our understanding of the practicality of operating these systems on a commercial
scale.
80
|
Colorado School of Mines
|
APPENDIX A
SUPPORTING INFORMATION FOR CHAPTERS 2, 3, AND 4
A.1 Supporting Information for Chapter 2
Two methods were employed to properly identify PEGs, PPGs, PEG-carboxylates, and PEG-
dicarboxylates through LC-qTOF analysis. The first method was based on the exact mass and
fragmentation of these compounds and their proton, ammonium, and sodium adducts (Rosenblum et al.,
2016). For example, the identification of PEG-EO9 in the 40 g/L sample, was determined based on the
exact mass of this compound, 414.2465. The mass for the proton adduct for this compound is 415.2538,
the ammonium adduct 432.2803, and the sodium adduct 437.2357. Figure A.1 shows the MS-MS of this
compound. The sodium adduct is shown as the most abundant with the other adducts at much lower
concentrations. As this compound gets fragmented, ions at these new masses are shown with a strong
intensity. When the sodium adduct at 437.2357 loses a water molecule (mass units of 18), this
corresponds with the ion at m/z 419.2275. Then for each ethylene oxide that fragments off, a subsequent
reduction in mass units by 44.0262 is seen. Thus, after 419.2275, strong intensity ions at m/z 375.1801,
331.1441, 287.1494, 243.1226, 199.0966, and 155.0704 are seen.
Figure A.1. The MS-MS of PEG-EO9 from sample taken with feed TDS concentration of 40 g/L. The
sodium adduct of this compound is strongest and shown on the right with a mass of 437.2357.
81
|
Colorado School of Mines
|
The second method used for identification was the Kendrick mass defect. The Kendrick mass
defect is the amount by which the measured mass multiplied by the scaling factor differs from the
nominal mass. When the scaling factor for PEGs, which is 0.99404559, is multiplied by the measured
mass of 437.2362, the Kendrick mass defect is 0.975. This same Kendrick mass defect is seen in this
sample for all sodium adducts of PEGs and is summarized in Table A.1.
Table A.1. Kendrick mass table for PEGs found in feed water at 40 g/L. The Kendrick mass scaling
factor used is 0.99404559. PEG-EO6 was not found in this sample.
Retention Measured Kendrick Molecular Identified Calculated
Kendrick mass
time (min) mass mass defect formula compound exact mass
7.1 261.1307 260.975212 0.975 C10H22O6 PEG-EO5 261.1309
10.8 349.1835 348.975581 0.976 C14H30O8 PEG-EO7 349.1833
11.4 393.2098 392.975666 0.976 C16H34O9 PEG-EO8 393.2095
11.8 437.2362 436.975851 0.976 C18H38O10 PEG-EO9 437.2357
12.3 481.2621 480.9755368 0.976 C20H42O11 PEG-EO10 481.2619
12.6 525.2886 524.9758216 0.976 C22H46O12 PEG-EO11 525.2881
12.9 569.3153 568.9763063 0.976 C24H50O13 PEG-EO12 569.3144
Another aspect of this study was the determination of system performance after substantial
changes in TDS concentrations were fed into the MBR. This was accomplished by feeding D-J basin PW
with NaCl added to increase the TDS concentration to 100 g/L into the MBR at the conclusion of the
month-long Permian basin study. With no acclimation period, the TDS was immediately dropped to 75
g/L on June 22nd . This concentration was held until August 15th when the concentration was again
dropped by 25 g/L to 50 g/L. And finally, on December 7th, the TDS level was raised by 50 g/L back up
to 100 g/L. The results of this experiment follow the trend of the rest of this study by showing a consistent
permeate DOC concentration, regardless of the influent DOC levels. The permeate DOC concentrations at
each phase of this experiment were 13, 14, and 13.6 mg/L, respectively. The MBR showed resiliency in
its performance to remove organics even during highly stressful environmental changes. Again, showing
that the MBR can be a viable PW treatment option regardless of the water chemistry of the PW it is
treating.
82
|
Colorado School of Mines
|
Table A.2. List of read counts and accession IDs for all samples sequenced in the study
Raw Filtered Denoised Final BioSample
Sample ID Sample Type Day
Reads reads reads reads Accession (SRA)
MBR-D0-1 Sludge 0 198994 183595 180789 178505
MBR-D0-2 Sludge 0 185903 175266 173384 170395 SAMN22072450
MBR-D0-3 Sludge 0 133790 122817 120085 114532
MBR-D10-1 Sludge 10 205503 193309 191065 186570
MBR-D10-2 Sludge 10 195993 184643 182668 178890 SAMN22072451
MBR-D10-3 Sludge 10 207770 195890 193912 190016
MBR-D27-1 Sludge 27 190234 179821 178434 175118
MBR-D27-2 Sludge 27 181722 172163 171073 167867 SAMN22072452
MBR-D27-3 Sludge 27 161878 152045 150581 146186
MBR-D56-1 Sludge 56 170097 161741 160863 158952
MBR-D56-2 Sludge 56 195341 185412 184499 182145 SAMN22072453
MBR-D56-3 Sludge 56 182819 173485 172428 170434
MBR-D94-1 Sludge 94 195936 185070 184178 181714
MBR-D94-2 Sludge 94 201793 190600 189683 186871 SAMN22072454
MBR-D94-3 Sludge 94 183857 173366 172518 169778
MBR-D118-1 Sludge 118 218542 206086 205004 201743
MBR-D118-2 Sludge 118 203673 192058 190973 187711 SAMN22072455
MBR-D118-3 Sludge 118 217044 205267 203965 200984
MBR-D140-1 Sludge 140 193412 182473 181554 178894
MBR-D140-2 Sludge 140 207366 196136 195146 192637 SAMN22072456
MBR-D140-3 Sludge 140 204682 192602 191349 188292
MBR-D154-1 Sludge 154 181575 171257 169937 167322
MBR-D154-2 Sludge 154 210449 197540 196134 192872 SAMN22072457
MBR-D154-3 Sludge 154 186760 175753 174589 171996
MBR-D168-1 Sludge 168 210448 198574 197655 195047
MBR-D168-2 Sludge 168 198556 187270 186444 184173 SAMN22072458
MBR-D168-3 Sludge 168 190856 179838 178946 176580
MBR-D174-1 Sludge 174 204846 193574 192612 189637
MBR-D174-2 Sludge 174 192469 182272 181484 179011 SAMN22072459
MBR-D174-3 Sludge 174 213139 201563 200403 197017
MBR-D185-1 Sludge 185 214101 201833 201006 198262
MBR-D185-2 Sludge 185 206380 194403 193489 191051 SAMN22072460
MBR-D185-2 Sludge 185 184967 174614 173835 171484
MBR-D194-1 Sludge 194 180680 170363 169533 167246
MBR-D194-2 Sludge 194 191916 181458 180664 178293 SAMN22072461
MBR-D194-3 Sludge 194 199468 188072 187144 184369
MBR-D206-1 Sludge 206 196013 185004 184006 181074
MBR-D206-2 Sludge 206 196629 185117 184035 181216 SAMN22072462
MBR-D206-3 Sludge 206 196367 185016 183974 181012
MBR-D222-1 Sludge 222 199150 188484 187518 184688
MBR-D222-2 Sludge 222 193835 183450 182479 179490 SAMN22072463
MBR-D222-3 Sludge 222 215918 204120 203192 200002
MBR-D235-1 Sludge 235 192986 182537 181513 178883
MBR-D235-2 Sludge 235 183858 174179 173206 170961 SAMN22072464
MBR-D235-3 Sludge 235 212172 200892 199792 196592
MBR-D250-1 Sludge 250 213873 202519 201144 197761
MBR-D250-2 Sludge 250 218105 205574 204196 200741 SAMN22072465
MBR-D250-3 Sludge 250 205873 194585 193502 190361
MBR-D263-1 Sludge 263 201244 185225 183774 180139
SAMN22072466
MBR-D263-2 Sludge 263 201340 185289 183605 179637
85
|
Colorado School of Mines
|
MBR-D263-3 Sludge 263 206922 190225 188373 184602
MBR-D278-1 Sludge 278 191621 176201 174476 171269
MBR-D278-2 Sludge 278 195331 179704 178282 174740 SAMN22072467
MBR-D278-2 Sludge 278 213200 196468 194903 191467
MBR-D291-1 Sludge 291 210830 193581 191760 187305
MBR-D291-2 Sludge 291 206096 189779 187995 184076 SAMN22072468
MBR-D291-3 Sludge 291 194237 179096 177688 174065
MBR-D307-1 Sludge 307 143531 132428 131280 128355
MBR-D307-2 Sludge 307 200713 184048 182458 178656 SAMN22072469
MBR-D307-3 Sludge 307 196556 180814 179079 175226
MBR-D316-1 Sludge 316 212542 196314 194414 190366
MBR-D316-2 Sludge 316 189604 174892 173233 169215 SAMN22072470
MBR-D316-3 Sludge 316 217514 200544 198674 194297
MBR-D326-1 Sludge 326 180256 166318 164317 160627
MBR-D326-2 Sludge 326 184849 170745 169184 165788 SAMN22072471
MBR-D326-3 Sludge 326 179881 166341 164915 161758
FD-D140 Feed 140 185676 171908 170159 166663 SAMN22072472
FD-D168 Feed 168 201315 185466 183552 179539 SAMN22072473
FD-D174 Feed 174 196488 181298 179177 174868 SAMN22072474
FD-D185-1 Feed 185 183024 166937 164730 160867
SAMN22072475
FD-D185-2 Feed 185 216578 197516 195127 190542
FD-D194 Feed 194 181955 165505 163162 158842 SAMN22072476
FD-D206-1 Feed 206 194101 175880 173001 168105
SAMN22072477
FD-D206-2 Feed 206 186887 169652 167415 162939
FD-D222 Feed 222 189969 172958 171130 166618 SAMN22072478
FD-D235 Feed 235 216481 196740 194610 190173 SAMN22072479
FD-D250 Feed 250 217393 198457 196166 191641 SAMN22072480
FD-D263 Feed 263 127132 116137 114465 111732 SAMN22072481
FD-D278 Feed 278 205204 186447 184257 180153 SAMN22072482
FD-D291 Feed 291 198733 181870 179938 175958 SAMN22072483
FD-D307 Feed 307 172697 159100 157087 153863 SAMN22072484
FD-D316 Feed 316 101935 93490 92091 90144 SAMN22072485
FD-D326 Feed 326 119223 109328 107677 105162 SAMN22072486
MBR-D370 Sludge 370 155891 136973 134281 108427 SAMN22072487
MBR-D370 Sludge 379 158992 141183 137835 108924 SAMN22072488
MBR-PR- Sludge-
386 170063 150423 145774 114051 SAMN22072489
D386 Permian PW
MBR-PR- Sludge-
393 182856 16163 158644 124473 SAMN22072490
D393 Permian PW
MBR-PR- Sludge-
400 152636 135253 132661 103644 SAMN22072491
D400 Permian PW
MBR-SS- Sludge-
452 166151 147423 144419 112560 SAMN22072492
D452 Salinity shock
MBR-SS- Sludge-
467 179357 15785 154662 129043 SAMN22072493
D467 Salinity shock
MBR-SS- Sludge-
488 163039 144356 142337 119884 SAMN22072494
D488 Salinity shock
MBR-SS- Sludge-
506 165010 147734 145417 115871 SAMN22072495
D506 Salinity shock
MBR-SS- Sludge-
618 155562 139068 137575 116754 SAMN22072496
D618 Salinity shock
86
|
Colorado School of Mines
|
Table A.4. Detailed results of sample analysis. Analyses was conducted by US EPA, Region 6. U result was below reporting limit, RL.
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
MBR FEED DUP EPA 8270 Acenaphthene U 1
MBR FEED DUP EPA 8270 Acenaphthylene U 1
MBR FEED DUP EPA 8270 Acetophenone U 1
MBR FEED DUP EPA 8270 Adamantane U 2
MBR FEED DUP EPA 8270 Anthracene U 1
MBR FEED DUP EPA 8270 Atrazine U 1
MBR FEED DUP EPA 8270 Azobenzene U 1
MBR FEED DUP EPA 8270 Benzaldehyde U 1
MBR FEED DUP EPA 8270 Benzoic acid U 5
MBR FEED DUP EPA 8270 Benzo (a) anthracene U 1
MBR FEED DUP EPA 8270 Benzo (a) pyrene U 1
MBR FEED DUP EPA 8270 Benzo (b) fluoranthene U 1
MBR FEED DUP EPA 8270 Benzo (g,h,i) perylene U 1
MBR FEED DUP EPA 8270 Benzo (k) fluoranthene U 1
MBR FEED DUP EPA 8270 Benzyl alcohol U 1
MBR FEED DUP EPA 8270 1,1'-Biphenyl U 1
MBR FEED DUP EPA 8270 Bis(2-chloroethoxy)methane U 1
MBR FEED DUP EPA 8270 Bis(2-chloroethyl)ether U 1
MBR FEED DUP EPA 8270 Bis(2-chloro-1-methylethyl)ether U 1
MBR FEED DUP EPA 8270 Bis(2-Ethylhexyl)adipate U 1
MBR FEED DUP EPA 8270 Bis(2-ethylhexyl)phthalate U 2
MBR FEED DUP EPA 8270 4-Bromophenyl phenyl ether U 1
MBR FEED DUP EPA 8270 2-Butoxyethanol U 1
MBR FEED DUP EPA 8270 2-Butoxyethanol Phosphate U 5
MBR FEED DUP EPA 8270 Butyl benzyl phthalate U 3
MBR FEED DUP EPA 8270 Carbazole U 3
MBR FEED DUP EPA 8270 Caprolactam U 1
90
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
MBR FEED DUP EPA 8270 4-Chloroaniline U 3
MBR FEED DUP EPA 8270 2-Chloronaphthalene U 1
MBR FEED DUP EPA 8270 2-Chlorophenol U 2
MBR FEED DUP EPA 8270 4-Chlorophenyl phenyl ether U 1
MBR FEED DUP EPA 8270 4-Chloro-3-methylphenol U 2
MBR FEED DUP EPA 8270 Chrysene U 1
MBR FEED DUP EPA 8270 Dibenzofuran U 1
MBR FEED DUP EPA 8270 Dibenz (a,h) anthracene U 1
MBR FEED DUP EPA 8270 1,2-Dichlorobenzene U 1
MBR FEED DUP EPA 8270 1,3-Dichlorobenzene U 1
MBR FEED DUP EPA 8270 1,4-Dichlorobenzene U 1
MBR FEED DUP EPA 8270 3,3´-Dichlorobenzidine U 2
MBR FEED DUP EPA 8270 2,4-Dichlorophenol U 2
MBR FEED DUP EPA 8270 Diethyl phthalate U 1
MBR FEED DUP EPA 8270 Dimethyl phthalate U 1
MBR FEED DUP EPA 8270 1,3-Dimethyladamantane U 2
MBR FEED DUP EPA 8270 2,4-Dimethylphenol 88.3 19.9
MBR FEED DUP EPA 8270 1,2-Dinitrobenzene U 1
MBR FEED DUP EPA 8270 1,3-Dinitrobenzene U 1
MBR FEED DUP EPA 8270 1,4-Dinitrobenzene U 1
MBR FEED DUP EPA 8270 2,4-Dinitrophenol U 3
MBR FEED DUP EPA 8270 2,4-Dinitrotoluene U 1
MBR FEED DUP EPA 8270 2,6-Dinitrotoluene U 1
MBR FEED DUP EPA 8270 4,6-Dinitro-2-methylphenol U 2
MBR FEED DUP EPA 8270 Di-n-butyl phthalate U 1
MBR FEED DUP EPA 8270 Di-n-octyl phthalate U 3
MBR FEED DUP EPA 8270 Fluoranthene U 1
MBR FEED DUP EPA 8270 Fluorene U 1
MBR FEED DUP EPA 8270 Hexachlorobenzene U 1
91
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
MBR FEED DUP EPA 8270 Hexachlorobutadiene U 3
MBR FEED DUP EPA 8270 Hexachlorocyclopentadiene U 1
MBR FEED DUP EPA 8270 Hexachloroethane U 1
MBR FEED DUP EPA 8270 Indeno (1,2,3-cd) pyrene U 1
MBR FEED DUP EPA 8270 Isophorone U 1
MBR FEED DUP EPA 8270 (R)-( + )-Limonene U 2
MBR FEED DUP EPA 8270 1-Methylnaphthalene U 1
MBR FEED DUP EPA 8270 2-Methylnaphthalene U 1
MBR FEED DUP EPA 8270 2-Methylphenol 36.4 19.9
MBR FEED DUP EPA 8270 3 &/or 4-Methylphenol 25.3 19.9
MBR FEED DUP EPA 8270 Naphthalene U 1
MBR FEED DUP EPA 8270 2-Nitroaniline U 1
MBR FEED DUP EPA 8270 3-Nitroaniline U 3
MBR FEED DUP EPA 8270 4-Nitroaniline U 3
MBR FEED DUP EPA 8270 Nitrobenzene U 1
MBR FEED DUP EPA 8270 2-Nitrophenol U 2
MBR FEED DUP EPA 8270 4-Nitrophenol U 3
MBR FEED DUP EPA 8270 N-Nitrosodimethylamine U 1
MBR FEED DUP EPA 8270 N-Nitrosodiphenylamine/Diphenylamine U 1
MBR FEED DUP EPA 8270 N-Nitrosodi-n-propylamine U 1
MBR FEED DUP EPA 8270 Pentachlorophenol U 2
MBR FEED DUP EPA 8270 Phenanthrene U 1
MBR FEED DUP EPA 8270 Phenol 63.5 19.9
MBR FEED DUP EPA 8270 Pyrene U 1
MBR FEED DUP EPA 8270 Squalene U 2
MBR FEED DUP EPA 8270 Terpiniol U 1
MBR FEED DUP EPA 8270 2,3,4,6-Tetrachlorophenol U 2
MBR FEED DUP EPA 8270 2,3,5,6-Tetrachlorophenol U 2
MBR FEED DUP EPA 8270 1,2,4-Trichlorobenzene U 1
92
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
MBR FEED DUP EPA 8270 2,4,6-Trichlorophenol U 2
MBR FEED DUP EPA 8270 2,4,5-Trichlorophenol U 2
MBR FEED DUP EPA 8270 2-Fluorophenol 5.84
MBR FEED DUP EPA 8270 Phenol-d5 5.75
MBR FEED DUP EPA 8270 2-Chlorophenol-d4 5.62
MBR FEED DUP EPA 8270 1,2-Dichlorobenzene-d4 3.12
MBR FEED DUP EPA 8270 Nitrobenzene-d5 4.28
MBR FEED DUP EPA 8270 2-Fluorobiphenyl 3.16
MBR FEED DUP EPA 8270 2,4,6-Tribromophenol 6.92
COAG PW TO MBR EPA 8270 4-Chlorophenyl phenyl ether U 1
COAG PW TO MBR EPA 8270 4-Chloro-3-methylphenol U 2
COAG PW TO MBR EPA 8270 Chrysene U 1
COAG PW TO MBR EPA 8270 Dibenzofuran U 1
COAG PW TO MBR EPA 8270 Dibenz (a,h) anthracene U 1
COAG PW TO MBR EPA 8270 1,2-Dichlorobenzene U 1
COAG PW TO MBR EPA 8270 1,3-Dichlorobenzene U 1
COAG PW TO MBR EPA 8270 1,4-Dichlorobenzene U 1
COAG PW TO MBR EPA 8270 3,3´-Dichlorobenzidine U 2
COAG PW TO MBR EPA 8270 2,4-Dichlorophenol U 2
COAG PW TO MBR EPA 8270 Diethyl phthalate U 1
COAG PW TO MBR EPA 8270 Dimethyl phthalate U 1
COAG PW TO MBR EPA 8270 1,3-Dimethyladamantane U 2
COAG PW TO MBR EPA 8270 2,4-Dimethylphenol 88.6 20.1
COAG PW TO MBR EPA 8270 1,2-Dinitrobenzene U 1
COAG PW TO MBR EPA 8270 1,3-Dinitrobenzene U 1
COAG PW TO MBR EPA 8270 1,4-Dinitrobenzene U 1
COAG PW TO MBR EPA 8270 2,4-Dinitrophenol U 3
COAG PW TO MBR EPA 8270 2,4-Dinitrotoluene U 1
COAG PW TO MBR EPA 8270 2,6-Dinitrotoluene U 1
93
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
COAG PW TO MBR EPA 8270 4,6-Dinitro-2-methylphenol U 2
COAG PW TO MBR EPA 8270 Di-n-butyl phthalate U 1
COAG PW TO MBR EPA 8270 Di-n-octyl phthalate U 3
COAG PW TO MBR EPA 8270 Fluoranthene U 1
COAG PW TO MBR EPA 8270 Fluorene U 1
COAG PW TO MBR EPA 8270 Hexachlorobenzene U 1
COAG PW TO MBR EPA 8270 Hexachlorobutadiene U 3
COAG PW TO MBR EPA 8270 Hexachlorocyclopentadiene U 1
COAG PW TO MBR EPA 8270 Hexachloroethane U 1
COAG PW TO MBR EPA 8270 Indeno (1,2,3-cd) pyrene U 1
COAG PW TO MBR EPA 8270 Isophorone U 1
COAG PW TO MBR EPA 8270 (R)-( + )-Limonene U 2
COAG PW TO MBR EPA 8270 1-Methylnaphthalene U 1
COAG PW TO MBR EPA 8270 2-Methylnaphthalene U 1
COAG PW TO MBR EPA 8270 2-Methylphenol 38.8 20.1
COAG PW TO MBR EPA 8270 3 &/or 4-Methylphenol 32.5 20.1
COAG PW TO MBR EPA 8270 Naphthalene U 1
COAG PW TO MBR EPA 8270 2-Nitroaniline U 1
COAG PW TO MBR EPA 8270 3-Nitroaniline U 3
COAG PW TO MBR EPA 8270 4-Nitroaniline U 3
COAG PW TO MBR EPA 8270 Nitrobenzene U 1
COAG PW TO MBR EPA 8270 2-Nitrophenol U 2
COAG PW TO MBR EPA 8270 4-Nitrophenol U 3
COAG PW TO MBR EPA 8270 N-Nitrosodimethylamine U 1
COAG PW TO MBR EPA 8270 N-Nitrosodiphenylamine/Diphenylamine U 1
COAG PW TO MBR EPA 8270 N-Nitrosodi-n-propylam U 1
COAG PW TO MBR EPA 8270 Pentachlorophenol U 2
COAG PW TO MBR EPA 8270 Phenanthrene U 1
COAG PW TO MBR EPA 8270 Phenol 84.6 20.1
94
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
COAG PW TO MBR EPA 8270 Pyrene U 1
COAG PW TO MBR EPA 8270 Squalene U 2
COAG PW TO MBR EPA 8270 Terpiniol U 1
COAG PW TO MBR EPA 8270 2,3,4,6-Tetrachlorophenol U 2
COAG PW TO MBR EPA 8270 2,3,5,6-Tetrachlorophenol U 2
COAG PW TO MBR EPA 8270 1,2,4-Trichlorobenzene U 1
COAG PW TO MBR EPA 8270 2,4,6-Trichlorophenol U 2
COAG PW TO MBR EPA 8270 2,4,5-Trichlorophenol U 2
COAG PW TO MBR EPA 8270 2-Fluorophenol 4.59
COAG PW TO MBR EPA 8270 Phenol-d5 4.55
COAG PW TO MBR EPA 8270 2-Chlorophenol-d4 4.63
COAG PW TO MBR EPA 8270 1,2-Dichlorobenzene-d4 2.4
COAG PW TO MBR EPA 8270 Nitrobenzene-d5 3.65
COAG PW TO MBR EPA 8270 2-Fluorobiphenyl 2.65
COAG PW TO MBR EPA 8270 2,4,6-Tribromophenol 6.67
COAG PW TO MBR EPA 8270 Terphenyl-d14 2.42
COAG PW TO MBR EPA 8270 1(2H)-Naphthalenone, 3,4-di... 3.8
COAG PW TO MBR EPA 8270 1H-Inden-1-one, 2,3-dihydro- 5.1
COAG PW TO MBR EPA 8270 1-Naphthalenol, 1,2,3,4-tet... 2.4
COAG PW TO MBR EPA 8270 2-Propanol, 1-(2-butoxy-1-m... 12.1
COAG PW TO MBR EPA 8270 Cyclohexaneacetic acid 4.1
COAG PW TO MBR EPA 8270 Dimethyl Benzoic Acid Isomer (1) 25
COAG PW TO MBR EPA 8270 Dimethyl Benzoic Acid Isomer (2) 3.7
COAG PW TO MBR EPA 8270 Ethylmethyl Phenol Isomer (1) 4
COAG PW TO MBR EPA 8270 Heptacosane 2.5
COAG PW TO MBR EPA 8270 Hexacosane 3.1
COAG PW TO MBR EPA 8270 Methyl Benzoic Acid Isomer (1) 33
COAG PW TO MBR EPA 8270 Methyl benzoic acid isomer (2) 10.1
COAG PW TO MBR EPA 8270 Methyl benzoic acid isomer (3) 2.8
95
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
COAG PW TO MBR EPA 8270 Methylethyl Phenol Isomer 7
COAG PW TO MBR EPA 8270 Nonacosane 1.7
COAG PW TO MBR EPA 8270 Octacosane 2.9
COAG PW TO MBR EPA 8270 Octadecanoic acid 10.2
COAG PW TO MBR EPA 8270 Pentacosane 2.6
COAG PW TO MBR EPA 8270 Tetracosane 2.1
COAG PW TO MBR EPA 8270 Trimethyl phenol isomer 1.6
AFTER MBR EPA 8270 Acenaphthene U 1.1
AFTER MBR EPA 8270 Acenaphthylene U 1.1
AFTER MBR EPA 8270 Acetophenone U 1.1
AFTER MBR EPA 8270 Adamantane U 2.1
AFTER MBR EPA 8270 Anthracene U 1.1
AFTER MBR EPA 8270 Atrazine U 2.1
AFTER MBR EPA 8270 Azobenzene U 1.1
AFTER MBR EPA 8270 Benzaldehyde U 1.1
AFTER MBR EPA 8270 Benzoic acid U 5.3
AFTER MBR EPA 8270 Benzo (a) anthracene U 1.1
AFTER MBR EPA 8270 Benzo (a) pyrene U 1.1
AFTER MBR EPA 8270 Benzo (b) fluoranthene U 1.1
AFTER MBR EPA 8270 Benzo (g,h,i) perylene U 1.1
AFTER MBR EPA 8270 Benzo (k) fluoranthene U 1.1
AFTER MBR EPA 8270 Benzyl alcohol U 1.1
AFTER MBR EPA 8270 1,1'-Biphenyl U 1.1
AFTER MBR EPA 8270 Bis(2-chloroethoxy)methane U 1.1
AFTER MBR EPA 8270 Bis(2-chloroethyl)ether U 1.1
AFTER MBR EPA 8270 Bis(2-chloro-1-methylethyl)ether U 1.1
AFTER MBR EPA 8270 Bis(2-Ethylhexyl)adipate U 1.1
AFTER MBR EPA 8270 Bis(2-ethylhexyl)phthalate U 2.1
AFTER MBR EPA 8270 4-Bromophenyl phenyl ether U 1.1
96
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
AFTER MBR EPA 8270 2-Butoxyethanol U 1.1
AFTER MBR EPA 8270 2-Butoxyethanol Phosphate U 5.3
AFTER MBR EPA 8270 Butyl benzyl phthalate U 3.2
AFTER MBR EPA 8270 Carbazole U 3.2
AFTER MBR EPA 8270 Caprolactam U 1.1
AFTER MBR EPA 8270 4-Chloroaniline U 3.2
AFTER MBR EPA 8270 2-Chloronaphthalene U 1.1
AFTER MBR EPA 8270 2-Chlorophenol U 2.1
AFTER MBR EPA 8270 4-Chlorophenyl phenyl ether U 1.1
AFTER MBR EPA 8270 4-Chloro-3-methylphenol U 2.1
AFTER MBR EPA 8270 Chrysene U 1.1
AFTER MBR EPA 8270 Dibenzofuran U 1.1
AFTER MBR EPA 8270 Dibenz (a,h) anthracene U 1.1
AFTER MBR EPA 8270 1,2-Dichlorobenzene U 1.1
AFTER MBR EPA 8270 1,3-Dichlorobenzene U 1.1
AFTER MBR EPA 8270 1,4-Dichlorobenzene U 1.1
AFTER MBR EPA 8270 3,3´-Dichlorobenzidine U 2.1
AFTER MBR EPA 8270 2,4-Dichlorophenol U 2.1
AFTER MBR EPA 8270 Diethyl phthalate U 1.1
AFTER MBR EPA 8270 Dimethyl phthalate U 1.1
AFTER MBR EPA 8270 1,3-Dimethyladamantane U 2.1
AFTER MBR EPA 8270 2,4-Dimethylphenol U 2.1
AFTER MBR EPA 8270 1,2-Dinitrobenzene U 1.1
AFTER MBR EPA 8270 1,3-Dinitrobenzene U 1.1
AFTER MBR EPA 8270 1,4-Dinitrobenzene U 1.1
AFTER MBR EPA 8270 2,4-Dinitrophenol U 3.2
AFTER MBR EPA 8270 2,4-Dinitrotoluene U 1.1
AFTER MBR EPA 8270 2,6-Dinitrotoluene U 1.1
AFTER MBR EPA 8270 4,6-Dinitro-2-methylphenol U 2.1
97
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
AFTER MBR EPA 8270 Di-n-butyl phthalate U 1.1
AFTER MBR EPA 8270 Di-n-octyl phthalate U 3.2
AFTER MBR EPA 8270 Fluoranthene U 1.1
AFTER MBR EPA 8270 Fluorene U 1.1
AFTER MBR EPA 8270 Hexachlorobenzene U 1.1
AFTER MBR EPA 8270 Hexachlorobutadiene U 3.2
AFTER MBR EPA 8270 Hexachlorocyclopentadiene U 1.1
AFTER MBR EPA 8270 Hexachloroethane U 1.1
AFTER MBR EPA 8270 Indeno (1,2,3-cd) pyrene U 1.1
AFTER MBR EPA 8270 Isophorone U 1.1
AFTER MBR EPA 8270 (R)-( + )-Limonene U 2.1
AFTER MBR EPA 8270 1-Methylnaphthalene U 1.1
AFTER MBR EPA 8270 2-Methylnaphthalene U 1.1
AFTER MBR EPA 8270 2-Methylphenol U 2.1
AFTER MBR EPA 8270 3 &/or 4-Methylphenol U 5.3
AFTER MBR EPA 8270 Naphthalene U 1.1
AFTER MBR EPA 8270 2-Nitroaniline U 1.1
AFTER MBR EPA 8270 3-Nitroaniline U 3.2
AFTER MBR EPA 8270 4-Nitroaniline U 6.3
AFTER MBR EPA 8270 Nitrobenzene U 1.1
AFTER MBR EPA 8270 2-Nitrophenol U 2.1
AFTER MBR EPA 8270 4-Nitrophenol U 3.2
AFTER MBR EPA 8270 N-Nitrosodimethylamine U 1.1
AFTER MBR EPA 8270 N-Nitrosodiphenylamine/Diphenylam U 1.1
AFTER MBR EPA 8270 N-Nitrosodi-n-propylamine U 1.1
AFTER MBR EPA 8270 Pentachlorophenol U 2.1
AFTER MBR EPA 8270 Phenanthrene U 1.1
AFTER MBR EPA 8270 Phenol U 2.1
AFTER MBR EPA 8270 Pyrene U 1.1
98
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
AFTER MBR EPA 8270 Squalene U 4.2
AFTER MBR EPA 8270 Terpiniol U 1.1
AFTER MBR EPA 8270 2,3,4,6-Tetrachlorophenol U 2.1
AFTER MBR EPA 8270 2,3,5,6-Tetrachlorophenol U 2.1
AFTER MBR EPA 8270 1,2,4-Trichlorobenzene U 1.1
AFTER MBR EPA 8270 2,4,6-Trichlorophenol U 2.1
AFTER MBR EPA 8270 2,4,5-Trichlorophenol U 2.1
AFTER MBR EPA 8270 2-Fluorophenol 5.06
AFTER MBR EPA 8270 Phenol-d5 5.46
AFTER MBR EPA 8270 2-Chlorophenol-d4 5.28
AFTER MBR EPA 8270 Nitrobenzene-d5 3.54
AFTER MBR EPA 8270 2-Fluorobiphenyl 2.97
AFTER MBR EPA 8270 2,4,6-Tribromophenol 6.21
AFTER MBR EPA 8270 Terphenyl-d14 3.18
AFTER MBR EPA 8270 2-Propanol, 1-(2-butoxy-1-m... 3.3
AFTER MBR EPA 8270 C13H9NO Isomer 3
AFTER MBR EPA 8270 C8H9NO2 Isomer 2.7
AFTER MBR EPA 8270 Heptacosane 2.8
AFTER MBR EPA 8270 Hexacosane 3
AFTER MBR EPA 8270 Hexadecanoic acid 15
AFTER MBR EPA 8270 Hexathiane 6.5
AFTER MBR EPA 8270 Methane, triiodo- 20.2
AFTER MBR EPA 8270 Methyl Acridone Isomer (1) 1.7
AFTER MBR EPA 8270 Methyl Acridone Isomer (2) 1.9
AFTER MBR EPA 8270 Methyl Benzoic Acid Isomer 3.9
AFTER MBR EPA 8270 Mono(2-ethylhexyl) phthalate 2.3
AFTER MBR EPA 8270 Nonacosane 2.1
AFTER MBR EPA 8270 Octadecanoic acid 12.9
AFTER MBR EPA 8270 Pentacosane 2.3
99
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
AFTER MBR EPA 8270 Sulfur, mol. (S8) 78
AFTER MBR EPA 8270 Triacontane 1.5
AFTER MBR EPA 8270 Unknown at 10.146 3.6
AFTER MBR EPA 8270 Unknown Sulfur Compound 5.5
RAW UNFILTERED PW EPA 8270 Acenaphthene U 4
RAW UNFILTERED PW EPA 8270 Acenaphthylene U 4
RAW UNFILTERED PW EPA 8270 Acetophenone U 4
RAW UNFILTERED PW EPA 8270 Adamantane U 8
RAW UNFILTERED PW EPA 8270 Anthracene U 4
RAW UNFILTERED PW EPA 8270 Atrazine U 4
RAW UNFILTERED PW EPA 8270 Azobenzene U 4
RAW UNFILTERED PW EPA 8270 Benzaldehyde U 4
RAW UNFILTERED PW EPA 8270 Benzoic acid U 20
RAW UNFILTERED PW EPA 8270 Benzo (a) pyrene U 4
RAW UNFILTERED PW EPA 8270 Benzo (b) fluoranthene U 4
RAW UNFILTERED PW EPA 8270 Benzo (g,h,i) perylene U 4
RAW UNFILTERED PW EPA 8270 Benzo (k) fluoranthene U 4
RAW UNFILTERED PW EPA 8270 Benzyl alcohol U 4
RAW UNFILTERED PW EPA 8270 1,1'-Biphenyl 11.6 4
RAW UNFILTERED PW EPA 8270 Bis(2-chloroethoxy)methane U 4
RAW UNFILTERED PW EPA 8270 Bis(2-chloroethyl)ether U 4
RAW UNFILTERED PW EPA 8270 Bis(2-chloro-1-methylethyl)ether U 4
RAW UNFILTERED PW EPA 8270 Bis(2-Ethylhexyl)adipate U 4
RAW UNFILTERED PW EPA 8270 Bis(2-ethylhexyl)phthalate 26.3 8
RAW UNFILTERED PW EPA 8270 4-Bromophenyl phenyl ether U 4
RAW UNFILTERED PW EPA 8270 2-Butoxyethanol U 4
RAW UNFILTERED PW EPA 8270 2-Butoxyethanol Phosphate U 20
RAW UNFILTERED PW EPA 8270 Butyl benzyl phthalate U 12
RAW UNFILTERED PW EPA 8270 Carbazole U 12
100
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
RAW UNFILTERED PW EPA 8270 Caprolactam U 4
RAW UNFILTERED PW EPA 8270 4-Chloroaniline U 12
RAW UNFILTERED PW EPA 8270 2-Chloronaphthalene U 4
RAW UNFILTERED PW EPA 8270 2-Chlorophenol U 8
RAW UNFILTERED PW EPA 8270 4-Chlorophenyl phenyl ether U 4
RAW UNFILTERED PW EPA 8270 4-Chloro-3-methylphenol U 8
RAW UNFILTERED PW EPA 8270 Chrysene U 4
RAW UNFILTERED PW EPA 8270 Dibenzofuran U 4
RAW UNFILTERED PW EPA 8270 Dibenz (a,h) anthracene U 4
RAW UNFILTERED PW EPA 8270 1,2-Dichlorobenzene U 4
RAW UNFILTERED PW EPA 8270 1,3-Dichlorobenzene U 4
RAW UNFILTERED PW EPA 8270 1,4-Dichlorobenzene U 4
RAW UNFILTERED PW EPA 8270 3,3´-Dichlorobenzidine U 8
RAW UNFILTERED PW EPA 8270 2,4-Dichlorophenol U 8
RAW UNFILTERED PW EPA 8270 Diethyl phthalate U 4
RAW UNFILTERED PW EPA 8270 Dimethyl phthalate U 4
RAW UNFILTERED PW EPA 8270 1,3-Dimethyladamantane U 8
RAW UNFILTERED PW EPA 8270 2,4-Dimethylphenol 136 40
RAW UNFILTERED PW EPA 8270 1,2-Dinitrobenzene U 4
RAW UNFILTERED PW EPA 8270 1,3-Dinitrobenzene U 4
RAW UNFILTERED PW EPA 8270 1,4-Dinitrobenzene U 4
RAW UNFILTERED PW EPA 8270 2,4-Dinitrophenol U 12
RAW UNFILTERED PW EPA 8270 2,4-Dinitrotoluene U 4
RAW UNFILTERED PW EPA 8270 2,6-Dinitrotoluene U 4
RAW UNFILTERED PW EPA 8270 4,6-Dinitro-2-methylphenol U 8
RAW UNFILTERED PW EPA 8270 Di-n-butyl phthalate U 4
RAW UNFILTERED PW EPA 8270 Di-n-octyl phthalate U 12
RAW UNFILTERED PW EPA 8270 Fluoranthene U 4
RAW UNFILTERED PW EPA 8270 Fluorene U 4
101
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
RAW UNFILTERED PW EPA 8270 (R)-( + )-Limonene U 8
RAW UNFILTERED PW EPA 8270 1-Methylnaphthalene 26.4 4
RAW UNFILTERED PW EPA 8270 2-Methylnaphthalene 47.3 4
RAW UNFILTERED PW EPA 8270 2-Methylphenol 75.2 40
RAW UNFILTERED PW EPA 8270 3 &/or 4-Methylphenol 69 40
RAW UNFILTERED PW EPA 8270 Naphthalene U 4
RAW UNFILTERED PW EPA 8270 2-Nitroaniline U 4
RAW UNFILTERED PW EPA 8270 3-Nitroaniline U 12
RAW UNFILTERED PW EPA 8270 4-Nitroaniline U 12
RAW UNFILTERED PW EPA 8270 Nitrobenzene U 4
RAW UNFILTERED PW EPA 8270 2-Nitrophenol U 8
RAW UNFILTERED PW EPA 8270 4-Nitrophenol U 12
RAW UNFILTERED PW EPA 8270 N-Nitrosodimethylamine U 4
RAW UNFILTERED PW EPA 8270 N-Nitrosodiphenylamine/Diphenylamine U 4
RAW UNFILTERED PW EPA 8270 N-Nitrosodi-n-propylamine U 4
RAW UNFILTERED PW EPA 8270 2,3,4,6-Tetrachlorophenol U 8
RAW UNFILTERED PW EPA 8270 2,3,5,6-Tetrachlorophenol U 8
RAW UNFILTERED PW EPA 8270 1,2,4-Trichlorobenzene U 4
RAW UNFILTERED PW EPA 8270 2,4,6-Trichlorophenol U 8
RAW UNFILTERED PW EPA 8270 2,4,5-Trichlorophenol U 8
RAW UNFILTERED PW EPA 8270 2-Fluorophenol 41.9
RAW UNFILTERED PW EPA 8270 Phenol-d5 46.4
RAW UNFILTERED PW EPA 8270 2-Chlorophenol-d4 48
RAW UNFILTERED PW EPA 8270 1,2-Dichlorobenzene-d4 31.8
RAW UNFILTERED PW EPA 8270 Nitrobenzene-d5 46.5
RAW UNFILTERED PW EPA 8270 2-Fluorobiphenyl 36.9
RAW UNFILTERED PW EPA 8270 2,4,6-Tribromophenol 74.4
RAW UNFILTERED PW EPA 8270 Terphenyl-d14 31.9
RAW UNFILTERED PW EPA 8270 C14H30 Branched Hydrocarbon 632
102
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
RAW UNFILTERED PW EPA 8270 C16H34 Branched Hydrocarbon 372
RAW UNFILTERED PW EPA 8270 Hexadecane, 2,6,10,14-tetra... 468
RAW UNFILTERED PW EPA 8270 Methylpentyl Cyclohexane Isomer 428
RAW UNFILTERED PW EPA 8270 Nonadecane 592
RAW UNFILTERED PW EPA 8270 Octadecane 578
RAW UNFILTERED PW EPA 8270 Pentacosane 386
RAW UNFILTERED PW EPA 8270 Pentadecane 978
RAW UNFILTERED PW EPA 8270 Pentadecane, 2,6,10,14-tetr... 562
RAW UNFILTERED PW EPA 8270 Pentadecane, 2,6,10-trimethyl- 410
RAW UNFILTERED PW EPA 8270 Tetradecane 906
RAW UNFILTERED PW EPA 8270 Tricosane 376
RAW UNFILTERED PW EPA 8270 Tridecane 1190
RAW UNFILTERED PW EPA 8270 Undecane, 2,6-dimethyl- 538
COAG PW TO MBR EPA 8270 Acenaphthene U 1
COAG PW TO MBR EPA 8270 Acenaphthylene U 1
COAG PW TO MBR EPA 8270 Acetophenone U 1
COAG PW TO MBR EPA 8270 Benzo (a) anthracene U 1
COAG PW TO MBR EPA 8270 Benzo (a) pyrene U 1
COAG PW TO MBR EPA 8270 Benzo (b) fluoranthene U 1
COAG PW TO MBR EPA 8270 Benzo (g,h,i) perylene U 1
COAG PW TO MBR EPA 8270 Benzo (k) fluoranthene U 1
COAG PW TO MBR EPA 8270 Benzyl alcohol U 1
COAG PW TO MBR EPA 8270 1,1'-Biphenyl U 1
COAG PW TO MBR EPA 8270 Bis(2-chloroethoxy)methane U 1
COAG PW TO MBR EPA 8270 Bis(2-chloroethyl)ether U 1
COAG PW TO MBR EPA 8270 Bis(2-chloro-1-methylethyl)ether U 1
COAG PW TO MBR EPA 8270 Bis(2-Ethylhexyl)adipate U 1
COAG PW TO MBR EPA 8270 Bis(2-ethylhexyl)phthalate U 2
COAG PW TO MBR EPA 8270 4-Bromophenyl phenyl ether U 1
103
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
COAG PW TO MBR EPA 8270 2-Butoxyethanol U 1
COAG PW TO MBR EPA 8270 2-Butoxyethanol Phosphate U 5
COAG PW TO MBR EPA 8270 Butyl benzyl phthalate U 3
COAG PW TO MBR EPA 8270 Carbazole U 3
COAG PW TO MBR EPA 8270 Caprolactam U 1
COAG PW TO MBR EPA 8270 4-Chloroaniline U 3
COAG PW TO MBR EPA 8270 2-Chloronaphthalene U 1
COAG PW TO MBR EPA 8270 2-Chlorophenol U 2
COAG PW TO MBR EPA 8270 4-Chlorophenyl phenyl ether U 1
COAG PW TO MBR EPA 8270 4-Chloro-3-methylphenol U 2
COAG PW TO MBR EPA 8270 Chrysene U 1
COAG PW TO MBR EPA 8270 Dibenzofuran U 1
COAG PW TO MBR EPA 8270 Dibenz (a,h) anthracene U 1
COAG PW TO MBR EPA 8270 1,2-Dichlorobenzene U 1
COAG PW TO MBR EPA 8270 1,3-Dichlorobenzene U 1
COAG PW TO MBR EPA 8270 1,4-Dichlorobenzene U 1
COAG PW TO MBR EPA 8270 3,3´-Dichlorobenzidine U 2
COAG PW TO MBR EPA 8270 2,4-Dichlorophenol U 2
COAG PW TO MBR EPA 8270 Dimethyl phthalate U 1
COAG PW TO MBR EPA 8270 1,3-Dimethyladamantane U 2
COAG PW TO MBR EPA 8270 2,4-Dimethylphenol 112 20.1
COAG PW TO MBR EPA 8270 1,2-Dinitrobenzene U 1
COAG PW TO MBR EPA 8270 1,3-Dinitrobenzene U 1
COAG PW TO MBR EPA 8270 1,4-Dinitrobenzene U 1
COAG PW TO MBR EPA 8270 2,4-Dinitrophenol U 3
COAG PW TO MBR EPA 8270 2,4-Dinitrotoluene U 1
COAG PW TO MBR EPA 8270 2,6-Dinitrotoluene U 1
COAG PW TO MBR EPA 8270 4,6-Dinitro-2-methylphenol U 2
COAG PW TO MBR EPA 8270 Di-n-butyl phthalate U 1
104
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
COAG PW TO MBR EPA 8270 Di-n-octyl phthalate U 3
COAG PW TO MBR EPA 8270 Fluoranthene U 1
COAG PW TO MBR EPA 8270 Fluorene U 1
COAG PW TO MBR EPA 8270 Hexachlorobenzene U 1
COAG PW TO MBR EPA 8270 Hexachlorobutadiene U 3
COAG PW TO MBR EPA 8270 Hexachlorocyclopentadiene U 1
COAG PW TO MBR EPA 8270 Hexachloroethane U 1
COAG PW TO MBR EPA 8270 Indeno (1,2,3-cd) pyrene U 1
COAG PW TO MBR EPA 8270 Isophorone U 1
COAG PW TO MBR EPA 8270 (R)-( + )-Limonene U 2
COAG PW TO MBR EPA 8270 1-Methylnaphthalene 1.2 1
COAG PW TO MBR EPA 8270 2-Methylnaphthalene 1.3 1
COAG PW TO MBR EPA 8270 2-Methylphenol 54.2 20.1
COAG PW TO MBR EPA 8270 3 &/or 4-Methylphenol 47.7 20.1
COAG PW TO MBR EPA 8270 Naphthalene U 1
COAG PW TO MBR EPA 8270 2-Nitroaniline U 1
COAG PW TO MBR EPA 8270 3-Nitroaniline U 3
COAG PW TO MBR EPA 8270 4-Nitroaniline U 3
COAG PW TO MBR EPA 8270 Nitrobenzene U 1
COAG PW TO MBR EPA 8270 2-Nitrophenol U 2
COAG PW TO MBR EPA 8270 4-Nitrophenol U 3
COAG PW TO MBR EPA 8270 N-Nitrosodimethylamine U 1
COAG PW TO MBR EPA 8270 N-Nitrosodiphenylamine/Diphenylamine U 1
COAG PW TO MBR EPA 8270 N-Nitrosodi-n-propylamine U 1
COAG PW TO MBR EPA 8270 Pentachlorophenol U 2
COAG PW TO MBR EPA 8270 Phenanthrene U 1
COAG PW TO MBR EPA 8270 Phenol 128 20.1
COAG PW TO MBR EPA 8270 Pyrene U 1
COAG PW TO MBR EPA 8270 Squalene U 2
105
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
COAG PW TO MBR EPA 8270 Terpiniol U 1
COAG PW TO MBR EPA 8270 2,3,4,6-Tetrachlorophenol U 2
COAG PW TO MBR EPA 8270 2,3,5,6-Tetrachlorophenol U 2
COAG PW TO MBR EPA 8270 1,2,4-Trichlorobenzene U 1
COAG PW TO MBR EPA 8270 2,4,6-Trichlorophenol U 2
COAG PW TO MBR EPA 8270 2,4,5-Trichlorophenol U 2
COAG PW TO MBR EPA 8270 Phenol-d5 5.76
COAG PW TO MBR EPA 8270 2-Chlorophenol-d4 5.74
COAG PW TO MBR EPA 8270 1,2-Dichlorobenzene-d4 2.97
COAG PW TO MBR EPA 8270 Nitrobenzene-d5 4.01
COAG PW TO MBR EPA 8270 2-Fluorobiphenyl 3.27
COAG PW TO MBR EPA 8270 2,4,6-Tribromophenol 7.2
COAG PW TO MBR EPA 8270 Terphenyl-d14 3.18
COAG PW TO MBR EPA 8270 1(2H)-Naphthalenone, 3,4-di... 3.7
COAG PW TO MBR EPA 8270 1H-Inden-1-one, 2,3-dihydro- 4.4
COAG PW TO MBR EPA 8270 1H-Inden-1-one, 2,3-dihydro... 3.1
COAG PW TO MBR EPA 8270 2-Propanol, 1-(2-butoxy-1-m... 11.2
COAG PW TO MBR EPA 8270 Cyclohexaneacetic acid 5
COAG PW TO MBR EPA 8270 Dimethyl Benzoic Acid Isomer (1) 18.3
COAG PW TO MBR EPA 8270 Dimethyl Benzoic Acid Isomer (2) 2
COAG PW TO MBR EPA 8270 Dimethyl Benzoic Acid Isomer (3) 2.2
COAG PW TO MBR EPA 8270 Dimethyl Naphthalene isomer (01) 1.8
COAG PW TO MBR EPA 8270 Dimethyl naphthalene isomer (02) 1.7
COAG PW TO MBR EPA 8270 Ethylmethyl Phenol Isomer (1) 1.5
COAG PW TO MBR EPA 8270 Ethylmethyl Phenol Isomer (2) 5.3
COAG PW TO MBR EPA 8270 Methyl Benzoic Acid Isomer (1) 38.6
COAG PW TO MBR EPA 8270 Methyl benzoic acid isomer (2) 11.5
COAG PW TO MBR EPA 8270 Methyl benzoic acid isomer (3) 2.4
COAG PW TO MBR EPA 8270 Methylethyl Benzoic Acid Isomer 2.4
106
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
COAG PW TO MBR EPA 8270 Methylethyl phenol isomer 8.4
COAG PW TO MBR EPA 8270 Sulfur, mol. (S8) 3.8
COAG PW TO MBR EPA 8270 Tolylacetic Acid Isomer 6.6
COAG PW TO MBR EPA 8270 Trimethyl phenol isomer 1.3
MBR PERMEATE EPA 8270 Acenaphthene U 1
MBR PERMEATE EPA 8270 Acenaphthylene U 1
MBR PERMEATE EPA 8270 Adamantane U 2.1
MBR PERMEATE EPA 8270 Anthracene U 1
MBR PERMEATE EPA 8270 Atrazine U 1
MBR PERMEATE EPA 8270 Azobenzene U 1
MBR PERMEATE EPA 8270 Benzaldehyde U 1
MBR PERMEATE EPA 8270 Benzoic acid U 5.2
MBR PERMEATE EPA 8270 Benzo (a) anthracene U 1
MBR PERMEATE EPA 8270 Benzo (a) pyrene U 1
MBR PERMEATE EPA 8270 Benzo (b) fluoranthene U 1
MBR PERMEATE EPA 8270 Benzo (g,h,i) perylene U 1
MBR PERMEATE EPA 8270 Benzo (k) fluoranthene U 1
MBR PERMEATE EPA 8270 Benzyl alcohol U 2.1
MBR PERMEATE EPA 8270 1,1'-Biphenyl U 1
MBR PERMEATE EPA 8270 Bis(2-chloroethoxy)methane U 1
MBR PERMEATE EPA 8270 Bis(2-chloroethyl)ether U 1
MBR PERMEATE EPA 8270 Bis(2-chloro-1-methylethyl)ether U 1
MBR PERMEATE EPA 8270 Bis(2-Ethylhexyl)adipate U 1
MBR PERMEATE EPA 8270 Bis(2-ethylhexyl)phthalate U 2.1
MBR PERMEATE EPA 8270 4-Bromophenyl phenyl ether U 1
MBR PERMEATE EPA 8270 2-Butoxyethanol U 1
MBR PERMEATE EPA 8270 2-Butoxyethanol Phosphate U 5.2
MBR PERMEATE EPA 8270 Butyl benzyl phthalate U 3.1
MBR PERMEATE EPA 8270 Carbazole U 3.1
107
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
MBR PERMEATE EPA 8270 Caprolactam U 1
MBR PERMEATE EPA 8270 4-Chloroaniline U 3.1
MBR PERMEATE EPA 8270 2-Chloronaphthalene U 1
MBR PERMEATE EPA 8270 2-Chlorophenol U 2.1
MBR PERMEATE EPA 8270 4-Chlorophenyl phenyl ether U 1
MBR PERMEATE EPA 8270 4-Chloro-3-methylphenol U 2.1
MBR PERMEATE EPA 8270 Chrysene U 1
MBR PERMEATE EPA 8270 Dibenzofuran U 1
MBR PERMEATE EPA 8270 Dibenz (a,h) anthracene U 1
MBR PERMEATE EPA 8270 1,2-Dichlorobenzene U 2
MBR PERMEATE EPA 8270 1,3-Dichlorobenzene U 2.1
MBR PERMEATE EPA 8270 1,4-Dichlorobenzene U 2.1
MBR PERMEATE EPA 8270 3,3´-Dichlorobenzidine U 2.1
MBR PERMEATE EPA 8270 2,4-Dichlorophenol U 2.1
MBR PERMEATE EPA 8270 Diethyl phthalate U 1
MBR PERMEATE EPA 8270 Dimethyl phthalate U 1
MBR PERMEATE EPA 8270 1,3-Dimethyladamantane U 2.1
MBR PERMEATE EPA 8270 2,4-Dimethylphenol U 2.1
MBR PERMEATE EPA 8270 1,2-Dinitrobenzene U 1
MBR PERMEATE EPA 8270 1,3-Dinitrobenzene U 1
MBR PERMEATE EPA 8270 1,4-Dinitrobenzene U 1
MBR PERMEATE EPA 8270 2,4-Dinitrophenol U 3.1
MBR PERMEATE EPA 8270 2,4-Dinitrotoluene U 1
MBR PERMEATE EPA 8270 2,6-Dinitrotoluene U 1
MBR PERMEATE EPA 8270 4,6-Dinitro-2-methylphenol U 2.1
MBR PERMEATE EPA 8270 Di-n-octyl phthalate U 3.1
MBR PERMEATE EPA 8270 Fluoranthene U 1
MBR PERMEATE EPA 8270 Fluorene U 1
MBR PERMEATE EPA 8270 Hexachlorobenzene U 1
108
|
Colorado School of Mines
|
Table A.5 Continued
SAMPLE NAME METHOD NAME ANALYTE Result (µg/L) RL (µg/L)
MBR PERMEATE EPA 8270 Hexachlorobutadiene U 6.2
MBR PERMEATE EPA 8270 Hexachlorocyclopentadiene U 1
MBR PERMEATE EPA 8270 Hexachloroethane U 2.1
MBR PERMEATE EPA 8270 Indeno (1,2,3-cd) pyrene U 1
MBR PERMEATE EPA 8270 Isophorone U 1
MBR PERMEATE EPA 8270 (R)-( + )-Limonene U 2.1
MBR PERMEATE EPA 8270 1-Methylnaphthalene U 1
MBR PERMEATE EPA 8270 2-Methylnaphthalene U 1
MBR PERMEATE EPA 8270 2-Methylphenol U 2.1
MBR PERMEATE EPA 8270 3 &/or 4-Methylphenol U 5.2
MBR PERMEATE EPA 8270 Naphthalene U 1
MBR PERMEATE EPA 8270 2-Nitroaniline U 1
MBR PERMEATE EPA 8270 3-Nitroaniline U 3.1
MBR PERMEATE EPA 8270 4-Nitroaniline U 3.1
MBR PERMEATE EPA 8270 Nitrobenzene U 1
MBR PERMEATE EPA 8270 2-Nitrophenol U 2.1
MBR PERMEATE EPA 8270 4-Nitrophenol U 3.1
MBR PERMEATE EPA 8270 N-Nitrosodimethylamine U 1
MBR PERMEATE EPA 8270 N-Nitrosodiphenylamine/Diphenylamine U 1
MBR PERMEATE EPA 8270 N-Nitrosodi-n-propylamine U 1
MBR PERMEATE EPA 8270 Pentachlorophenol U 2.1
MBR PERMEATE EPA 8270 Phenanthrene U 1
MBR PERMEATE EPA 8270 Phenol U 2.1
MBR PERMEATE EPA 8270 Pyrene U 1
MBR PERMEATE EPA 8270 Squalene 3.9 2.1
MBR PERMEATE EPA 8270 Terpiniol U 1
MBR PERMEATE EPA 8270 2,3,4,6-Tetrachlorophenol U 2.1
MBR PERMEATE EPA 8270 2,3,5,6-Tetrachlorophenol U 2.1
MBR PERMEATE EPA 8270 1,2,4-Trichlorobenzene U 2.1
109
|
Colorado School of Mines
|
Table A.6. Continued
Bor Calci Potas Lithi Magne Sodi Stron Chlo Bro Nitr Sulf Sample Time Taken
Sample
on um sium um sium um tium ride mide ate ate number during Test
7.8 181. 1440 2420 109 519.
Stage 3 reject-1 207.2 7.05 780 2.18 0.78
2 8 2 5 1 1
16. 226. 3352 5635 263
Stage 3 reject-4 444.5 16.2 558.7 2.27 3.52 293
73 2 8 0 0
11. 73.1 1659 2788 204 52.7
Stage 3 reject-3 212.5 8.24 121.4 0.64 0.8
3 5 2 6 5 3
10. 928. 21.3 5087 8550 154 121.
Brine from Cycle 1 755.7 121.4 12.19 5.15
95 9 2 3 1 9 4
13. 21.5 4666 7843 211 706.
Brine from Cycle 2 393 677.5 121.4 4.63 5.26
16 7 8 4 6 5
Final brine from 22. 48.3 22.1 4506 7574 319 427.
649.2 877.6 0.6 5.29
Cycle 3 17 8 4 8 5 1 3
2X Seawater
Initial Feed Cycle 1- 1742 2928 525 Cycle 1, Step 1,
9 676 638.5 0.6 2212 14.9 5.3 0 1
1st T100A 0 3 4 84 mins in
10. 369. 1015 1706 285. Cycle 1, Step 2,
150 1 377.6 0.4 1156 8 2.2 0 2
3 2 2 6 8 43 mins in
861. Cycle 2, Step 1,
Perm1 1 8.2 22.2 37.3 0.3 7 0.3 1448 0 2.8 7.4 3
3 30 mins in
Cycle 3, Step 5,
Perm2 1 8.2 26.2 60.5 0.3 16 1839 0.3 3091 0.3 2.1 7 4
5 min in
Perm3 1 8.7 32.6 84.5 0.1 21.3 2640 0.4 4439 0.7 2.1 7.2
571. 1590 2673 442.
Stage 1 reject 1 9.8 587.7 0.6 1873 12.8 4.3 0
9 6 9 6
10. 144. 1612
200 1 340.7 0.4 175.5 9593 2.3 2.4 0 5.6
1 1 6
Perm4 1 8.6 26.5 85.8 0.2 22 2576 0.4 4330 0 2.1 6.5
Perm5 1 9.2 36 128.3 0.2 35.8 3806 0.5 6398 1.6 1.8 4.4
Perm6 1 9.7 49 169.7 0.3 48.7 4927 0.7 8282 2.3 2 4.8
654. 1728 2905 476.
Stage 2 reject 1 9.7 673.4 0.5 1897 14.1 0
7 4 4 9
115
|
Colorado School of Mines
|
Table A.6. Continued
Bor Calci Potas Lithi Magne Sodi Stron Chlo Bro Nitr Sulf Sample Time Taken
Sample
on um sium um sium um tium ride mide ate ate number during Test
21. 515. 3421 5751
Stage 3 reject 4 1257 2 1087 10.1 15.2 233
7 8 5 6
26. 596. 3728 6267 285.
Cycle 3 Brine 1386 2.2 1244 11.3 20.1
4 4 4 4 3
RO Perm 4 3.5 0.1 0.5 0 0 16 0 26.8 0 0.2 0
Lithium Mining Water
576 388. 2674 Cycle 1, Step 1,
Initial T100A 225.5 2710 781.6 2591 2.9 0 270 121 1
.5 8 8 32 mins in
607 588. 583. Cycle 1, Step 2,
Sample 3 Perm 6 23.2 40.3 9.8 0.1 5440 0 210 43.9 2
.4 8 7 35 mins in
665 1410 Cycle 2, Step 1,
Sample 3 Perm 9 39.1 88 1409 27.9 1273 0.2 0 14.7 39.1 3
.9 9 10 mins in
Sample 3 stage 3 848 183. 3504 Cycle 2, Step 6,
240.5 3373 225.3 4542 1.2 0 9.1 58 4
reject .4 5 3 5 mins in
898 617. 5619
Sample 4 T100A 445.3 5481 1133 6942 4.6 0 5 179
.4 5 1
894 369. 3675 510.
Sample 4 T150 (Feed) 292.7 3548 692.9 4679 2.8 0 6.9
.6 3 5 9
259. 293.
Sample 4 Perm 1 599 11 24.5 5.6 0.1 2827 0 227 41.3
8 6
733 494. 483.
Sample 4 Perm 2 21 40.9 8.6 0.1 5325 0 26 40.5
.9 4 1
703. 653.
Sample 4 Perm 3 844 23 45.9 10.4 0.1 7267 0 15.8 40.7
5 3
Sample 4 stage 1 100 5056
513 386.1 4764 937.2 6103 3.8 0 1.5 152
reject 8 8
836 655. 645.
Sample 4 Perm 4 15.9 44.7 10.4 0.1 7165 0 12.8 39.6
.9 5 8
907 982. 882.
Sample 4 Perm 5 22.3 61 15.4 0.1 9896 0 14.5 39.4
.6 1 1
947 1344
Sample 4 Perm 6 28.4 79.3 1260 20.5 1139 0.2 0 14.5 39.1
.2 8
119
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.