UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An investigation into the controls and variability of the flowback water inorganic geochemistry of the… Owen, Jennifer Nicole 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2018_february_owen_jennifer.pdf [ 6.2MB ]
Metadata
JSON: 24-1.0362574.json
JSON-LD: 24-1.0362574-ld.json
RDF/XML (Pretty): 24-1.0362574-rdf.xml
RDF/JSON: 24-1.0362574-rdf.json
Turtle: 24-1.0362574-turtle.txt
N-Triples: 24-1.0362574-rdf-ntriples.txt
Original Record: 24-1.0362574-source.json
Full Text
24-1.0362574-fulltext.txt
Citation
24-1.0362574.ris

Full Text

AN INVESTIGATION INTO THE CONTROLS AND VARIABILITY OF THE FLOWBACK WATER INORGANIC GEOCHEMISTRY OF THE MONTNEY FORMATION, NORTHEASTERN BRITISH COLUMBIA AND NORTHWESTERN ALBERTA, CANADA  by  JENNIFER NICOLE OWEN  B.Sc., Simon Fraser University, 2009  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Geological Sciences)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  December 2017 © Jennifer Nicole Owen, 2017 ii  Abstract  The Montney Formation is the principal unconventional hydrocarbon reservoir currently being developed in Canada.  The flowback water from 31 wells located on 9 well pads was sampled over time and analyzed for major ions, key minor ions, and δ18O and δ2H isotopes. The injected hydraulic fracturing fluids and produced waters, if available, were analyzed for the same parameters.  The results of the study are used to compare the flowback water chemistry between wells and investigate the variables that have a significant influence on the chemistry.  When comparing the flowback water chemistry between multiple wells, consideration must be given to the length of the flowback period, as the major ion concentrations typically increase over time. The dominant influence on the increasing concentrations is mixing between hydraulic fracturing fluid and formation water.  Cl and stable water isotopes were used as conservative tracers to calculate the increasing proportions of formation water.  These proportions were used with geochemical models to determine that mixing explains the Na and K concentrations, while mixing with ion exchange is influencing Ca, Mg, and Sr concentrations. Sulfate concentrations are influenced by pyrite oxidation and sulfate reduction.  The rate of increase of the major ions varies between wells, although it is often, but not always, similar between wells completed at the same site, due to similarities in reservoir properties and well completion.  The inconsistency is due to the many variables that may impact the flowback water chemistry.  A multiple regression analysis identified shut-in time as an important variable, with longer shut-in correlating to higher concentrations.  The chemistry of hydraulic fracturing fluids and formation waters were found to be important variables for some ions.  The minor ions included in the study are Ba, B, and Li.  Ba concentrations are likely related to barite dissolution/precipitation and are highest where iii  sulfate concentrations are low.  B and Li concentrations are both dominantly influenced by mixing and may vary due to differences in formation water chemistry.  Overall, the results are expected to contribute to the growing knowledge on flowback water chemistry and its use in investigating the processes occurring in the reservoir during hydraulic fracturing.              iv  Lay Summary  After hydraulic fracturing and before production, high salinity water flows from oil and gas wells for a period of time ranging from days to weeks.  The salinity of the water increases over time and can be used to investigate processes that are occurring in the subsurface. The main influence on the chemistry of the water is mixing between the injected fluid and the water in the formation.  This has been found in previous studies and was supported by the results from the study wells, located in northeastern British Columbia and northwestern Alberta.  The amount of time that the injected fluid remains in contact with the rock was also found to increase the salinity.  Looking at the chemistry in more detail and using statistical methods and geochemical models assisted in understanding the complexity of the system, identifying the variability between different sites, and determining what this means for future studies. v  Preface  This thesis is based on research conducted by the author, Jennifer Owen.  The development of the specific research questions were determined with input from Dr. R. Marc Bustin (supervisor).  The fluid samples used in the study were collected and sent to The University of British Columbia (UBC) by several companies – ARC Resources Ltd., Black Swan Energy Ltd., Unconventional Gas Resources, and other companies that wish to remain anonymous.  The majority of the sample analyses were conducted in laboratory facilities at UBC, in the Department of Earth, Ocean, and Atmospheric Sciences, by the author with assistance from Scott Hazell for sample preparation and general chemistry tests, Maureen Soon for ICP-OES analysis, and Vivian Lai for ICP-MS analysis. Vivian Lai also provided input on the sample preparation method for the dissolved metals analysis.  Fluid samples for the anion analysis and rock samples for cation exchange capacity were sent to the BC Ministry of Environment and Climate Change Strategy Analytical Chemistry Services Laboratory in Victoria, BC. Publically available XRD data was compiled by Daniel Laird.  Rock chip and core samples were obtained from the BC Oil and Gas Commission Core Research Facility in Fort St. John, BC.  These samples were prepared for X-Ray Diffraction (XRD) and Scanning Electron Microscopy (SEM) at UBC by the author.  The quantitative XRD analysis was conducted by Dr. Marc Bustin.  Assistance for developing the linear fixed effects models was provided by the Statistical Opportunity for Students (SOS) program through the Department of Statistics at UBC. Parts of the research were presented as a poster at the American Geophysical Union (AGU) Fall Meeting in San Francisco, California in December 2016, on which Dr. Marc Bustin vi  was a co-author.  Initial findings of the research were included in a paper in Geoscience BC Summary of Activities 2016 [Owen, J.N. and Bustin, R.M. (2017): Implications of the inorganic geochemistry of flowback water from the Montney Formation, northeastern British Columbia and northwestern Alberta: progress report; in Geoscience BC Summary of Activities 2016, Geoscience BC, Report 2017-1, p. 49–54.] I was the lead author for both the poster and the paper and received feedback and edits from Dr. Marc Bustin. I was responsible for writing the manuscript for all chapters of this thesis, with manuscript suggestions and edits from Supervisor Dr. Marc Bustin, Committee Members Dr. Roger Beckie and Dr. Uli Mayer, and External Examiner Dr. Ken Hickey. vii  Table of Contents  Abstract .......................................................................................................................................... ii  Lay Summary ............................................................................................................................... iv  Preface .............................................................................................................................................v  Table of Contents ........................................................................................................................ vii  List of Tables ................................................................................................................................ xi  List of Figures ............................................................................................................................. xiii  Acknowledgements ................................................................................................................... xvii  Chapter 1: Introduction ................................................................................................................1  1.1 Introduction ..................................................................................................................... 1  1.2 Background Information ................................................................................................. 2 1.2.1 Hydraulic Fracturing Fluids ........................................................................................ 2 1.2.2 Formation Water ......................................................................................................... 4 1.2.3 Flowback Water .......................................................................................................... 6  1.2.3.1 Inorganic chemistry ............................................................................................ 6 1.2.3.2 Organic chemistry ............................................................................................... 7 1.2.3.3 Influences on flowback water chemistry ............................................................ 7 1.2.3.3.1 Mixing between hydraulic fracturing fluid and formation water .................. 8 1.2.3.3.2 Water-rock interactions ............................................................................... 10 1.3 Organization of thesis ................................................................................................... 12 Chapter 2: Investigating the Montney Formation flowback water chemistry through mixing calculations and geochemical modelling .......................................................................14 2.1 Introduction ................................................................................................................... 14  2.2 Methods......................................................................................................................... 20  2.2.1 Sample collection ...................................................................................................... 20 2.2.1.1 Hydraulic fracturing fluid ................................................................................. 20 2.2.1.2 Flowback water ................................................................................................. 20 2.2.1.3 Produced water.................................................................................................. 20 viii  2.2.2 Sample analysis ......................................................................................................... 21 2.2.2.1 Major ion chemistry .......................................................................................... 21 2.2.2.2 Oxygen and hydrogen isotopes ......................................................................... 22 2.2.3 PHREEQC mixing model ......................................................................................... 23 2.3 Montney Formation flowback water chemistry ............................................................ 25 2.4 Flowback water mixing ratios ....................................................................................... 32 2.5 Variability in mixing ratios ........................................................................................... 40 2.5.1 Mixing ratio variability between tracers used ........................................................... 40 2.5.2 Mixing ratio variability over the flowback period for individual wells.................... 41 2.5.3 Mixing ratio variability between sites....................................................................... 43 2.5.3.1 Potential complications with the different tracers used .................................... 45 2.6 Insight from major ion chemistry ................................................................................. 47 2.6.1 Sodium and potassium .............................................................................................. 50 2.6.2 Calcium, magnesium, and strontium ........................................................................ 53 2.6.3 Sulfate ....................................................................................................................... 60  2.7 Discussion and Conclusions ......................................................................................... 64  Chapter 3: Flowback water chemistry from the Montney Formation: Part I - Stratigraphic and areal variability .....................................................................................................................69  3.1 Introduction ................................................................................................................... 69  3.2 Methods......................................................................................................................... 75  3.2.1 Hydraulic Fracturing Fluid Chemistry ...................................................................... 75 3.2.2 Flowback Water Chemistry ...................................................................................... 76 3.2.3 Produced Water Chemistry ....................................................................................... 77 3.2.4 Linear Regression and Linear Mixed Effects Models .............................................. 77 3.3 Challenges of comparing flowback data from different wells ...................................... 79 3.4 Montney Formation flowback water chemistry ............................................................ 83 3.4.1 Overview of Montney Formation flowback water chemistry ................................... 88 3.4.2 Flowback water chemistry by formation member .................................................... 91 3.4.2.1 Upper Montney member ................................................................................... 91 3.4.2.2 Middle Montney member ................................................................................. 93 ix  3.4.2.3 Lower Montney member................................................................................... 94 3.4.3 Causes of variability in flowback water.................................................................... 94 3.4.4 Comparison of upper and middle Montney member Flowback Water ..................... 99 3.5 Flowback water trend analysis .................................................................................... 102 3.5.1 Linear regression and linear mixed effects models ................................................. 102 3.5.2 Results of flowback water chemistry trend analysis ............................................... 104 3.5.3 Results of flowback water volume trend analysis................................................... 106 3.5.4 Implications of the trend analyses........................................................................... 108 3.6 Conclusions ................................................................................................................. 113 Chapter 4: Flowback water chemistry from the Montney Formation: Part II – Controls on variability ....................................................................................................................................115  4.1 Introduction ................................................................................................................. 115  4.2 Methods....................................................................................................................... 119  4.2.1 Fluid chemistry ....................................................................................................... 119 4.2.3 Montney Formation Mineralogy ............................................................................. 121 4.3 Montney Formation flowback water chemistry .......................................................... 122 4.4 Dominant variables on flowback water chemistry ...................................................... 126 4.4.1 Initial sample flowback water chemistry ................................................................ 131 4.4.2 Intermediate to late flowback water chemistry ....................................................... 136 4.4.3 Important variables influencing flowback water chemistry .................................... 140 4.4.3.1 Shut-in time ..................................................................................................... 140 4.4.3.2 Hydraulic fracturing fluid chemistry .............................................................. 140 4.4.3.3 Formation water chemistry ............................................................................. 142 4.4.3.4 Additional variables ........................................................................................ 145 4.4.3.5 Montney Formation Mineralogy ..................................................................... 146 4.4.3.6 Overall correlation .......................................................................................... 148 4.5 Conclusions ................................................................................................................. 149 Chapter 5: Key minor elements in Montney Formation flowback water: Barium, boron, and lithium ..................................................................................................................................152  5.1 Introduction ................................................................................................................. 152  x  5.2 Methods....................................................................................................................... 155  5.2.1 Hydraulic fracturing fluid, flowback water, and produced water chemistry .......... 155 5.2.2 Montney Formation mineralogy ............................................................................. 157 5.3 Major ion chemistry of the Montney Formation flowback water ............................... 158 5.4 Key Minor Elements in Flowback Water ................................................................... 162 5.4.1 Barium..................................................................................................................... 163 5.4.2 Boron....................................................................................................................... 176 5.4.3 Lithium .................................................................................................................... 182  5.5 Conclusions ................................................................................................................. 186 Chapter 6: Conclusions and Recommendations .....................................................................189 6.1 Conclusions ................................................................................................................. 189 6.2 Recommendations ....................................................................................................... 193 References ...................................................................................................................................195  Appendices ..................................................................................................................................208  Appendix A Detailed sample preparation method .................................................................. 208 Appendix B Example of code used for PHREEQC geochemical mixing model ................... 211 Appendix C Percentages of formation water for all analyzed samples using δ2H isotopes, δ18O isotopes, and Cl concentrations as conservative tracers ......................................................... 213   xi  List of Tables  Table 1.1: List of categories of chemical additives used in hydraulic fracturing fluids (modified from FracFocus, 2017). ................................................................................................................... 4 Table 2.1: Summary of the four PHREEQC mixing models used for the analysis. ..................... 24 Table 2.2 (following page): Summary of pH, conductivity, total alkalinity, TDS, and major ion concentrations in the flowback waters collected in the study. ...................................................... 26 Table 2.3: Summary of the δ2H values, δ18O values, and Cl concentrations of the hydraulic fracturing fluids and formation water used for the mixing model. ............................................... 36 Table 2.4: Range of the δ2H values, δ18O values and Cl concentrations over the flowback period for flowback waters collected from the study wells. .................................................................... 37  Table 2.5: Summary of proportions of formation water contributing to the flowback water for the conservative tracers used in the study. .......................................................................................... 39  Table 2.6 (following page): Summary of the hydraulic fracturing fluid chemistry and the formation water chemistry. ........................................................................................................... 48 Table 2.7: Summary of CEC and mineralogy of selected Montney Formation samples. ............ 53 Table 3.1: Summary of some of the variables that may influence flowback water geochemistry........................................................................................................................................................ 80  Table 3.2 (following page): Summary of conductivity, TDS, and the major ion concentrations for each of the 9 study sites. ............................................................................................................... 86  Table 3.3: Groups of sites based on similarities in slope coefficient.......................................... 105 Table 4.1: List of several of the variables that may potentially be impacting the flowback water chemistry.. ................................................................................................................................... 116  Table 4.2 (following page): Summary of the flowback water TDS and major ion chemistry for the sampled wells. ....................................................................................................................... 123  Table 4.3: Summary of data for the variables considered in the regression analysis. ................ 128 Table 4.4(following page): Summary of hydraulic fracturing fluid chemistry and formation water chemistry for each of the wells. .................................................................................................. 129 Table 4.5: Summary of the Montney Formation mineralogy used in the regression analysis. ... 131 xii  Table 4.6: R2 values for the initial sample flowback water chemistry and the variables of interest.  ..................................................................................................................................................... 133  Table 4.7: Summary of the datasets used in the step-wise regression analysis in order to considered all sites and all variables of interest. ......................................................................... 134  Table 4.8: R2 values for the intermediate-late flowback water chemistry and the variables. ..... 137 Table 4.9: Cation exchange capacity (CEC) of samples from the Montney Formation. ............ 142 Table 5.1(following page): Summary of the general chemistry and the major ion, Ba, B, and Li concentrations for the study wells............................................................................................... 160  Table 5.2: Summary of the hydraulic fracturing fluid Ba, B, and Li concentrations for each of the study wells. ................................................................................................................................. 163  Table 5.3: Summary of produced water Ba, B, and Li concentrations compiled as part of the study. ........................................................................................................................................... 168  Table 5.4: Summary of the compiled mineralogy for the study sites. ........................................ 181  xiii  List of Figures  Figure 2.1: Schematic cross-section of the Montney Formation. ................................................. 18 Figure 2.2: Location of the 8 study sites (A, B, and D-I). ............................................................ 19 Figure 2.3: Flowback water TDS for the study sites plotted as a function of the cumulative volume of flowback collected.  a) Site A wells – upper Montney Formation; b) Site B wells – upper Montney Formation; c) Site D wells – upper (well D-1 and well D-2) and middle (well D-3 and well D-4) Montney Formation; d) Site E and Site F wells – middle Montney Formation; e) Site G and Site H wells – middle (well G-1) and upper (well H-1) Montney Formation; f) Site I wells – lower Montney Formation. ............................................................................................... 30  Figure 2.4: Flowback water TDS plotted as a function of percent recovered.  a) Site A wells – upper Montney Formation; b) Site B wells – upper Montney Formation; c) Site D wells – upper (well D-1 and well D-2) and middle (well D-3 and well D-4) Montney Formation; d) Site E and Site F wells – middle Montney Formation; e) Site G and Site H wells – middle (well G-1) and upper (well H-1) Montney Formation; f) Site I wells – lower Montney Formation. .................... 31 Figure 2.5: Examples of the linear relationship between selected major ions in the Montney Formation flowback water. a) Na-Cl plot; b) Ca-Mg plot. ........................................................... 33 Figure 2.6: Cross plots between the conservative tracers used to calculate the mixing ratios for this study.  a) Cl concentrations versus δ18O values; b) Cl concentrations versus δ2H values; and c) δ2H values versus δ18O values. ................................................................................................. 35 Figure 2.7: Fraction of formation water calculated for well D-4 using the three conservative tracers – Cl concentrations, δ18O, and δ2H. .................................................................................. 42 Figure 2.8: a) Correlations between the TDS of the initial flowback (FB) water samples from each well and the shut-in time following hydraulic fracturing and before the beginning of the flowback period; b) Correlation between the minimum percent of formation water at a well and the shut-in time using the proportion calculated with all three of the tracers. .............................. 45 Figure 2.9: δ2H vs δ18O plot for the site H well. ........................................................................... 47 Figure 2.10: Modelled and measured results for (a) Na; and (b) K concentrations for well I-1, provided as an example. ................................................................................................................ 51  xiv  Figure 2.11: Modelled and measured results for (a) Ca; (b) Mg; and (c) Sr concentrations for well I-1. ......................................................................................................................................... 56  Figure 2.12: The modelled Mg concentrations for different ion exchange values in the geochemical model using well I-1 as an example. ....................................................................... 58 Figure 2.13: Modelled and measured results for (a) Ca; and (b) Mg concentrations with the modelled results for mineral dissolution.. ..................................................................................... 60  Figure 2.14: Sulfate concentrations in the injected hydraulic fracturing (HF) fluid, the flowback water, and the formation water for each of the sites. .................................................................... 62 Figure 2.15: The SO4 concentrations over the flowback period at the site I wells.. ..................... 63 Figure 3.1: Schematic cross-section of the Montney Formation showing divisions between the lower, middle, and upper informal members. ............................................................................... 73  Figure 3.2: Location of the sites (A-I) included in the study.. ...................................................... 74 Figure 3.3: Different potential variables for graphical representation of flowback water chemical data from one well (data from well I-1). a) Na concentrations versus flowback day; b) Na concentrations versus cumulative flowback volume; c) Na concentrations versus percent recovered3; d) Na concentrations versus percent of total flowback water4; e) Na concentrations versus Cl concentrations; f) Na/Cl mass ratio versus percent recovered. ..................................... 82 Figure 3.4: TDS at the study wells over the flowback period as cumulative flowback volume. a) Site A wells – upper Montney Formation; b) Site B wells – upper Montney Formation; c) Site C wells – upper (wells 1-5) and middle (wells 6 & 7) Montney Formation; d) Site D wells – upper (wells 1 & 2) and middle (wells 3 & 4) Montney Formation; e) Site E wells – middle Montney Formation; f) Site F wells – middle Montney Formation; g) Site G well – middle Montney Formation; h) Site H well – upper Montney Formation; i) Site I wells – lower Montney Formation. ..................................................................................................................................... 85  Figure 3.5: Different sources of variability in the studied Montney Formation flowback water.  TDS is used as an example, as the major ions show similar trends to TDS. a) All data are grouped into early (day 1-2), middle (day 2-7), and late (> day 7) flowback water and produced water; b) The results for all wells are grouped by the Montney Formation member where hydraulic fracturing occurred; c) The results are grouped by site; d) The results for all wells are grouped by region. ........................................................................................................................ 90 xv  Figure 3.6: The flowback water results as TDS for all samples from all wells divided based on a) the percent of total flowback and b) the percent recovered. ......................................................... 91 Figure 3.7: a) Na-Cl plot separated by site; b) Ca-Cl plot separated by site. ............................... 96 Figure 3.8: Na concentrations over the flowback period for (a) site C; and (b) site D wells. .... 101 Figure 3.9: An example of the linear fixed effects model with data from site E. ....................... 103 Figure 3.10: Slope coefficients with confidence intervals for a) TDS; and b) Sr concentrations over flowback time. .................................................................................................................... 106  Figure 3.11: Summary of the slope coefficients for the cumulative flowback volume as a function of flowback time for each of the wells for the early (day 0-3), mid (day 3-15), and late (> day 15) portions of the flowback period. ............................................................................... 108 Figure 3.12: Sr concentrations in produced water samples, used to approximate the formation water chemistry.. ......................................................................................................................... 111  Figure 4.1: Location of the 31 study sites in BC and Alberta, Canada....................................... 118 Figure 4.2: Schematic cross-section through the Montney Formation from the western margin in BC to the eastern margin Alberta................................................................................................ 119 Figure 4.3: Relationship between the initial sample flowback water TDS values and (a) the HF (hydraulic fracturing) fluid TDS; (b) the shut-in time; and (c) the breakdown pressure for the study wells. ................................................................................................................................. 132  Figure 4.4: Relative importance of variables found to be significant (p-value < 0.05) in the step-wise regression analysis. a) Results based on dataset I. Results using dataset II are similar; b) Results based on dataset III when hydraulic fracturing (HF) fluid was excluded in order to include the site C samples.  Results from dataset IV are similar. ............................................... 135  Figure 4.5: Relative importance of variables for intermediate to late flowback water. a) Dataset I – site C and breakdown pressure excluded; b) Dataset II – sites C, F, and H excluded; c) Dataset III – HF fluid chemistry and breakdown pressure excluded; d) Dataset IV – Site F, site H, and HF fluid chemistry excluded. ...................................................................................................... 139  Figure 4.6: Relative importance of parameters, including the last sample chemistry as a proxy for the formation water chemistry for a) the initial flowback water samples; and b) the intermediate-late flowback water samples. ...................................................................................................... 145  xvi  Figure 4.7: Correlations between (a) Na concentrations; (b) Ca concentrations; (c) Na/Cl; and (d) Ca/Cl with median clay. .............................................................................................................. 148  Figure 5.1: Schematic cross-section of the Montney Formation in BC and Alberta. ................. 154 Figure 5.2: Location of the study sites in northeastern BC and northwestern Alberta. .............. 155 Figure 5.3: Ba concentrations over the flowback period plotted as cumulative flowback volume for a) Site A wells; b) Site B wells; c) Site C wells; d) Site D wells; e) Site E-H wells; and f) Site I wells. ......................................................................................................................................... 165  Figure 5.4: Ba concentrations in produced water samples from wells completed in the Montney Formation and located up to 20 km from the study sites. ........................................................... 167  Figure 5.5: The relationship between Ba and SO4 concentrations for the sampled wells. ......... 170 Figure 5.6: Barite in a core sample from a well located near site A. .......................................... 170 Figure 5.7: a) Ba and SO4 concentrations for well I-1 and well I-2.  b) The saturation indices (SI) for barite for well I-1 and well I-2. ............................................................................................. 175 Figure 5.8: Boron concentrations over the flowback period, plotted as cumulative flowback volume. a) upper Montney member wells; b) middle and lower Montney member wells. ........ 177 Figure 5.9: Correlation between the maximum B concentration in flowback water for each of the wells and the total vertical depth (TVD) of the well. ................................................................. 179 Figure 5.10: Boron concentrations across the formation plotted as cumulative flowback volume...................................................................................................................................................... 179  Figure 5.11: Li-B plot showing the three regions where the sampled wells plot. ...................... 186    xvii  Acknowledgements Many people have provided support during my time at UBC and made this research possible.  First of all, I would like to thank my supervisor, Dr. Marc Bustin, for making this project possible by arranging access to the samples, allowing me to have firsthand experience on several analytical instruments at UBC, and reviewing the draft versions of the thesis chapters many times.  The recommendations have greatly improved this thesis.  I am also thankful for feedback provided by my committee members, Dr. Roger Beckie and Dr. Uli Mayer, and external examiner Dr. Ken Hickey.   This thesis would not have been possible without the willingness of several companies to provide the samples.  The sample collection and shipment conducted by ARC Resources Ltd., Black Swan Energy Ltd., Unconventional Gas Resources, and other companies that wish to remain anonymous was appreciated.  Financial support from Trican, Encana Corporation, Husky Energy, Canadian Natural Resources (CNRL), Chevron Corporation, and Geoscience BC made this research project possible.  I am thankful for the assistance provided by many people, including Maureen Soon, Vivian Lai, and Dirk Kirste, for training me to use various analytical instruments and helping run samples.  I also appreciate Mati Raudsepp stopping in at my office sometimes to check that I am working.  I would like to thank the undergraduate research students who helped with the sample preparation and the basic analytical tests, in particular Scott Hazell.  Without them I would not have been able to analyze all of the samples in a reasonable amount of time.   I am grateful for the encouragement from my parents and their unfailing belief that I could complete this thesis.  I am also thankful for my sister, her new husband, and my friends for reminding me of life outside of school.     Finally, I would like to thank my fiancé, Sean Stevenson, for supporting me in so many ways.  I am looking forward to the next chapter of our life together.     1  Chapter 1: Introduction 1.1 Introduction The combined technologies of hydraulic fracturing and horizontal drilling are used to develop oil and gas reservoirs in low permeability rock. The injection of the hydraulic fracturing fluids in multiple stages under high pressures causes fractures to form in clusters in targeted locations along the horizontal portion of the well to access the hydrocarbons in the formation.  Of the injected fluid volume, which is typically between approximately 5,000 to 50,000 m3 of water in total per well (Vidic et al., 2013; Alessi et al., 2017), approximately 75% remains in the subsurface (Haluszczak et al., 2013). However, this varies between wells and between formations and in some cases a volume of fluid equivalent to 100% of the injected fluid volume is returned to surface (Rivard et al., 2014).  The fluid collected at surface is referred to as flowback water.  Flowback water is typically high in total dissolved solids (TDS) and various other ions and compounds both from the hydraulic fracturing process and from the formation, including inorganic ions (e.g., Chapman et al., 2012; Barbot et al., 2013; Haluszczak et al., 2013; Rowan et al., 2015), organic ions (e.g., Abualfaraj et al., 2014; Akob et al., 2015; Lester et al., 2015), and naturally occurring radioactive material (NORMS) (e.g., Nelson et al., 2014, 2015, 2016).  The TDS of the flowback water generally increases over the flowback period due to the increasing concentrations of the major ions, although in some wells the TDS can stabilize in later samples (e.g., Haluszczak et al., 2013; Zolfaghari et al., 2016).  Characterization of the chemistry of this water is important as it provides insight into geochemical processes that are occurring in the reservoir, and in addition, can assist with flowback water management at future wells to minimize impacts on the freshwater resources around oil and gas development.   2  The subject of this thesis is the inorganic flowback water chemistry from wells completed in the Lower Triassic Montney Formation in northeastern British Columbia and northwestern Alberta.  This formation includes shale, dolomitic-siltstone and fine grained sandstone lithologies (Edwards et al., 1994; Dixon, 2000).  Conventional oil and gas exploration in the Montney Formation targeted turbidite deposits, sandstone units, and coquina dolomite units (Bird et al., 1994); however, current exploration is focused on the unconventional resources in the fine grained deposits (Zonneveld et al., 2011). There is no Montney Formation flowback water chemistry available in the literature as the majority of the previous work has been completed in other formations, most notably the Marcellus Shale Formation in the eastern United States (e.g., Barbot et al., 2013; Capo et al., 2013; Haluszczak et al; 2013; Marcon et al.; 2017).  The present study characterizes the major ion and selected minor ion chemistry of the Montney Formation flowback water from nine well pads located across the formation.  The variability between sites and the dominant factors impacting the flowback water chemistry are investigated. 1.2 Background Information The following subsections provide relevant background information on hydraulic fracturing fluids, formation water, and flowback water.  Hydraulic fracturing fluid and formation water are discussed as the chemistry of these fluids likely impact the flowback water chemistry through mixing in the reservoir.  This summary is designed to provide an initial background understanding of these fluids to support the more detailed discussion in the subsequent papers. 1.2.1 Hydraulic Fracturing Fluids Hydraulic fracturing fluids are composed of water, proppant, and chemical additives.  The water and proppant make up approximately 99.5% of the volume of the fracturing fluid, with the additives making up the remaining 0.5-1% (Gregory et al., 2011; McLaughlin et al., 3  2016).  The volume of water required to hydraulically fracture a well varies between wells and formations.  For the Montney Formation, the average volume used per well is between about 8,000 and 11,000 m3 (Alessi et al., 2017).  The base fluid used to make up the hydraulic fracturing fluid can be surface water, groundwater, or a mixture of freshwater and treated flowback water from previously fractured wells (Shipman et al., 2013).  If flowback water is recycled, the high salinity and presence of organic matter need to be taken into consideration for hydraulic fracturing fluid performance (Esmaeilirad et al., 2016).  There is ongoing research on chemical additives that can be used with higher salinity water, including recycled flowback water, to improve the water management related to hydraulic fracturing operations (e.g., Fontenelle et al., 2013; Shipman et al., 2013).   In addition to the makeup water, the other components of the hydraulic fracturing fluid are the proppants and the chemical additives.  Proppants are sand sized grains of quartz with or without a resin coat or can be made of a synthetic material including ceramic, bauxite, resin, or plastic.  The purpose of these grains is to keep the fractures created during the hydraulic fracturing process ‘propped’ open once the pressure in the well has been decreased through production.  The proppant is approximately 9% of the fracturing fluid, by weight (Vidic et al., 2013).  There are a multitude of chemical additives that can be used in hydraulic fracturing fluid.  Additives can include acids, friction reducers, surfactants, salts, scale inhibitors, pH adjustors, iron controls, corrosion inhibitors, and biocides (Gregory et al., 2011; Table 1.1).  In general, not all additives listed are required at each hydraulic fracturing site (Gregory et al., 2011).  In general, between 4 and 28 additives are used in the hydraulic fracturing fluid for a well (McLaughlin et al., 2016).  The composition of the hydraulic fracturing fluid is based on site-specific characteristics, including the water chemistry used to make up the hydraulic fracturing 4  fluid, the formation water chemistry, and the formation lithology.  In some wells, chemical tracers, such as tritiated water and thiocyanates, are injected with the hydraulic fracturing fluid to study the extent of the fractures and potential hydraulic communication between nearby wells (Salman and Kurtoglu, 2014).  These studies show that different hydraulic fracturing stages are contributing to the flowback water at different times throughout the flowback period (Sullivan et al., 2004).      Table 1.1: List of categories of chemical additives used in hydraulic fracturing fluids (modified from FracFocus, 2017). Chemical Additive Example Purpose Acid Hydrochloric acid To dissolve minerals and help crack the rock Biocide Glutaraldehyde Removes bacteria Breaker Ammonium persulphate Breaks down gel additives Clay stabilizer Choline chloride Decreases swelling of clays Corrosion inhibitor Acetaldehyde Decreases corrosion of the pipe Crosslinker Potassium metaborate Maintains viscosity of the fluid Friction reducer Polyacrylamide Decreases friction in the fluid Gelling agent Guar gum Acts to thicken the fluid Iron control Acetic acid Decreases precipitation of iron oxides Non-emulsifier Isopropanol Decreases formation of emulsions in fluid pH adjusting agent Sodium hydroxide Helps with maintaining the optimal pH for the fluid components Scale inhibitor Sodium polycarboxylate Decreases scale in the pipe Surfactant Methanol Increases fluid viscosity and reduces surface tension  1.2.2 Formation Water Formation water can have TDS greater than 300,000 mg/L (Bagheri et al., 2014) and is classified as Na-Cl, Cl-Ca or Na-Ca-Cl type water (Connolly et al., 1990; Stueber and Walter, 1991). The TDS range of the formation water from the Montney Formation investigated to date, ranges from about 50,000 to 250,000 mg/L TDS and it is classified mainly as Na-Cl type (Kirste et al., 1997).  The formation water chemistry can change over time compared to the chemistry of 5  the original connate water deposited with the formation due to mixing with meteoric water or other formation waters and from water-rock interactions, such as albitization, dolomitization, interactions with clay minerals, and mineral precipitation or dissolution (e.g., Bagheri et al., 2014; Connolly et al., 1990).  In order to determine if the elevated concentrations are due solely to evaporation or if other processes are occurring to alter the formation water chemistry, the ion concentrations in the formation water can be compared to the seawater evaporation pathway (Dresel and Rose, 2010).  Seawater evaporation curves have been developed for the major ions and account for several stages of mineral precipitation during the evaporation process, including calcite precipitation, gypsum precipitation, and halite precipitation (e.g., Carpenter et al., 1978; McCaffrey et al., 1987).  The Cl/Br ratio is often used in studies on the formation water chemistry of sedimentary basins as this ratio remains constant until halite saturation is reached at which point Cl is preferentially incorporated over Br into the halite crystal lattice while the aqueous Br concentrations continue to increase (Carpenter et al., 1978; McCaffrey et al., 1987, Freeman 2007; Stueber and Walter 1991).  The Alberta Basin formation waters generally plot on or near the seawater evaporation curve indicating that the elevated salinity is due to the evaporation of seawater rather than halite dissolution (Connolly et al., 1990).  However, numerical models indicate that there may be some contribution to the Alberta basin brines from halite dissolution in addition to contributions from evaporated seawater and freshwater (Gupta et al., 2012).  Minor depletions in the Na and Cl concentrations in formation water can result from albitization and chlorite formation, respectively (Rowan et al., 2015). Kirste et al. (1997) plotted major cation concentrations with Br concentrations for formation water brines from the Montney, Halfway, and Doig Formations.  They found that the Montney Formation waters had Na concentrations that plotted close to the seawater evaporation 6  curve while K and Mg concentrations were depleted and Ca concentrations were enriched relative to evaporated seawater.  A similar enrichment in Ca concentrations and depletion in Mg concentrations relative to seawater evaporation was observed by Dresel and Rose (2010) for western Pennsylvanian brines, by Haluszczak et al. (2013) for brines from the Marcellus Shale and by Michael et al. (2003) for brines in Devonian aquifers in the Alberta Basin.  This was interpreted to be a result of dolomitization.  Ca concentrations in formation waters may also increase due to albitization (Michael et al., 2003).  The depletion in K relative to evaporated seawater may be due to the conversion of smectite to illite during diagenesis (Chaudhuri and Clauer, 1993; Cooley and Donnelly, 2012) or due to adsorption on clay minerals (Chan et al., 2002).  1.2.3 Flowback Water 1.2.3.1 Inorganic chemistry Flowback water is generally highly saline although concentrations vary over the flowback period and between wells and formations.  TDS values can be as high as 345,000 mg/L (Kolesar Kohl et al., 2014) and Cl concentrations can be greater than 100,000 mg/L (Haluszczak et al., 2013).  Of the major cations, Na concentrations are highest (commonly greater than 30,000 mg/L), while Ca, Mg, and K concentrations are lower (e.g., Gregory et al., 2011; Haluszczak et al., 2013).  Additional ions that are often elevated in flowback water include Ba and Sr (Barbot et al., 2013).  In general, the TDS and the concentrations of the major ions typically increase over the flowback period, in some wells reaching a plateau and in other wells showing a continual increase over the flowback period (e.g., Blauch et al., 2009; Haluszczak et al., 2013).  The variable response may be due to the complexity of the fracture system, with the continuous increase being associated with more secondary fractures and a greater surface area for exposure 7  to formation water (Bearinger, 2013; Ghanbari et al., 2013; Zolfaghari et al., 2015a). In contrast to Cl and the major cations, sulfate (SO4) concentrations typically decrease (Haluszczak et al., 2013) or remain relatively stable (Ziemkiewicz and He, 2015).  1.2.3.2 Organic chemistry Organic compounds and their degradation products are found in flowback water (Lester et al., 2015).  The majority of the organic compounds can typically be characterized as aliphatic organics, followed by cycloaliphatic and aromatic organics, and only a small proportion (<5%) of polycyclic aromatic hydrocarbons (Strong et al., 2014).  Organic compounds in Marcellus Shale flowback water can include volatile organic carbon, 1,2-dichloroethane, benzene, benzo(a)pyrene, dibromochloromethane, ethylbenzene, naphthalene, pentachlorophenol, styrene, toluene, tetrachloroethylene, vinyl chloride, and xylenes (Abualfaraj et al., 2014; Strong et al., 2014; Ziemkiewicz and He, 2015).  Some of these compounds are additives in hydraulic fracturing fluid; however, others, such as benzene and toluene, can be found naturally near oil and gas reservoirs (Lester et al., 2015).  Acetate, formate, and pyruvate have also been detected in flowback water and indicate that organic compounds in the hydraulic fracturing fluid are being degraded by bacteria (Akob et al., 2015).  In their study, they found that the organic chemistry of the flowback water varied between wells even within the same formation.   1.2.3.3 Influences on flowback water chemistry Different hypotheses have been proposed to explain the changes in the inorganic ion concentrations over the flowback period.  These include mixing of the formation water with the hydraulic fracturing fluid (e.g., Haluszczak et al., 2013; Engle and Rowan, 2014; Olsson et al., 2013; Vengosh et al., 2017), diffusion into the fractures followed by mixing (Balashov et al., 2015), and water-rock interactions (e.g., Barbot et al., 2013; Marcon et al., 2017; Seales et al., 8  2016; Zolfaghari et al., 2016).  The processes influencing the flowback water chemistry may vary between sites and over the flowback period.  In addition, the processes affecting the ion concentrations may not be the same for all ions.  1.2.3.3.1 Mixing between hydraulic fracturing fluid and formation water Several studies suggest that the increase in the ion concentrations observed in flowback water is due to mixing between the relatively low salinity injected hydraulic fracturing fluid and the higher salinity formation water (e.g., Haluszczak et al., 2013; Engle and Rowan, 2014; Olsson et al., 2013; Vengosh et al., 2017).  Other potential sources of high salinity water include nearby units and higher porosity lenses or fractures within the low permeability unit (Stewart et al., 2015).  The elevated ion concentrations in flowback water are not related to the hydraulic fracturing fluid chemistry as these fluids typically have much lower concentrations of the majority of ions relative to flowback waters, indicating that the high concentrations are not related to the chemical additives (Ziemkiewicz and He, 2015).  In wells where recycled water is used to make up the hydraulic fracturing fluid, the major ion concentrations are higher in the injected fluid and may be similar to the initial flowback water from the well.  Inorganic chemistry results that support the mixing hypothesis include:  Haluszczak et al. (2013) compiled water chemistry from several sources within the Marcellus Shale, including flowback water samples and formation water brine samples.  Based on their dataset, they concluded that as flowback progressed, the effect of the formation water chemistry on the chemistry of the flowback water increased. This is based on the observation that the flowback water chemistry approaches that of the formation water chemistry over the flowback period; 9   Haluszczak et al. (2013) also compared the Cl and Br concentrations to the seawater evaporation curves and found that the source of salinity could be explained as evaporated seawater that has been evaporated past the point of halite saturation and has been subsequently diluted with fresher water, rather than a halite dissolution source;  Olsson et al. (2013) showed that an increased proportion of formation water mixed with the lower salinity hydraulic fracturing fluid can be used to explain the increase in Cl concentration over the flowback period;  Stewart et al. (2015) found that the ion ratios measured during sequential extractions of shales were not consistent with the ion ratios observed in flowback water samples, indicating that the water-rock interactions were not the dominant source of TDS in the flowback water; and  Kondash et al. (2017) used a mass balance calculation for TDS based on mixing to determine the percentage of hydraulic fracturing fluid recovered at different times during flowback. Isotopic data provides additional support for the mixing hypothesis.  A study examining the stable isotopes found that the δ18O values of flowback water increase and move away from the meteoric water line, indicating a hydraulic fracturing fluid derived from freshwater and an increasing proportion of formation water derived from evaporated seawater (Rowan et al., 2015).  Sr isotopic ratios can also be used to look at fluid mixing between the formation water and hydraulic fracturing fluid (Chapman et al., 2012).  The 87Sr/86Sr ratio is low in the injected fluid and the ratio initially increases in the flowback water before reaching stable values (Stewart et al., 2015).      10  Rather than mixing between the formation water and the hydraulic fracturing fluid, another possibility is diffusion of the salts in the formation water brine into the hydraulic fracturing fluid (Balashov et al., 2015).  They developed a model that shows that only 2% of the shale, by volume, is required to have brine as either free brine or capillary bound brine in order to explain the elevated ion concentrations observed in flowback water.  The model estimates the stimulated reservoir volume at 1.5 x 107 m3 based on typical values for hydraulic fracturing in the Marcellus Shale.  Salt diffusion out of the formation and towards the fractures was also modelled by Wang et al. (2016, 2017).  They found that diffusion of ions acts in the opposite direction to water osmosis, a process which should be taken into consideration as a mechanism for reducing flowback water volume.   1.2.3.3.2 Water-rock interactions Geochemical processes in addition to mixing may be required to more completely explain the flowback water chemistry.  If only mixing were involved in the increasing ion concentrations, there would be a 1:1 correlation between all ions, whereas there is often an enrichment or depletion of certain ions at different times throughout the flowback period (Barbot et al., 2013).  The additional geochemical processes impacting the flowback water chemistry may be water-rock interactions in the subsurface, including mineral precipitation or dissolution, and ion exchange with clay.   Halite dissolution has been suggested as a potential source of salinity in flowback water and halite has been observed in Marcellus Shale core (Blauch et al., 2009); however, there has been some discussion on if halite is only found in some locations in the Marcellus Shale or if it was present as a precipitate formed during the drilling process (e.g., Haluszczak et al., 2013).  A numerical model developed for Na and Cl concentrations indicates that some halite dissolution is 11  necessary to explain the elevated concentrations of Na and Cl measured in flowback water and that mixing alone does not adequately predict the Na and Cl concentrations (Seales et al., 2016).   Other mineral reactions can involve precipitation and dissolution of carbonate, sulfate, and phosphate minerals.  Several laboratory experiments have observed the dissolution of carbonates in shale samples in synthetic hydraulic fracturing fluids under elevated temperature and pressure conditions (Wilke et al., 2015; Dieterich et al., 2016; Marcon et al., 2017; Lu et al., 2017).  Clay dissolution was also observed in one of the experiments (Marcon et al., 2017).  The oxidation of pyrite by the oxic hydraulic fracturing fluid was observed by Wilke et al. (2015) and Harrison et al. (2017), which resulted in a decreased pH if sufficient carbonate was not available to buffer the system.  Buffering by carbonate was found to be an important control on the release of trace metals from the shale samples (Wilke et al., 2015). Various secondary mineral phases were observed to precipitate on the core samples in the experiments.  These included gypsum, barite, strontianite, celestite, apatite (Dieterich et al., 2016), smectite, anhydrite (Marcon et al., 2017), and iron oxides and hydroxides (Lu et al., 2017).   Ion exchange may also have an impact on flowback water chemistry.  For flowback water samples collected from wells in the Horn River Basin in BC, the Na and K to Cl ratios were initially high while these ratios decreased later in the flowback period (Zolfaghari et al., 2016).  This was interpreted as ion exchange early on in the flowback period releasing Na and K ions followed by dissolution of halite and potassium chloride, which would increase Cl concentrations (Zolfaghari et al., 2016).  Na and K ions may be released from clays as these ions are the interlayer cations for smectite and illite, respectively (Zolfaghari et al., 2015b). The sources of the halite and potassium chloride salts could be evaporated pore water in the pore spaces (Ghanbari et al., 2013).  Sequential extraction experiments show that the monovalent 12  cations are more loosely bound to the rock and are likely from ion exchange or formation water or salts, while divalent cations are more tightly bound and may have a greater impact later in the flowback period (Zolfaghari et al., 2015b).  Barium (Renock et al., 2016), boron, and lithium (Warner et al., 2014) may also be released through ion exchange facilitated by the injection of the low salinity hydraulic fracturing fluid into the formation which is in equilibrium with the higher salinity formation water.        1.3 Organization of thesis   The introduction to this thesis provided a brief overview of relevant background information and previous work completed on flowback water chemistry from other formations.  The aim of the present series of studies is to characterize the Montney Formation flowback water to provide information on the variability and controls on the flowback water chemistry from this formation.  Although the results of this study are specific to the Montney Formation, the overall conclusions can be applied to other formations where hydraulic fracturing is occurring. Chapters 2 through 5 are a series of four separate, standalone research papers.  The research papers are designed to each cover different aspects of the study as a whole.  Chapter 2 focusses on determining the mixing ratios between the injected hydraulic fracture fluid and the formation water.  The mixing ratios are then used with geochemical models to determine if the major ion concentrations can be explained by mixing alone or if other geochemical processes also have a significant effect.  Chapters 3 and 4 present related aspects of a larger study.  Chapter 3 is part I of the study and discusses comparing the flowback water chemistry between different sites and presents a method for making a comparison.  Chapter 4 is part II of the study and investigates the dominant variables influencing the flowback water chemistry.  Chapter 5 includes a discussion on selected minor elements, including barium, boron, and lithium, present at elevated 13  concentrations in flowback water.  Chapter 2 uses the geochemical results from 24 wells, as the hydraulic fracturing fluids were not collected for the wells from one site (7 wells) and the hydraulic fracturing fluid chemistry was required for the analysis.  Chapters 3 through 5 use the results from 31 wells.  The four papers are followed by the major conclusions of the studies and proposed further areas of research in Chapter 6.  14  Chapter 2: Investigating the Montney Formation flowback water chemistry through mixing calculations and geochemical modelling  2.1 Introduction Production of oil and gas from unconventional reservoirs involves the combination of horizontal drilling and hydraulic fracturing to access hydrocarbons in low permeability formations.  The hydraulic fracturing process entails the injection of large volumes of fluid into the target formation in order to fracture the rock and increase the permeability to allow for the production of oil and/or gas from the reservoir.  The volume of fluid required is on the order of 1,000's to 10,000's of cubic meters (m3) per well (Vidic et al., 2013; Alessi et al., 2017).  Prior to the production stage, a large volume of variably saline water is recovered at the surface.  This water, referred to as flowback water, is volumetrically approximately 25% of the injected fluid volume (Haluszczak et al., 2013); however, there is a large amount of variability and the volume of flowback water can equal (or exceed) the volume of fluid that was injected into the reservoir (Rivard et al., 2014).  The flowback water, even initially, is chemically different than the injected fluid.  The total dissolved solids (TDS) of the flowback water generally increases over the flowback period and can reach levels greater than 300,000 mg/L (Barbot et al., 2013).  In several previous flowback water studies, the increasing TDS has been attributed, at least in part, to mixing between the injected hydraulic fracturing fluid and the connate formation water (e.g., Haluszczak et al., 2013; Olsson et al., 2013; Capo et al., 2014; Engle and Rowan, 2014; Kolesar Kohl et al., 2014; Kondash et al., 2017).  Mixing is considered since the major ion concentrations and isotopic values (e.g., δ18O, δ2H, and 87Sr/86Sr) approach those in formation water over the 15  flowback period, indicating an increasing proportion of formation water over time (e.g., Rostron and Arkadakskiy, 2014).  Examining the change in the major ions and the isotopes can be used to determine if mixing alone can adequately explain the major ion concentrations and the isotopic values observed in flowback water. In order to determine mixing ratios between different water sources, conservative tracers are required.  A conserved tracer is one that is not involved in any chemical or biological reactions in the studied system and consequently any changes observed in the tracer concentration can be inferred to be due to mixing between different endmembers.  Chemical tracers, such as tritiated water and methanol, can be injected with the hydraulic fracturing fluid and are used to look at the extent of the fracture network (Salman and Kurtoglu, 2014).  However, there are also ions (e.g., chloride [Cl]) and stable isotopes (e.g., δ18O and δ2H) that are naturally found in flowback water and act as conservative tracers in the system.  Cl is the dominant ion in flowback water and increases over the flowback period, often reaching concentrations greater than 100,000 mg/L (Haluszczak et al., 2013).  Cl is a suitable conservative tracer in flowback water studies where halite dissolution is not occurring and where the water remains undersaturated in halite.  The stable water isotopes1 are also useful tracers for the mixing proportions in flowback water as the formation temperature and length of time for water-rock interactions between the injected hydraulic fracturing fluid and the rock are not sufficient to significantly change the isotopic signature of water in the system (Anderson and Chai, 1974;                                                  1 Isotopic values are reported as delta values (δ) which are calculated as: 𝛿 = ൬𝑅௦௔௠௣௟௘  𝑅௦௧௔௡ௗ௔௥ௗ  − 1൰ 𝑥 1000‰, 𝑤ℎ𝑒𝑟𝑒 𝑅 =  𝑅𝑎𝑟𝑒 𝑖𝑠𝑜𝑡𝑜𝑝𝑒 𝑎𝑏𝑢𝑛𝑑𝑎𝑛𝑐𝑒𝐶𝑜𝑚𝑚𝑜𝑛 𝑖𝑠𝑜𝑡𝑜𝑝𝑒 𝑎𝑏𝑢𝑛𝑑𝑎𝑛𝑐𝑒 The water isotopic values are measured relative to Vienna Standard Mean Ocean Water (VSMOW).  16  O'Neil and Kharaka, 1975; Rowan et al., 2015).  Water produced from deep formations typically has elevated δ18O and δ2H values relative to meteoric water and results that plot to the right of the Global Meteoric Water Line (GMWL) (e.g., Sharma et al., 2014; Rowan et al., 2015).  This shift in the isotopic signature of the formation water may be due to water-rock interactions, such as gypsum dehydration and exchange between water and carbonates, which have occurred since the time of deposition (Sharma et al., 2014). Simple mixing calculations using conservative tracers can be completed to calculate the percentages of the endmember source waters (i.e., the hydraulic fracturing fluid and the formation water) in the mixed flowback water.  The expected concentrations of different elements based on mixing alone can be compared to the measured concentrations in flowback water to provide information on the geochemical processes occurring during the flowback period.   The current study focusses on flowback water from the Lower Triassic Montney Formation in northeast British Columbia (BC) and northwest Alberta, Canada.  The formation unconformably overlies the Permian Belloy Formation and is overlain by the Doig Phosphate Zone of the Middle Triassic Doig Formation (Edwards et al., 1994).  The Montney Formation consists of fine grained sediments, principally composed of carbonate, quartz, and feldspar, with minimal clay, and is dominated by dolo-quartz-siltstone and very fine grained quartz sandstone (Zonneveld et al., 2011; Zonneveld and Moslow, 2014).  The formation generally varies from shale in the west-northwest to interbedded sandstone and shale in the east-southeast (Edwards et al., 1994), representing distal shelf to deltaic and shoreline environments (Chalmers and Bustin, 2012).  Very fine grained sandstone deposits of the Montney Formation located mainly in the region near the BC-Alberta border, are interpreted as turbidite deposits (Zonneveld et al., 2010).  Past studies on the Montney Formation have focused in either BC or Alberta.  In Alberta, the 17  formation has been divided into a lower very fine grained sandstone and siltstone member, a middle coquina dolomite member and an upper siltstone member (Davies et al., 1997).  In BC, the coquina dolomite member is absent and there is an overlying shale member (Dixon, 2000).  A recent stratigraphic study of the Montney Formation by Davies and Hume (2016) applied across the basin divides the formation based on a combination of stratigraphy and biostratigraphy into lower (Griesbachian-Dienerian), middle (Smithian), and upper (Spathian) members (Fig. 2.1).  The upper member in this framework includes the siltstone unit, which is sometimes referred to as the Lower Doig siltstone, and is located below the Doig Phosphate Zone.  The present study follows Davies and Hume's (2016) recent subdivision of the Montney Formation into lower, middle, and upper members.       18    Figure 2.1: Schematic cross-section of the Montney Formation.  This division into lower, middle, and upper Montney Formation is used in the present study.  Each of these members represents a Third-Order sequence. The upper Montney Formation includes the lower Doig siltstone unit which is below the Doig Phosphate Zone (modified from Davies and Hume, 2016).  The study area includes wells completed in the Montney Formation in both BC and Alberta.  Fluid samples were obtained from 24 wells located at 8 sites (Fig. 2.2).  The wells were completed in the upper (13 wells), middle (9 wells), and lower (2 wells) Montney Formation.  Both the Cl concentrations and δ18O and δ2H isotopes are used to determine mixing proportions between the injected hydraulic fracturing fluid and the formation water. Geochemical mixing models are also used to determine if mixing alone can be used to explain the major ion concentrations in the flowback water or if there are additional process such as ion exchange or mineral precipitation and dissolution occurring to produce the observed flowback water 19  chemistry.  The major ions included in the analysis are sodium (Na), potassium (K), calcium (Ca), magnesium (Mg), strontium (Sr), and sulfate (SO4).       Figure 2.2: Location of the 8 study sites (A, B, and D-I). Wells completed in the upper Montney Formation are located at site A (8 wells), site B (2 wells), site D (2 wells), and site H (1 well).  Middle Montney Formation wells are located at site D (2 wells), site E (4 wells), site F (2 well), and site G (1 well).  The two wells located at site I were completed in the lower Montney Formation.  The extent of the Montney Formation is shown in green (modified from Edwards et al., 1994). The map area in relation to British Columbia and Alberta is show in the inset figure.  20  2.2 Methods 2.2.1 Sample collection 2.2.1.1 Hydraulic fracturing fluid Hydraulic fracturing fluids, as the base fluid with additives, were collected at each site in 500 mL or 1 L containers.  For the majority of wells, a separate sample used in each of the hydraulic fracturing stages was collected (14 to 26 samples per well).  The fluids used in the different stages were combined into two or more composite samples per well based on similar electrical conductivity readings.  Overall values for the major ion concentrations in the hydraulic fracturing fluid were derived for each well using the relative volumes of fluid injected in each stage.  For example, if two composite samples were analyzed for a well and one represents 75% of the injected fluid volume, the chemistry of this sample will represent ¾ of the chemistry in the overall values for the well, while the other sample will represent ¼ of the chemistry. A single representative hydraulic fracturing fluid sample was collected for the site F wells and each of the site I wells. 2.2.1.2 Flowback water Flowback water samples were collected in 500 mL or 1 L containers throughout the flowback period at each of the 24 sites.  A higher sampling frequency of 2 to 3 samples per day was used in the first week and then the frequency was reduced to 1 sample per day for the remainder of the flowback period. The variable frequency was used in order to capture the higher variability in chemistry earlier in the flowback period when the flowrate is higher.     2.2.1.3 Produced water For our study, the produced water is considered to be the water that is collected after the well has begun producing oil or gas.  The chemistry of the produced water is used to 21  approximate the formation water chemistry for the study wells with the recognition that the impact of the injected fluid on the produced water chemistry may extend well into the productive life of a well.  Several produced water samples were collected from the wells at site A and site B following the beginning of production.  The formation water for the two site I wells is approximated based on a single produced water sample collected from a producing well approximately 20 km away from site I.  No produced water samples were obtained for the wells from sites D through H.   The formation water chemistry for these wells was derived from the major ion concentrations from publically available data for produced water from nearby wells completed in the Montney Formation.  Samples with anomalously low TDS or potential contamination, were removed from consideration and the median values of the remaining results were used to approximate the formation water chemistry for the remaining wells on sites D through to H. The isotopic values for the formation water for all sites were estimated based on the analysis of the produced water samples collected as part of the present study as isotope values were not included in the publically available geochemical results used for the major ion chemistry. 2.2.2 Sample analysis 2.2.2.1 Major ion chemistry All samples were stored at 4°C prior to analysis.  As the samples were purposely not preserved or filtered on site, the flowback water and produced water samples were heated in 120 mL acid-washed Teflon® containers in a hot water bath to reservoir temperature (75-80°C) for a 24 hour period.  After heating, the samples were filtered through 0.45µm filters and divided into three sub-samples for analyses of: 1) pH, conductivity, and total alkalinity; 2) anions; and 3) dissolved metals.  The hydraulic fracturing fluid samples were also filtered and sub-divided but 22  were not heated.  The anion samples were sent to a laboratory in Victoria, BC and analyzed by ion chromatography.  The remaining analyses were conducted at The University of British Columbia in Vancouver, BC.  The pH was measured with an OMEGA® PHB21 portable pH probe and conductivity was measured with a HACH® CDC401 conductivity probe. Total alkalinity was measured by titration with sulfuric acid (H2SO4) using a HACH® digital titrator (model 16900).  The dissolved metal samples were preserved to pH < 2 with trace metal grade nitric acid (HNO3).   For dissolved metal analyses, the preserved samples were acid digested to remove organics prior to analysis, as the presence of organics may interfere with the analysis.  The first step in the digestion was adding concentrated trace metal grade HNO3 and concentrated trace metal grade hydrochloric acid (HCl) in a 2:1 ratio to each sample and evaporating the mixture on a hot plate.  This step was followed by adding additional HNO3 to the sample and re-evaporating the remaining fluid.  The evaporated residue was re-diluted with a 1% HNO3 deionized water solution.  The digested samples were analyzed for the major cations by inductively coupled plasma-optical emission spectrometry (ICP-OES) with a Varian 725-ES ICP-OES using indium (In) as the internal standard.  The ICP-OES analysis included Na, K, Ca, Mg, and Sr ions. TDS was calculated by summing the ion concentrations.  This TDS value was compared to an estimated TDS value based on the conductivity measurement as a method to check the result.  2.2.2.2 Oxygen and hydrogen isotopes The isotope samples were prepared by taking a sub-sample of unheated, filtered fluid and mixing it with activated charcoal for a minimum of 48 hours.  The activated charcoal acts to remove organic compounds through adsorption, as the organics may interfere with the isotopic measurements.  Following the mixing period, the sample was filtered a second time to remove 23  the activated charcoal.  The stable oxygen and hydrogen isotopic analysis was completed using an LGR® DLT-100 with an LC-PAL liquid autosampler.  High salinity samples were diluted with deionized water prior to the analysis in order to prevent the salts present in the sample from precipitating and plugging the instrument tubing.  The dilution factor was recorded and the isotopic values of the deionized water were determined with the instrument in order to complete a mixing calculation to determine the isotopic values of the diluted sample. 2.2.3 PHREEQC mixing model A series of geochemical models were developed with PHREEQC, version 3.2.0 (Parkhurst and Appelo, 2013).  Each of the models used the Pitzer database due to the high ionic strength of the formation water. Four models increasing in complexity were run with the addition of different geochemical processes at each stage (Table 2.1).  The modelled results were compared to the measured flowback water results in order to determine which geochemical processes are impacting the flowback water chemistry.  An understanding of the Montney Formation mineralogy to support the geochemical model was obtained through a combination of publically available X-Ray Diffraction (XRD) results and additional samples collected at the BC Oil and Gas Commission core research facility in Fort St. John, BC and analyzed at UBC.  The additional samples were analyzed by XRD following the modified smear mount method (Munson et al., 2016) and the results were interpreted using the Rietveld method (Rietveld, 1967, 1969).      24  Table 2.1: Summary of the four PHREEQC mixing models used for the analysis.  The model results were compared to the measured concentrations and the calculated proportions of formation water to provide an understanding of the geochemical processes that are impacting the flowback water major ion concentrations.  Model I is a simple mixing model between the injected hydraulic fracturing fluid and the formation water.  The proportion of formation water is increased in 5% increments and as no other geochemical processes are considered in the initial model, the resulting ion concentrations show a linear increase.  Model II adds possible secondary mineral precipitation and allows the mixed fluid to reach equilibrium (i.e., if a secondary mineral is oversaturated in the fluid, it is allowed to precipitate until the saturation index (SI) is equal to 0).  This model represents the maximum amount of precipitation that would occur when the two fluids are mixed, as it does not consider the kinetics of mineral precipitation.  Model III adds ion exchange to the model based on the exchange sites being in equilibrium with the formation water.  Setting the exchange sites to be in equilibrium with the formation water determines which ions are on the exchange sites so they do not need to be entered into the model manually.  The amount of exchangeable ions was 25  estimated at 0.5 mole per kilogram of water.  This value is a reasonable estimate based on an average density (2.65 g/cm3) and the median cation exchange capacity (CEC) for 15 Montney Formation samples that were analyzed for CEC.   The 15 samples were selected for varying clay content.  A median value is used for all wells since there was limited correlation with mineralogy and a more precise value is not available.  Model IV adds carbonate dissolution to the model.  Based on the ions considered at the level of detail in the model and the mineralogy of the formation, only the carbonate minerals are both abundant in the Montney Formation and comprised solely of the major ions included in the model.  The inclusion of quartz, feldspar, and clay would require the addition of silicon (Si) and aluminum (Al) (Al for feldspar and clay only) to the model and in addition, due to the stability of these minerals, they would not be expected to dissolve significantly during hydraulic fracturing.  Up to 1% (by wt.) of calcite and 1.8% (by wt.) of dolomite are permitted to dissolve if the fluid is undersaturated in these minerals, which is only a fraction of the amount of calcite and dolomite that are available in the Montney Formation.  These percentages were selected to represent a small amount of dissolution and are equivalent to 1 mole per liter of water for each of the minerals.       2.3 Montney Formation flowback water chemistry The Montney Formation flowback waters collected in the study had elevated concentrations of the major ions with TDS up to approximately 228,000 mg/L (Table 2.2), similar to what has previously been reported for flowback waters from other formations where hydraulic fracturing is occurring (e.g., Marcellus Shale – Barbot et al., 2013; Haluszczak et al., 2013; Ziemkiewicz and He, 2015).  Together Cl and Na ions account for 75 to 95% (median: 92%) of the TDS in the sampled Montney Formation flowback water.  The other ions that are elevated in the flowback water are Ca, K, Mg, Sr, and, in some wells, SO4 (Table 2.2).  The 26  concentrations of the major ions, and therefore TDS, increase over the flowback period, since all of the study wells used injected fluids with major ion concentrations that were lower relative to those in formation water (Fig. 2.3, Fig. 2.4).  In contrast, the SO4 concentrations do not show a continuous increase and are often constant over the duration of the flowback period that the fluids were sampled.  The pH remains neutral for the majority of the flowback water samples (median: 7.2) and most samples have moderate total alkalinity (median: 184 mg CaCO3/L), although both the pH and the total alkalinity show a decreasing trend over the flowback period.  The organic compounds were not analyzed in the flowback water from the study wells.  The exclusion of these molecules may contribute to the higher charge balance error (generally ±15%) seen in some of the samples.  The necessity to dilute the samples for the analysis to reach the range of the standards may introduce additional error and contribute to higher charge balance errors.          Table 2.2 (following page): Summary of pH, conductivity, total alkalinity, TDS, and major ion concentrations in the flowback waters collected in the study. Data for wells at the same site and completed within the same member of the Montney Formation are grouped together for this summary.  For the TDS and the major ions, lower values for all wells are from earlier in the flowback period, while higher values are from later in the flowback period.       27  Montney Formation member upper middle lower Site A (8 wells) B (2 wells) D (2 wells) H (1 well) D (2 wells) E (4 wells) F (2 wells) G (1 well) I (2 wells) Parameter n 22 30 19 18 22 26 12 8 28 pH min 7.5 6.3 2.3 6.1 6.6 6.0 3.4 6.8 6.4 median 7.8 7.0 6.9 6.3 7.2 7.4 6.3 7.3 7.1 max 8.1 7.9 9.2 8.1 9.5 7.8 6.7 7.6 7.7 Conductivity (mS/cm) min 60.8 47.6 50.4 33.8 7.12 22.1 29.1 27.9 11.1 median 102 123 86.4 113 36.1 35.7 37.5 64.8 56.1 max 131 190 107 135 74.6 51.8 50.4 95.4 130 Total Alkalinity (mg CaCO3/L) min 115 71 <10 67 135 222 <10 205 172 median 195 119 104 80 236 337 51 228 232 max 405 188 685 128 475 395 91 260 403 TDS (mg/L) min 41,358 30,135 31,522 20,713 3,609 11,105 18,655 13,972 4,587 median 61,220 89,949 62,520 83,807 21,631 18,672 24,009 38,963 31,998 max 92,731 228,259 80,944 112,226 56,800 32,864 34,139 65,052 90,178 Cl (mg/L) min 25,722 18,541 19,303 12,308 1,893 6,671 10,295 8,186 2,260 median 43,437 55,235 38,355 49,006 12,510 12,251 14,455 24,790 17,678 max 65,777 164,018 50,963 66,998 34,565 21,611 21,495 41,783 51,038 Na (mg/L) min 8,264 9,162 9,960 6,451 1,344 2,787 6,849 3,895 1,991 median 12,759 25,609 18,489 28,509 7,884 5,256 9,066 10,999 12,112 max 17,818 51,027 23,974 37,637 18,872 9,119 10,575 18,244 33,782 Ca (mg/L) min 1,616 1,789 1,169 629 13 350 560 174 110 median 3,059 5,950 2,874 2,798 628 641 702 845 1,032 max 5,389 11,705 4,058 3,677 2,112 1,126 795 2,229 4,081 K (mg/L) min 553 284 466 414 49 187 166 186 105 median 796 1,035 860 1,534 257 295 196 410 558 max 1,239 1,920 1,171 1,806 466 474 251 650 1,296 Mg (mg/L) min 296 154 171 102 9 62 93 74 32 median 547 619 418 650 79 107 117 273 166 max 941 1,369 611 808 223 220 136 586 524 Sr (mg/L) min 298 60 123 30 4 38 33 29 13 median 630 404 264 141 55 72 44 73 206 max 1,219 924 358 195 268 161 69 101 1,011 SO4 (mg/L) min 51 <0.1 134 575 98 90 379 1308 0.7 median 91 59 189 899 126 113 560 1370 19 max 234 174 645 1084 189 202 824 1407 48  28  There is variability in the flowback water chemistry between the different wells (Table 2.2, Fig. 2.3, Fig. 2.4).  For example, the TDS in the initial flowback water samples from the study wells range from 3,610 mg/L (well D-4) to 123,000 mg/L (well C-7) while TDS of the final samples range from 23,800 mg/L (well E-2) to 228,000 mg/L (well B-1).  In general, the TDS and major ions are higher in the initial flowback water from the upper Montney Formation wells relative to the initial flowback water from both the lower and middle Montney Formation wells.  In the late flowback period water, the major ion concentrations reach similar levels in the samples from the upper and lower Montney Formation wells whereas the major ion concentrations in the middle Montney Formation flowback water remain lower (Table 2.2, Fig. 2.3, Fig. 2.4), although there is overlap in the results for flowback water from the different members.  The wells completed on the same site and within the same member of the Montney Formation most commonly have similar flowback water chemistry (e.g., well D-3 and well D-4; well E-1 and well E-2), although there are exceptions (e.g., well I-1 and well I-2). Comparisons between the flowback water chemistry produced from the different members of the formation or even between wells completed in the same member are complicated as the chemistry is impacted by variables related to well completion, including the hydraulic fracturing fluid chemistry, the number of stages, and shut-in time, in addition to the properties of the formation, the response of the well to hydraulic fracturing, and the length of the flowback period.  Longer shut-in periods in particular were found to correlate with higher TDS and major ion concentrations in the flowback water (see Chapter 4) which would contribute to the variability between wells, unrelated to the member of the formation where the well was completed.    The rate of increase in TDS and the major ion concentrations during the flowback period can vary between different wells.  For some wells the flowback water major ion concentrations 29  do not reach a stable plateau concentration by the end of the sampled flowback period as has been observed in the flowback water from other formations (e.g., Marcellus Shale – Blauch et al., 2009; Haluszczak et al., 2013; Horn River Basin – Zolfaghari et al., 2016).  The increase in ion concentrations over the full length of the flowback period for some wells has been attributed to the development of a more extensive fracture network due to the access to a greater volume of formation water (Bearinger, 2013; Ghanbari et al., 2013; Zolfaghari et al., 2015a).  However, the major ion concentrations must eventually reach a plateau as the formation water acts as an endmember in the system or potentially decline, if water of condensation comprises a significant component of the flowback.    30   Figure 2.3: Flowback water TDS for the study sites plotted as a function of the cumulative volume of flowback collected.  a) Site A wells – upper Montney Formation; b) Site B wells – upper Montney Formation; c) Site D wells – upper (well D-1 and well D-2) and middle (well D-3 and well D-4) Montney Formation; d) Site E and Site F wells – middle Montney Formation; e) Site G and Site H wells – middle (well G-1) and upper (well H-1) Montney Formation; f) Site I wells – lower Montney Formation. These plots do not include the produced water samples collected at site A and site B.  a. b. c. d. e. f. 31   Figure 2.4: Flowback water TDS plotted as a function of percent recovered.  The overall trend of the TDS vs. percent recovered is comparable to the trend in TDS vs. cumulative flowback volume (Fig. 2.3). a) Site A wells – upper Montney Formation; b) Site B wells – upper Montney Formation; c) Site D wells – upper (well D-1 and well D-2) and middle (well D-3 and well D-4) Montney Formation; d) Site E and Site F wells – middle Montney Formation; e) Site G and Site H wells – middle (well G-1) and upper (well H-1) Montney Formation; f) Site I wells – lower Montney Formation. These plots do not include the produced water samples collected at site A and site B.  a. b. c. d. e. f. 32  2.4 Flowback water mixing ratios Mixing between the injected hydraulic fracturing fluid and the formation water is the most probable mechanism for the increasing TDS over the flowback period.  The approximately linear relationship between the major ion concentrations provides support for mixing as a dominant influence since a linear relationship suggests mixing between higher concentration and lower concentration fluid sources (Fig. 2.5).  Although the Montney Formation generally has low water saturation (e.g., Wood, 2015), the high surface area accessed by the created fracture network and any natural fractures in the formation increases the area of contact and thus increases the accessible formation water.  Mixing between the formation water and the hydraulic fracturing fluid would occur in the fracture space.  Additional access to formation water may be from the process of countercurrent imbibition.  Countercurrent imbibition is an important consideration in hydraulically fractured wells, as it can increase oil production (e.g., Dehghanpour et al., 2013; Rokhforouz and Amiri, 2017).  When the hydraulic fracturing fluid is injected into the formation, a large portion of the fluid is imbibed into the formation (Engelder et al., 2014).    Imbibition experiments have shown that shale samples placed in deionized water result in an increase in conductivity in the water over the length of the experiment (e.g., Zolfaghari et al., 2016).  The increase in conductivity could be inferred to be a consequence of the movement of formation water, in addition to oil and/or gas, out of the rock through countercurrent imbibition.  Physical mixing would occur between the formation water and the remaining hydraulic fracturing fluid in the fractures.  Other potential sources of the high TDS are diffusion of the ions in the formation into the injected fluid and osmosis, where water would move into the formation and the ions in the remaining fluid in the fractures would become more concentrated (Balashov et al., 2015; Zolfaghari et al., 2016; Wang et al., 2016; 2017).  Based on 33  the available data, the influence of diffusion and/or osmosis cannot be separated from the influence of mixing with formation water accessed in the fracture spaces and through countercurrent imbibition.         Figure 2.5: Examples of the linear relationship between selected major ions in the Montney Formation flowback water. a) Na-Cl plot; b) Ca-Mg plot.  The relationship between the major ions supports mixing between two endmembers – the injected hydraulic fracturing fluid with relatively low TDS and the high TDS formation water.  The three samples with elevated Cl concentrations at site B are from well B-1 at the end of the flowback period, following a 24 day shut-in period. The flowback water results are grouped by site although the site D wells are divided into the upper (u) and middle (m) Montney Formation wells. ▲- upper Montney Formation wells;  - middle Montney Formation wells;  - lower Montney Formation wells.  In order to calculate the relative percentages of hydraulic fracturing fluid and formation water contributing to the flowback water over time for the study wells, the Cl concentrations, δ18O values, and δ2H values are used as conservative tracers.  The use of the Cl concentrations and the isotopic values as conservative tracers for mixing between the injected hydraulic a. b. 34  fracturing fluid and the formation water endmembers is supported by cross plots of these parameters plotting along a straight mixing line (Fig. 2.6).  The mixing between the hydraulic fracturing fluid and the formation water to produce flowback water can be represented by the following mass balance equation:    𝐶ி஻𝑓ி஻ =  𝐶ுி𝑓ுி +  𝐶ிௐ𝑓ிௐ    (1) Where CFB is the value in the flowback water, fFB is the proportion of flowback water, CHF is the value in the hydraulic fracturing fluid, fHF is the proportion of the hydraulic fracturing fluid, CFW is the value in the formation water, and fFW is the contribution of formation water as a percentage. For the equation, the term 'value' represents either the Cl concentration in mg/L, or the δ18O and δ2H values in per mille (‰).  As the conservative tracers can be explained through mixing between two fluid sources, without any influence from mineral precipitation and/or dissolution, it can be assumed that fFB = fFW + fHF = 1.  Equation 1 can then be transformed to calculate the percentage of formation water based on one of the tracers: 𝑓ிௐ =  ஼ಷಳି஼ಹಷ஼ಷೈି஼ಹಷ × 100% (2) The use of Cl concentrations in combination with the isotopic values to calculate the mixing ratios allows for some verification of the results and can indicate where variability in the data is real or due to potential analytical errors.  Mixing percentages calculated using conservative tracers can be used to determine if other potentially non-conserved ions are being influenced by geochemical processes other than mixing by comparing the analytical results to the modelled results.   35   Figure 2.6: Cross plots between the conservative tracers used to calculate the mixing ratios for this study.  The linear trend between the injected hydraulic fracturing fluid and the formation water indicates mixing between the two endmembers.  Data from Site I (well I-1 and well I-2) are provided as an example. a) Cl concentrations versus δ18O values; b) Cl concentrations versus δ2H values; and c) δ2H values versus δ18O values.  The input parameters for calculating mixing include the Cl concentrations and isotopic values for the hydraulic fracturing fluid and the formation water (Table 2.3).  These parameters are measured in the flowback water (Table 2.4) and can be compared to the mixing model.  Based on the TDS of the injected hydraulic fracturing fluid, both freshwater-based fluids (TDS < 2,000 mg/L) and blended fluids composed of a mixture of freshwater and recycled flowback water (TDS > 3,000 mg/L) were used for the study wells.  Freshwater-based fluids were used for well A-3, the site D wells (1 to 3), well H-1, and the site I wells.  Blended fluids were used for the remaining site A wells, well D-4, and all wells on sites B, E, F, and G.       a. b. c. 36  Table 2.3: Summary of the δ2H values, δ18O values, and Cl concentrations of the hydraulic fracturing fluids and formation water used for the mixing model. Site Well Hydraulic fracturing fluid Formation water δ2H (‰) δ18O (‰) Cl (mg/L) δ2H (‰) δ18O (‰) Cl (mg/L) A 1 -131.9 -15.4 10,276 -49.9 1.9 90,395 2 -132.7 -15.8 10,423 -38.2 2.5 3 -141.7 -17.8 71 -30.0 3.9 4 -131.1 -15.0 18,108 -25.5 2.5 5 -131.6 -14.3 19,850 -34.8 0.2 6 -131.9 -15.4 11,365 -22.1 5.2 7 -133.1 -16.4 966 -33.3 3.8 8 -129.2 -14.5 17,495 -28.2 3.3 B 1 -117.5 -11.6 6,276 -16.4 8.9 128,802 2 -122.1 -7.0 6,432 -46.6 2.2 D 1 -151.0 -18.5 160 -33.3 3.8 60,215 2 -145.7 -18.2 639 3 -146.9 -18.9 289 4 -139.3 -15.9 2,044 E 1 -114.3 -7.9 6,414 2 -120.5 -7.8 3,150 3 -116.2 -6.1 3,842 4 -111.7 -6.5 4,740 F 1, 2 -139.0 -16.3 7,448 G 1 -124.3 -13.3 1,841 -33.3 3.8 47,585 H 1 -148.0 -19.2 7 -33.3 3.8 77,735 I 1 -154.1 -20.3 1,204 -41.3 7.6 89,168 2 -156.4 -20.5 760  Notes: Isotopic values for the hydraulic fracturing fluids are averages of several samples for all except site F and site I wells, which are based on one sample for each of these sites. Cl concentrations for the hydraulic fracturing fluids are calculated based on the volumetric proportions of the composite samples. Isotopic values for the formation water for well A-7, site D-G wells, and well H-1 are average values based on the produced water samples collected at nearby sites in this study. Cl concentrations for the formation water for site A and site B wells are averages of the produced water samples collected at each of these sites. Cl concentrations for other sites are based on publically available data for nearby wells.  37  Table 2.4: Range of the δ2H values, δ18O values and Cl concentrations over the flowback period for flowback waters collected from the study wells. Site Well δ2H (‰) δ18O (‰) Cl (mg/L) Initial sample Final sample Initial sample Final sample Initial sample Final sample A 1 -66.9 -62.0 0.4 3.0 53,604 57,000 2 -103.4 -91.3 -12.5 -7.8 30,678 55,970 3 -109.0 -100.9 -12.1 -8.8 38,368 43,747 4 -96.1 -101.5 -9.1 -8.8 37,585 45,281 5 -86.6 -80.2 -3.1 -4.3 48,828 55,995 6 -70.5 -45.4 -3.6 5.9 58,196 65,777 7 -115.4 -109.7 -11.5 -12.1 25,722 34,270 8 -68.3 -85.6 -2.2 -5.1 48,745 43,128 B 1 -116.0 -77.0 a 28,838 164,018 2 -119.9 -79.7 18,541 83,998 D 1 -123.2 -95.9 -11.7 0.9 27,693 50,963 2 -123.0 -119.0 -14.1 -10.2 28,655 38,355 3 -127.7 -111.4 -8.1 -10.7 3,255 20,450 4 -142.8 -108.2 -14.4 -11.3 1,893 30,803 E 1 -112.8 -94.9 -4.3 -4.7 8,201 21,611 2 -106.3 -93.3 -4.2 0.2 7,341 16,605 3 -103.5 -83.9 -3.4 3.6 12,670 18,576 4 -110.8 -108.1 -5.4 -10.7 6,688 16,914 Fb 1 -105.0 -104.4 -11.7 -12.4b 14,915 16,003 2 -102.0 -102.1 -9.7 -12.2b 10,295 21,495 G 1 -122.9 -71.8 -10.9 -2.4 8,186 41,783 H 1 -138.4 -112.8 -17.3 -7.5 12,308 66,998 I 1 -145.7 -119.1 -18.2 -11.3 2,260 30,245 2 -140.9 -88.4 -16.1 -9.4 4,850 51,038   a – Values are excluded as they were determined to be contaminated.   b – The isotopic values may be contaminated.  Overall, the Cl concentrations and the stable isotopes used as conservative tracers in the study show that the percentage of formation water increases approximately linearly over the flowback period at each of the wells, although the rate of increase varies between wells.  The 38  range in the contribution of formation water at the beginning of the flowback period for the study wells is generally approximately 10 to 35% and increases to approximately 40 to 60% at the end of the flowback period (Table 2.5).  Each of these ranges represent the 25th to 75th percentiles of the proportion of formation water calculated for the initial and final flowback water samples, respectively, for the study wells.  Well A-1, well A-6, well B-1, and well D-1 show over a 100% contribution from formation water, depending on which conservative tracer is used, indicating that the late flowback water is in some cases similar to produced water and that the values for the formation water endmember were underestimated or that the linear mixing model does not account for all the variance due to other geochemical processes such as mineral dissolution.                39  Table 2.5: Summary of proportions of formation water contributing to the flowback water for the conservative tracers used in the study.  The minimum value is early in the flowback period, while the maximum value is late in the flowback period.   Site Well Percentage of Formation water (%) Based on δ2H Based on δ18O Based on Cl Min Max Min Max Min Max A 1 79 85 92 107 54 58 2 31 44 11 44 25 57 3 29 37 26 42 42 48 4 28 33 34 44 27 38 5 46 53 69 77 41 51 6 56 79 57 104 59 69 7 29 39 53 61 28 37 8 43 60 52 69 31 43 B 1 1 47 a 18 129 2 3 67 10 63 D 1 24 69 29 121 32 85 2 20 29 19 54 45 81 3 16 31 17 36 5 34 4 0 29 0 38 0 56 E 1 2 24 6 27 3 28 2 16 31 31 69 7 24 3 6 39 28 46 16 26 4 1 8 8 11 3 22 Fb 1 32 43 12 50 13 19 2 31 49 17 72 5 27 G 1 2 58 14 64 14 87 H 1 8 33 8 51 16 86 I 1 8 31 7 32 1 33 2 14 59 16 43 5 57 Notes:  a – The δ18O values were excluded as the values were interpreted to be contaminated by organics that were not removed by the activated charcoal treatment.  b – The isotopic values are potentially contaminated, as the mixing ratios calculated using the isotopes do not agree with those calculated using the Cl concentrations.   40  2.5 Variability in mixing ratios The variability in the mixing ratios between the different conservative tracers, over the flowback period at individual wells, and between sites provides insight to the reservoir mechanics.  In the following section each of these sources of variability are discussed and the implications for the mixing analysis are presented. 2.5.1 Mixing ratio variability between tracers used The mixing ratios show some variability depending on whether Cl concentration or δ18O and δ2H isotopic values are used to calculate the percentage of formation water (see Table 2.5 above).  There is no indication that one of the tracers is consistently over- or under- estimating mixing as the calculated mixing percentage is not regularly higher or lower when using a particular isotope or the Cl concentrations.  However, when examining the data for a single well, one of the parameters used to calculate the mixing ratios often estimates the highest or lowest ratio for all of the samples from the well.  Part of the variability in determining the mixing ratios may be due to the uncertainty in estimating the values to use for the formation water, as the exact values are unknown, and potentially due to variations in formation water chemistry within the fractured zone which is not taken into account as a constant value is used to represent the formation water chemistry at each well.  In some cases, the isotopic values may be representative of the true formation water chemistry values whereas the Cl concentrations may be inaccurate resulting in variability in the ratios calculated with the different tracers.  Geochemical processes such as carbonate dissolution or leaching of clays could potentially have some influence on the Cl concentrations but not the isotopic values.  Produced water samples were not available for the majority of the study wells and even when available are not expected to represent 100% formation water as complete mixing would likely not occur.  In general, using the produced 41  water chemistry for the formation water chemistry should underestimate the TDS in formation water as the produced water is likely somewhat diluted by the injected fluids from pure formation water.  However, the produced water Cl concentrations and isotopic values used in this study are within the range expected for formation water (e.g., Wei et al., 2010 [Cl]; Rostron and Holmden, 2000 [isotopes]) and are considered reasonable estimates given the difficulty of directly sampling formation water from the Montney Formation as it is near irreducible water saturation (e.g., Wood, 2015).  Calculating the mixing ratios using Cl concentrations, δ18O values, and δ2H values provides a method to evaluate the calculated ratios three times for the same data point.  The range in the proportion of formation water using the three tracers generally falls within a standard deviation of 5% and 15%, based on the 25th and 75th percentiles for all standard deviations in the studied samples.  In cases where the three mixing ratios show some disagreement, the results can be used to provide a range of mixing ratios for the sample.  2.5.2 Mixing ratio variability over the flowback period for individual wells Although the percentage of formation water contributing to flowback water increases approximately linearly over the time sampled, there are occurrences where the fraction of formation water shows a brief positive or negative spike (Fig. 2.7).  Positive spikes indicate an increase in the proportion of formation water as reflected by an increase in TDS and major ion concentrations.  The increases and decreases in the contributions from formation water are consistent for all 3 tracers (Fig. 2.7) indicating that they are due to mixing inasmuch as Cl concentrations and the isotopes would not both be affected by the same analytical errors or the same geochemical reactions. Halite dissolution is one example of a potential reaction which would only cause an increase in Cl concentrations and not affect the isotopes.   42   Figure 2.7: Fraction of formation water calculated for well D-4 using the three conservative tracers – Cl concentrations, δ18O, and δ2H. The ratios show good agreement early in the flowback period, with lower ratios calculated using the two isotopes later in the flowback period.  The discrepancy later in the flowback period may be due to the error around estimating the values of the formation water or a source of Cl from another geochemical process.  The brief increase in the contribution from formation water near 800m3 of cumulative flowback is recorded in all three tracers; however, the magnitude of the change in mixing ratio is not consistent for all three tracers.  This well shows an increase from 9.9 to 16.3% formation water in the sample collected at approximately 630 m3 of cumulative flowback water to 22.9 to 38.4% formation water in the sample collected near 800 m3 of cumulative flowback water.  This increase in the percentage of formation water is not related to a shut-in period during the flowback period.      A variable pattern in tracer concentration is also commonly seen in hydraulic fracturing studies where chemical tracers are used (e.g., Sullivan et al., 2004).  This variability may be due to opening or closing of the induced fracture systems at different times during flowback as the pore pressure declines.  For example, an increase in the overall proportion of formation water would represent the contribution of flowback water from a stage where fracturing was more extensive and there is a higher surface area exposed resulting in a greater input from formation water relative to other stages.  Variability in formation water chemistry along the lateral could 43  also contribute to the brief changes in the flowback water chemistry as different areas begin to flow back.  Another potential explanation for the spike in concentrations is a brief shut-in period during the flowback period, which would be expected to cause an increase in TDS and the percentage of formation water due to more time for mixing, as shown by an increase in the contribution from formation water.  An example of an increase in TDS following a shut-in period occurs near the end of the flowback period at well B-1 when there was a shut-in period of 24 days near 3,000 m3 cumulative flowback volume, which corresponds to the spike late in flowback period of this well (see Fig. 2.3b above).  There were short periods of flow of up to 2 days for the first 15 days of the 24 day shut-in period, followed by no flow over the final 9 days. Sites where the increase in formation water is more constant may have more constant mixing for the different stages or the sampling frequency may miss the pulses of water from certain stages.        2.5.3 Mixing ratio variability between sites In addition to variability in the mixing ratios over the flowback period of a single well, the percentage of formation water contributing to flowback water throughout the flowback period as well as the rate of increase in the proportion of formation water fluctuate between wells.  Since the proportion of formation water is expected to increase over the flowback period, wells with longer flowback periods would be anticipated to have a higher proportion of formation water at the end of the flowback period as more time would allow for more mixing to occur in the newly created fractures.  However, the maximum percentage of formation water for a well and the length of the flowback period only show low to moderate correlations, depending on if the Cl concentrations (R2 < 0.1) or the isotopes (R2 ≈ 0.3) are used to calculate the proportion of flowback water.  The low to moderate correlations indicate that the amount of formation water contributing to flowback water does not increase at the same rate for all wells.  44  Factors that could be contributing to the inconsistency include the shut-in time prior to flowback, and the complexity of the fracture network.  There is a moderate positive correlation (R2: 0.68) between the length of the shut-in period and the TDS in the initial flowback water (Fig. 2.8a).  A longer shut-in time prior to the commencement of the flowback period allows for more time for mixing within the reservoir and produces an initial liquid with a higher proportion of formation water and therefore higher major ion concentrations (Fig. 2.8b).  In addition to mixing during the shut-in period, there would be more time for imbibition, diffusion, and/or osmosis which would all increase the TDS of the flowback water.  Another important variable is the complexity of the fracture network.  A more complex fracturing network would also allow for more mixing and thus a higher proportion of formation water in the flowback due to the greater surface contact area (Bearinger, 2013; Ghanbari et al., 2013; Zolfaghari et al., 2015a).  The extent of the fracture network, or the stimulated reservoir volume, was not quantified in the present study since no microseismic was collected during hydraulic fracturing.       45   Figure 2.8: a) Correlations between the TDS of the initial flowback (FB) water samples from each well and the shut-in time following hydraulic fracturing and before the beginning of the flowback period. There is a moderate correlation which indicates that a longer shut-in time is associated with higher TDS in the early flowback samples.  This may be related to a higher percentage of formation water; b) Correlation between the minimum percent of formation water at a well and the shut-in time using the proportion calculated with all three of the tracers.  The moderate correlation indicates that wells with a longer shut-in time have a higher proportion of formation water in the early flowback water samples.  2.5.3.1 Potential complications with the different tracers used The majority of the mixing proportions show some variability in the percentages calculated using the different tracers; however, greater inconsistencies are seen in the samples from wells from sites B, E, and F.  The reason for the greater differences observed for the wells from these sites are unknown but one of the possibilities is that the presence of organic contaminants in some samples were not removed effectively by the activated charcoal treatment, interfering with the isotopic values and producing anomalous mixing ratios.  Sites B and E have unexpectedly high mixing ratios when the δ18O values were used in the calculations.  At site F, a. b. 46  the mixing ratios calculated using the isotopic values were relatively invariant or decreasing.  The flowback water samples from the site F wells were treated with a sodium hypochlorite (NaClO) solution to remove hydrogen sulfide (H2S) prior to sampling.  This solution contains water; however, the volume of the NaClO solution used in the treatment only represents approximately 1% of the total fluid flowback volume and would thus not be expected to impact the isotopic values significantly.  The isotopic values are therefore likely contaminated. The Cl ions produced as part of the reaction between NaClO and H2S did not have a detectable influence on the overall Cl concentrations as increasing concentrations were still measured over the flowback period from the site F wells.  The percentages of formation water calculated using the Cl concentrations for the potentially contaminated samples provide a check for the isotope data and will be used for the geochemical modelling.     An additional potential source of variability in the mixing ratios is the addition of a third fluid endmember.  Only one of the study wells shows some potential indication of another water source.  The site H well initially shows good agreement between the ratios calculated using the different tracers until approximately 400 m3 of flowback water had been recovered from the well. Later samples show a higher proportion of formation water when Cl concentrations are used (55-86% formation water) relative to the ratios calculated with the isotopic values (15-51% formation water).  The inconsistency may be due to a third source of water with higher Cl concentrations and lower isotopic values.  The other source of water could be due to fracturing out of zone or accessing high permeability lenses with a different formation water chemistry.  The isotopes show some indication of an additional water source as the trend of the flowback samples does not follow a mixing line from the injected hydraulic fracturing fluid to the formation water (Fig. 2.9).  47   Figure 2.9: δ2H vs δ18O plot for the site H well.  The mixing line between the hydraulic fracturing fluid and the estimated formation water values is shown as a dashed line.  The later flowback samples begin to plot away from this line, indicating a potential third fluid source.  2.6 Insight from major ion chemistry Further to mixing between the injected hydraulic fracturing fluid and the formation water, potential additional geochemical processes, including mineral precipitation and dissolution (Blauch et al., 2009; Wilke et al., 2015; Dieterich et al., 2016; Harrison et al., 2017; Marcon et al., 2017) and ion exchange (Renock et al., 2016; Zolfaghari et al., 2016), may influence the concentrations of some of the major ions over the flowback period.  In order to determine if the concentrations of the major ions can be explained based solely on mixing, the major ion concentrations of the analyzed flowback water samples are plotted with the calculated mixing ratios and compared to the major ion concentrations predicted by the geochemical models.  Model I looks only at mixing between the injected fluid and the formation water, using the chemistry results summarized in Table 2.6.  In wells where only mixing between these two fluid sources does not explain the measured concentrations of the major ions in flowback water, more 48  complex models including secondary mineral precipitation (Model II), ion exchange (Model III) and/or carbonate dissolution (Model IV) are used to determine other factors influencing the flowback water chemistry.    Table 2.6 (following page): Summary of the hydraulic fracturing fluid chemistry and the formation water chemistry.  The hydraulic fracturing fluid chemistry includes a summary of the calculated chemistry from the composite samples for each well.  The formation water chemistry is based on the median values for the produced water samples collected in the present study (sites A, B, and I) and the produced waters from nearby wells with publically available results (sites D-H). The temperature used in the model for the hydraulic fracturing fluids was 25°C and the temperature used for the formation water was 75°C.  Units are mg/L.              49  Site Well pH Total Alkalinity TDS Cl Na Ca Mg K Sr SO4 Hydraulic Fracturing Fluid               A 1 6.8 108 15,285 10,276 3,212 1,004 162 181 161 8.5 2 6.5 96 16,802 10,423 3,873 1,428 227 226 245 25 3 7.3 76 239 71 69 42 22 21 <10 18 4 6.7 86 25,297 18,108 4,555 1,566 256 255 309 15 5 6.7 94 28,072 19,850 5,365 1,791 274 273 332 18 6 6.7 89 16,824 11,365 3,486 1,052 205 224 217 15 7 7.2 81 3,386 966 1,733 420 59 63 59 20 8 6.7 110 25,382 17,495 5,046 1,724 261 280 295 4 B 1 7.5 157 22,018 6,276 11,973 2,828 299 379 160 45 2 7.3 134 23,459 6,432 12,980 3,067 284 417 175 41 D 1 7.3 64 549 160 208 46 10 11 1 78 2 7.1 87 1,284 639 398 58 12 46 2 98 3 7.2 66 711 289 209 50 10 10 1 104 4 7.1 258 4,717 2,044 1,605 142 29 241 6 273 E 1 6.5 171 11,246 6,414 3,659 608 96 187 59 129 2 6.6 107 5,052 3,150 1,237 322 53 96 28 95 3 6.4 152 6,059 3,842 1,546 323 53 89 16 111 4 6.8 161 8,464 4,740 2,788 453 79 150 47 112 F 1, 2 7.0 83 12,218 7,448 3,948 372 60 105 25 165 G 1 7.2 150 5,604 1,841 3,047 305 64 137 40 74 H 1 4.9 3 786 5.1 573 42 16 11 <1 28 I 1 7.7 242 1,849 1,204 66 74 16 8 <1 <0.1 2 7.7 258 1,863 760 67 77 16 8 <1 <0.1 Formation Water          A  7.1 201 125,443 90,395 25,191 7,638 1,073 1,365 1,297 82 B  5.1 84 181,321 128,802 32,591 14,835 1,469 1,982 1,037 155 D-F  6.9 238 97,354 60,215 29,050 5,375 778 1,575 869 73 G  6.8 166 67,172 47,585a 18,850 3,897 600 578 119 1,320 H  6.2 116 129,744 77,735 34,800 5,635 1,100 2,010 115 489 I   5.9 63 91,254 89,168 54,000 10,000 1,140 2,040 1,620 <0.1 Note:  a – The formation water Cl concentration for site G uses the 75th percentile instead of the median since the median Cl concentrations from the produced waters from nearby wells was lower than the measured Cl concentrations in the last flowback water sample from this well.  50  2.6.1 Sodium and potassium Model I provides a moderate to good fit for the measured Na and K concentrations in flowback water, indicating that mixing is the dominant process influencing these ions (Fig. 2.10).  There are some sites where the concentration of Na and/or K still follow a linear increase but are slightly lower or higher than those predicted by the mixing model indicating that the concentration used for formation water is over- or under-estimated in the model.  Since a linear increase is apparent, the changing concentrations are in most cases not interpreted to be significantly influenced by geochemical processes other than mixing.   The site B wells in particular have high Na concentrations in late flowback water with the last three samples having higher major ion concentrations relative to the formation water endmember.  The percentage of formation water calculated based on the δ2H values and the Cl concentrations are not in agreement for the well B-1 samples, indicating other geochemical processes (e.g., mineral dissolution) impact the flowback water chemistry.  The elevated Na concentrations in the later samples at well B-2 are higher than predicted by the mixing model; however, the percentage of formation water is generally in agreement between the δ2H values and the Cl concentrations, supporting mixing between endmembers.  Such results indicate that there was an increase in the contribution from formation water rather than mineral dissolution, as both the Cl concentrations and the δ2H values were similarly impacted. Well A-1, well D-2, well E-2, well E-3, and well H-1 have a good fit between the modelled Na and K concentrations and the mixing ratios based on Cl concentrations but not the isotopes.  The difference in model fit for different ions may be related to the uncertainty in the formation water chemistry and/or potential changes in the formation water over the flowback period as different regions of the formation are accessed.   51   Figure 2.10: Modelled and measured results for (a) Na; and (b) K concentrations for well I-1, provided as an example.  There is little difference in the modelled results with only mixing (Model I) compared to mixing and ion exchange (Model III) for these parameters.  The initial mixing ratios calculated using Cl concentrations are lower than the mixing ratios calculated using the isotopes; however the mixing ratios near the end of the flowback period show good agreement. The mixing ratios calculated using Cl concentrations show a better fit for Na concentrations, while the mixing ratios calculated using the isotopes show a better fit for K concentrations.  The addition of secondary mineral precipitation in Model II does not influence the Na or K concentrations as no Na or K bearing mineral phases are included in the model nor anticipated based on chemistry.  In Modell III, the addition of ion exchange changes the modelled concentrations slightly for these two ions (Fig. 2.10).  The CEC of the Montney Formation was estimated from a series of 15 samples with varying clay content (Table 2.7).  The results indicate that the CEC is low.  There is some indication that lower CEC is associated with lower clay content; however, the overall correlation is weak (R2 = 0.01) due to the limited range in both clay and CEC in the Montney Formation.  This makes it difficult to estimate the CEC without a. b. 52  conducting a CEC analysis for samples in each of the study areas.  A median value was selected for the model, which may overestimate the CEC for some wells and underestimate it in other wells.  The addition of ion exchange to the model increases the Na concentrations slightly relative to mixing alone.  The increase in Na concentrations in the model when ion exchange is added is likely related to the addition of the relatively low TDS hydraulic fracturing fluid into the formation which causes Na ions to desorb from the cation exchange sites on clays present in the formation.  This process would be similar to ion exchange which occurs when injecting freshwater into a saline aquifer, which shows characteristic breakthrough curves as the Na ions are initially mobilized in exchange for the less mobile divalent ions (e.g., Appelo, 1994).   The effect of ion exchange on K concentrations in the flowback water varies with the site. Ion exchange was observed to produce an increase and a decrease in K concentrations relative to mixing alone although the change in K concentrations is often minimal and not discernable when graphically comparing models with mixing only and with mixing coupled with ion exchange.             53  Table 2.7: Summary of CEC and mineralogy of selected Montney Formation samples.  These samples were selected to represent a range in clay contents.  The results are presented with increasing CEC. Sample CEC (cmol (+)/kg) % Quartz (by wt.) % Clay (by wt.) % Feldspar (by wt.) % Carbonate (by wt.) % Pyrite (by wt.) 1 9.3 40.9 7.4 20.5 25.1 6.1 2 13.3 40.4 6.2 21 29.3 3 3 13.8 28.8 17.5 15.8 34.9 3 4 14.1 35.1 11.7 23.3 27.6 2.3 5 14.5 35.7 17.7 20.1 24.4 2.2 6 14.7 36.7 17.1 23.6 21.3 1.3 7 15.8 25.4 20.3 29.3 23.8 1.2 8 17.2 40.9 18.5 19.7 18.6 2.3 9 17.8 35.8 4.2 12.1 46.2 1.7 10 18.2 37.5 8.5 19.6 33 1.6 11 18.4 30.7 10.7 18.4 36.5 3.7 12 18.7 31.4 6.4 15.8 45.0 1.4 13 20.3 19.7 8.5 15.4 53.8 2.5 14 20.3 34.4 13.5 14.3 35.9 1.9 15 20.4 26.6 10.9 25 35.5 2  2.6.2 Calcium, magnesium, and strontium The measured Ca, Mg, and Sr concentrations in flowback water are predominantly lower than predicted by mixing alone indicating that these divalent ions are removed from solution (Fig. 2.11).  Potential processes that could remove Ca, Mg, and Sr ions from solution are secondary mineral precipitation and adsorption through ion exchange.  When only mixing is considered in Model I, calcite, dolomite, and aragonite are generally oversaturated (SI > 0) in the fluid produced by mixing the hydraulic fracturing fluid and the formation water.  The oversaturation of these minerals is in agreement with what is observed in the sampled flowback water.  When the percentage of formation water is low (< 15%) in the model, calcite, dolomite, and aragonite are undersaturated (SI < 0) in the mixed fluids for many of the wells.  These minerals are undersaturated due to the chemistry of the injected fluid.  Initially, calcite has a negative SI for modelled results from wells A-2, A-3, A-6, D-1, D-2, D-3, the site E wells, the 54  site F wells, and well H-1, while dolomite and aragonite are initially undersaturated in the model for all wells excluding the site I wells.   Permitting secondary mineral precipitation in Model II indicates that calcite can precipitate during the flowback period, and in doing so, will decrease the SI of dolomite and aragonite causing the mixed fluids to be undersaturated in these two minerals for all wells.  The addition of calcite precipitation does not cause a significant decrease in the modelled Ca concentrations and the concentrations generally continue to remain above the Ca concentrations observed in the flowback water samples.  The relative Ca concentrations decrease by up to 12% when comparing the results of Model I (mixing only) and Model II (calcite precipitation), with the vast majority of Ca concentrations decreasing by less than 5% (median: 0.8% relative decrease) when calcite precipitation is added.  The minor change in Ca concentrations due to calcite precipitation indicates that secondary precipitation does not have a large effect on Ca concentrations in flowback water even when calcite precipitates to equilibrium. The flowback water samples are generally oversaturated with respect to calcite (SI > 0) which indicates that the mixed fluids have not reached equilibrium with calcite and that there would be less calcite precipitation occurring in the reservoir than predicted in the Model II which was set to reach equilibrium and does not take the reaction kinetics into consideration.   Calcium carbonate precipitation can be inferred in flowback water by examining the Ca/Mg molar ratio over the flowback period.  A decreasing Ca/Mg molar ratio indicates that calcite is precipitating (Barbot et al., 2013).  In our study, a decrease in the Ca/Mg ratio is only observed in the flowback water from two of the wells – well B-1, which shows only a slight decrease, and well H-1.  The calcite saturation indices for well B-1 indicate that calcite is oversaturated in the flowback water although the saturation indices are relatively stable over the 55  flowback period, while well H-1 flowback water shows an increase in calcite SI over the flowback period from near equilibrium to slightly oversaturated under reservoir temperature.  The increasing calcite saturation indices may indicate that equilibrium through calcite precipitation was reached in the initial samples but not later in the flowback period.  The Ca/Mg ratios for the remaining study wells are generally constant or show an increase over the flowback period, indicating that significant calcite precipitation is not occurring.   In addition to calcite, celestite is oversaturated (SI > 0) for the mixed fluids for well G-1 when the proportion of formation water contributing to the flowback water reaches 75% or greater, resulting in a decrease in Sr concentrations when the water is in equilibrium.  The decrease in Sr concentrations is not observed in the late flowback water samples from well G-1 (approximately 50-85% formation water) indicating that celestite is not precipitating in the well due to the kinetics of the reaction.  Celestite is slightly oversaturated (0.1 < SI < 0.3) at reservoir temperature (75°C) in the flowback water samples obtained from this well; however, the SI decreases at lower temperatures and is close to equilibrium (-0.1 < SI < 0.1) at 15°C.     The addition of ion exchange in Model III predicts a decrease in Ca, Mg, and Sr concentrations in the flowback for the majority of sites (Fig. 2.11).  The relative decrease in the concentrations of the divalent ions is greatest when the proportion of formation water is low.  Model III shows the process of the hydraulic fracturing fluid coming into contact with the ion exchange sites in the formation.  As discussed in Section 2.6.1, the divalent cations would adsorb on ion exchange sites in place of the more mobile Na ions, causing a decrease in the aqueous concentrations of the divalent ions.  The addition of ion exchange improves the fit of the geochemical mixing model to the experimental results for the divalent major ions in most wells (Fig. 2.11).  The site B wells are the only wells which do not show a significant decrease in the 56  modelled Ca, Mg, and Sr concentrations when ion exchange is added to the geochemical model.  The hydraulic fracturing fluid used for the two wells at site B includes a high proportion of recycled flowback water with elevated TDS (> 15,000 mg/L) as well as higher δ18O and δ2H values compared to those measured in the hydraulic fracturing fluids used at other sites (see Table 2.5).  The higher TDS and isotopic values are attributed to a greater proportion of recycled water.  As the injected fluid shows a greater degree of similarity to the formation water, less ion exchange would occur when it is injected into the formation.   Figure 2.11: Modelled and measured results for (a) Ca; (b) Mg; and (c) Sr concentrations for well I-1.   The modelled results are represented by the lines, while the measured results with the percentages of formation water calculated with Cl, δ18O, and δ2H are represented as data points.  The results from Model I and Model II show very little difference for Ca concentrations.  When mineral precipitation is included, calcite precipitates until SI = 0 in order to represent the maximum amount of precipitation that could occur. There is no difference between Model I and Model II for Mg and Sr concentrations as no minerals containing Mg or Sr were predicted to precipitate for this well.  The addition of ion exchange in Model III decreases the Ca, Mg, and Sr concentrations.  As was observed for Na and K, the proportions of formation water are initially lower when calculated with Cl concentrations compared to δ18O and δ2H isotopes.  The Sr concentrations show the best fit to the modelled concentrations.  The calculated proportion of formation water for the Ca and Mg concentrations are higher than those predicted by the geochemical model, possibly relating to uncertainties and assumptions made in the model.  a. b. c. 57  The Ca, Mg, and Sr concentrations predicted by Model III remain higher than the measured concentrations in flowback water from some wells.  Some sites also show differences in the fit of the model depending on if Cl concentrations or one of the stable isotopes are used to calculate mixing.  For example, for well I-1 the mixing ratios based on the Cl concentrations show a better fit to the model than the mixing ratios calculated with the isotopic values for the early flowback period (Fig. 2.11a).  Over- or under-estimating the formation water chemistry or isotopic values in the model would change the slope of the mixing line, causing an offset in the modeled and the measured values.  The uncertainties in the values for formation water are derived from not being able to directly obtain formation water samples from the study wells as well as from the possibility of changes in the formation water chemistry as different zones of the fractured area are accessed during the flowback period.  One example where the formation water value appears to be too low is for the Sr concentrations used for well H-1.  The measured values for Sr concentrations at this well are higher than the modelled values for the percentage of formation water calculated using both the Cl concentrations and the stable isotopes.  Only one produced water sample from a well nearby site H included Sr concentrations in the analysis.  The concentrations of the other major ions in this produced water sample were below the 25th percentile for produced water results from the southeast region, indicating that the Sr concentration in this sample would underestimate the concentration in formation water.  Further studies are necessary to characterize the variability in formation water chemistry across the Montney Formation.   In addition to potential variability in the formation water chemistry, additional explanations for a model that does not accurately calculate the divalent ion concentrations include uncertainty in the amount of ion exchange, the exclusion of organics in the model, or a 58  combination of these factors.  The ion exchange value of 0.5 moles of exchangeable ions per kg of water included in the model may be underestimated for some sites and overestimated for others.  A constant value was used to approximate the median CEC of 17.3 cmol (+)/kg.  As there was limited variability in the CEC for the 15 Montney Formation samples, the values for exchange in the model would only vary from 0.25 to 0.55 moles of exchangeable ions per kg of water, hence the use of the median value is not expected to influence the fit of the model significantly (Fig. 2.12).  A change in the ion exchange value in the model would impact all three of these ions and in some circumstances, one of the ions may show a good fit with the current model parameters (e.g., Sr concentrations in Fig. 2.11 above) whereby changing the ion exchange value would weaken the fit of the model.    Figure 2.12: The modelled Mg concentrations for different ion exchange values in the geochemical model using well I-1 as an example.  Using 0.25 moles of exchangeable ions per kg of water presents the lower limit based on the CEC results, while the upper limit (0.55) is close to the median value (0.5).  A value of 1 mole of exchangeable ions per kg of water is presented as a comparison above what is expected based on the CEC measurements.  There is not a great amount of variability between the different values for ion exchange.   59  Organic molecules can influence the mineral saturation indices in geochemical modelling (Marcon et al., 2017) and have been shown to be involved in ion exchange in soils (e.g., Droge and Goss, 2012).  The organic compounds present in the hydraulic fracturing fluid and the formation water were not considered in the present study but may influence the flowback water geochemistry.  More complex geochemical models with a greater understanding of the formation water chemistry, the CEC, and the effect of organics on the CEC may further improve the model fit.   The addition of carbonate dissolution to the geochemical model was investigated as a potential process that may improve the fit of the model for the Ca and Mg concentrations.  Calcite has the potential to dissolve in some wells when the proportion of formation water is low, due to the chemistry of the injected fluid.  Negative calcite SI values were modelled for wells A-2, A-3, A-6, D-1, D-2, D-3, the site E wells, the site F wells, and well H-1.  As dolomite was found to be undersaturated in the fluid following mixing and calcite precipitation, the chemistry indicates that dolomite has the potential to dissolve in all wells. When both calcite and dolomite are permitted to dissolve in the model, there is a significant increase in the Mg concentrations above the observed flowback water concentrations due to dolomite dissolution (Fig. 2.13).  In this scenario, additional calcite precipitation occurs which causes a further decrease of the Ca concentrations in the mixed fluid.  If only calcite is permitted to dissolve, the Ca concentrations do increase slightly when there is a negative calcite SI; however, the contribution of Ca ions from calcite dissolution to bring the system to equilibrium (SI = 0) is minor relative to the contribution from mixing with formation water.  For example, at well H-1, the maximum amount of calcite dissolution occurs in the fluid with 25% formation water.  The difference in Ca concentrations for a fluid with this mixing proportion between the models with and without 60  calcite dissolution is 18 mg/L, which is < 5% of the Ca concentration measured for a fluid with this mixing ratio.      Figure 2.13: Modelled and measured results for (a) Ca; and (b) Mg concentrations with the modelled results for mineral dissolution.  The addition of mineral dissolution results in dolomite dissolution, which causes an increase in Mg concentrations as well as a decrease in Ca concentrations due to increased calcite precipitation.  For the flowback water from this well, the Ca concentrations show a better fit; however, the Mg concentrations show a significant increase which is not seen in the experimental results.  2.6.3 Sulfate In contrast to the Cl and major cation concentrations, the measured SO4 concentrations in the Montney Formation flowback water do not show a consistent increasing trend over the flowback period for the majority of the study wells, indicating geochemical processes other than mixing.  The measured SO4 concentrations are generally higher than the concentrations used to approximate formation water in the model (Fig. 2.14).  The elevated SO4 concentrations may be due to oxidation of pyrite (Wilke et al., 2015; Harrison et al., 2017) or the oxidation of H2S, if present, both processes potentially stimulated by the injection of the oxic hydraulic fracturing a. b. 61  fluid into the anoxic formation.  Laboratory experiments have observed pyrite oxidation on time scales relevant to hydraulic fracturing (Wilke et al., 2015; Harrison et al., 2017).  Pyrite is present in the Montney Formation at an average abundance of approximately 2% by weight, based on the XRD results compiled for samples from the areas near the study wells.  H2S gas is found in elevated concentrations in some regions of the Montney Formation (Kirste et al., 1997; Desrocher et al., 2004).  H2S in the flowback water was not measured as part of the study, although, the flowback waters from site A and site F contain detectable H2S.  The samples from site A were collected prior to treatment for H2S.  The SO4 concentrations in the site A flowback waters are relatively stable over the flowback period and range from 51 to 234 mg/L for the different wells.  The SO4 concentrations are greater than the SO4 concentrations in the hydraulic fracturing fluids (median: 18 mg/L) and often greater than the estimated value for formation water at site A (82 mg/L).  The site F samples were collected following treatment with NaClO to remove H2S.  Both site F wells show increasing SO4 concentrations over the flowback period up to 824 mg/L, which is likely related to the production of H2SO4 during treatment with NaOCl.  The presence of the acid is supported by the decreasing pH in the flowback waters from these wells (pH < 4 in late flowback water samples).    62   Figure 2.14: Sulfate concentrations in the injected hydraulic fracturing (HF) fluid, the flowback water, and the formation water for each of the sites.  The SO4 concentrations in the flowback water are often higher than the SO4 concentrations estimated for formation water, indicating that mixing is not the dominant process on SO4 concentrations in flowback water.  For sites with multiple wells, a median value is presented for the HF fluid and the minimum, median, and maximum values for flowback water are calculated for the combined dataset.    The SO4 concentrations in the flowback waters from site I are lower than almost all other flowback water samples collected as part of the study (< 50 mg/L) and decrease over the flowback period (Fig. 2.15).  The relatively elevated SO4 concentrations in the early flowback waters are not related to SO4 concentrations in the injected hydraulic fracturing fluid which remain below the detection limit (< 0.1 mg/L) but may be due to pyrite oxidation (Wilke et al., 2015; Harrison et al., 2017).  Bacterial SO4 reduction likely contributes to the decreasing SO4 63  concentrations in the flowback water from the site I wells.  The injection of the relatively fresh hydraulic fracturing fluid, containing organic additives, reduces the salinity in the formation making a more favorable environment for the bacteria (Engle and Rowan, 2014).  The reduction of SO4 could in turn increase the solubility of barite which would result in higher barium (Ba) concentrations in areas where SO4-reducing bacteria are more active.  The highest Ba concentrations among the study wells were measured in the late flowback samples from well I-1 (215 mg/L) and well I-2 (467 mg/L), supporting this possibility.  The Ba concentrations in the Montney Formation flowback water from the study wells are presented elsewhere (see Chapter 5).  Figure 2.15: The SO4 concentrations over the flowback period at the site I wells.  The elevated SO4 concentrations measured earlier in the flowback period may be due to pyrite oxidation (Harrison et al., 2017; Wilke et al., 2015), as they are not related to the SO4 concentrations in the injected hydraulic fracturing fluid, which were less than 0.1 mg/L.  The decreasing SO4 concentrations are interpreted to be due to bacterial SO4 reduction which may vary between wells due to the differences in the bacterial assemblage found in the subsurface in different regions (Engle and Rowan, 2014).      64  The SO4-bearing mineral celestite is slightly oversaturated (0 < SI < 1) in the flowback water from only a few wells, including the well G-1 samples and some samples from the site A wells.  The potential for precipitation of celestite was considered in the geochemical model and was discussed above in relation to the flowback water Sr concentrations.  The geochemical model predicts celestite precipitation in the later flowback water samples from well G-1 when the model reaches equilibrium, which should correspond to a decrease in both Sr and SO4 concentrations.  However, the SO4 concentrations at well G-1 remain relatively constant (range: 1308 to 1407 mg/L) over the flowback period.  The lack of a decrease in concentrations in the flowback water results indicates that equilibrium has not been reached in the flowback water and celestite precipitation is not occurring in this well.   2.7 Discussion and Conclusions The major ion chemistry of the Montney Formation flowback water from the study wells provides insight into the geochemical processes that are occurring in the reservoir.  The overview of the major ion flowback water chemistry showed that the major ion concentrations, and therefore the TDS, increase over the flowback period, while the SO4 concentrations are generally invariant.  These results are in agreement with studies on flowback water chemistry in other formations (e.g., Haluszczak et al., 2013; Ziemkiewicz and He, 2015; Rosenblum et al., 2017).  There is some variability in the ion concentrations between wells; however, wells completed at the same site and within the same member of the Montney Formation generally show greater similarity in flowback water chemistry relative to those completed at different sites, which is interpreted to be due to the wells on the same site having comparable completion parameters and reservoir properties.    65  The increasing major ion concentrations in the flowback water over the flowback period indicate that mixing between the relatively low TDS hydraulic fracturing fluid and the high TDS formation water is the dominant geochemical process that is influencing the flowback water chemistry.  In addition to physical mixing between the injected fluid and formation water accessed in the fracture network and through countercurrent imbibition, other potential contributors to high TDS are ion diffusion (Balashov et al., 2015; Zolfaghari et al., 2016; Wang et al., 2016; 2017), and/or osmosis (Wang et al., 2016; 2017).  With the current data, it was not possible to differentiate between these processes and all may be adding to the increasing TDS. Through the use of Cl, δ18O, and δ2H as conservative tracers, the proportion of formation water contributing to flowback water was calculated for each of the samples analyzed for the study wells.  There was variability in the proportion of formation water in the flowback water at different wells over the flowback period.  For the initial flowback water samples, the contribution from formation water generally was in the range of 10-35%, while for the final flowback water samples the formation water proportion was generally 40-60%.  Since there is a significant contribution from formation water, the actual volume of the injected hydraulic fracturing fluid recovered during the flowback period is much less than the volume of total fluid recovered.  For the study wells, the calculated volume of hydraulic fracturing fluid recovered based on the mixing proportions is typically < 10% of the injected fluid.   Studies conducted with flowback water in other formations have used the increasing Cl concentrations (Olsson et al., 2013; Vengosh et al., 2017) and TDS (Kondash et al., 2017; Rosenblum et al., 2017) to calculate the proportion of formation water in flowback water.  The percentage of formation water calculated in these studies for a flowback time similar to the late flowback water in our study wells show either a comparable percentage of formation water at 66  approximately 50% (Kondash et al., 2017; Vengosh et al., 2017) or a higher percentage of formation water (>90%; Olsson et al., 2013; Rosenblum et al., 2017).  In the Rosenblum et al. (2017) study, the higher proportion of formation water was attributed to a 30 day shut-in period prior to flowback for their study well.   Our study also shows that the proportion of formation water is impacted by additional factors.  The low correlation between the highest proportions of formation water and the length of the flowback period indicates that many variables are influencing mixing, including the length of the shut-in period and the extent and complexity of the fracture system.  The uncertainty in the formation water chemistry due to not being able to directly sample the Montney Formation flowback water may lead to some variability in the mixing ratio calculated with the different tracers, as the values used may be accurate for one of the tracers but not others.  Using three tracers provided a method to check the results through a comparison of the three mixing ratios obtained and allowed for a representative range in ratios to be determined for each flowback water sample.  In the site B, site E, and site F wells, the comparison between the mixing ratios calculated with the three tracers indicated that some of the isotopic results were contaminated.  In well H-1, the use of multiple tracers allowed for the identification of potential mixing with a third fluid source with a different geochemical signature, in addition to the hydraulic fracturing fluid and the formation water. Potential fluid sources in addition to the injected fluid and the formation water include high permeability lenses and contiguous formations.  The second portion of the study used geochemical modelling to determine if processes other than mixing were required to explain the flowback water major ion chemistry.  The Na and K concentrations in the studied flowback water can be explained by mixing for the vast majority of the study wells, while the Ca, Mg, and Sr concentrations were influenced by a combination of 67  mixing and ion exchange.  Ion exchange causes the concentrations of the divalent ions to be lower over the course of the flowback period, relative to if only mixing was occurring.  Sequential extraction experiments completed with Marcellus Shale samples by Stewart et al. (2015) found that although Ca, Mg, and Sr ions were primarily associated with carbonates, these ions were also present on the exchange sites indicating the potential for ion exchange.  Ion exchange has also previously been suggested as a contributing source of Na ions to flowback water in hydraulically fractured shales (Zolfaghari et al., 2014).  The Ca concentrations in flowback water may be minimally impacted by calcite precipitation; however, the majority of wells do not show any evidence that calcite is precipitating in the formation, even though the mixed fluids generally have a SI above 0.  Experimental evidence from laboratory experiments conducted by others does not support calcite precipitation in the reservoir and has demonstrated that calcite dissolution is occurring (Lu et al., 2017).  These experiments examined shale samples by SEM before and after placing the sample in saline fluids under reservoir temperature and pressure over a three week period.  The geochemical models for wells A-2, A-3, A-6, D-1, D-2, D-3, the site E wells, the site F wells, and well H-1 support the potential for calcite dissolution when the hydraulic fracturing fluid is injected into the formation and the proportion of formation water in the mixed fluid is low since the calcite SI is below 0.  In the scenario which permitted calcite dissolution without dolomite dissolution, the contribution of Ca ions from calcite dissolution remained minor relative to the contribution from formation water, indicating that even if calcite dissolution is occurring in the reservoir, mixing with formation water remains the dominant source of Ca in flowback water.   The SO4 concentrations were not accurately modelled by mixing and are often higher than the values predicted in the mixing model.  The elevated SO4 concentrations may be related 68  to the oxidation of pyrite or H2S due to the injection of the hydraulic fracturing fluid as proposed by Wilke et al. (2015) and Harrison et al. (2017).  The decreasing SO4 concentrations seen in the flowback water from the site I wells may be related to SO4 reduction by bacteria as suggested elsewhere by Engle and Rowan (2014).   Overall, the results for the current study support mixing between the injected hydraulic fracturing fluid and the formation water as the dominant process affecting the flowback water chemistry.  A more thorough examination of the flowback water chemistry; however, provides supplementary information on geochemical processes occurring in the reservoir.  Our results show that the flowback water chemistry is influenced by processes in addition to mixing, including ion exchange, pyrite oxidation, and SO4 reduction, through comparison of the chemistry and the geochemical models.  Further, more detailed studies looking at the variability in the formation water chemistry, the effect of organics, and the importance of ion exchange between different sites would improve the fit of the geochemical models and contribute to their use as a prediction tool for flowback water chemistry.    69  Chapter 3: Flowback water chemistry from the Montney Formation: Part I - Stratigraphic and areal variability  3.1 Introduction Economic production of gas and liquid from ultralow permeability strata has been made possible due to horizontal drilling and completions through hydraulic fracturing.  Hydraulic fracturing involves the injection of a large volume of hydraulic fracturing fluid (range: approximately 2,000 to 50,000 m3 in total [Vidic et al., 2013; Alessi et al., 2017]) over up to 50 isolated intervals (stages) in order to create a complex fracture network thereby increasing the system permeability of the reservoir.  Following well completion, the well is put on production resulting initially in the flow back of liquids (the flowback water2) that are a mixture of the injected fluids and the reservoir fluids.  The total volume of the flowback water for most unconventional reservoirs is approximately 25% of the injected fluid volume (Haluszczak et al., 2013) but the volume is highly variable between wells and formations.  Flowback water is generally high in total dissolved solids (TDS) from the order of 1,000's to 100,000's of milligrams per liter (e.g., Haluszczak et al., 2013; Kolesar Kohl et al., 2014).  In addition to major ions and trace metals, flowback water can also contain organics (e.g., Strong et al., 2014; Lester et al., 2015) and radium (e.g., Nelson et al., 2016).  Most commonly the TDS and the major ion concentrations in flowback water increase during the flowback period, although this depends on the chemistry of the injected water. Over time the concentrations in flowback water                                                  2 In this paper, flowback water refers to water produced by a well prior to production, while the water returned to the surface during production will be considered produced water.  The transition from flowback water to produced water is not necessarily distinct. 70  approach those in formation water (e.g., Blauch et al., 2009; Haluszczak et al., 2013).  The current explanation for the source of the TDS and the major ions is that the flowback water chemistry is a product of mixing between the injected hydraulic fracturing fluid and the formation water (e.g., Haluszczak et al., 2013; Olsson et al., 2013; Engle and Rowan, 2014; Vengosh et al., 2017), and potentially influenced by water-rock interactions (e.g., Barbot et al., 2013; Seales et al., 2016; Zolfaghari et al., 2016; Marcon et al., 2017).   Previous research on flowback water geochemistry has focused mainly on the environmental implications of hydraulic fracturing (e.g., Cooley and Donnelly, 2012; Vengosh et al., 2013, 2014; Vidic et al. 2013; Connor et al., 2015; Reible et al., 2016; Ward et al., 2016) and understanding the processes that are occurring in the reservoir (e.g., Barbot et al., 2013; Bearinger, 2013; Ghanbari et al., 2013; Zolfaghari et al., 2015a).  The environmental effects studies often look at water management associated with hydraulic fracturing operations including water use, flowback water disposal, and flowback water treatment for reuse (e.g., Fontenelle et al., 2013; Vidic et al., 2013; Goss et al., 2015; Alessi et al., 2017).  Geochemical studies that identify unique chemical tracers (e.g., strontium isotopes, Cl/Br ratio) in flowback water have been used to differentiate between the sources of the salinity between flowback water and other saline water sources (e.g., Chapman et al., 2012; Capo et al., 2014; Johnson et al., 2015; Kolesar Kohl et al., 2014; Ziemkiewicz and He, 2015; Zheng et al., 2017).  Mineral precipitation and dissolution in the reservoir can be inferred by examining ion ratios and the change in mineral saturation indices in flowback water over time (Barbot et al., 2013).  Laboratory observation of both mineral precipitation (e.g., gypsum, barite, and anhydrite) and carbonate dissolution have been recorded in studies with synthetic hydraulic fracturing fluids and shale samples under 71  elevated temperatures and pressures (Wilke et al., 2015; Dieterich et al., 2016; Marcon et al., 2017).   The rate of increase of TDS and the major ion concentrations in flowback water over time has been used to infer the complexity of the fracture system (Bearinger, 2013; Ghanbari et al., 2013; Zolfaghari et al., 2015a).  A continuous increase in TDS during flowback has been suggested to be an indication of the progressively greater number of secondary fractures and a higher surface area exposed to hydraulic fracturing fluid whereas a plateau in TDS in the late flowback period supports a system dominated by primary fractures.  Flowback water chemistry can vary even between wells completed in the same formation.  The study of the source of TDS in flowback water and the causes for variability in chemistry between different sites is an important area of research since the flowback water chemistry and rheology can provide insight on processes taking place in the subsurface during hydraulic fracturing, including the nature of the stimulated reservoir volume (SRV), relative merits of flowback control (i.e., slow-back vs fast-back), and propensity for scaling or corrosion of tubulars or reservoir salting.  A comparison of the flowback water chemistry between wells is important in order to determine if there are trends in how the flowback water chemistry varies laterally or stratigraphically within a formation.  Characterizing lateral variability in flowback water can assist in optimizing storage and/or treatment strategies and determining if recycling (reuse) of completion fluid for future wells is practical and if so optimizing liquid blends.  The use of appropriate treatment and disposal methods is essential to minimize the risk of surface and groundwater contamination related to incorrect storage and accidental spills of flowback water (e.g., Vengosh et al., 2013).   72  In our study the flowback water from the Lower Triassic Montney Formation in northeastern British Columbia and northwestern Alberta, Canada is investigated.  The Montney Formation is the most important producing unconventional hydrocarbon reservoir in Canada and is currently being actively developed.  The Montney Formation varies in lithology and reservoir properties both aerially and stratigraphically.  This formation is composed of fine grained sediments including shale, dolo-siltstones, and fine-grained sandstone and overall becomes coarser grained from BC to Alberta (Edwards et al., 1994; Dixon, 2000; Zonneveld et al., 2011; Zonneveld and Moslow, 2014).  Turbidite deposits are prominent near the BC-Alberta border (Zonneveld et al., 2010) and dolocoquina subunits are found in the eastern regions in Alberta (Davies et al., 1997; Davies and Hume, 2016). Overall, the Montney Formation becomes slightly coarser with increasing depth, varying from siltstone to very fine grained sandstone-siltstone (Davies et al., 1997; Dixon, 2000).  Past stratigraphic studies on different regions of the Montney Formation have focused on either BC (Dixon, 2000) or Alberta (Davies et al., 1997) but did not cover the whole formation.  The most recent subdivision of the Montney Formation into (informal) lower, middle, and upper members is based on lithostratigraphy and biostratigraphy (Davies and Hume, 2016).  This subdivision is used in the present study (Fig. 3.1).   73   Figure 3.1: Schematic cross-section of the Montney Formation showing divisions between the lower, middle, and upper informal members.  Each member represents a Third-Order sequence. The lower Montney Formation (Griesbachian-Dienerian stage) unconformably overlies the Permian Belloy Formation and extends to the base of the lowstand wedge.  The middle Montney Formation (Smithian stage) extends from this lowstand wedge to the base of a second lowstand wedge. The upper Montney Formation (Spathian stage) includes the second lowstand wedge and the lower Doig siltstone. This member thins to the east and is absent in much of Alberta (modified from Davies and Hume, 2016).  In the study we collected and analyzed flowback water from 9 sites that include 31 wells completed in the upper (18 wells), middle (11 wells), and lower (2 wells) members of the Montney Formation (Fig. 3.2) as defined in figure 3.1.  The study wells were completed at total vertical depths between 1,775 and 3,230 m with the horizontal laterals of the wells ranging from between 1,720 and 3,330 m long.  The TDS and major ion concentrations of the flowback water from the study wells are the focus of this paper.  The major ions include chloride (Cl), sodium 74  (Na), potassium (K), calcium (Ca), magnesium (Mg), and strontium (Sr). The major ion chemistry of the flowback water collected from the upper, middle, and lower Montney Formation study wells is summarized and potential issues that may arise in comparing flowback water chemistry between different locations are discussed.  A method to compare the flowback water chemistry from different wells using linear regression and linear mixed effects models is presented.    Figure 3.2: Location of the sites (A-I) included in the study.  Wells completed in the upper Montney Formation are located at sites A through D, and site H, wells completed in the middle Montney Formation are located at sites C through G, and wells completed in the lower Montney Formation are located at site I.  The lateral extent of the Montney Formations is shown in green (modified from Edwards et al., 1994). 75  3.2 Methods 3.2.1 Hydraulic Fracturing Fluid Chemistry Samples of hydraulic fracturing fluid (base liquid plus additives) were collected for the majority of the wells sampled for flowback water.  For wells completed on site A, site B, site D, site E, site G, and site H, several hydraulic fracturing fluid samples were collected that are representative of the fluids used in completing the multiple stages along the lateral.  The samples from the different stages of a well were combined into composite samples based on similar electrical conductivity readings used as quick proxy for total dissolved solids. For the site F and site I wells, one sample of the hydraulic fracturing fluid was collected to represent the fluids used for each of these wells. The composite samples and the samples from site F and site I were filtered through a 0.45 µm filter and analyzed for pH, conductivity, total alkalinity, anions, and dissolved metals.  All analyses were conducted at The University of British Columbia (UBC) in Vancouver, BC, excluding the anion samples which were sent to a laboratory in Victoria, BC for the ion chromatography analysis.  The pH and conductivity were measured using an Omega® PHB21 potable pH meter and a HACH® CDC401 conductivity probe, respectively. The total alkalinity was measured by titrating the solution with sulfuric acid (H2SO4) using a HACH® digital titrator (Model 16900).  The dissolved metals samples were initially preserved with trace metal grade nitric acid (HNO3) to pH < 2.  In order to remove organic compounds that may interfere with the inductively coupled plasma-optical emission spectrometry (ICP-OES), the dissolved metal samples were then acid digested using trace metal grade HNO3 and hydrochloric acid (HCl) in a 2:1 ratio and evaporated in an acid-washed glass beaker on a hot plate.   The digested samples were treated a second time with concentrated HNO3, re-evaporated, and then re-diluted to the initial volume with a 1% HNO3 solution.  The ICP-OES analysis was performed 76  on a Varian® 725-ES ICP-OES with indium (In) as the internal standard.  The ion concentrations were added together to calculated the TDS.   The overall hydraulic fracturing fluid chemistry for the wells with composite samples was calculated from the geochemical analysis of the composite samples and the relative proportion of the fluid volumes used in each stage.  For example, if 4 composite samples were analyzed for a well and the hydraulic fracturing stages combined in each sample each represent 25% of the total fluid volume, then the chemistry of the 4 composite samples are given equal weight in the overall fluid chemistry calculation.  No hydraulic fracturing fluid samples were obtained for wells completed at site C.      3.2.2 Flowback Water Chemistry Flowback water samples were collected from the 31 study wells.  Two to three samples were collected each day for the first week of the flowback period then sampling was decreased to one sample per day in wells where the flowback period extended beyond one week.  The sampling schedule allowed for more sampling at the beginning of the flowback period when the recovered fluid volume is typically greater and the fluid chemistry is expected to change more rapidly. The samples were stored at 4°C and were then heated to reservoir temperature (75-80°C) for 24 hours prior to analysis to re-dissolve any precipitates that formed due to cooling of the flowback water after sampling.  The heating step was included since it was logistically impractical for the samples to be filtered and preserved with acid at the time of sample collection.  The samples were filtered and the same suite of analyses completed for the hydraulic fracturing fluids were completed for the heated flowback water samples. 77  3.2.3 Produced Water Chemistry Produced water samples were included in the water sampling at site A, site B, and a well nearby to site I.  The produced water samples were prepared and analyzed following the same steps as the flowback water samples.  The results for the produced water samples were used to provide a measure of formation water chemistry for the wells from these three sites.  For the remaining wells an approximation of the formation water chemistry was acquired from publically available data3 from nearby wells with produced water results from the Montney Formation. The publically available fluid data was evaluated and samples with anomalously low conductivity values or with noted errors (e.g., contaminated sample) were removed.  The suite of parameters available typically includes calculated TDS, and measured pH, Cl, Na, K, Ca, and Mg; Sr was available for the majority of samples.   3.2.4 Linear Regression and Linear Mixed Effects Models Simple linear regression and linear mixed effects models were combined to conduct a trend analysis as a method to investigate the relationship of both flowback water chemistry and flowback water volume with flowback time for the different sites.  All regression analyses in support of this work were completed in R using the base package for linear regression (R Core Team, 2017) and the lme4 package for the linear mixed effects models (Bates et al., 2015).  The initial step in the analysis was looking at the simple linear regression between the parameters of interest for each of the individual wells.  The simple linear regression with one covariate is in the form of:                                                  3 The publically available data is from data compiled by the BC Oil and Gas Commission and accessed through geoSCOUT®.  78  𝑦 = 𝑥𝛽 + 𝜀 Where, y is the response variable, x is a covariate, β is the slope coefficient, and ε is the error term.  In our analysis, y is either the TDS, the concentration of one of the major ions, or the flowback volume, while x is flowback time.  If the slope coefficients in the linear regression are sufficiently similar and the data contains a categorical variable, data can be combined and a linear mixed effects model can be used in the analysis.  A linear mixed effects model includes an additional term which is used to group the results within a given category in order to look at another level within the data (Bates, 2010). In its simplest form it includes one covariate for the fixed effects:     𝑦 = 𝑥𝛽 + 𝑢𝛾 +  𝜖 Where, y is the response variable, x is a covariate for the fixed effects, β is the slope coefficient for the fixed effects, u is a covariate for the random effects, γ is the coefficient for the random effects, and ε is the error term.   To further explain how these models apply to the flowback water dataset, consider an example where TDS results are available over the flowback period for 4 wells completed at the same site.  If a simple linear regression was conducted for TDS over time, the slope coefficient (β) could be determined either for each individual well or for all results from the site combined: 𝑇𝐷𝑆 ~ 𝐹𝑙𝑜𝑤𝑏𝑎𝑐𝑘 𝑡𝑖𝑚𝑒 This notation indicates that TDS is determined as a function of flowback time.  When the goal of the analysis is to examine the data at the site level, including the results for all the wells together does not take into account that the TDS results for each individual well are related and are expected to differ somewhat from the results from other wells at the site.  Linear mixed effects models will take the differences between wells into account while allowing the data for the site 79  to be examined together.  In order to determine if a linear mixed effects model is appropriate for grouping the wells at a site, a simple linear regression needs to be completed for each individual well to verify that the slope coefficients show a high enough degree of similarity.  For our analysis, the slope coefficients were taken to be comparable when the slope coefficients with their confidence intervals have overlapping values.  A linear mixed effects model provides a method to divide the combined results for the site using an additional level in the data, which in this example are the individual wells: 𝑇𝐷𝑆 ~ 𝐹𝑙𝑜𝑤𝑏𝑎𝑐𝑘 𝑡𝑖𝑚𝑒 + (𝐹𝑙𝑜𝑤𝑏𝑎𝑐𝑘 𝑡𝑖𝑚𝑒 | 𝑊𝑒𝑙𝑙) This representation shows that TDS is determined as a function of flowback time for the site but that the data are also grouped together by well.  The linear mixed effects model is in the form to allow for each well to have both a different slope and intercept from the other wells at the site based on the TDS as a function of the flowback time.  Using the linear mixed effects model will give the overall fixed effects slope coefficient for the site, as well as the random effects coefficient to take the differences between the individual wells into account.  The random effects are the differences within the data for each well that are not explained by the fixed effects term.  The fixed effects slope coefficient can then be used to compare the study sites, by grouping sites with similar fixed effects slope coefficients.    3.3 Challenges of comparing flowback data from different wells  Flowback water geochemistry is influenced by a plethora of different variables, many of which are correlated, which complicates comparisons between different wells and interpretation of the geochemistry of the fluids.  The variables include those that are dependent on the properties of the reservoir, the completion design, and the response of the reservoir to the completion (Table 3.1).  The factors that are not directly related to the properties of the reservoir 80  can create variability between different wells, even at the same site, thus making it difficult to compare sites or determine the cause of any variability.   Table 3.1: Summary of some of the variables that may influence flowback water geochemistry. Many of the variables are interrelated. Reservoir properties Completion design Response of reservoir to completion   Mineralogy   Porosity, permeability, and fabric of reservoir   Amorphous precipitates   Presence and complexity of natural or induced fractures   Water saturation (Sw), Oil saturation (So)   Reservoir wettability   Reservoir capillary pressure   Imbibition potential   Reservoir temperature and pressure   Formation water chemistry and rheology   Depth of burial   Paleo depth of burial   Amount of uplift and degassing   Inherent and induced stress and stress anisotropy    Volume of hydraulic fracturing fluid pumped   Chemistry and rheology of hydraulic fracturing fluid   Length of horizontal well   Number of stages   Clusters per stage   Perforations per cluster   Orientation of perforations   Shut-in time   Effectiveness of proppant   Proppant embedment     Length of flowback period   Rate of flowback   Volume imbibed into reservoir   Complexity of induced fractures   Surface area contacted by the completions   Stimulated Reservoir Volume   Water-rock interactions   Reservoir damage (chemical + physical)   Stress sensitivity of reservoir matrix and natural and induced fracture system    Breakdown pressure   Fracture closure pressure(s)    Due to the multiple variables that have the potential to alter flowback water geochemistry, different methods of presenting and interpreting the results must be utilized and are discussed in this paper. The parameter selected for the graphical representation of flowback water geochemical results is an important consideration in flowback water studies as this can impact the interpretation of the data (Fig. 3.3).  Flowback water results are often plotted as the 81  concentration of the element of interest versus the number of days (i.e., time) since the beginning of flowback (i.e., flowback day; e.g., Haluszczak et al., 2013; Capo et al., 2014; Ziemkiewicz and He, 2015). However, flowback day cannot be studied independently of the shut-in time prior to flowback, or percentage of flowback per day, to name a few of the plethora of parameters that need be considered.  Examples of different parameters for plotting flowback water chemistry illustrate how the interpretation of the variation in the flowback water results can vary depending on the parameter selected.  For example, a plot of Na concentrations from one of the study wells appear to approach a stable maximum value near 20,000 mg/L when cross plotted against flowback day (Fig. 3.3a).  However, when Na concentrations are plotted against cumulative flowback volume, the percent recovered4, the percent of the total flowback5, or Cl concentrations, the Na concentrations continue to increase through to the end of the flowback period (Fig. 3.3b-e).  Ion ratios can also be used to look at geochemical results (Fig. 3.3f), where in the presented example the Na/Cl ratio shows a slight decrease and then stabilizes.                                                  4 The percent recovered is defined as the percentage of the injected fluid volume that has been recovered from the well as flowback water. This does not represent recovery of only hydraulic fracturing fluid as there is mixing with formation water occurring in the reservoir.  5 The percent of total flowback is the relative percentage of flowback water collected over the flowback period (i.e., the volume at the end of the flowback period is 100% of the total flowback volume).  82   Figure 3.3: Different potential variables for graphical representation of flowback water chemical data from one well (data from well I-1). The parameters selected to represent the data influences the interpretation of the flowback water chemistry. a) Na concentrations versus flowback day; b) Na concentrations versus cumulative flowback volume; c) Na concentrations versus percent recovered3; d) Na concentrations versus percent of total flowback water4; e) Na concentrations versus Cl concentrations; f) Na/Cl mass ratio versus percent recovered.   Yet further complications in comparing flowback water geochemistry are that the studied wells were flowed back for variable lengths of time prior to being put on production and that the cumulative flowback volumes vary. As the major ions typically increase with both time and cumulative flowback volume, any differences in these variables can impact the comparison between wells.  Wells with longer flowback periods and greater volumes of fluid recovered tend to have higher concentrations of the major ions, relative to wells with a shorter flowback period or lower flowback water recovery, everything else being equal.  Due to the variability in the a. b. c. d. e. f. 83  length of the flowback period, using minimum, median, or maximum concentrations is not an effective way to characterize the flowback water chemistry for different wells.  A well to well comparison could be made by picking an arbitrary flowback day or flowback volume and comparing the concentrations of the closest sample for each well.   This method would require the selection of a sample early on in the flowback period or with low cumulative flowback volume in order to select a flowback time or volume that was represented by all wells in the study.  Due to the factors summarized above, in-depth statistical analysis is required in order to compare the flowback water geochemistry from different wells.  Of the statistical methods tested in our study, linear regression and linear mixed effects models proved to be the most promising method and will be used in the present study to compare the different sites and wells. 3.4 Montney Formation flowback water chemistry The flowback water chemistry in the studied wells varies over the flowback period.  In general, the concentrations of the major ions, and therefore TDS, increase over the flowback period (Fig. 3.4).  The magnitude of the increase in TDS over the flowback period and the maximum major ion concentrations attained are different between wells.  Even though the concentrations of the major ions vary, in all sampled waters Cl is the dominant ion, followed by Na, Ca, and K (Table 3.2).  Mg and Sr concentrations, although important, are lower than the other major ions.  The flowback water pH remains near neutral in the majority of samples (5th to 95th percentiles: 6.1 to 8.0, median: 7.0) and generally shows a slight decrease over the flowback period in the study wells.  The site F flowback waters show a greater decrease in pH to slightly acidic values at the end of the flowback period (pH < 4), due to the production of H2SO4 during the treatment of the flowback water for hydrogen sulfide (H2S) with sodium hypochlorite 84  (NaClO).  The site D flowback waters show more variability in pH relative to other wells in the study with values ranging from 2.3 to 9.5.  The total alkalinity also generally decreases over the flowback period for the flowback water from the study wells, with the majority of samples remaining above 100 mg/L as CaCO3 (range: < 10 to 685 mg/L as CaCO3, median: 149 mg/L as CaCO3).  In general, the charge balance error for the flowback water samples was within ±15%.  The higher charge balance error relative to low TDS waters may be due to not measuring the organic molecules in the flowback water as well as the higher error introduced by the high dilution factors required to analyze the fluids.    85   Figure 3.4: TDS at the study wells over the flowback period as cumulative flowback volume. a) Site A wells – upper Montney Formation; b) Site B wells – upper Montney Formation; c) Site C wells – upper (wells 1-5) and middle (wells 6 & 7) Montney Formation; d) Site D wells – upper (wells 1 & 2) and middle (wells 3 & 4) Montney Formation; e) Site E wells – middle Montney Formation; f) Site F wells – middle Montney Formation; g) Site G well – middle Montney Formation; h) Site H well – upper Montney Formation; i) Site I wells – lower Montney Formation.  a. b. c. d. e. f. g. h. i. 86  Table 3.2 (following page): Summary of conductivity, TDS, and the major ion concentrations for each of the 9 study sites. Site C and site D are divided into upper and middle Montney Formation wells. Lower concentrations are associated with earlier flowback times and higher concentrations are from later in the flowback period.  The data from multiple wells are combined in cases where flowback water samples were obtained from multiple wells at a site.                    87  Site n   Conductivity TDS Cl Na Ca K Mg Sr mS/cm mg/L mg/L mg/L mg/L mg/L mg/L mg/L upper Montney Formation               A (8 wells) 22 min 60.8 41,358 25,722 8,264 1,616 553 296 298 median 102 61,220 43,437 12,759 3,059 796 547 630 max 131 92,731 65,777 17,818 5,389 1,239 941 1,220 B (2 wells) 30 min 47.6 30,135 18,541 9,162 1,789 284 154 60 median 123 89,949 55,235 25,609 5,950 1,035 619 404 max 190 228,259 164,018 51,027 11,705 1,920 1,369 924 C (5 wells) 52 min 56.1 34,137 21,344 9,932 1,360 569 191 194 median 102 70,449 44,439 19,779 3,088 1,025 402 477 max 157 147,839 104,897 33,212 7,333 1,500 681 916 D (2 wells) 19 min 50.4 31,522 19,303 9,960 1,169 466 171 123 median 86.4 62,520 38,355 18,489 2,874 860 418 264 max 107 80,944 50,963 23,974 4,058 1,172 611 358 H (1 well) 18 min 33.8 20,713 12,308 6,451 629 414 102 30 median 113 83,807 49,006 28,509 2,798 1,534 650 141 max 135 112,226 66,998 37,637 3,677 1,806 809 195 middle Montney Formation               C (2 wells) 20 min 120 82,519 52,570 23,802 3,638 1,065 441 573 median 147 110,450 71,218 30,960 5,310 1,461 669 893 max 171 140,538 93,007 38,460 7,332 1,865 945 1,263 D (2 wells) 22 min 76 3,609 1,893 1,344 13 49 10 4 median 99 21,631 12,510 7,884 628 257 79 55 max 123 56,800 34,565 18,872 2,112 466 223 268 E (4 wells) 26 min 22.1 11,105 6,671 2,787 350 187 62 38 median 35.7 18,672 12,252 5,256 641 295 107 72 max 51.8 32,864 21,611 9,119 1,126 474 220 161 F (2 wells) 12 min 29.1 18,655 10,295 6,849 560 166 93 33 median 37.5 24,009 14,455 9,066 702 196 117 44 max 50.4 34,139 21,495 10,575 795 251 136 69 G (1 well) 8 min 27.9 13,972 8,186 3,896 174 186 74 29 median 64.8 38,963 24,790 10,999 845 410 273 73 max 95.4 65,052 41,783 18,244 2,229 650 586 101 lower Montney Formation                I (2 wells) 28 min 11.1 4,587 2,260 1,991 110 105 32 13 median 56.1 31,998 17,678 12,112 1,032 558 166 206 max 130 90,178 51,038 33,782 4,081 1,296 524 1,011  88  3.4.1 Overview of Montney Formation flowback water chemistry Different methods of grouping the TDS and major ion results are used to provide a high level overview of the flowback water data from the current study.  In the following section a series of plots are provided to summarize and compare the flowback water results.  The flowback day (time), the formation member where the well was completed, and the well location all have an influence on the flowback water chemistry and are some of the factors that need to be considered when interpreting flowback water geochemistry.  The total flowback and percent recovered are also considered. Flowback time: The importance of flowback time is investigated by dividing the data into early (day 1-2), middle (day 2-7), late (> day 7), and produced water samples (Fig. 3.5a).  An increasing trend in TDS is apparent over the flowback period; however, there is overlap between the different groups of data due to the variability between the different wells that is not taken into account in the plot.  Stratigraphy: On average, the wells completed in the upper Montney Formation have higher major ion concentrations in the flowback water relative to the wells completed in either the middle or the lower Montney Formation (Fig. 3.5b).  The higher concentrations in the upper Montney Formation flowback water are most apparent in the early time flowback water chemistry.  Aerial distribution: Grouping the results by site shows that the flowback water TDS varies considerably between different sites and that the variation is not linked solely to the member of the Montney Formation where the well was completed (Fig. 3.5c).  There is also variation across the study region, with the overall TDS values appearing slightly higher in the 89  southeastern region (Fig. 3.5d); however, data is limited to the results from only one well in this region.  Total flowback and Percent recovered: There is limited variation in TDS over the flowback period when the flowback water results for all the wells are compiled and divided based on the percent of total flowback water (Fig. 3.6a).  The division does not take the length of the flowback period or the volume of flowback water produced from a well into consideration.  The low variability in this case is due to the data for all wells being averaged out over the flowback period for all wells.  By dividing the results by the percentage of flowback water, the last sample for a well with a shorter flowback period would be included as 90-100% flowback and decrease the overall value, for example.  Grouping the data based on the percent recovered (Fig. 3.6b) shows that the range in flowback water TDS is similar up to 25% recovered, comparable to the low variability in the total percent recovered plot.  The TDS in flowback water for samples with a higher percent recovered (25-35%) appear to be higher; however, these results are based on limited samples dominantly from well B-1.   90   Figure 3.5: Different sources of variability in the studied Montney Formation flowback water.  TDS is used as an example, as the major ions show similar trends to TDS. a) All data are grouped into early (day 1-2), middle (day 2-7), and late (> day 7) flowback water and produced water; b) The results for all wells are grouped by the Montney Formation member where hydraulic fracturing occurred; c) The results are grouped by site.  Sites with multiple wells have all data from those wells grouped together.  Sites C and D include wells from both the upper and the middle Montney Formation; d) The results for all wells are grouped by region.  The northwest region includes sites A and I (upper and lower Montney Formation), the central region includes sites B through G (upper and middle Montney Formation) and the southeast region includes site H (upper Montney Formation). Note: b-d summarize all flowback water results from early to late flowback combined but do not include the produced water results. The box portion of the boxplot shows the 25th percentile, the median, and the 75th percentile.  The whiskers show the interquartile range (IQR).  The points represent potential outliers which fall outside the IQR. c. d. a. b. 91   Figure 3.6: The flowback water results as TDS for all samples from all wells divided based on a) the percent of total flowback and b) the percent recovered.  When the results are considered as percent of total flowback, there is no significant variability when the results are averaged out for all wells combined.  The results for percent recovered do not vary significantly up to 25% recovered.  The variability for samples with a higher percent recovered (> 25%) is due to the data being from four wells (25-30%: well D-2 [n=1], well B-1 [n=5], and well B-2 [n=2]; 30-35%: well B-1 [n=2] and well A-2 [n=1]).  These higher TDS results would mainly reflect the chemistry of the flowback water results from site B and do not indicate that there is a significant increase in TDS after 25% recovery is reached.  The box portion of the boxplot shows the 25th percentile, the median, and the 75th percentile.  The whiskers show the interquartile range (IQR).  The points represent potential outliers which fall outside the IQR.  3.4.2 Flowback water chemistry by formation member 3.4.2.1 Upper Montney member Eighteen of the study wells were completed in the upper Montney member.  These wells are located in the northwestern (site A), central (sites B through D), and southeastern (site H) regions of the study area.  The flowback water TDS, major ion concentrations, and rate of a. b. 92  increase in TDS during flowback for the upper Montney member study wells all fall within a similar range of values over the flowback period; there is no significant variation in the flowback water chemistry between the different regions of the study area for the majority of the major ions.  The exception is the concentrations in the flowback water from well H-1 initially increased at a more rapid rate relative to the other wells (see Fig. 3.4h).  Overall, the flowback water TDS ranges from a minimum of 20,700 mg/L in the initial flowback water samples from well H-1 to a maximum of 228,000 mg/L in the late flowback water samples from well B-1 (see Table 3.2). The upper Montney member wells were completed with freshwater-based hydraulic fracturing fluids for wells A-3, D-1, D-2, and H-1 (TDS < 1,500 mg/L) and fluids composed of a blend of freshwater and recycled flowback water for the remaining site A wells and the site B wells (TDS > 3,000 mg/L).  The hydraulic fracturing fluids for the site C wells were not sampled but freshwater was used to make up these fluids.   The flowback water major ion concentrations for the upper Montney member wells are similar when samples collected at a similar flowback volume are compared.  There are exceptions: low Ca and Sr concentrations occur for well H-1 and high Sr concentrations occur for the site A wells.  The maximum major ion concentrations occur in flowback water from well B-1 in the samples following a 9 day shut-in period preceded by a 15 day low flow period.  The one exception is the maximum Sr concentration (1,220 mg/L), which is measured in well A-6 in the northwestern region of the study area.  The overall lowest Sr concentrations in late flowback water were measured at the southeastern site (well H-1).  Accepting these data at face value suggests the Sr concentrations in upper Montney member flowback water show an increasing trend across the study area from the southeast (site H) to the northwest (site A).  In sequential extraction experiments completed on shale samples by Stewart et al. (2015), a large proportion of 93  Sr was found in the carbonate fraction, therefore, the variability in Sr concentrations may be due to dissolution of carbonate with differing amounts of Sr.  Some carbonate dissolution may occur during hydraulic fracturing however, dissolution could also occur over the depositional history of the formation thus influencing the formation water chemistry which would in turn influence the flowback water chemistry through mixing.  Further flowback water sampling of upper Montney member wells in the southeastern and northwestern regions is required to further investigate the trend. 3.4.2.2 Middle Montney member All wells completed in the middle Montney member (n=11) are located in the central region of the study area.  The site D through F wells generally have similar major ion concentrations although the major ion concentrations in flowback water from the site D wells increase at a more rapid rate and reach higher concentrations by the end of the flowback period relative to the site E and site F wells (see Fig. 3.4d-f).  Within the site E wells, wells E-1, E-2, and E-4 have similar TDS while well E-3 has higher TDS at a lower flowback volume.  The major ion concentrations are higher for flowback water from well E-3 except for Na concentrations which are similar to the Na concentrations in flowback water from other site E wells.  The one well located at site G has slightly higher major ion concentrations and the two wells at site C have significantly higher major ion concentrations relative to flowback water from the site D-F wells.  The two site C wells have higher major ion concentrations relative to the flowback from the other middle Montney member wells throughout the flowback period.  The hydraulic fracturing fluids for the middle Montney member wells are interpreted to be a blend of freshwater and recycled water based on the elevated TDS (> 4,500 mg/L), excluding the hydraulic fracturing fluid for the site C wells and well D-3 which are dominantly freshwater. 94  3.4.2.3 Lower Montney member Only two studied wells were completed in the lower Montney member.  These wells are located at a site in the northwestern region of the study area (site I).  The initial flowback water major ion concentrations for the two lower Montney Formation wells are low.  The major ion concentrations increase more rapidly for well I-2 than well I-1 and higher concentrations are recorded for the flowback water from well I-2 for all times.  For example, the TDS increases from 4,590 to 53,800 mg/L for well I-1 and from 9,370 to 88,700 mg/L for well I-2 (Fig. 3.4i).  Both of these wells were sampled over 22 days, although, well I-1 had a greater total cumulative flowback water volume.  The major ion concentrations in the flowback water from the two site I wells reach similar concentrations to those seen in the flowback water from the upper and middle Montney Formation wells. 3.4.3 Causes of variability in flowback water In a companion paper (Chapter 4) we show the most probable cause of the increasing major ion concentrations is mixing between the normally low salinity hydraulic fracturing fluid and more saline formation water, with an increasing proportion of formation water over the flowback period.  Mixing between two end members is illustrated by the linear relationship observed with ion-ion plots (Fig. 3.7).  More efficient mixing between the hydraulic fracturing fluid and the formation water and possible variability in formation water chemistry between wells would contribute to differences in the rate of increase in TDS in flowback water. Potential variables that result in more mixing are a more complex fracture network and a higher SRV, which would result in more fractures and a more extensive fracture network.  The complexity of the fractures would in turn expose a greater surface area for water-rock interactions between the injected hydraulic fracturing fluid and the reservoir, as well as increase the access to formation 95  water.  Many of the study wells do not reach a plateau by the end of the flowback period and continue to show increasing ion concentrations for the full flowback period.  This behavior in flowback water from other formations has been attributed to a greater fracture complexity with a greater surface area exposed to the hydraulic fracturing fluid (Bearinger, 2013; Ghanbari et al., 2013; Zolfaghari et al., 2015a).  The smaller diameter of the secondary fractures would result in a more drawn out contribution from the fractures due to slower flow, while the higher surface area of these fractures would result in higher TDS, thus contributing to a continual increase in the TDS in flowback water.  A 100% contribution from formation water presents an upper limit to the major ion concentrations that can be attainted; however, based on the calculations using conservative tracers the majority of the study wells do not reach concentrations that are equivalent to 100% formation water (see Chapter 2).          96   Figure 3.7: a) Na-Cl plot separated by site; b) Ca-Cl plot separated by site.  The approximately linear relationship between the major ions indicates mixing between two endmembers – the injected hydraulic fracturing fluid and the formation water.  These plots show that ion concentrations in flowback water do not increase at a constant rate at all sites.  The 3 site B samples with high concentrations were collected from well B-1 following a shut-in period within the flowback period. ▲- upper (u) Montney Formation wells;  - middle (m) Montney Formation wells;  - lower Montney Formation wells.  In addition to the formation water accessed directly in the fractures created during hydraulic fracturing, formation water produced through countercurrent imbibition, as well as the processes of diffusion and osmosis (Balashov et al., 2015; Zolfaghari et al., 2016; Wang et al., 2016; 2017), may contribute to the high TDS.  The production of formation water through countercurrent imbibition is inferred from the increasing salinity in the water used in laboratory imbibition experiments with core samples, such as those conducted by Zolfaghari et al. (2016).  Imbibition undoubtedly is the main mechanism accounting for the low fluid recovery from hydraulically fractured wells (also see Roychaudhuri et al., 2013; Engelder et al., 2014).  A greater imbibed volume is correlated to longer shut-in times, higher clay content, and lower total a. b. 97  organic carbon content (Lan et al., 2014).  These variables are expected to vary between sites and potentially between wells, depending on the heterogeneity of the reservoir.  At sites where more imbibition occurred, a higher proportion of the injected hydraulic fracturing fluid would be sequestered in the formation and the flowback water would have a higher TDS due to the greater contribution from formation water accessed through countercurrent imbibition.  Mixing would occur in the fractures between the fluid that was not imbibed and the formation water that was drawn out of the formation. The influence of diffusion, osmosis, and countercurrent imbibition cannot be separated based on the current results and may all be contributing to the increasing TDS in flowback water. The results from the present study show that there is similarity between the flowback water chemistry of wells located at the same site and within the same member of the Montney Formation (e.g., wells D-3 and D-4 or wells E-1 and E-2; Fig. 3.4 d and e above).  Similarity in flowback water chemistry between nearby wells has previously been observed in Marcellus Shale flowback water (Barbot et al., 2013).  This result is likely due in large part to similarities in the reservoir properties between nearby wells, including formation water chemistry and formation mineralogy, which impact the flowback water chemistry through mixing and water-rock interactions.  In addition, wells from the same site are more likely to have similar completion design parameters, such as shut-in time, which influence the time and fluid volumes available for mixing.  It is intuitive that wells from the same site should have similar flowback water chemistry; however, based on the results from the study wells this is not always the case due to the influences from many variables.   One notable example of where wells located at the same site show significant differences in flowback water chemistry is site I.  The two site I wells were completed at the same site within 98  the lower Montney member with similar injected fluid chemistry and injected fluid volume.  The TDS and major ion concentrations increase more rapidly at well I-2 and reach higher concentrations by the end of the flowback period.  Well I-2 had 30 hydraulic fracturing stages and a 5 day shut-in period compared to well I-1 which had only 24 stages and was shut-in for 1 day.  The higher concentrations at well I-2 are interpreted to be primarily due to more extensive mixing between the hydraulic fracturing fluid and the formation water, relating to the longer shut-in time, additional hydraulic fracturing stages, and a more complex fracture network at this well (Chapter 2).   There are several examples in the study in addition to the site I wells which support the correlation between a longer shut-in period and a higher initial flowback water TDS.  The two middle Montney member wells at site C had high initial TDS in the flowback water (> 80,000 mg/L) and shut-in periods of 22 and 24 days, relative to the lower TDS (< 25,000 mg/L) and < 10 day shut-in periods for the other middle Montney member wells in our study.  The initial high concentrations at site C cannot be related to the injected fluid, which is made up of relatively low TDS freshwater with chemical additives.  Within the site E wells, the higher major ion concentrations in well E-3 flowback water may be due to the slightly longer shut-in period (8 days relative to 5 or 6 days for the other site E wells) and the lower flowback water recovery for this well, which was 600 m3 relative to > 1,000 m3 cumulative flowback water for the other three site E wells.  A longer shut-in period would result in more time for mixing to occur within the formation.  A greater amount of time would also allow for more diffusion, osmosis, and imbibition, all of which may be contributing to higher TDS.        The increase in major ion concentrations is not always constant and there are occurrences at some sites where there is a spike in major ion concentration or TDS (e.g., well C-4; Fig. 3.4c).  99  The spike in concentration tends to be consistent for all major ions.  One possible explanation for an increase in major ion concentrations is a shut-in period within the flowback period, as was observed near the end of the flowback period for well B-1 and during the flowback period for well C-4.  Other possible factors causing a brief change in the major ion concentrations are changes in the flowrate, where an increase in the flowrate can be related to a decrease in the major ion concentrations (e.g., well C-1), and changes in pH, where a decrease in pH correlates with a decrease in major ion concentrations (e.g., well D-1).  In some wells the cause of the spike in concentration is not apparent (e.g., well D-4) and presumably due to factors that were not quantified in the present study.  The additional variables not measured as part of our study include the type of fracture network, fracture surface area, the SRV, and the rate of imbibition, among others (see Table 3.1).   3.4.4 Comparison of upper and middle Montney member Flowback Water Due to the lateral variability in flowback water chemistry across the study area, it is difficult to compare or isolate the variation in flowback water between the different Montney members. Of the sites included in the study, only sites C and D have wells completed in both the upper and middle Montney members and can be used to directly compare the flowback water from these members of the formation. For site C wells, there is some overlap between the major ion concentrations in the flowback water of the different wells but the flowback water from the middle Montney member generally has higher concentrations of the major ions than that from the upper Montney member  (Fig. 3.8a).  However, the major ion concentrations for flowback water from well C-5 (upper Montney member) are within the range of or higher than the major ion concentrations measured at the two site C middle Montney member wells, well C-6 and well C-7.  The longer shut-in time for well C-5 of 30 days may contribute to the higher major ion 100  concentrations in the flowback water from this well.  The shut-in times for the remaining site C wells are between 21 and 28 days.  The other completion parameters (e.g., number of stages, volume of injected fluid) are similar for wells at site C.  Site D shows a more distinct difference in chemistry between flowback waters collected from the wells completed in the upper and middle Montney members (Fig. 3.8b).  Conversely to the site C wells, the upper Montney member flowback water from wells D-1 and D-2 have higher TDS and higher major ion concentrations relative to the middle Montney member flowback water from wells D-3 and D-4.  The two upper Montney member wells at site D have slightly longer shut-in times of 4 and 7 days relative to the middle Montney member wells at this site, which have 1 and 2 day shut-in periods.  The difference in the length of the shut-in period likely contributes to the variability in flowback water chemistry, making it difficult to discern if the member of the formation where the well was completed has any effect.   A comparison of the upper Montney member flowback water major ion chemistry between sites C and D shows that the chemistry is similar between the wells at these sites, with the site D TDS and major ion concentrations generally being slightly lower overall (see Table 3.2).  In addition, the rate of increase of the major ion concentrations over the flowback period is similar for these wells.  The middle Montney member flowback water major ion chemistry, as well as the rate of increase in major ion concentrations, is notably different between the two sites, indicating lateral variability in the flowback water chemistry from the middle Montney member.  The site C middle Montney member wells have higher major ion concentrations and a slower rate of increase over the flowback period, while the site D wells have lower major ion concentrations but show a more rapid increase over time.  The comparison between the site C and site D wells indicates that there is variability in flowback water chemistry 101  between sites even from the same formation member, thus making it difficult to make predictions on flowback water chemistry based on the member where the well is completed. None of the study sites have wells completed in both the lower member and either the upper or the middle member of the Montney formation so a similar comparison with the lower Montney member flowback is not possible.     Figure 3.8: Na concentrations over the flowback period for (a) site C; and (b) site D wells. TDS and the other major ions follow a similar pattern of increasing concentrations over the flowback period.  Wells C-1 through C-5, well D-1, and well D-2 are completed in the upper Montney member.  Wells C-6, C-7, D-3, and D-4 are completed in the middle Montney member. The wells completed in the upper Montney member have similar Na concentrations (site C median: 19,800 mg/L, site D median: 18,500 mg/L), while the Na concentrations for the wells completed in the middle Montney member are significantly different (site C median: 31,000 mg/L, site D median: 7,880 mg/L). Note the difference in the y-axis scales.  a. b. 102  3.5 Flowback water trend analysis 3.5.1 Linear regression and linear mixed effects models A statistical analysis method beyond simply looking at median or maximum values is required to provide an overall comparison of the flowback water results between sites due to the differences in the lengths of the flowback periods and the variable rate of increase of TDS and the major ion concentrations for different wells.  A combination of linear regression and linear mixed effects models were selected to complete a trend analysis of the results for the study wells in order to provide a method for the comparison.  The first analysis investigates the increase in TDS or the major ion concentrations over time since the beginning of the flowback period.  The increasing major ion concentrations do not always fit an exact linear increase; however, a linear model does provide a reasonable approximation over the duration of the flowback period for which the wells were sampled.  For example, the results for the correlation between TDS and flowback day show an overall moderate positive correlation for the study wells (median R2 = 0.85).  The major ion concentrations in later samples would be expected to eventually reach a plateau as the flowback chemistry approaches the formation water chemistry unless mixing is not the dominant influence on the chemistry.  If water of condensation is the dominant liquid source later in the production life of the well the slope would theoretically become negative.  A non-linear relationship would require a different model.  An examination of the slope coefficients derived from the linear regressions of TDS or the major ion concentrations as a function of flowback time showed enough similarity to group the wells by site and use a linear mixed effects model to derive an overall fixed effects slope coefficient for each site (Fig. 3.9).  The one exception was the two site I wells which were found to have significantly different slopes for the 103  parameters of interest as a function of time and are thus considered separately for the analysis using the slope coefficients from the linear regression for each of the wells.    Figure 3.9: An example of the linear fixed effects model with data from site E.  The fixed effects line represents the overall fixed effects for the 4 wells at this site and the slope of this line can be used to compare different sites.  The differences in the slope for the individual wells can be described by the random effects.    The second analysis looks at the increase in flowback volume over time.  Some of the wells (e.g., site C wells) show a linear increase in volume over the sampled flowback period; however, other wells show an increase in flowback that is more rapid early in the flowback period and begins to slow down over time.  Due to the change in behavior of the increasing cumulative flowback volume over the flowback period and the lack of either a consistently linear or non-linear trend for the study wells, the results were divided into three periods for the analysis: 1) early (day 0-3); 2) mid (day 3-15); and 3) late (> day 15) flowback.  Each of these segments was treated as a linear regression to derive the slope coefficient for the flowback volume as a function of time.  The differences in slope coefficient when each well was examined 104  individually did not allow for the wells to be grouped by site as the variation was too great.  Simple linear regression was therefore used for the analysis.  The site A wells were the only wells that were grouped together and analyzed with a linear mixed effects model to obtain a single value for the site, although the use of this model was due to limited samples available for each well.   3.5.2 Results of flowback water chemistry trend analysis The slope coefficients derived from the linear regressions for the two site I wells and from the linear mixed effects models for the remaining sites are used to compare the different sites by considering the change in TDS or the concentration of a particular major ion as a function of time.  The analysis does not directly consider other variables, such as differences in injected fluid chemistry or shut-in time, and it is only used to provide a method to compare the sites based on the increase in TDS and the major ion concentrations over the flowback period.  For some of the sites similarity in the overall fixed effects slope coefficients indicates that TDS and the major ions increase at the same rate over the flowback period. The similarity in slopes are hence a variable by which the sites can be grouped (Table 3.3).  The wells located in the central region of the study area can consistently be divided based on the rate of increase into: Group I – sites E, F, and G with a slower rate of increase; and Group II – sites B, C, and D with a more rapid rate of increase over the flowback period for the parameters of interest (Fig. 3.10a).  There is not a distinct difference in the slope coefficients for sites B through G for Mg concentrations, indicating that some elements can increase at relatively similar rates for all the sites in the central region.  Of the major ions included in the analysis, the Sr concentrations show the greatest variability in the rate of increase between the wells (Table 3.3, Fig. 3.10b) and will be discussed in Section 3.5.4.  The wells located in the northwestern (sites A and I) and the 105  southeastern (site H) regions of the study area do not consistently fall within one of the groups in Table 3.3, indicating that there may be some regional effect on the flowback water chemistry.  Since the slope coefficient is based on the increase of TDS or major ion concentrations over time, differences in the length of the flowback period do not impact this value as long as a linear fit is observed; however, eventually the concentrations are expected to reach a plateau and a non-linear fit would be required.   Table 3.3: Groups of sites based on similarities in slope coefficient.  Group I – slowest rate of increase, Group II – intermediate rate of increase, and Group III – fastest rate of increase. No distinction could be made between the sites for the rate of increase in Mg over the flowback period to divide the majority of wells between Group I and II. Parameter Group I Group II Group III TDS E-G, I-1 A-D, I-2 H Cl A, E-G, I-1 B-D, I-2 H Na E-G A-D, I-1, I-2 H Ca E-G, I-1 B-D, I-2 A, H K E-G A-D, I-1, I-2 H Mg A-G, I-1, I-2 - H Sr A, E-G B-D, H, I-1 I-2      106   Figure 3.10: Slope coefficients with confidence intervals for a) TDS; and b) Sr concentrations over flowback time.  For TDS, the sites can be divided by the rate of increase into slowest – Sites E, F, G, and I-2; intermediate – sites A, B, C, D, and I-2; and fastest – site H.  These divisions are similar for most of the major ions (see Table 3.3).  b) For Sr, the sites can be divided into slow – Sites A, E, F, and G; intermediate – Sites B, C, D, H, and well I-1; and fast – Well I-2.  The error bars show the confidence intervals of the slope coefficients. Wells are considered to have similar slope coefficients when their confidence intervals overlap with the calculated value of a different site.  If only the confidence intervals overlap it does not indicate a great degree of similarity between sites.  The large confidence interval for the site A slope coefficient is due to the limited data for wells from this site.    3.5.3 Results of flowback water volume trend analysis In order to complement the trend analysis of the TDS and the major ion concentrations over flowback time, the change in cumulative flowback volume over time was investigated.  Several studies have used the production rates of flowback water to gain insight into fluid flow and the fracture system in the reservoir (e.g., Clarkson, 2013; Kanfar and Clarkson, 2016; Williams-Kovacs and Clarkson, 2016).  An analysis using these methods to gain additional a. b. Slow Slow Intermediate Intermediate Fast Fast 107  information from the flowback water volume is beyond the scope of this paper and the present study only includes a preliminary discussion on flowback production over time. The results of the analysis show that the slope coefficient for the regression of the volume of flowback water as a function of time is variable both between wells and for the different portions of the flowback period for the same well (Fig. 3.11).  For the majority of wells, the rate of increase in flowback production is most rapid in the early flowback period (day 0-3).  However, well B-1, well C-2, well C-4, well H-1, and well I-1 are exceptions where the rate of cumulative flowback increases more rapidly in the mid flowback period (day 3-15).  In all cases where the flowback period extends beyond 15 days, the rate of change in flowback production is reduced, although only the site B, site G, and site I wells were sampled for longer than 15 days.    108   Figure 3.11: Summary of the slope coefficients for the cumulative flowback volume as a function of flowback time for each of the wells for the early (day 0-3), mid (day 3-15), and late (> day 15) portions of the flowback period.  A lower slope coefficient indicates a slower flowrate over the portion of the flowback period relative to a higher slope coefficient.  In general the rate of change in cumulative flowback production is most rapid in the early flowback period.  The site A wells are grouped together using a linear mixed effects model to obtain an overall slope coefficient for this site as the data from this site was limited.  The remaining wells are presented separately as the slope coefficients (derived from a simple linear regression model) were generally not sufficiently similar to group together to use a linear mixed effects model.  Wells without a slope coefficient for one of the periods had two or fewer data points that fell within that period.  3.5.4 Implications of the trend analyses The analysis comparing the increase of TDS and the major ion concentrations as a function of time provided a method to group the data by site using the slope coefficients from 109  linear mixed effects models.  The similarity in slope coefficients for the groups of sites summarized in Table 3.3 argues for similar changes in the rate of mixing between the injected hydraulic fracturing fluid and the formation water for the sites within each group.  The similar change in rate could be related to a combination of the formation water chemistry, the permeability of the formation, and the extent of both natural and induced fractures at the sites. Greater similarity in slope coefficients would be expected between nearby wells since there would likely be less variability in the properties of the formation over a shorter distance.  Sites E, F, and G are located in close proximity (< 15 km from each other), were completed in the middle Montney Formation, and have comparable slope coefficients; however, site D, located within 15 km of these sites, has slope coefficients more similar to sites B and C.  The difference may be caused by variation in the fracture networks that developed at the different sites, including how rapidly the microfractures close as reservoir pressure declines during flowback.  The variability between the site D and the sites E through G wells shows that the increase in TDS and the major ions over the flowback period can vary even between nearby wells.  It is of interest to note that site D includes two wells each completed in the upper and middle Montney Formation.  All four wells from site D had similar slope coefficients, indicating a comparable rate of mixing which would be consistent with a similar formation water chemistry and fracture network.  The higher concentrations in the flowback water from wells D-1 and D-2 (upper Montney member) were interpreted to be due to a higher proportion of formation water in the initial flowback water (see Chapter 2), likely related to the longer shut-in times of 4 and 7 days for these wells relative to the wells completed in the middle Montney member (wells D-3 and D-4; 1 and 2 days).   As mentioned above, the major ion which showed the greatest variability in the trend analysis is Sr. The Sr slope coefficients were highest for well I-2 (Fig. 3.10b), whereas the slope 110  coefficients for the other major ions for this well were similar to the Group II wells (Table 3.3).  The inconsistency may be due to variability in the Sr concentrations in formation water. The Sr concentrations in the formation water near sites A and I (northwestern region) are higher (median: 1350 mg/L) relative to the central and southeastern areas in the study (Fig. 3.12), accounting for the rapid increase in Sr concentrations for well I-2.  The lower slope coefficients indicating a slower increase in Sr concentrations at site A and well I-1, which may be due to more limited mixing between the injected fluid and the formation water.     The Sr slope coefficient for well H-1 in the southeastern region was also notable as the rate of increase is slower relative to the increase in TDS and the other major ion concentrations at this well over the flowback period.  The slope coefficient for Sr at well H-1 is similar to sites B-D and well I-1 (Group II).  The Sr concentrations in formation water near the site H well (115 mg/L) are much lower than in the central and northwestern areas which may result in a slower rate of increase over the flowback period; however, there are only results available for one produced water sample from the southeast region.  The concentrations of the other major ions in formation water do not show as great a variation as the Sr concentrations, which likely explains the more consistent results for the other major ions. 111   Figure 3.12: Sr concentrations in produced water samples, used to approximate the formation water chemistry.  Median Sr concentrations are higher in produced water samples from wells in the northwestern region near site A and site I.  However, the Sr concentrations from site A show overlap with the Sr values in wells nearby sites B-F.  Sr concentrations are significantly lower in produced water samples nearby sites G (central region) and H (southeastern region). Note: Only one sample was available with Sr concentrations nearby sites H and I. The box portion of the boxplot shows the 25th percentile, the median, and the 75th percentile.  The whiskers show the interquartile range (IQR).  The points represent potential outliers which fall outside the IQR.  The amount of variability in the increase in flowback volume over time at a single well and between different wells does not lead to any obvious correlations between the flowback water production rate and the flowback water chemistry.  The initial rapid increase in the volume of flowback water produced over time observed for most wells is likely related to the initial flowback from the larger aperture fractures.  When the production rate increases faster later in the flowback period, as was observed for well B-1, well C-2, well C-4, well H-1, and well I-1, it 112  indicates that areas of greater permeability, such as more permeable lenses, were accessed later in the flowback period.   A higher slope coefficient for the flowback volume as a function of time would indicate a more rapid increase in flowback water production from a well.  The differences in the rate of flowback volume produced are likely related to the permeability in the subsurface due to both the created and existing fracture networks, which were not measured as part of the current study.  The results of the current analysis do not consistently support a correlation between a more rapid rate of flowback water production and a faster increase in TDS as similar changes in TDS were observed for both a more rapid and a slower rate of increase in flowback water volume over time.  During the early flowback period, for example, wells with a higher rate of production, with slope coefficients greater than 300, were found to have TDS increase over this period by a factor of between 1.1 and 3.0.  Wells with a slower rate of production and slope coefficients between 100 and 250 had a similar increase in TDS over the same portion of the flowback water period, with an increase by a factor of 1.0 to 3.6.  A lack of a distinction in the rate of flowback water production and the increase in TDS suggests that the relative rate of flowback water produced over time cannot be used to predict the increase in TDS at a well.  The variability between the different wells illustrates the complexity of the hydraulic fracturing process and the numerous variables that can impact both the flowback chemistry and the flowback volume.   A further, more detailed analysis of the data at the individual well level with consideration of the different variables that may be impacting the flowback water volume and chemistry is needed to further investigate the relationship between the flowback water chemistry and flowrate.   113  3.6 Conclusions The challenges surrounding the interpretation of flowback water chemistry are a result of the changes in the TDS and the major ion concentrations over the flowback period as well as due to the many variables that may impact the chemistry.  A simple comparison of the maximum flowback water ion concentrations between wells is therefore not possible without considering the length of the flowback period and the volume of flowback water collected, among other factors.  Some of the variables (e.g., shut-in time) were presented here as an explanation of the differences in flowback water chemistry between wells.  A more detailed analysis of the relative importance of the different variables on flowback water chemistry will be considered in a subsequent part of the thesis (Chapter 4). By taking the potential concerns with comparing flowback water chemistry into consideration, we were able to analyze the Montney Formation water chemistry and present a method to compare the changing chemistry between wells.  The main conclusions from our study are as follows:      The concentrations of the major ions increase over the flowback period, although the rate of increase varies between different sites and sometimes between the wells on the same site.  The increasing major ion concentrations are likely due to mixing between the relatively low TDS hydraulic fracturing fluid and the high TDS formation water with several potential variables affecting the rate of increase at different sites.    The flowback water major ion concentrations were generally similar for wells completed at the same site and within the same member of the Montney Formation.  The two site I wells were an exception to this with well I-2 showing a more rapid increase in major ion concentrations over the flowback period relative to well I-1.  The difference in the 114  flowback water between these two wells may be due to the longer shut-in time, additional hydraulic fracturing stages, and/or a more extensive fracture network at well I-2 causing more efficient mixing between the injected fluid and the formation water at this well.    The trend analysis provides a method for grouping the wells based on the increase in the major ion concentrations over flowback time.  The majority of the major ions fall within the following groups: slow rate of increase – sites E-G, well I-1, site A (Cl, Sr); moderate rate of increase – sites B-D, well I-2, site A (TDS, Na, K); and rapid rate of increase – site H, site A (Ca).  The groups of wells based on the Sr concentrations differed from these groups with the most rapid rate of increase observed for well I-2.  This may be related to more variability in Sr concentrations in the formation water across the study area as well as the efficiency in mixing at different sites.  When the upper Montney Formation flowback water was examined, a possible trend in the Sr concentrations increasing from southeast to northwest was identified; however, additional sampling of wells in the southeast and northwest regions is necessary to confirm the trend.  There was no obvious correlation with the rate of production of flowback water and the flowback water chemistry.  However, a more detailed analysis is recommended to further investigate the relationship between these two variables.    115  Chapter 4: Flowback water chemistry from the Montney Formation: Part II – Controls on variability  4.1 Introduction Unconventional reservoirs represent an important hydrocarbon resource.  These reservoirs are developed with a combination of horizontal drilling and hydraulic fracturing, which requires several thousand cubic meters of fluid per well to be pumped into the formation (Alessi et al., 2017) in order to fracture the rock and increase the permeability to access the hydrocarbons in the formation.  Prior to the production phase of the well, the well produces a fluid with high total dissolved solids (TDS) – the flowback water6.  In some wells the flowback period begins immediately after hydraulic fracturing; however, in many wells there is a time lag where the well is shut-in before flowing back.  The shut-in period can last for less than a day up to several months (e.g., Makhanov et al., 2014).  The volume of flowback water varies between different formations but typically only represents a fraction (approximately 25%) of the injected fluid volume (Haluszczak et al., 2013).  The flowback water represents a mixture of the injected hydraulic fracturing fluid and the formation water (e.g., Haluszczak et al., 2013; Engle and Rowan, 2014; Vengosh et al., 2017), with the proportion of formation water increasing over the flowback period.  The increase in the fraction of formation water is reflected in the increase in the TDS and the major ion concentrations in flowback water, including chloride (Cl), sodium (Na), potassium (K), calcium (Ca), magnesium (Mg), and strontium (Sr).  The chemistry, as well                                                  6 For the purposes of this study, the flowback water includes the fluids produced from the well following hydraulic fracturing and up until the well is placed on production. 116  as the rate of increase in the TDS and the major ions, of flowback water can provide insight into processes occurring in the reservoir.  However, there are many different variables that are related to the well completion, the formation properties, and the response of the well to hydraulic fracturing that can impact the flowback water chemistry (Table 4.1).    Table 4.1: List of several of the variables that may potentially be impacting the flowback water chemistry. Several of the variables listed are interrelated (modified from Chapter 3).  Reservoir properties Completion design Response of reservoir to completion   Mineralogy   Porosity, permeability, and fabric of reservoir   Amorphous precipitates   Presence and complexity of natural or induced fractures   Water saturation (Sw), Oil saturation (So)   Reservoir wettability   Reservoir capillary pressure   Imbibition potential   Reservoir temperature and pressure   Formation water chemistry and rheology   Depth of burial   Paleo depth of burial   Inherent and induced stress and stress anisotropy   Amount of uplift and degassing    Volume of hydraulic fracturing fluid pumped   Chemistry and rheology of hydraulic fracturing fluid   Length of horizontal well   Number of stages   Clusters per stage   Perforations per cluster   Orientation of perforations   Shut-in time   Proppant embedment   Effectiveness of proppant      Length of flowback period   Rate of flowback   Volume imbibed into reservoir   Complexity of induced fractures   Surface area contacted by the completions   Stimulated Reservoir Volume (SRV)   Water-rock interactions   Reservoir damage (chemical + physical)   Stress sensitivity of reservoir matrix and natural and induced fracture system    Fracture closure pressure(s)   Breakdown pressure     In order to determine the utility of flowback water as a ‘window’ into the reservoir or as a tool for interpreting completion efficiency, it is critical to understand the numerous variables that contribute to the composition and volume of flowback. Examining the flowback water chemistry from several wells can help determine which variables have the greatest influence on chemistry. 117  Most previous studies have considered flowback water from single wells which does not provide the opportunity for understanding or deciphering the contribution of the plethora of variables that contribute to the chemistry of the flowback fluids.  Many variables that may impact the flowback water chemistry should be considered, most notably the variables that influence the mixing process (i.e., hydraulic fracturing fluid chemistry and formation water chemistry) and the water-rock interactions (i.e., formation mineralogy and fabric).  The following paper acts as a continuation of the work presented in Chapter 3.  The study includes flowback water samples from 31 wells completed in the Montney Formation in the Peace River Basin in British Columbia (BC) and Alberta, Canada (Fig. 4.1).  The wells are located on 9 sites, with between 1 and 8 sampled wells per site.  The Montney Formation was deposited along the northwestern coast of Pangea during the Early Triassic epoch (Davies, 1997).  The deposits represent a marine inner to distal shelf environment and are dominantly dolomitic-siltstones and very fine grained sandstones (Chalmers and Bustin, 2012; Zonneveld et al., 2011; Zonneveld and Moslow, 2014).  The most recent informal subdivision of the Montney Formation, which is used in our study, divides the formation into three members based on a combination of biostratigraphy and lithostratigraphy (Davies and Hume, 2016; Fig. 4.2).  Of the study wells, 18 wells were completed in the upper member, 11 wells in the middle member, and 2 wells in the lower member.  The flowback water from these wells was analyzed for the major ions – Cl, Na, K, Ca, Mg, and Sr.  Chapter 3 focused on comparing the flowback water chemistry between the different wells.  As multiple wells are sampled, the second part of the study presented here uses the flowback water data to investigate the dominant variables that are impacting the flowback water chemistry using regression analyses and to discuss the inferences 118  that can be made with reference to the reservoir properties, completions, and formation water chemistry.    Figure 4.1: Location of the 31 study sites in BC and Alberta, Canada.  The Montney Formation is shown in purple (modified from Edwards et al., 1994).  Site A – 8 wells, site B – 2 wells, site C – 7 wells, site D – 4 wells, site E – 4 wells, site F – 2 wells, site G – 1 well, and site H – 1 well.        119   Figure 4.2: Schematic cross-section through the Montney Formation from the western margin in BC to the eastern margin Alberta.  The Montney Formation is divided into upper, middle, and lower informal members based on lithology and biostratigraphy.  The upper Montney member includes the lower Doig siltstone (modified from Davies and Hume, 2016).  4.2 Methods 4.2.1 Fluid chemistry The fluids analyzed for the study wells include the injected hydraulic fracturing fluid, the flowback water, and the produced water.  Hydraulic fracturing fluids were either collected as one sample per fracturing stage (sites A, B, D, E, G, and H wells) or as a representative sample for the well (sites F and I wells).  For the wells where several hydraulic fracturing fluid samples were collected, composite samples were created by mixing samples with similar electrical conductivity readings.  An overall chemistry for the hydraulic fracturing fluid for a well was calculated from the composite sample chemistry based on the volumetric proportions of the 120  fluids used in the different fracturing stages along the lateral of the well.  For example, if two composite samples were analyzed and each represents an equal volume of fluid used in the combined stages then the results from each of the samples will be given equal weight in the overall chemistry of the hydraulic fracturing fluid for the well.  The hydraulic fracturing fluid samples were not available for the site C wells. Flowback water samples were collected at all sites.  The flowback water was sampled at a higher frequency in the initial flowback period (two to three samples per day) then the sampling frequency was decreased to one sample per day.  The sampling schedule was designed to sample the flowback water more often at the beginning of the flowback period when the flowrate is typically higher and the chemistry is expected to show more variability.  Produced water samples were collected at site A, site B, and a well located in the same region as site I.  These samples represent the fluids produced by the well following the flowback period once the well is in the production stage.  For the remaining wells, the produced water chemistry was estimated using publically available data7 for nearby wells completed in the Montney Formation.  The produced water is used in our study to provide an approximation of the formation water chemistry of the Montney Formation as it is not possible to directly sample the formation water.  The produced water does not represent 100% formation water as there would be some influence on the water from the injected fluid as well as from potential geochemical reactions (e.g., mineral precipitation and dissolution).                                                   7 The publically available data was obtained from the information accessed through the geoSCOUT® fluid analysis database. 121  All fluid samples were stored at 4°C.  The flowback and produced water samples were heated to 75-80°C for a period of 24 hours to represent reservoir temperatures.  Heating the samples was conducted to re-dissolve any precipitate that formed after sampling as sample filtering and preservation was not conducted at the time of sample collection.  All samples were filtered with a 0.45 µm filter and subdivided for: 1) anion analysis by ion chromatography; 2) pH (Omega® PHB21 potable pH meter), conductivity (HACH® CDC401 conductivity probe), and alkalinity by titration with sulfuric acid (HACH® digital titrator, Model 16900); and 3) dissolved metals by inductively coupled plasma-optical emission spectrometry (ICP-OES).  The dissolved metal samples were preserved with trace metal grade nitric acid (HNO3) to a pH < 2.  Prior to analysis with the ICP-OES (Varian® 725-ES) the dissolved metal samples were acid digested to remove organics using a 2:1 ratio of HNO3 and hydrochloric acid (HCl) followed by a second digestion step with HNO3 only.  A more detailed description of the acid digestion process is included in Chapter 3.  TDS is calculated as a sum of the ion concentrations.  The anion analysis was conducted at a laboratory in Victoria, BC and the remaining analyses were conducted at laboratory facilities at The University of British Columbia in Vancouver, BC.   4.2.2 Hydraulic fracturing metrics The volume of hydraulic fracturing fluid, shut-in time, number of stages, and breakdown pressure used for each well were obtained from the completion programs for the study wells provided by the operators. 4.2.3 Montney Formation Mineralogy None of the completed wells in our study were cored and hence mineralogy was not directly obtainable from the completed zones.  An estimate for the mineralogy of the completed strata for each well was made by compiling publically available mineralogy determined by 122  X-Ray Diffraction (XRD) from nearby wells which were sampled at similar depths to the fractured zones of the study wells. In areas where public data were absent, additional samples were obtained from the BC Oil and Gas Commission Core Research Facility in Fort St. John, BC for XRD analyses.  The supplementary samples included both drill cutting samples and core samples and were analyzed at UBC.  The samples were crushed in ethanol using a mortar and pestle in preparation for smear mounts on glass slides for XRD analysis on a Bruker D8 Focus, following the modified smear mount method in Munson et al. (2016). The mineralogy was quantified by the Rietveld method (Rietveld, 1967, 1969). 4.3 Montney Formation flowback water chemistry A detailed description of the Montney Formation flowback water chemistry was presented in Chapter 3 and will be summarized briefly here.  The dominant ions in the sampled Montney Formation flowback water are Cl and Na (Table 4.2).  The concentrations of Ca, K, Mg, and Sr are also high.  The concentrations of these six major ions show an increasing trend over the flowback period; however, the rate of increase often changes over the flowback period, typically with a slower rate of increase as the flowback period progresses.  The increase in flowback water TDS is due in large part to mixing between the injected hydraulic fracturing fluid and the formation water (e.g., Haluszczak et al., 2013; Olsson et al., 2013; Engle and Rowan, 2014; Vengosh et al., 2017).  The rate of increase is dependent on the efficiency of the mixing process which is likely related to the development of the fracture system, as a higher permeability and more complex fracture network would result in a greater contribution from formation water and a more rapid increase in TDS in the flowback water.  Diffusion and/or osmosis between the formation and the injected fluid may also contribute to the high TDS (Balashov et al., 2015; Zolfaghari et al., 2016; Wang et al., 2016; 2017) as well as formation 123  water expelled from the formation due to countercurrent imbibition.  The possibility of countercurrent imbibition is considered as imbibition of the hydraulic fracturing fluids into the formation has been shown to be an important process in hydraulically fractured, low permeability reservoirs (Engelder et al., 2014).  Experiments have shown that a shale placed in water results in an increase in water salinity (Zolfaghari et al., 2016).  From these results, we infer that as the hydraulic fracturing fluid is imbibed into the rock, there is movement of the formation water out of the rock into the fractures, in addition to the movement of oil and/or gas.  Fluid mixing would occur in the fractures between the remaining hydraulic fracturing fluid and the formation water which was expelled from the rock.  The mixing process would be ongoing from the initial stage of hydraulic fracturing and throughout the flowback period. The impact of diffusion and osmosis on flowback water chemistry cannot be differentiated from the imbibition process and mixing based on the available data.  Table 4.2 (following page): Summary of the flowback water TDS and major ion chemistry for the sampled wells. Wells at the same site and that were completed in the same member of the Montney Formation are included together.  The organic compounds were not measured in the flowback water samples which may contribute to higher charge balance errors.  The charge balance error for the majority of samples is ±15%.        124  Site n   TDS Cl Na Ca K Mg Sr mg/L mg/L mg/L mg/L mg/L mg/L mg/L upper Montney Formation               A (8 wells) 22 min 41,358 25,722 8,264 1,616 553 296 298 median 61,220 43,437 12,759 3,059 796 547 630 max 92,731 65,777 17,818 5,389 1,239 941 1,220 B (2 wells) 30 min 30,135 18,541 9,162 1,789 284 154 60 median 89,949 55,235 25,609 5,950 1,035 619 404 max 228,259 164,018 51,027 11,705 1,920 1,369 924 C (5 wells) 52 min 34,137 21,344 9,932 1,360 569 191 194 median 70,449 44,439 19,779 3,088 1,025 402 477 max 147,839 104,897 33,212 7,333 1,500 681 916 D (2 wells) 19 min 31,522 19,303 9,960 1,169 466 171 123 median 62,520 38,355 18,489 2,874 860 418 264 max 80,944 50,963 23,974 4,058 1,172 611 358 H (1 well) 18 min 20,713 12,308 6,451 629 414 102 30 median 83,807 49,006 28,509 2,798 1,534 650 141 max 112,226 66,998 37,637 3,677 1,806 809 195 middle Montney Formation               C (2 wells) 20 min 82,519 52,570 23,802 3,638 1,065 441 573 median 110,450 71,218 30,960 5,310 1,461 669 893 max 140,538 93,007 38,460 7,332 1,865 945 1,263 D (2 wells) 22 min 3,609 1,893 1,344 13 49 10 4 median 21,631 12,510 7,884 628 257 79 55 max 56,800 34,565 18,872 2,112 466 223 268 E (4 wells) 26 min 11,105 6,671 2,787 350 187 62 38 median 18,672 12,252 5,256 641 295 107 72 max 32,864 21,611 9,119 1,126 474 220 161 F (2 wells) 12 min 18,655 10,295 6,849 560 166 93 33 median 24,009 14,455 9,066 702 196 117 44 max 34,139 21,495 10,575 795 251 136 69 G (1 well) 8 min 13,972 8,186 3,896 174 186 74 29 median 38,963 24,790 10,999 845 410 273 73 max 65,052 41,783 18,244 2,229 650 586 101 lower Montney Formation                I (2 wells) 28 min 4,587 2,260 1,991 110 105 32 13 median 31,998 17,678 12,112 1,032 558 166 206 max 90,178 51,038 33,782 4,081 1,296 524 1,011 125  There is considerable variability between sites and even between wells at the same site (e.g., the site I wells).  A comparison between multiple wells is difficult due to the variability in the lengths of the flowback periods for different wells as well as the variables related to the completion and formation properties that may be influencing the flowback water chemistry (Table 4.3). The length of the shut-in period in particular was observed to impact the flowback water chemistry, with longer shut-in periods correlated to higher ion concentrations in the initial flowback water.  The many factors that may be influencing the flowback water chemistry also pose a challenge in comparing the flowback water chemistry between wells completed in the three different members of the Montney Formation.  Lateral variability in lithology within a single member also occurs.  The number of wells available for sampling as part of the present study was not sufficient to provide a detailed analysis of lateral variability within formation members, as the majority of sites were in the central region of the formation (6 of 9 sites).   As a direct comparison between wells was not ideal, a method to assess the rate of increase in TDS and the major ions in flowback water was presented in Chapter 3 and is summarized here.  The method used linear regression and linear mixed effects models to compare the flowback water chemistry at different sites using slope coefficients to group wells.  It was found that in general, the wells in the central region could be divided into two groups – sites E, F, and G with a slower rate of increase in major ion concentrations as a function of time and sites B, C, and D with a faster rate of increase.  There was some indication that nearby sites show greater similarity (sites E, F, and G); however, site D is near these sites and shows a more rapid increase in the major ion concentrations.  The rate of increasing concentrations of the different ions for wells in the northwestern region (sites A and I) did not consistently fall into either of these groups. Well H-1 in the southeastern region generally had the most rapid rate of 126  increase for TDS and all ions studied except for Sr concentrations.  The difference in the rate of increase in Sr concentrations was interpreted to be due to more variability in the Sr concentrations in formation water across the study area relative to the other major ions.  A preliminary interpretation of an increasing trend in Sr from the southeast to the northwest was suggested which causes the rate of increase at site H (southeast) to be slower relative to the other major ions, while the rate of increase for well I-2 (northwest) was more rapid than for other major ions.  An examination of the rate of production of flowback water over time did not reveal any obvious correlation with the flowback water chemistry.   4.4 Dominant variables on flowback water chemistry The trend analysis in Chapter 3 provided an initial examination of the Montney Formation flowback water chemistry results; however, the additional variables which may influence the flowback water chemistry were not taken into account.  In the second part of our study, a further assessment of selected variables is conducted.  The variables that were initially intuitively considered to have an effect on flowback water chemistry were assessed by looking at their correlation with TDS and the major ions (Cl, Na, K, Ca, Mg, and Sr).  The variables of interest include:  Hydraulic fracturing fluid volume;  Hydraulic fracturing fluid chemistry;  Number of stages;  Breakdown pressure;  Shut-in time;  Formation water chemistry; and  Formation mineralogy. 127   The data for these variables for each of the wells are summarized in Tables 4.3, 4.4, and 4.5.  The effects of imbibition on the flowback water chemistry are not included in the analysis. The closest proxy for imbibition of the results collected in the present study is the percent of the injected fluid recovered; however, the use of this parameter in the current analysis is not appropriate due to the division of the data based on the cumulative flowback volume.  The division of the data in this way holds the cumulative flowback volume constant; therefore, a sample with a higher total injected volume would have a lower percent recovered due to the nature of the data rather than as a reflection of the amount of imbibition.  A brief discussion on the use of percent recovered as a proxy for imbibition is included in Section 4.4.3.6.   In order to examine the importance of the selected parameters on flowback water, two sets of flowback data were examined: 1) initial samples; and 2) the samples collected closest to 1,000 m3 of cumulative flowback water for each well.  The second dataset includes samples collected between approximately 800 to 1,200 m3 of cumulative flowback volume and includes the majority of wells (n=26).  Well A-1, well A-8, well E-3, and the two site F wells produced less than 800 m3 of flowback water at the end of the flowback period and are not included.  The initial samples represent early flowback water while the samples collected near 1,000 m3 of cumulative flowback water are taken to represent intermediate to late flowback water.  Many of the wells had over 1,000 m3 of flowback water; however, only 15 wells had greater than 1,500 m3 of cumulative flowback and of the 15 wells 6 are from a single site (site C).  Therefore a robust regression analysis for the higher flowback water volumes was not possible with the limited data from the present study.          128  Table 4.3: Summary of data for the variables considered in the regression analysis. Site Well Vol. of HF fluid injected (m3) No. of Stages HF fluid  vol. / stage (m3/stage) Breakdown pressure (MPa) Shut-in time (days) Vol. of flowback water (m3) Percent of injected vol. recovered A 1 10,650 15 710.0 44.3 26 607 6% 2 9,439 20 472.0 39.4 4 2,421 26% 3 10,192 22 463.3 41.9 12 811 8% 4 9,879 20 493.9 37.3 12 1,130 11% 5 11,027 20 551.3 40.4 13 1,037 9% 6 11,762 15 784.1 46.5 24 923 8% 7 10,579 20 528.9 42.8 7 1,492 14% 8 11,209 22 509.5 35.0 14 720 6% B 1 11,070 24 461.2 36.0 4 3,678 33% 2 11,791 24 491.3 36.9 2 3,488 30% C 1 16,222 30 540.7 49.9 28 1,884 12% 2 16,243 30 541.4 47.9 23 1,583 10% 3 13,892 29 479.0 42.4 26 2,114 15% 4 16,114 29 555.7 47.1 21 2,696 17% 5 13,972 29 481.8 49.9 30 1,447 10% 6 18,088 30 602.9 49.1 24 2,177 12% 7 20,175 29 695.7 53.1 22 2,115 10% D 1 6,902 16 431.4 36.2 4 954 14% 2 7,501 14 535.8 40.8 7 2,040 27% 3 6,573 16 410.8 36.5 2 988 15% 4 6,899 14 492.8 37.7 1 1,286 19% E 1 7,743 20 387.2 37.7 5 1,735 22% 2 7,636 20 381.8 35.7 5 1,630 21% 3 7,254 20 362.7 37.4 8 601 8% 4 7,440 20 372.0 36.1 6 1,124 15% F 1 4,841 14 345.8 n/a 2 390 8% 2 4,789 14 342.1 n/a 1 257 5% G 1 8,642 25 345.7 37.5 1 1,743 20% H 1 8,623 26 331.7 n/a 3 1,488 17% I 1 23,078 24 961.6 44.9 1 5,126 22% 2 26,292 30 876.4 45.5 5 4,269 16%   129  Table 4.4 (following page): Summary of hydraulic fracturing fluid chemistry and formation water chemistry for each of the wells. The majority of the hydraulic fracturing fluid samples are Na-Cl type fluid, which is the same as the flowback water, although some are Ca-Na-Cl type.  Two of the hydraulic fracturing fluid samples from site H are Na-HCO3 type. There was often a charge imbalance for the hydraulic fracturing fluid samples which may be related to interference from the additives. The formation water results are medians from publically available produced water data for nearby wells and produced water samples collected in our study.  Units are mg/L, except for pH.                  130  Site Well pH TDS Cl Na Ca Mg K Sr Hydraulic Fracturing Fluid              A 1 6.8 15,285 10,276 3,212 1,004 162 181 161 2 6.5 16,802 10,423 3,873 1,428 227 226 245 3 7.3 239 71 69 42 22 21 <10 4 6.7 25,297 18,108 4,555 1,566 256 255 309 5 6.7 28,072 19,850 5,365 1,791 274 273 332 6 6.7 16,824 11,365 3,486 1,052 205 224 217 7 7.2 3,386 966 1,733 420 59 63 59 8 6.7 25,382 17,495 5,046 1,724 261 280 295 B 1 7.5 22,018 6,276 11,973 2,828 299 379 160 2 7.3 23,459 6,432 12,980 3,067 284 417 175 D 1 7.3 549 160 208 46 10 11 1 2 7.1 1,284 639 398 58 12 46 2 3 7.2 711 289 209 50 10 10 1 4 7.1 4,717 2,044 1,605 142 29 241 6 E 1 6.5 11,246 6,414 3,659 608 96 187 59 2 6.6 5,052 3,150 1,237 322 53 96 28 3 6.4 6,059 3,842 1,546 323 53 89 16 4 6.8 8,464 4,740 2,788 453 79 150 47 F 1, 2 7.0 12,218 7,448 3,948 372 60 105 25 G 1 7.2 5,604 1,841 3,047 305 64 137 40 H 1 5.2 710 7 548 57 23 11 <1 I 1 7.7 1,849 1,204 66 74 16 8 <1 2 7.7 1,863 760 67 77 16 8 <1 Formation Water                 A  7.1 123,912 84,352 27,713 7,715 1,088 1,347 1,365 B  5.1 181,416 128,159 33,203 15,182 1,492 1,980 1,048 C  6.4 121,840 74,512 37,467 6,260 822 1,373 1,094 D - F  6.9 100,545 61,735 27,799 6,569 917 2,211 918 G  6.7 78,323 47,476 21,764 5,186 727 1,759 128 H  6.0 153,025 95,568 42,120 10,551 1,877 2,232 115 I   5.9 91,254 89,168 54,000 10,000 1,140 2,040 1,620    131  Table 4.5: Summary of the Montney Formation mineralogy used in the regression analysis. Sites with multiple wells completed at different depths are divided in this table.  Carbonates include ankerite, calcite, dolomite, ferrodolomite, and siderite.  Clays include chlorite, illite/mica, and kaolinite.  Feldspars include albite, microcline, and orthoclase.  Other minor minerals not included in the regression analysis are fluorapatite (present in wells nearby sites C-F <2%) and pyrite (present at all sites ~2%). Site Well Median (Bulk wt. %) Quartz Carbonate Clay Feldspar A 1-8 34.1 34.9 6.9 22.3 B 1 32.3 30.5 15.3 19.5 2 32.8 30.6 16.3 19.1 C 1-3 34.4 28.6 3.8 23.5 4, 5 38.7 30.7 7.5 20.2 6, 7 41.7 28.5 9.9 16.0 D 1 39.3 35.9 6.3 15.5 2 41.3 24.8 8.4 15.9 3, 4 39.6 18.8 16.2 21.0 E 1-4 39.6 18.8 16.2 21.0 F 1-2 40.3 18.0 16.3 20.2 G 1 35.4 17.7 17.3 26.2 H 1 45.1 19.3 6.8 24.4 I 1-2 44.0 15.0 25.0 12.0  4.4.1 Initial sample flowback water chemistry Sampling was initiated within a day of the beginning of the flowback period for the study wells; however, as the flowrate differed between wells, the initial samples were collected at different flowback volumes.  The volume of flowback water produced when the initial samples were collected varies between 0 and 562 m3 which represents 0 to 54% (median: 14%) of the total flowback water from the wells.  The initial flowback water samples are suitable to use to represent the initial flowback water chemistry as there is no trend (R2 < 0.05) between the TDS and the flowback volume for these samples which could create spurious correlations in the analysis.  Positive correlations (R2 > 0.2) were found between the flowback water TDS of the 132  initial samples and the hydraulic fracturing fluid TDS, the shut-in time, and the breakdown pressure (Fig. 4.3).  The major ions in flowback water show similar relationships with these variables (Table 4.6). The hydraulic fracturing fluid volume per stage and the number of stages have a weak correlation (R2 < 0.2) with the initial flowback water chemistry for the parameters included in the analysis.    Figure 4.3: Relationship between the initial sample flowback water TDS values and (a) the HF (hydraulic fracturing) fluid TDS; (b) the shut-in time; and (c) the breakdown pressure for the study wells.  TDS is provided as an example of the parameters included in the study.  The site C wells are excluded from (a) as the HF fluid was not sampled at these wells.  The site F and site H wells are excluded from (c) as the breakdown pressure data was not available for wells from these sites.       a. b. c. 133  Table 4.6: R2 values for the initial sample flowback water chemistry and the variables of interest.  R2 values greater than 0.2 are in bold italic text, values greater than 0.5 are shaded in grey.  Negative correlations are denoted by (-). Parameter TDS Cl Na Ca Mg Sr K HF fluid chemistry 0.31 0.42 0.08 0.32 0.34 0.53 0.04 HF fluid volume per stage 0.09 0.10 0.05 0.08 0.11 0.15 0.08 Shut-in time 0.59 0.55 0.59 0.57 0.47 0.55 0.69 Number of stages 0.10 0.08 0.18 0.07 0.00 0.01 0.12 Breakdown pressure 0.41 0.37 0.50 0.31 0.17 0.24 0.42 Formation water chemistry 0.11 0.03 0.01 0.00 0.00 0.21 0.55 (-) Chemistry of last flowback sample 0.26 0.25 0.21 0.48 0.23 0.67 0.38 Median carbonate 0.47 0.47 0.34 0.53 0.67 0.57 0.52 Median clay 0.37 (-) 0.34 (-) 0.35 (-) 0.36 (-) 0.43 (-) 0.37 (-) 0.50 (-) Median feldspar <0.01 (-) <0.01 (-) 0.01 (-) <0.01 (-) 0.01 0.01 <0.01 (-) Median quartz 0.05 (-) 0.05 (-) 0.02 (-) 0.12 (-) 0.20 (-) 0.16 (-) 0.05 (-)  A step-wise regression was completed in order to identify the variables in the multiple linear regression that were significantly contributing to the correlation with the flowback water chemistry.  This was completed using the step function in R (R Core Team, 2017) using the linear equation: 𝑦 =  𝛽ଵ𝑥ଵ + 𝛽ଶ𝑥ଶ + ⋯ +  𝛽௡𝑥௡ + 𝜀 Where, y is the flowback water chemistry as TDS or a major ion concentration, x1 to xn are the values of the variables of interest, β1 to βn are the regression coefficients of the variables of interest and ε is the error term.  The variables included in the regression analysis are those from Table 4.6, excluding the chemistry of the last sample and the mineralogy which are both considered in separate analyses (see Sections 4.4.3.3 and 4.4.3.5).  As a result of the lack of data for some of the variables at different sites, the analysis was completed on four datasets (Table 4.7).  Once the significant 134  variables were identified for each of the datasets, the relative importance of the variables was determined using the Lindeman et al. (1980; lmg) method in the relaimpo package in R (Grömping, 2006).  This package takes into account the different scales of the variables in order to compare them.  A similar analysis looking at the relative importance of variables impacting the flowback water volume was completed by Zhou et al. (2015).   Table 4.7: Summary of the datasets used in the step-wise regression analysis in order to considered all sites and all variables of interest. For the initial flowback water samples, the results of the step-wise regression were similar for datasets I and II as well as datasets III and IV, due in part to the inclusion of the hydraulic fracturing fluid chemistry in datasets I and II. Dataset Sites included Sites excluded Variables excluded I A, B, D-I C Breakdown pressure II A, B, D, E, G, I C, F, H None III A-I none HF fluid chemistry, Breakdown pressure IV A-E, G, I F, H HF fluid chemistry  The regression analysis indicates that shut-in time is the most important of the variables considered in the study on the initial flowback water chemistry, accounting for between 38 and 69% of the variance in the flowback water chemistry (Fig. 4.4).  The hydraulic fracturing fluid chemistry was also found to be correlated with flowback water concentrations of Cl, Ca, Mg, and Sr (Fig. 4.4a).  The formation water TDS was found to have some influence on the flowback water TDS; however, this variable was only found to be important for datasets I and II (Fig. 4.4a).  Formation water and the number of hydraulic fracturing stages were found to have minor importance for Ca and Mg concentrations when datasets III and IV were used (Fig. 4.4b). The number of stages has a negative coefficient in the multiple regression model for Mg and Sr concentrations in flowback water; however, a simple linear regression with this parameter shows no correlation with the flowback water chemistry (R2 < 0.05; see Section 4.4.3.4 for discussion).   135   Figure 4.4: Relative importance of variables found to be significant (p-value < 0.05) in the step-wise regression analysis. a) Results based on dataset I. Results using dataset II are similar; b) Results based on dataset III when hydraulic fracturing (HF) fluid was excluded in order to include the site C samples.  Results from dataset IV are similar.  Shut-in time was determined to be the most important variable influencing the initial sample flowback water chemistry.  The change in the R2 value for Na concentrations between a and b is due to including the site C samples which all have a longer shut-in period resulting in a strengthening of the correlation.  These results were calculated using the R package "relaimpo" (Grömping, 2006).  The total height of each bar corresponds to the R2 value.    The relationship between the initial flowback water chemistry and the formation mineralogy is complex since the mineralogical data is compositional.  Spurious correlations exist within a compositional dataset due to the data being closed (i.e. adds to 100%); resulting in a change in one parameter causing a change in the other parameters (e.g., Chayes, 1960; Aitchison, 1982; Pawlowsky-Glahn and Egozcue, 2006).  For example, in the mineralogy dataset used in the current study, the median percentages of carbonate and clay are negatively correlated (R2: 0.62) making it difficult to determine if the observed correlations are related to having more a. b. 136  carbonate or less clay in the formation.  Since the positive correlation with percent carbonate (Table 4.6) are seen for all of the examined parameters and not only the elements found in carbonates (Ca, Mg), the positive correlation of the initial flowback water chemistry and the median carbonate content is not interpreted to be due to carbonate dissolution.  The correlation between clay and the major ion concentrations in particular is influenced heavily by the results for the two site I wells which have relatively low major ion concentrations in the initial flowback water and the highest proportion of clay.  The median quartz and feldspar values are not strongly correlated with the flowback water chemistry (R2 ≤ 0.2).    4.4.2 Intermediate to late flowback water chemistry The flowback water sample closest to 1,000 m3 of cumulative flowback water at each of the wells was selected to represent intermediate to late flowback water chemistry for the regression analysis.  The R2 values for the simple linear regressions are presented in Table 4.8.  The parameters with higher R2 values (> 0.2) are generally the same as those for the initial samples although overall the correlations are slightly lower (see Table 4.6).  The correlations with the formation water chemistry and the intermediate-late flowback water chemistry remain low (R2 ≤ 0.30) but are higher than the correlation with the initial sample flowback water chemistry.        137  Table 4.8: R2 values for the intermediate-late flowback water chemistry and the variables. R2 values greater than 0.2 are in bold italic text, values greater than 0.5 are shaded in grey.  Negative correlations are denoted by (-).  The step-wise regression was completed following the same method that was used for the initial sample flowback water chemistry.  The same four datasets were used, without well A-1, well A-8, well E-3, or the site F wells due to low flowback water recovery from these wells (Table 4.7).  The results of the analysis indicate that the shut-in time remains an important variable impacting flowback water chemistry (Fig. 4.5); however, its relative importance is less than was found with the initial flowback sample dataset.  The hydraulic fracturing fluid chemistry remains an important variable for Sr, Na, and Mg concentrations (Mg in dataset II only) and the formation water chemistry becomes an important variable for TDS, Cl, Ca, and Mg concentrations.  There is some correlation (R2 > 0.2) between the flowback water chemistry and breakdown pressure for most of the parameters examined but this variable was only found to be significant (p-value < 0.05) in the multiple regression with Ca concentrations when dataset IV was used (Fig. 4.5d).  Other parameters that are significant but have a low relative importance in Parameter TDS Cl Na Ca Mg Sr K HF fluid chemistry 0.20 0.20 0.14 0.47 0.31 0.55 0.02 HF fluid volume per stage <0.01 <0.01 <0.01 <0.01 <0.01 0.06 <0.01 Shut-in time 0.41 0.43 0.29 0.39 0.24 0.56 0.38 Number of stages 0.13 0.10 0.21 0.08 0.02 0.04 0.16 Breakdown pressure 0.29 0.27 0.32 0.21 0.14 0.34 0.38 Formation water chemistry 0.30 0.09 0.05 0.05 0.09 0.14 0.21 (-) Chemistry of last flowback sample 0.44 0.38 0.57 0.66 0.64 0.68 0.70 Median carbonate 0.36 0.41 0.17 0.48 0.57 0.57 0.32 Median clay 0.35(-) 0.35(-) 0.25 (-) 0.32 (-) 0.44 (-) 0.39 (-) 0.45 (-) Median feldspar <0.01 (-) <0.01 <0.01 <0.01 0.03 0.01 <0.01 Median quartz 0.03 (-) 0.04 (-) <0.01 0.12 (-) 0.12 (-) 0.16 (-) <0.01 138  certain datasets are the number of stages and the volume of hydraulic fracturing fluid per stage (Fig. 4.5).  Both of these variables have a weak positive correlation (R2 < 0.1) when considered alone with the flowback water concentrations and have negative coefficients in the multiple regressions (see Section 4.4.3.4 for discussion). The correlation between the major ion concentrations in flowback water and the formation mineralogy for the intermediate to late flowback water dataset is similar to the correlations for the initial sample dataset. The major ions are positively correlated with median percent carbonate (0.17 ≤ R2 ≤ 0.57), negatively correlated with median percent clay (0.25 ≤ R2 ≤ 0.45) and show weak to no correlation with median percent quartz (0.01 ≤ R2 ≤ 0.16) or feldspar (R2 < 0.05).    139   Figure 4.5: Relative importance of variables for intermediate to late flowback water. a) Dataset I – site C and breakdown pressure excluded; b) Dataset II – sites C, F, and H excluded. The importance of the hydraulic fracturing (HF) fluid chemistry for Mg concentrations in flowback water in this dataset may be due to some correlation between the Mg concentration in the HF fluid and the Mg concentration in the formation water (R2: 0.22), which changes the relative importance of these two parameters for this dataset; c) Dataset III – HF fluid chemistry and breakdown pressure excluded; d) Dataset IV – Site F, site H, and HF fluid chemistry excluded. The relative importance of the shut-in time is higher when the site C wells are included (c, d) which is likely due to the longer shut-in time for these wells having an impact on the correlation.  Correlation between flowback water chemistry and the number of stages as well as the HF fluid volume per stage (d) are negative.  These results were calculated using the R package "relaimpo" (Grömping, 2006).  The total height of each bar corresponds to the R2 value. a. b. c. d. 140  4.4.3 Important variables influencing flowback water chemistry 4.4.3.1 Shut-in time Shut-in time was found to be an important variable influencing flowback water chemistry and although the relative importance is higher early in the flowback period, it continues to be correlated with the later flowback water chemistry even after 1,000 m3 of flowback water has been produced from a well.  The elevated TDS and major ion concentrations in the initial flowback water from wells with longer shut-in periods are interpreted to be related to a longer time for mixing between the injected hydraulic fracturing fluid and the formation water as well as for water-rock interactions prior to initiating flowback from the well. Higher concentrations at the beginning of the flowback period result in a greater volume of flowback water with higher ion concentrations over the course of the flowback period.  Having higher TDS and higher major ion concentrations in the flowback water is an important consideration for flowback water management as it may require additional treatment for reuse or disposal (e.g., Fontenelle et al., 2013; Notte et al., 2016). The correlation between the ion concentrations and the shut-in time in the later flowback period would be expected to decrease over time as the ion concentrations are expected to be influenced by other variables throughout the flowback period, including the flowrate, the fracture complexity, and the formation water chemistry.   4.4.3.2 Hydraulic fracturing fluid chemistry Another consideration for flowback water management is the use of blended freshwater and recycled flowback water for use for hydraulic fracturing fluids.  The higher TDS of the blended fluids influences the initial flowback water chemistry.  A positive correlation with the hydraulic fracturing fluid chemistry was anticipated as the initial flowback water is expected to be composed of a relatively high proportion of the injected fluid due to less time for fluid mixing 141  and any water-rock interactions that are occurring in the subsurface.  The hydraulic fracturing fluid chemistry was identified as a significant variable (p-value < 0.05) for Cl, Ca, Mg, and Sr concentrations in the initial flowback water samples collected for the present study.  The lack of a correlation between the Na and K in the hydraulic fracturing fluid and the initial flowback water is unexpected (R2: 0.08 and 0.04 for Na and K, respectively).  Several of the initial flowback water samples plot close to a 1:1 line between the hydraulic fracturing fluid chemistry and the flowback water chemistry for these ions, indicating that there may be a correlation at some sites.  The majority of the initial flowback water samples have higher Na and K concentrations that are not correlated to the chemistry of the injected fluid, as the concentrations in the initial flowback water are often high when the concentrations in the injected fluid are low.  This relationship may indicate that Na and K concentrations are increasing due to cation exchange, which would increase the concentrations of these ions in the initial flowback water due to exchange with the divalent ions on the exchange sites of the clay minerals.  The cation exchange process would be initiated by the injection of the relatively fresh hydraulic fracturing fluid into the high TDS fluid present in the formation (e.g., Appelo, 1994).  The Montney Formation has low cation exchange capacity (CEC; Table 4.9).  There is some variability in CEC in the available results (range: 9.30 to 20.4 cmol(+)/kg, median: 17.3 cmol(+)/kg), indicating that the relative importance of the cation exchange process could vary between wells due to differences in formation mineralogy and fluid chemistry.  Overall, the CEC is not correlated to the percentage of clay in samples from the Montney Formation (R2 = 0.01); however, when examined separately both the exchangeable Na and K ions correlate with the percentage of clay (R2 = 0.71) indicating a dependence on mineralogy for these two ions.     142  Table 4.9: Cation exchange capacity (CEC) of samples from the Montney Formation. These samples were selected for the CEC analysis based on the variable clay content in order to determine if the CEC is correlated to the percentage of clay. Samples are ordered by increasing clay.  Sample % Clay (by wt.) Exchangeable Cations and Effective CEC by 0.1 N Barium Chloride Extraction (cmol(+)/kg) CEC Ca Na K Mg 1 4.2 17.8 16.3 0.6 0.2 0.7 2 6.2 13.3 11.3 1.1 0.3 0.5 3 6.4 18.7 17.1 0.8 0.2 0.5 4 7.4 9.3 7.5 1.2 0.2 0.4 5 8.5 20.3 16.8 1.4 0.4 1.7 6 8.5 18.2 14.3 1.5 0.8 1.6 7 10.7 18.4 16.5 1.2 0.3 0.4 8 10.9 20.4 18.1 0.9 0.5 0.9 9 11.7 14.1 11.0 1.8 0.6 0.8 10 13.5 20.3 17.4 1.4 0.6 0.8 11 17.1 14.7 10.6 2.9 0.6 0.5 12 17.5 13.8 9.4 2.6 0.8 0.9 13 17.7 14.5 11.3 1.8 0.7 0.7 14 18.5 17.2 13.5 1.8 0.9 1.0 15 20.3 15.8 11.4 2.7 0.9 0.7  4.4.3.3 Formation water chemistry  The formation water chemistry TDS, Cl, Ca, and Mg concentrations have significant correlations with the intermediate-late flowback water chemistry in the multiple regressions while the other major cations in flowback water do not show significant correlations with the formation water chemistry (Fig. 4.5).  Intuitively, the formation water chemistry is expected to be one of the main variables impacting the flowback water chemistry since mixing between the injected fluid and the formation water is anticipated to be a dominant process influencing the flowback water chemistry (e.g., Haluszczak et al., 2013; Olsson et al., 2013; Engle and Rowan, 2014; Vengosh et al., 2017), particularly in the later flowback period when the proportion of formation water is higher. The lack of a correlation between the flowback water chemistry and the formation water chemistry for Na, K, and Sr could be related to a greater importance of other 143  parameters, such as shut-in time, or other processes not considered in the present study, including cation exchange or a stronger dependence on lateral variability between sites.  The low variability in the formation water chemistry between sites and the uncertainty in the estimated values for the formation water chemistry used in the analysis may also contribute to the low correlation between the formation water and the flowback water for TDS and the major ion concentrations. The chemistry of the last flowback water sample collected at each of the wells is used in a secondary analysis as a proxy for formation water, in as much as this sample has the highest percentage of formation water for each of the wells.  The correlation between the formation water proxy and the flowback water chemistry is higher relative to the correlation between the formation water chemistry estimated from the produced water results and the flowback water chemistry for both the initial and intermediate-late data sets (see Tables 4.6 and 4.8).  For the secondary analysis looking at the initial sample chemistry, a multiple regression was completed with the shut-in time, the hydraulic fracturing fluid chemistry, and the last sample chemistry as the variables. The shut-in time remains the variable with the highest relative importance (> 40%; Fig. 4.6a).  The relative importance of the injected fluid and the last sample chemistry is lower (< 25%). The intermediate to late flowback water chemistry, approximated by the 1,000 m3 sample for each well, shows a higher relative importance for the last sample chemistry (30% to 58%; Fig. 4.6b). There is a significant increase in the relative importance of this variable from the initial sample chemistry, supporting the increasing importance of formation water chemistry on the flowback water chemistry as the flowback period progresses.  However, there is some complication added to the analysis due to using samples from the same site as both an 144  independent variable (the 1,000 m3 sample chemistry) and as one of the dependent variables (the last sample chemistry).  An earlier and a later sample from the same site would be expected to have some correlation since the processes affecting the early sample would also have some effect on the late sample.  The shut-in time remains an important variable for this analysis (18 to 35% relative importance). In order to investigate the importance of the formation water chemistry in the absence of the influence of the shut-in time, the initial sample major ion concentrations were subtracted from the intermediate-late (1,000 m3) flowback water major ion concentrations.  However, the correlation with formation water was not significantly increased for the modified data.            145   Figure 4.6: Relative importance of parameters, including the last sample chemistry as a proxy for the formation water chemistry for a) the initial flowback water samples; and b) the intermediate-late flowback water samples. The low correlation seen for Na concentrations (adjusted R2: 0.42) in (a) is due to low correlations between the initial sample flowback water Na concentrations and both the hydraulic fracturing fluid (R2: 0.08) and Na concentrations in the last samples from each of the wells (R2: 0.03), which is due to several samples with relatively low initial Na concentrations and high Na concentrations in the last sample.  For some of the other major ions, these samples also have low concentrations in the initial sample and high concentrations in the last sample, although they do not have as great of an effect on the overall correlation.  4.4.3.4 Additional variables The breakdown pressure was only found to have a significant correlation with Ca concentrations in intermediate to late flowback water when dataset IV is used (Fig. 4.5d).  A higher breakdown pressure may create a more extensive and more complex fracture network due to higher pressure conditions.  The more extensive fracture system would allow for increased mixing and a higher surface area exposed for more potential reactions with the injected hydraulic fracturing fluid (e.g., Bearinger, 2013; Zolfaghari et al., 2015a).  A higher breakdown pressure a. b. 146  would also relate to higher stress in the formation so that during the flowback period more microfractures would close and decrease the water to rock ratio.  When considered alone, there is a positive correlation between the breakdown pressure and both the initial and intermediate-late flowback water chemistry (Tables 4.6 and 4.8).   This correlation may be due in part to the moderate correlation between breakdown pressure and shut-in time (R2: initial: 0.56; intermediate-late: 0.62), rather than a significant correlation with the flowback water chemistry since this variable was not significant in the majority of the multiple regressions.  The correlations between the Ca and Mg concentrations with the number of hydraulic fracturing stages and the hydraulic fracturing fluid volume per stage are negative in cases where these variables are significant (p-value < 0.05) (Fig. 4.4, 4.5).  When considered individually, these parameters show a weak positive correlation or no correlation with the flowback water chemistry for the initial samples or 1,000 m3 of flowback water samples (see Tables 4.6 and 4.8).  The low correlation indicates that it is not a relevant parameter and that the change in sign of the correlation with the number of hydraulic fracturing stages and the hydraulic fracturing fluid volume per stage in the multiple regression may be due to instability in the model due to the weak correlation or due to minor correlation with other variables.  For example, both the shut-in time with the number of stages and the breakdown pressure with the hydraulic fracturing fluid volume per stage showed some correlation (R2 > 0.2).       4.4.3.5 Montney Formation Mineralogy The relationship between the Montney Formation mineralogy and the flowback water chemistry is discussed separately due to the added complexity of using closed compositional data in multiple regressions. To obtain a better understanding of the relationship between the formation mineralogy and the flowback water chemistry, both the concentrations of the major 147  ions and the ratios of the major cations to Cl were considered. The monovalent ion (Na, K) concentrations have a negative correlation with median percent (by weight) clay (Fig. 4.7a); while the Na/Cl and K/Cl ratios have a positive correlation with median percent clay (Fig. 4.7c).  The correlation with median percent carbonate has the opposite sign for both the concentrations and the ratios.  The divalent ions (Ca, Mg, and Sr) have a negative correlation with median percent clay for both the concentration of the ion and the ratio with Cl (Fig. 4.7b, d).  The correlations with median percent carbonate are positive. The increasing monovalent cation to Cl ratios indicate that the monovalent cation concentrations are increasing more rapidly than the Cl concentrations as the clay content increases whereas the divalent cation concentrations are increasing at a slower rate than the Cl concentrations.  These relationships indicate the occurrence of cation exchange.  When the relatively low salinity hydraulic fracturing fluid is injected into the formation containing the saline formation water, the relatively mobile Na ions would be displaced from the cation exchange sites for Ca ions (e.g., Appelo, 1994) producing an increase in Na concentrations and a decrease in Ca concentrations.  An increase in exchangeable Na and K ions with increasing percentage of clay is supported by the cation exchange results for the Montney Formation (Table 4.9).  The relationship between the ion ratios and the carbonate content could be due to the inverse relationship between clay content and carbonate content rather than a result of a relationship between the ratios and the carbonate content.   148   Figure 4.7: Correlations between (a) Na concentrations; (b) Ca concentrations; (c) Na/Cl; and (d) Ca/Cl with median clay. The Na, Ca, and Cl concentrations are from the initial sample data.  The 1000 m3 data shows similar results.  The ion ratios are molar ratios.  Median clay was calculated based on existing data from nearby wells completed near the same stratigraphic interval of the Montney Formation.  4.4.3.6 Overall correlation The multiple regressions do not fully characterize the flowback water system (adjusted R2 values < 0.85) indicating that there are other variables that are not considered in the analysis and/or the relative importance of studied variables varies by well.  Several additional variables were not evaluated in the present study but are likely important (see Table 4.1) including the extent and complexity of both the natural and induced fracture systems, the imbibition rate, and a. b. c. d. 149  the reservoir porosity, permeability, and fabric. Percent recovered could act as a proxy for the amount of imbibition, as a higher percent recovered would correlate with a lower imbibition rate in the flowback period.  A lower imbibition rate would be expected to produce less formation water through countercurrent imbibition and result in flowback water with a lower salinity.  The current dataset was examined at different times throughout the flowback period and shows only a weak negative correlation between percent recovered and TDS for the flowback water samples (e.g., correlation for the day 3 dataset [R2 = 0.09] and for the day 7 dataset [R2 = 0.10]).  These results indicate that the amount of imbibition may not have an important influence on the flowback water chemistry or its importance may be obscured within the data for percent recovered by additional factors, including the length of the shut-in period and the flowrate.   A potential source of uncertainty in the study is that particular variables may only be important at certain sites or certain wells.  For example, the number of stages may be important at one site but not all sites.  If this is the case, there would be no correlation for the full dataset and the variable would not be identified in this type of analysis. The potential for inconsistency in the variables influencing the flowback water chemistry illustrates the complexity of the controls on the system. More in-depth studies may identify additional variables that should be considered and would provide a more complete understanding of the influences on flowback water chemistry.   4.5 Conclusions  The analysis of the flowback water from multiple wells from the same formation allows for an investigation into the variables that impact the flowback water chemistry.  The influence of these variables complicates the interpretation of flowback water chemistry data although working towards a more complete understanding of the controls on flowback water chemistry 150  will permit further interpretations to be made on the processes that are occurring in the formation during and following hydraulic fracturing.  The main conclusions from the multiple regression analysis are:  Of the variables considered in the current study, those identified as important to the flowback water chemistry were the shut-in time, the hydraulic fracturing fluid chemistry (Cl, Ca, Mg, Sr – for the initial flowback), and the formation water chemistry (TDS, Cl, Ca, Mg – for the intermediate to late flowback).  Using the last sample flowback water chemistry to approximate the formation water chemistry shows that the shut-in time decreases in importance over the flowback period, while the relative importance of the last sample flowback water chemistry, as a proxy for formation water, increases.  The positive correlation observed between the Ca and Mg concentrations in flowback water and the percentage of carbonates in the formation is not interpreted to be due to carbonate dissolution as the other major ions (e.g., Na) are also positively correlated with percent carbonate.    The ratios of the monovalent and divalent cations to Cl show opposing trends with the percentage of clay.  The monovalent cation ratio shows a positive correlation while the divalent ratio shows a negative correlation.  This result indicates that cation exchange is occurring and increasing in importance with higher proportions of clay.  The regression analysis provides an initial step in identifying important variables impacting the flowback water chemistry.  Further work should focus on identifying and quantifying additional variables that may influence the flowback water chemistry in order to develop more complex models.   In addition, some parameters may only be affecting the flowback water chemistry at certain wells and would therefore not show a correlation 151  when the full dataset is examined. More in-depth studies at an individual well level or controlled laboratory experiments may identify other variables.  152  Chapter 5: Key minor elements in Montney Formation flowback water: Barium, boron, and lithium  5.1 Introduction Following hydraulic fracturing of unconventional oil and gas wells, fluid returns to the surface.  The initial liquid is referred to as flowback water8 and has elevated total dissolved solids (TDS) due to the high concentrations of the major ions, including chloride (Cl), sodium (Na), calcium (Ca), potassium (K), magnesium (Mg), and strontium (Sr) (e.g., Haluszczak et al., 2013; Rowan et al., 2015; Ziemkiewicz and He, 2015).  If the well is hydraulically fractured with a freshwater based fluid, the ion concentrations normally increase from the beginning of the flowback period due to mixing between the injected hydraulic fracturing fluid and the formation water, possibly with some influence from mineral precipitation and dissolution (Wilke et al., 2015; Dieterich et al., 2016; Harrison et al., 2017; Marcon et al., 2017) and ion exchange (Renock et al., 2016; Zolfaghari et al., 2016).  The majority of the discussion surrounding the inorganic chemistry of flowback water has focused on the major ions (e.g., Haluszczak et al., 2013; Rowan et al., 2015; Rosenblum et al., 2017).  However, additional ions are commonly included as part of the geochemical analysis and can be used to gain further information from the flowback water.  Barium (Ba), for example, is often high in flowback water (> 100 mg/L) and shows more variability between wells in a particular formation relative to the major cations (e.g., Chapman et al., 2012; Barbot et al., 2013).  Other ions, including boron (B) and lithium (Li),                                                  8 Flowback water is defined in our study as the water that returns to surface after hydraulically fracturing a well and prior to the production phase of the well.  153  may be useful in distinguishing the fluid from hydraulically fractured wells from surface water or from water produced from conventional oil and gas wells, which can be accomplished by examining the B/Cl and Li/Cl ratios and the B and Li isotopic values of the fluids (Warner et al., 2014; Vengosh et al., 2015).  Additional studies characterize the naturally occurring radioactive materials (NORM) (e.g., Nelson et al., 2015, 2016) and trace elements in flowback water (e.g., Phan et al., 2015), both of which may be important to consider for the storage of flowback water as there is potential for impacts on freshwater resources.  The present study focusses on Ba, B, and Li concentrations in flowback water from the Lower Triassic Montney Formation in northeast British Columbia (BC) and northwest Alberta.   These three minor ions were selected for further investigation as these ions are often elevated (> 10 mg/L) in the Montney Formation flowback water sampled in our study and may show more variability between different regions or between different members of the Montney Formation compared to the major ions.  High concentrations of these elements are not unique to Montney Formation flowback water as they are commonly elevated in flowback water from other hydraulically fractured formations (e.g., Haluszczak et al., 2013; Rowan et al., 2015; Phan et al., 2016).   The Montney Formation is dominantly dolo-siltstone and fine-grained sandstone (Zonneveld et al., 2011; Zonneveld and Moslow, 2014), deposited along the northwestern coast of Pangea in a marine inner to distal shelf environment (Davies, 1997; Chalmers and Bustin, 2012).  The most recent stratigraphic study on the Montney Formation divides the formation into lower, middle, and upper informal members, each representing a Third-Order sequence (Davies and Hume, 2016; Fig. 5.1).  The division into these three informal members is used in the present study.  Flowback water samples for this study were obtained from two lower Montney member 154  wells, eleven middle Montney member wells, and eighteen upper Montney member wells.  The wells are located on nine well pads across the formation with one to eight wells sampled per pad (Fig. 5.2).  The current study includes a discussion of Ba, B, and Li concentrations in Montney Formation flowback water in support of the analysis of the major ion chemistry of the flowback water from these wells provided in Chapters 2, 3, and 4.  Figure 5.1: Schematic cross-section of the Montney Formation in BC and Alberta. The formation is divided into the lower, middle, and upper Montney Formation informal members based on three Third-Order sequences, determined by lithostratigraphy and biostratigraphy. As defined by Davies and Hume (2016), the upper Montney member includes the lower Doig siltstone and underlies the Doig Phosphate Zone. This member thins to the east and is absent in eastern locations in Alberta.  Overall, the formation becomes coarser grained moving from west to east, towards the paleo-shoreline.  Modified from Davies and Hume (2016). 155   Figure 5.2: Location of the study sites in northeastern BC and northwestern Alberta.  Fluid samples were collected from 9 sites in the Montney Formation. Sites with upper Montney member wells include site A (8 wells), site B (2 wells), site C (5 wells), site D (2 wells), and site H (1 well).  The sites with middle Montney member wells include site C (2 wells), site D (2 wells), site E (4 wells), site F (2 wells), and site G (1 well).  The lower Montney member wells include the two wells at site I. The outline of the Montney Formation is modified from Edwards et al. (1994). 5.2 Methods 5.2.1 Hydraulic fracturing fluid, flowback water, and produced water chemistry Hydraulic fracturing fluids, comprised of the base fluid with additives, were collected from all sites, except for site C.  For sites A, B, D, E, G, and H a separate sample of the fluid used in each of the stages of the fracturing program was collected (14 to 26 samples per well).  Samples from consecutive stages with similar electrical conductivity readings were combined in equal proportions to make composite samples.  Combining samples with similar electrical conductivity reduces the number of analyses required while still allowing for an assessment of 156  the variability in chemistry between the fluids used in different stages of the hydraulic fracturing process.  The Ba, B, and Li concentrations for each well were determined by calculating the volumetric proportion represented by the stages in the composite samples and using this to determine the overall concentration. For example, if three composite samples were analyzed and each represents an equal volume of hydraulic fracturing fluid, the geochemical results from each of these samples is given equal weight in calculating the overall value.  Only one hydraulic fracturing fluid sample was obtained for the site F wells and each of the site I wells. Flowback water samples were collected from all sites over the flowback period, which varies between the wells from one day up to 33 days.  A higher sampling frequency of two to three samples per day was utilized early in the flowback period in order to capture the variation in fluid chemistry during this time.  After the first week of the flowback period one sample per day was collected.  Produced water samples, from after the well has begun producing oil or gas, were collected from the site A wells, the site B wells, and a well approximately 20 km away from site I.  No produced water samples were collected from the wells at the remaining sites.  Publically available produced water results9 were compiled for wells completed in the Montney Formation within approximately 20 km of the study wells in order to provide an estimate of the produced water chemistry in these regions.  However, these results are generally limited to the major cations and only occasionally included Ba concentrations. The B and Li concentrations were not included in the publically available analyses.                                                  9 The data was compiled by the BC Oil and Gas Commission and accessed using geoSCOUT®. 157  All samples were stored at 4°C prior to analysis.  As filtering and sample preservation was not conducted at the time of sampling, the flowback and produced water samples were heated in Teflon® containers in a hot water bath back to reservoir temperature (75-80°C) prior to the analysis.  This step was conducted to approximate reservoir temperatures and re-dissolve any precipitate that had formed during sample transportation and storage.  The hydraulic fracturing fluids were not heated.  All samples were filtered and subsampled for the separate analyses: 1) anions; 2) pH, electrical conductivity, and alkalinity; and 3) dissolved metals.  The anion samples were analyzed by ion chromatography at an external laboratory in Victoria, BC.  The remaining analyses were conducted at The University of British Columbia (UBC) in Vancouver, BC.  The pH was measured using an OMEGA® PHB21 portable pH meter, the electrical conductivity was measured using a HACH® CDC401 conductivity probe, and the alkalinity was determined by titrating the sample with sulfuric acid (H2SO4) using a HACH® Model 16900 digital titrator.  The dissolved metal samples were preserved with trace metal grade nitric acid (HNO3) to pH < 2.  The preserved samples were acid digested with HNO3 and hydrochloric acid (HCl) as outlined in Chapters 2, 3, and 4 to eliminate interference by the organic molecules in the analysis.  The major cation, Ba, B, and Li concentrations were determined by inductively coupled plasma-optical emission spectrometry (ICP-OES) using a Varian 725-ES ICP-OES. 5.2.2 Montney Formation mineralogy In order to consider the potential impact of water-rock interactions on the Montney Formation flowback water, the Montney Formation mineralogy was investigated.  The study wells were not cored therefore mineralogy is not available directly from these wells.  To obtain an approximation of the mineralogy of the fractured zones of the study wells, publically available X-Ray Diffraction (XRD) data was compiled for nearby wells completed in the same 158  stratigraphic interval of the Montney Formation.  Gaps in the available data were identified near site A and for the upper Montney member at site C and site D.  Additional samples from wells nearby these sites were obtained from the BC Oil and Gas Commission core research facility in Fort St. John, BC.  These samples included both core and drill cutting samples.  The additional samples were prepared following the modified smear mount method (Munson et al., 2016) and analyzed at UBC using a Bruker D8 Focus X-ray powder diffractometer.  The results were processed using the Rietveld method (Rietveld, 1967, 1969).  Identification of trace minerals observed in the core samples from a well near site A was completed on core fragments with approximate dimensions of 1cm x 1cm x 0.5cm using a Philips XL30 scanning electron microscope (SEM) at UBC.     The potential for barite precipitation or dissolution was investigated by examining the barite saturation indices (SI) of the flowback water samples.  No minerals containing B or Li ions were close to saturation in the flowback water.  A negative SI indicates that the mineral is undersaturated and will dissolve in the solution, while a positive SI suggests that the mineral is oversaturated and will precipitate.  Values close to zero show that the mineral is in equilibrium with the solution.  All SI values were calculated using the flowback water chemistry in PHREEQC, version 3.2.0 (Parkhurst and Appelo, 2013) at a reservoir temperature of 75°C. 5.3 Major ion chemistry of the Montney Formation flowback water A brief overview of the major ion chemistry of the Montney Formation flowback water from the study wells is provided in this paper, while a more in-depth discussion of these results has been presented in Chapters 2, 3, and 4 of this thesis. The dominant ions in Montney Formation flowback water are Cl, Na, Ca, K, Mg, and Sr (Table 5.1).  Wells completed at the same site and within the same member of the Montney Formation generally have similar major 159  ion chemistry; however, the many different variables, such as fracture complexity and amount of imbibition, that can impact the flowback water chemistry and result in differences even between wells at the same site with similar completion parameters (see Chapter 3; Chapter 4). Overall, the major ion concentrations are slightly higher in the upper Montney member flowback water, although this division is not distinct between all of the study wells.  Sr concentrations in the upper Montney member flowback water range from high concentrations in the northwestern region to low concentrations in the southeastern region of the study area, although sampling additional wells in the northwestern and the southeastern regions is required to verify the trend.  The other major ions do not show geographic trends across the study area.  The concentrations of the major ions increase over the flowback period for all of the study wells, which is mainly due to mixing between the injected hydraulic fracturing fluid and the formation water (see Chapter 2).  Mixing has previously been suggested as the source of increasing TDS and ion concentrations in flowback water by others (e.g., Haluszczak et al., 2013; Engle and Rowan, 2014; Kondash et al., 2017).  The injected hydraulic fracturing fluids for the wells in our study were composed of a freshwater-based fluid for well A-3, the site C wells, wells D-1 through D-3, well H-1, and the site I wells and a blend of freshwater with recycled flowback water for the remaining site A wells, the site B wells, well D-4, and the sites E, F, and G wells.  The TDS10 of the initial flowback water is either close to or higher than the injected hydraulic fracturing fluid TDS for all wells.  The chemistry of the injected fluid was found to have some influence on the initial flowback water chemistry, with higher injected fluid major ion concentrations resulting in higher initial flowback water concentrations for Cl, Ca, Mg,                                                  10 The ion concentrations are added together to calculate the TDS. 160  and Sr (see Chapter 4).  The length of the shut-in period was determined to be the dominant influence on the flowback chemistry with longer shut-in periods correlating to higher initial TDS and major ion concentrations, likely due to a longer time for mixing in the fractures (see Chapter 4).  The increasing contribution from formation water over the flowback period was calculated for the study wells using Cl and the stable water isotopes (δ18O and δ2H) as conservative tracers (see Chapter 2).  In general, the proportion of formation water for the study wells increases to approximately 60% by the end of the flowback period, based on the 75th percentile calculated using the proportions from the last sample for each well (see Chapter 2).  Although mixing between the hydraulic fracturing fluid and the formation water is interpreted to be the dominant influence on Montney Formation flowback water chemistry, the divalent ions (Ca, Mg, and Sr) are also influenced by ion exchange (see Chapter 2).  This interpretation is based on the lower concentrations of these ions in the early time flowback water due to exchange with Na ions, relative to the concentrations predicted with mixing alone. In contrast to the Cl and major cation concentrations, sulfate (SO4) concentrations are not elevated in all wells (Table 5.1).  In addition, the SO4 concentrations do not show the constant rate of increase that was observed for other ions.  For the study wells, the SO4 concentrations were interpreted to be influenced by pyrite oxidation as well as bacterial SO4 reduction as suggested by others (Engle and Rowan, 2014; Wilke et al., 2015; Harrison et al., 2017).  SO4 concentrations will be discussed further in relation to Ba concentrations in Section 5.4.1. Table 5.1 (following page): Summary of the general chemistry and the major ion, Ba, B, and Li concentrations for the study wells.  The results for wells located on the same site and within the same member of the Montney Formation are grouped together.  In general, the charge balance error falls within ±15% for the flowback water samples from the study wells.  The higher charge balance errors are interpreted to be due to excluding organic compounds from the analysis. 161  Montney Formation member upper middle lower Site A B C D H C D E F G I Parameter No. of wells 8 2 5 2 1 2 2 4 2 1 2 n 22 30 52 19 18 20 22 26 12 8 28 pH min 7.5 6.3 6.6 2.3 6.1 6.5 6.6 6.0 3.4 6.8 6.4 median 7.8 7.0 6.9 6.9 6.3 6.8 7.2 7.4 6.3 7.3 7.1 max 8.1 7.9 7.4 9.2 8.1 7.0 9.5 7.8 6.7 7.6 7.7 Conductivity (mS/cm) min 60.8 47.6 56.1 50.4 33.8 120 7.12 22.1 29.1 27.9 11.1 median 102 123 102 86.4 113 147 36.1 35.7 37.5 64.8 56.1 max 131 190 157 107 135 171 74.6 51.8 50.4 95.4 130 Total Alkalinity (mg CaCO3 per L) min 115 71 95 <10 67 76 135 222 <10 205 172 median 195 119 144 104 80 99 236 337 51 228 232 max 405 188 190 685 128 123 475 395 91 260 403 TDS (mg/L) min 41,358 30,135 34,137 31,522 20,713 82,519 3,609 11,105 18,655 13,972 4,587 median 61,220 89,949 70,449 62,520 83,807 110,450 21,631 18,672 24,009 38,963 31,998 max 92,731 228,259 147,839 80,944 112,226 140,538 56,800 32,864 34,139 65,052 90,178 Cl (mg/L) min 25,722 18,541 21,344 19,303 12,308 52,570 1,893 6,671 10,295 8,186 2,260 median 43,437 55,235 44,439 38,355 49,006 71,218 12,510 12,251 14,455 24,790 17,678 max 65,777 164,018 104,897 50,963 66,998 93,007 34,565 21,611 21,495 41,783 51,038 Na (mg/L) min 8,264 9,162 9,932 9,960 6,451 23,802 1,344 2,787 6,849 3,895 1,991 median 12,759 25,609 19,779 18,489 28,509 30,960 7,884 5,256 9,066 10,999 12,112 max 17,818 51,027 33,212 23,974 37,637 38,460 18,872 9,119 10,575 18,244 33,782 Ca (mg/L) min 1,616 1,789 1,359 1,169 629 3,638 13 350 560 174 110 median 3,059 5,950 3,088 2,874 2,798 5,310 628 641 702 845 1,032 max 5,389 11,705 7,333 4,058 3,677 7,332 2,112 1,126 795 2,229 4,081 K (mg/L) min 553 284 569 466 414 1,065 49 187 166 186 105 median 796 1,035 1,025 860 1,534 1,461 257 295 196 410 558 max 1,239 1,920 1,500 1,171 1,806 1,865 466 474 251 650 1,296 Mg (mg/L) min 296 154 191 171 102 441 9.5 62 93 74 32 median 547 619 402 418 650 669 79 107 117 273 166 max 941 1,369 680 611 808 945 223 220 136 586 524 Sr (mg/L) min 298 60 194 123 30 573 3.7 38 33 29 13 median 630 404 477 264 141 893 55 72 44 73 206 max 1,219 924 916 358 195 1,263 268 161 69 101 1,011 SO4 (mg/L) min 51 <0.1 <0.1 134 575 19.3 98 90 379 1,308 0.7 median 91 59 0.6 189 899 21 126 113 560 1,370 19 max 234 174 26 645 1,084 24 189 202 824 1,407 48 Ba (mg/L) min 1.7 1.9 12 1.0 <1 142 <1 2.1 0.1 <1 1.9 median 13 5.8 52 5.2 1.2 213 3.1 3.9 0.6 <1 46 max 20 7.5 202 7.3 1.4 311 4.9 6.4 1.4 <1 467 B (mg/L) min 6.0 3.1 11 12 13 19 3.1 7.2 6.0 7.4 2.1 median 8.0 10 14 18 33 21 10 11 9.0 13 8.7 max 10 14 20 21 38 24 15 15 12 16 11 Li (mg/L) min 15 11 10 8.3 4.0 34 <1 4.0 4.5 4.5 1.1 median 19 28 16 14 13 43 5.0 6.5 6.3 12 10 max 23 47 31 17 18 49 13 11 7.4 17 22 162  5.4 Key Minor Elements in Flowback Water Similarly to the major ions, the concentrations of Ba, B, and Li are likely significantly influenced by mixing between the injected hydraulic fracturing fluid and the formation water and possibly by water-rock interactions (e.g., Phan et al., 2016; Renock et al., 2016).  The minor ion concentrations are not directly related to the injected hydraulic fracturing fluid chemistry as the concentrations of these ions are generally low (< 10 mg/L) in this fluid and for sites where freshwater is used as the base fluid, the Ba, B, and Li concentrations are < 3 mg/L (Table 5.2).  The variability between sites and the potential sources of Ba, B, and Li in flowback water are discussed in the following subsections.                 163  Table 5.2: Summary of the hydraulic fracturing fluid Ba, B, and Li concentrations for each of the study wells.  These values are derived from the composite fluid samples and are estimates of the overall fluid chemistry based on the volumetric proportion of fluid that was used in each of the hydraulic fracturing stages.  Only one hydraulic fracturing fluid sample was collected for the site F wells and for each of the site I wells.  No hydraulic fracturing fluids were collected for the site C wells.  Site Well Ba (mg/L) B (mg/L) Li (mg/L) A 1 11 2.0 7.1 2 8.8 2.3 8.8 3 <0.2 <0.2 <0.2 4 27 2.4 10 5 32 2.6 11 6 15 2.3 7.8 7 4.9 <1 2.5 8 31 2.5 9.8 B 1 4.8 5.1 13 2 3.1 4.4 10 D 1 1.3 2.2 <1 2 2.0 0.4 <1 3 1.4 0.5 <1 4 2.1 1.3 <1 E 1 4.7 5.0 3.2 2 2.6 2.3 1.5 3 3.0 2.5 1.4 4 4.5 4.3 2.5 F 1, 2 1.4 3.0 2.9 G 1 4.5 5.0 2.9 H 1 0.5 0.4 0.01 I 1 1.0 0.7 <0.2 2 0.7 0.6 <0.2  5.4.1 Barium Within the sampled flowback waters, the Ba concentrations span from below the detection limit (< 1 mg/L) up to 467 mg/L (Table 5.1).  There is generally an increase in Ba in flowback water over the flowback period for each of the study wells with considerable variability between different sites (Fig. 5.3).  The shut-in time has some effect on the Ba concentrations, with a longer shut-in period correlating to a higher Ba concentration in the initial flowback water 164  samples (R2 = 0.37).  The length of the shut-in period for the study wells would influence the chemistry by controlling the amount of time for mixing and geochemical reactions prior to the initiation of flowback from the well.  The overall geographic trend for Ba concentrations across the study area is a decrease from the northwestern region of the study area near sites A and I to the southeastern region near site H.  The site C wells in the central region of the study area are an exception to the trend with higher Ba concentrations (12 to 311 mg/L) relative to wells at nearby sites, which have concentrations that remain < 10 mg/L.  The higher Ba concentrations in flowback water from site C are likely related, at least in part, to the length of the shut-in period for these wells.  Regional variability in Ba concentrations in flowback water has previously been observed in samples from the Marcellus Shale in Pennsylvania, although in the Marcellus Formation flowback water the Ba concentrations often reach concentrations > 1,000 mg/L after one week of flowback (Chapman et al., 2012; Barbot et al., 2013) which is considerably higher than those measured in our study on the Montney Formation. 165   Figure 5.3: Ba concentrations over the flowback period plotted as cumulative flowback volume for a) Site A wells; b) Site B wells; c) Site C wells; d) Site D wells; e) Site E-H wells; and f) Site I wells.  The Ba concentrations increase over the flowback period for the majority of wells.  The upper Montney member wells include the site A wells, the site B wells, wells C-1 through C-5, well D-1, well D-2, and well H-1; the middle Montney member wells include well C-6, well C-7, well D-3, well D-4, the site E and site F wells, and well G-1; the lower Montney member wells include the site I wells. Note the variability in the scales. a. b. c. d. e. f. 166  There is some indication of differences in Ba concentrations in flowback water from the upper and middle members of the Montney Formation when comparing wells completed at the same site.  At site C, the upper Montney member flowback water generally has lower Ba concentrations compared to the middle Montney member flowback water; however, well C-5 which was completed in the upper Montney member is an exception with similar Ba concentrations to the two middle Montney member wells (Fig. 5.3c).  The site D wells show the opposite trend with the upper Montney member flowback water having higher Ba concentrations relative to the middle Montney member flowback water, although there is some overlap over the flowback period for these wells (Fig. 5.3d).  The upper Montney member wells at site D also have longer shut-in periods which likely contribute to the higher concentrations.  The only two wells completed in the lower Montney member are located at site I.  Both of these wells had high Ba concentrations (> 100 mg/L) near the end of the flowback period; however as there are no other wells in this region with comparable cumulative flowback volumes an evaluation of the Ba concentration in flowback water from the lower Montney member relative to another member of the formation in the same region was not possible with the current data.    The regional variability in the flowback water Ba concentrations may be related to spatial variability in Ba concentrations in formation water.  The formation water chemistry in this study is approximated by the produced water results (Table 5.3), as the produced water concentrations are expected to approach those in formation water.  Ba concentrations are high in the publically available produced water results from wells completed within the Montney Formation near site C, as well as in the produced water sample near site I that was analyzed as part of our study (Fig. 5.4), corresponding to the locations of the two sites where the highest flowback water Ba concentrations were measured.  The Ba concentrations in the hydraulic fracturing fluid are low 167  (< 5 mg/L) for the majority of wells and would not have a large impact on the flowback water chemistry.  There are some elevated Ba values (up to 32 mg/L) in the fluids used at site A (Table 5.2).  These higher Ba concentrations could result in higher concentrations in the initial flowback water, although, the later stage flowback water is expected to be influenced mainly by the formation water chemistry and any water-rock interactions occurring.    Figure 5.4: Ba concentrations in produced water samples from wells completed in the Montney Formation and located up to 20 km from the study sites. Results for sites A, B, and I are for samples collected in the present study, results for sites D-H were compiled from publically available produced water results. The box portion of the boxplot shows the 25th percentile, the median, and the 75th percentile.  The whiskers show the interquartile range (IQR).  Site H and site I appear as a line as only one sample which included Ba results was available for these sites.      168  Table 5.3: Summary of produced water Ba, B, and Li concentrations compiled as part of the study.  The results for site A and site B are from produced water samples collected from the study wells.  The results for sites C-H are from publically available results, which did not include B and Li concentrations.  The concentrations for site I were from a produced water sample obtained from a nearby well.  Where multiple samples were analyzed, a median value is given. Site Well n Ba (mg/L) B (mg/L) Li (mg/L) A 1 2 13 10 28 2 1 19 10 32 3 2 18 11 31 4 2 24 10 31 5 3 14 11 30 6 2 14 11 31 7 7 18 13 39 8 1 16 11 28 B 1 4 3 11 33 2 4 2 12 32 C 1-7 5 427 no results D-F all 14 63 G 1 15 0.6 H 1 1 52 I 1, 2 1 321 15 61  The Ba concentrations in both the formation water and the flowback water may be impacted by different geochemical processes in different regions.  Overall, Ba concentrations in the study wells are negatively correlated with SO4 concentrations (Fig. 5.5).  A negative correlation has also been observed in previous studies with Marcellus Shale flowback water and was explained by the occurrence of bacterial SO4 reduction, which would be accelerated by creating a more favorable, lower salinity environment for the bacteria as a result of the injection of the hydraulic fracturing fluid into the formation (Engle and Rowan, 2014).  The presence and abundance of SO4 reducing bacteria in different regions would vary the amount of SO4 reduction occurring in the formation.  If the bacteria are present the onset of SO4 reduction is rapid 169  (Machel, 2001).The removal of SO4 would increase the barite solubility in the formation, thereby increasing the Ba concentrations. Barite dissolution has previously been suggested as a source of Ba in flowback water in which case Ba concentrations in flowback water would be positively correlated to the amount of barite in the formation (Renock et al., 2016; Zolfaghari et al., 2016).  Barite is not found in the Montney Formation at a high enough percentage to be detected by an XRD analysis; however, small (< 0.5 cm) crystals were observed in the core examined from the region near site A and identified as barite using SEM (Fig. 5.6). A Ba precipitate has also been observed along natural fractures in other formations (e.g., the Lower Keg River formation; Zolfaghari et al., 2016).  Additional factors contributing to higher barite solubility in the formation in comparison to the solubility in freshwater at surface include the high ionic strength of the water involved in the water-rock interactions during hydraulic fracturing combined with the reducing conditions in the formation (Renock et al., 2016), the elevated temperatures and pressures in the formation (Blount, 1977), and the presence of organic matter (He et al., 2014), which may also act to inhibit barite precipitation (Marcon et al., 2017).    170   Figure 5.5: The relationship between Ba and SO4 concentrations for the sampled wells.  Results for individual wells are grouped by site. These parameters are negatively correlated, which is related to SO4 reduction and barite solubility.  Note the logarithmic scale is used due to the large variability in concentrations between wells. ▲- upper Montney member wells; ● – upper and middle Montney member;  - middle Montney member wells;  - lower Montney member wells.     Figure 5.6: Barite in a core sample from a well located near site A.  100 µm Barite 171  The results of our study do not directly support barite dissolution as a dominant geochemical process impacting the Ba concentrations given that the Ba and SO4 concentrations in flowback water do not increase together as would be expected if only barite dissolution was occurring.  The Ba and SO4 concentrations at each of the sites were investigated in more detail.  The SI for barite in the flowback water remains close to equilibrium or slightly over/under-saturated for site A, site B, and site D with SI values generally remaining between -0.4 and 0.4.  The flowback water from the wells at sites A and D typically have increasing Ba concentrations and stable SO4 concentrations over time.  The increase in Ba may be due primarily to mixing with formation water as the SI values do not show an increasing trend which would indicate barite dissolution.  Site B is notable in that it has relatively stable (well B-1) or decreasing (well B-2) Ba concentrations over the flowback period, differing from the increasing Ba concentrations that are observed in the flowback water from the majority of the other wells.  The flowback water from the site B wells also show stable or decreasing SO4 concentrations and negative SI values for barite.  Well B-1 has barite SI values that generally remain between -0.4 and -0.1 indicating that the flowback water from this well is close to equilibrium with barite.  The decreasing Ba and SO4 concentrations at well B-2, along with the decreasing barite SI values which decrease from -0.1 to -1.0, support barite precipitation in this well.   The site C wells have high Ba in flowback water, reaching concentrations > 20 mg/L by the middle of the flowback period, and low (< 30 mg/L) SO4 concentrations.  As these wells were shut-in for 21 to 30 days prior to beginning flowback, bacterial SO4 reduction along with barite dissolution may have taken place in the reservoir during the shut-in time.  These two processes would in combination explain the low SO4 and high Ba in the flowback water from these wells. Mixing with formation water also likely contributes to the Ba concentrations.  The 172  SI for barite in several of these wells is negative (site C, wells 1 through 5) and often below -1.  If SO4 reduction was occurring at a faster rate than barite dissolution, the flowback water could still remain undersaturated with respect to barite in this scenario.  Flowback water from well C-6 and well C-7 is oversaturated in barite, although the SI is stable over time, which may indicate that SO4 reduction is occurring at a slower rate for these wells.       Flowback water from well E-1, well E-2, and well E-4 have barite SI values that are slightly oversaturated early in the flowback period (SI = 0.6-0.7) and remain positive but show a decreasing trend towards equilibrium.  The decreasing SI indicates barite precipitation.  The decreasing SO4 concentrations in the flowback water from these wells may be due to Ba precipitation rather than SO4 reduction, while the increasing Ba concentrations may be due to mixing with formation water with higher Ba concentrations and not Ba dissolution.  Well E-3 has barite SI values that increase over the flowback period from 0.2 to 0.4 and also shows a more rapid increase in Ba concentrations relative to the other site E well (see Fig. 5.3e).  The increasing SI, along with the increase in Ba and SO4 concentrations supports barite dissolution in this well.  SO4 reduction may be occurring but at a slower rate relative to SO4 production by barite dissolution.    The site F wells have increasing SO4 and decreasing Ba over the flowback period.  The barite SI for both wells decreases over the flowback period from slightly oversaturated (SI = 0.3 to 0.5) to either equilibrium (-0.04; well F-2) or until undersaturated (-0.4; well F-1).  The decreasing trend for SI indicates barite precipitation and is supported by the decreasing Ba concentrations over the flowback period. The site F flowback waters show increasing SO4 concentrations; however, this is likely a result of the chemical reaction that occurs due to the addition of sodium hypochlorite (NaOCl) to treat hydrogen sulfide (H2S) in the flowback water 173  at surface.  SO4 concentrations would continue to increase as it is continually produced by the oxidation of the H2S present in the flowback water at this site.   The site G and site H wells have high sulfate (> 600 mg/L) and among the lowest Ba in the study with Ba concentrations that remain below 1.5 mg/L and often below the detection limit (< 1 mg/L).  The elevated SO4 is likely related to pyrite oxidation which has been suggested as a source of SO4 in both formation water (Dresel, 1985) and in flowback water, as pyrite oxidation may increase during the hydraulic fracturing process due to the injection of oxic hydraulic fracturing fluid into the formation (Wilke et al., 2015; Harrison et al., 2017).  There is no indication that SO4 reduction is a dominant process in these wells as the SO4 concentrations remain high and relatively stable. The stable SO4 concentrations may indicate that SO4 reducing bacteria are not present or are inactive in the site G and site H wells.  The decreasing SI of barite over the flowback period at well G-1 (SI = 0.8 to 0.1) and well H-1 (SI = 0.6 to 0.2) indicates that the flowback water is moving towards equilibrium as any barite introduced through mixing with formation water would precipitate with the excess SO4.    The flowback water from the two site I wells attain among the highest Ba concentrations of the wells in our study with maximum values of 215 mg/L and 467 mg/L for well I-1 and well I-2, respectively.  The initial Ba concentrations in the flowback water from these wells however are low (< 10 mg/L) and the highest SO4 concentrations are measured during this period of flowback (Fig. 5.7a).  The elevated SO4 may be due to pyrite oxidation caused by the injection of the hydraulic fracturing fluid (Wilke et al., 2015; Harrison et al., 2017).  In the early flowback water, the SI for barite is near equilibrium (-0.1 < SI < 0.1).  Later in the flowback period for both wells, the SO4 concentrations decrease, which may be related to barite precipitation and SO4 reduction.  The barite SI initially increases due to the increase in Ba from from formation water 174  (up to 0.6 and 0.8, for well I-1 and well I-2, respectively) then decreases (to 0.2 and -0.4, for well I-1 and well I-2, respectively), showing that the system is moving back towards equilibrium (Fig. 5.7b).  The decreasing SI values suggest barite precipitation; however, the continually increasing Ba concentrations do not support precipitation and may indicate that the source of the high Ba in these wells is from mixing with formation water with high Ba concentrations.  The variable behavior of Ba and SO4 concentrations in the study wells shows that these ions are influenced by both barite precipitation and dissolution, as well as SO4 reduction and mixing with formation water, depending on the well and on the time during flowback.       175   Figure 5.7: a) Ba and SO4 concentrations for well I-1 and well I-2.  Ba concentrations are initially low in the flowback water from both wells.  As the Ba concentrations begin to increase, the SO4 concentrations decrease.   b) The saturation indices (SI) for barite for well I-1 and well I-2.  The flowback water is initially close to equilibrium with barite.  The increasing SI is due to increasing Ba from formation water, while the decreasing SI supports barite precipitation.  Both barite dissolution and mixing with high Ba formation water are interpreted to contribute to the Ba concentrations, while bacterial SO4 reduction and barite precipitation contribute to the removal of SO4 from the water.  An additional process that has been suggested as a possible source of Ba in flowback water in previous studies is cation exchange with clay (Renock et al., 2016). In sequential extraction experiments on Marcellus Shale samples, the majority of the Ba ions were found in the exchangeable fraction (Phan et al., 2015; Stewart et al., 2015; Renock et al., 2016) providing support for the release of Ba ions through cation exchange.  The cation exchange capacity (CEC) a. b. 176  for the Montney Formation is low (< 25 cmol(+)/kg) and does not correlate with the percentage of clay in the rock (R2 < 0.05) (see Chapter 2; Chapter 4).  Consequently cation exchange may not be as important in the Montney Formation compared to other formations where hydraulic fracturing is occurring.  Additionally, geochemical modelling with the major ions in our study indicated that ion exchange would produce a decrease in the concentrations of the divalent ions rather than an increase, due to these ions displacing Na ions on the exchange sites when the relatively low TDS hydraulic fracturing fluid is injected into the formation (see Chapter 2).   5.4.2 Boron The B concentrations increase with TDS over the flowback period.  Several of the site C wells (wells 1, 5, 6, and 7) and well D-2 are exceptions and have B concentrations that remain relatively stable over the course of the flowback period (Fig. 5.8).  The B concentrations are typically similar between wells from the same site and which are completed in the same member of the Montney Formation (Fig. 5.8).  In upper Montney member flowback water, the concentrations are lower for the site A and site B wells (3.1 to 14 mg/L), moderate for site C and site D wells (11 to 21 mg/L), and highest at well H-1 (13 to 38 mg/L). The middle Montney member wells can be divided into sites D-G with lower B (3.1 to 16 mg/L) and site C with higher B (19 to 24 mg/L).   The lower Montney member wells have similar or slightly lower B values relative to the upper and middle Montney member flowback water with maximum values of approximately 11 mg/L.  The B concentrations in the Montney Formation flowback water are comparable to or slightly higher than the concentrations in flowback water from other hydraulically fractured formations, including the Marcellus Shale which typically reaches flowback water B concentrations of approximately 20 mg/L (Haluszczak et al., 2013; Rowan et al., 2015).  The high B concentrations in the flowback water from the study wells are not related 177  to the hydraulic fracturing fluid chemistry as B concentrations are low in this fluid (≤5.1 mg/L; Table 5.2).  The initial B concentrations in the study wells show some correlation with the length of the shut-in period (R2 = 0.37).  Figure 5.8: Boron concentrations over the flowback period, plotted as cumulative flowback volume. a) upper Montney member wells. The produced water samples from sites A and B are not included; b) middle and lower Montney member wells. Individual wells are grouped by site. ▲- upper Montney member wells;  - middle Montney member wells;  - lower Montney member wells.  The available B results for Montney Formation produced waters, which are expected to approach the B concentrations in formation water, were limited to the produced water samples collected as part of the present study from site A wells, site B wells, and the well nearby to site I.  There is low variability in the B concentrations in the produced water from these wells with values ranging from 9.3 to 15 mg/L, although this is expected as the flowback waters from these three sites all had low (< 15 mg/L) and similar B concentrations.  Further sampling of produced waters from the Montney Formation and the inclusion of B concentrations in the analysis is required to determine the variability in formation water between the different regions in order to a. b. 178  determine if the variability in the concentrations in flowback water is related to differences in formation water chemistry across the study area.    A potential source of B in formation water is kerogen breakdown (Engle et al., 2016). The B content in kerogen is high and can be released under high temperature conditions (~150°C) (Williams et al., 2001).  The current temperature of the Montney Formation is 75-80°C; however, the maximum paleo-temperature is estimated at 175°C (Desrocher et al., 2004).  There was a moderate correlation (R2 = 0.58; Fig. 5.9) between the completion depth and the maximum flowback water B concentration from the study wells.  This relationship was investigated as greater depths correlate with higher formation temperatures; however, as the formation temperature for the Montney Formation for the study wells remains close to 75°C at all sites, this trend may be indicative of variation in the paleo-temperature across the formation rather than the current temperature conditions. There is also an observable regional geographic variation in B concentrations in flowback water across the study sites related to differences in depth as the completion depth increases from the northwest to the southeast (Fig. 5.10).     179   Figure 5.9: Correlation between the maximum B concentration in flowback water for each of the wells and the total vertical depth (TVD) of the well.  There is some correlation with increasing B concentrations and increasing depth; however, this relationship could also be impacted by the percentage of illite due to substitution of B in illite during diagenesis.  Figure 5.10: Boron concentrations across the formation plotted as cumulative flowback volume. The highest concentrations are observed in the well to the southeast, while the lowest concentrations are observed in the wells to the northwest, although there is some overlap with some of the central wells.  The produced water samples from site A, site B, and nearby to site I are not included in this plot. 180  In cases where the temperature of the formation is lower (< 150°C), B can be removed from formation water during smectite-illite diagenesis by substitution of Si with B in the tetrahedral sites (Williams and Hervig, 2005; Engle and Rowan, 2014).  This process applies to longer timescales for water-rock interactions altering the formation water chemistry rather than occurring on the timescale of the hydraulic fracturing process.  In the portion of the Montney Formation in BC, there is a decrease in clay content from the northwest to the southeast (BC OGC, 2012). The mineralogy results compiled in the regions near the study wells indicate that the clays present include illite, kaolinite, and chlorite, with illite making up the majority of the clay (Table 5.4).  A negative correlation between the percentage of illite and the B concentrations may indicate that a greater amount of B has been removed from the formation water by illitization during diagenesis.  However, based on the wells included in our study, there is only a weak negative correlation (R2 = 0.12) between maximum B concentrations and median weight percent illite due to the large amount of variability in B concentrations in the flowback water from the study wells.  For example, the regions near site A, site C, site D – upper Montney member wells, and site H all have relatively low clay (< 10%, by weight) although the flowback water from these wells have B concentrations spanning from low (< 10 mg/L) to high (> 30 mg/L).  There may also be further substitution and leaching of the B from the clay over time which would minimize the correlation between illite and B in the water.  Additional mineralogical analyses focusing specifically on the hydraulically fractured zones are necessary to further investigate the trend between B concentrations in flowback water and illite.   181  Table 5.4: Summary of the compiled mineralogy for the study sites.  The values provided are medians. Results are divided based on the member of the Montney Formation where the study wells were completed.  The data are from a combination of publically available results and analyses completed for the present study where gaps in the existing data were present. Mineral (wt. %) upper Montney member Site A (n=14) Site B (n=75) Site C (n=40) Site D (n=32) Site H (n=21) Carbonates 34.9 29.5 30.3 30.9 19.3 Ankerite trace 12.9 trace trace - Calcite 16.5 6.9 9.7 12.2 - Dolomitea 18.0 20.7 17.1 16.8 19.3 Siderite trace - 0.8 0.9 - Clays 6.9 16.5 5.1 7.7 6.8 Chlorite trace 3.4 1.2 trace 2.7 Illite&mica 6.4 10.0 4.0 5.6 2.6 Kaolinite - 2.7 1.0 - - Feldspar 22.3 19.4 23.1 14.6 24.4 K-Feldspar 10.1 5.5 11.8 7.1 11.4 Plagioclase 11.5 8.7 11.1 7.7 12.0 Other           Quartz 34.1 32.3 37.0 40.6 45.1 Fluorapatite - - 0.3 1.4 - Pyrite 2.1 1.5 1.6 2.4 1.2    Mineral (wt. %) middle Montney member lower Montney member Site C  (n=31) Site D (n=19) Site E (n=17) Site F (n=22) Site G (n=10) Site I (n=29) Carbonates 28.5 18.8 18.8 18.0 17.7 15.0 Ankerite 2.6 - - - - - Calcite 7.2 5.0 5.0 4.0 2.0 7.0 Dolomitea 18.3 13.0 13.0 14.0 14.8 8.0 Siderite trace - - - - - Clays 9.9 16.2 16.2 16.3 17.3 25.0 Chlorite 1.0 - - - 4.7 3.0 Illite&mica 7.3 14.0 13.5 15.5 12.3 19.5 Kaolinite 1.0 2.0 2.0 2.0   2.1 Feldspar 16.0 21.0 21.0 20.2 26.2 12.0 K-Feldspar 7.6 7.7 7.7 7.4 14.6 4.0 Plagioclase 8.2 11.5 11.5 11.0 12.2 7.0 Other             Quartz 41.7 39.6 39.6 40.3 35.4 44.0 Fluorapatite 0.3 trace trace trace - - Pyrite 1.8 2.2 2.0 2.7 1.5 3.0 Note: a – includes ferrodolomite 182  In addition to the impact of the formation water B concentrations on the flowback water chemistry through mixing, there is also the potential for ion exchange during the hydraulic fracturing process as B that had been adsorbed onto the clay surface could be released due to the injection of the relatively low salinity hydraulic fracturing fluid (Williams and Hervig, 2002; Warner et al., 2014).  As previously discussed in relation to Ba concentrations, the CEC of the Montney Formation is low and therefore cation exchange is likely not the main source of B ions in flowback water. 5.4.3 Lithium Lithium concentrations increase slightly over the flowback period with some variability between wells sampled across the formation.  The majority of the flowback water samples from upper and middle Montney member wells have Li concentrations < 20 mg/L; with the site B wells and wells C-5, C-6, and C-7 reaching slightly higher concentrations (> 30 mg/L; Table 5.1).  Li concentrations in lower Montney member flowback water remain low for well I-1 with a maximum concentration of 15 mg/L and moderate for well I-2 with a maximum of 22 mg/L.  Both maximum values for the site I wells were measured at the end of the flowback period.  The Montney Formation flowback water Li concentrations are lower relative to the concentrations in flowback water from other formations, such as the Marcellus Shale where the concentrations are often > 50 mg/L within the first week and > 90 mg/L within the first 30 days of the flowback period (Rowan et al., 2015; Phan et al., 2016).  The produced water samples collected from Montney Formation wells as part of the present study remain lower with Li concentrations between 27 and 41 mg/L for the site A and site B wells and 61 mg/L for the well nearby site I (Table 5.3).      183  The dominant control on the increasing Li concentrations in flowback water is likely mixing between the injected fluid and the formation water (Phan et al., 2016).  Mixing is supported by the increasing Li concentrations observed over the flowback period in our study.  The hydraulic fracturing fluids used in the study wells have low Li concentrations (≤ 11 mg/L); therefore, the injected fluid does not produce the elevated concentrations measured in flowback water.  The Li concentrations in the initial flowback water samples are impacted by the length of the shut-in period and have a moderate positive correlation (R2 = 0.43).  In sequential extraction experiments on shale samples conducted by Phan et al. (2016), the greatest proportion of Li in the rock was found in the silicate mineral fraction, most of which is likely in clays, although the oxidizable fraction, including organic matter, could account for up to 20% of the total Li.  The organic matter is a potential source of Li in formation water, as Li can be released from organic matter under high temperatures (Clauer et al., 2014).  Variation in formation temperature has been used to explain the regional variation in Li concentrations in other regions since a trend between formation water Li concentrations and depth has been observed (Macpherson et al., 2014; Macpherson, 2015; Phan et al., 2016).  In these studies, the maximum depth of burial was positively correlated with the maximum formation temperature.  The results of the current study show no correlation (R2 < 0.01) between the maximum Li concentration in flowback water and the completion depth; however, this may be due to differences in the current depth and the maximum depth of burial at the study sites, as well as due to the varying lengths of the flowback periods with differing contributions from formation water.  The concentration of Li in formation water may also be affected by smectite-illite diagenesis (Macpherson et al., 2014; Macpherson, 2015; Phan et al., 2016). During diagenesis, Li ions can substitute for the cations in the octahedral layers of the clay (Williams and Hervig, 184  2005; Clauer et al., 2014).  The incorporation of Li into illite during diagenesis would be expected to produce a negative correlation between illite and Li concentrations in formation water, which would be observable in the late stage flowback water chemistry.  The results from the present study only show a weak negative correlation (R2 = 0.14) between the maximum Li concentration and the median percent illite for the study wells.  The low correlation may be due to the release of Li ions during diagenesis as Li can also substitute for the interlayer cation in smectite (Williams and Hervig, 2005). Another potential source of Li in flowback water is ion exchange occurring during hydraulic fracturing, through the same process as B ions (Warner et al., 2014).  However, a sequential extraction study conducted on shale samples found only a minor amount of Li (< 2%) on the exchangeable sites (Phan et al., 2016).  The ion exchange capacity of the Montney Formation is likely less than the shale used in the Phan et al. (2016) study due to the low clay content of the Montney Formation (Zonneveld and Moslow, 2014), relative to the average clay content of the Marcellus Shale (30%) used in the investigation by Phan et al. (2016).  Consequently ion exchange is not expected to be a dominant contributor to Li concentrations in Montney Formation flowback water.   Due to the potential impact of formation temperature and smectite-illite diagenesis on both Li and B concentrations, the relationship between the two ions is examined in more detail.  The correlation between these ions is consistent between wells from the same site and overall there are three general regions in which the B-Li flowback water values plot for our study (Fig. 5.11).  The first region has relatively high Li and relatively low B values. The flowback water samples from sites A and B plot in this region, with later samples from site I approaching this region.  The second region can be divided into two sub-regions.  The first of these sub-185  regions has moderate Li and moderate B values and includes sites C through G and early samples from site I.  The second sub-region includes wells C-5, C-6, and C-7, which have high Li and moderate B values. These wells are included as a sub-region as they plot in a distinct area but still follow the general trend of other wells from site C.  The third region is relatively low Li and relatively high B values.  Only the samples from site H fall within this region.  There is an indication of a geographic trend with region 1 being comprised of wells from the northwestern region with the site B wells, region 2 being the remaining wells located in the central region, and region 3 being the well in the southeastern region of the study area (Fig. 5.11).  Site B is the northern most well in the central region; however, it is much closer to the other central region wells relative to the northwestern region wells (sites A and I, see Fig. 5.2).  The similarity between site A and site B may be due to the factors which are impacting the formation water chemistry in these two regions.  These two sites may have similarities in the B and Li content of the organic matter and the amount of substitution of these ions in illite.  Differences in the correlation between these ions may also be related to differences in paleo-temperature across the formation, as B and Li are released from organic matter at different temperatures (Clauer et al., 2014).  A more detailed study focusing on the local mineralogy and organic matter content within the fractured units is necessary to provide further conclusions on the impact of these variables on the flowback water chemistry and the correlations seen between B and Li ions.    186   Figure 5.11: Li-B plot showing the three regions where the sampled wells plot. Wells from the same site consistently fall within the same region. ▲- upper (u) Montney member wells;  - middle (m) Montney member wells;  - lower Montney member wells.    5.5 Conclusions The selected key minor ions in flowback water – Ba, B, and Li – show greater variability compared to the major ions across the study area.  More variation in the concentrations of the minor ions in formation water between regions is likely a contributing factor to the differences in the concentrations of Ba, B, and Li in flowback water as mixing between the hydraulic fracturing fluid and the formation water is an important geochemical process influencing the flowback water chemistry.  In addition to mixing, these three ions show some indication of the impact of other geochemical processes.   The Ba concentrations in flowback water generally show a decreasing trend over the study area from the northwestern sites (sites A and I) to the southeastern site (site H), although high Ba concentrations were also measured in the flowback water from site C in the central region.  The negative correlation observed between Ba and SO4 concentrations for the flowback water from the study wells supports the influence of SO4 reduction on the concentrations of these 187  ions, as suggested for flowback waters from the Marcellus Formation by Engle and Rowan (2014).  SO4 reduction would in turn increase the barite solubility due to the removal of SO4 ions from solution. If only barite dissolution were occurring a positive correlation would be expected between Ba and SO4 concentrations for all wells.  A detailed examination of the data for each of the wells indicates that both barite dissolution and/or precipitation may be occurring in different wells.  Barite precipitation has been observed in laboratory experiments by others where a rock sample was exposed to synthetic hydraulic fracturing fluid under formation temperatures and pressures (Dieterich et al., 2016; Paukert Vankeuren et al., 2017).  The observation of barite precipitation indicates that the kinetics of the reaction are rapid enough to occur during hydraulic fracturing.  The increasing Ba concentrations in flowback water may be impacted by both barite precipitation/dissolution and mixing with formation water, while the SO4 concentrations are impacted by a combination of pyrite oxidation, bacterial SO4 reduction, and barite precipitation/dissolution.   The B concentrations in the studied flowback water show an increasing trend across the study area from the northwestern sites to the southeastern site, with site B in the central region having B concentrations similar to the northwestern wells.  Li concentrations in the Montney Formation flowback water show less variability between wells relative to both Ba and B concentrations and remain < 20 mg/L for most sites.  Higher Li concentrations > 30 mg/L were measured in the flowback water from the site B wells and wells C-5, C-6, and C-7.  No B or Li containing minerals are close to saturation in the flowback water.  The controls on the formation water chemistry are expected to have an increasing influence on the flowback water chemistry over the flowback period due to a greater fraction of formation water contributing to flowback water over time.  Both B and Li ions are present in organic matter and can be released from the 188  organic matter into the formation water under high temperatures (Williams et al., 2001; Clauer et al., 2014).  As higher temperatures are associated with greater depths, a correlation between the maximum B and Li concentrations in flowback water and completion depth was expected.  However, only B concentrations showed a correlation (R2 = 0.58). The lack of correlation for Li concentrations (R2 < 0.01) may be due to differences between the current depth and the paleo-depth of burial or due to a greater importance of other processes on the concentration of this ion.  Substitution of B and Li ions in smectite-illite diagenesis and ion exchange during hydraulic fracturing were not clearly supported by the results of the flowback water chemistry from the current study in the Montney Formation.   The relationship between B and Li was consistent for wells from the same site.  Overall, there is a trend from the northwestern region (high Li, low B) to the southeastern region (low Li, high B), except for site B in the central region which has B/Li ratios closer to those in the flowback water from the northwestern region wells.  These differences may be due to the release of B and Li ions from organic matter at different temperatures, variation in the B and Li content of the organic matter, and/or variation in the amount of substitution in illite during diagenesis.  The addition of these three minor elements to the flowback water geochemical analysis provides additional information on the variability in flowback water chemistry in the Montney Formation that was not evident from the analysis of the major ions alone.  Further detailed studies focusing on the variability in formation water chemistry and the mineralogy of the fractured zones, specifically the abundance of barite and the clay mineralogy, will improve the understanding of the source of Ba, B, and Li in flowback water.  189  Chapter 6: Conclusions and Recommendations  6.1 Conclusions The inorganic geochemistry of flowback water can provide important insights into the processes that are occurring in the reservoir during hydraulic fracturing.  Prior to being of use for more in-depth studies on reservoir processes, the flowback water chemistry must be characterized for a formation and the variability between wells within that formation examined.  Following the initial characterization, the controls on the flowback water chemistry need to be determined.  The series of studies presented in this thesis focused on the inorganic geochemistry of flowback water from the Montney Formation.  An examination on how the chemistry varied between wells in the study area was completed and an analysis of some of the potential variables that were impacting the chemistry was conducted.   The results of the study provide support for mixing as the most likely geochemical process that is impacting the flowback water chemistry.  Through the use of Cl concentrations, δ18O isotopes, and δ2H isotopes as conservative tracers, it was possible to calculate the mixing proportions of hydraulic fracturing fluid and formation water in the flowback water samples.  The results show that the proportion of formation water increased over the flowback period and in some cases attained Cl concentrations and isotopic values similar to those that were estimated for formation water.  However, for the majority of the study wells the contribution of formation water by the end of the flowback period remained below 60%.  The mixing ratios calculated with the conservative tracers were then used to compare the concentrations of the other major ions with geochemical models to investigate potential processes impacting the chemistry.  The simplest model included only mixing between the hydraulic fracturing fluid and the formation 190  water.  This model was found to provide a reasonable approximation for the Na and K concentrations in flowback water.  The addition of ion exchange to the mixing model was required to improve the fit of the model for Ca, Mg, and Sr concentrations.  The analytical results for the divalent ions displayed lower concentrations in the flowback water than those predicted by mixing alone, which is supported by ion exchange since the divalent ions would replace the Na ions on the exchange sites due to the relative mobility of the ions.  As such, ion exchange would also impact the Na concentrations; however, the relative increase in Na concentrations is not as significant due to the high Na concentrations in the system.        Although mixing is the dominant process influencing the Montney Formation flowback water chemistry, the rate of increase in the proportion of formation water in flowback water was observed to vary between wells.  The consequence of the variability between wells is that the major ions increase at different rates and higher concentrations can be attained earlier in the flowback period after less time has elapsed or at a lower cumulative flowback volume for some wells relative to others.  In general wells completed at the same site and within the same informal member of the Montney Formation had similar flowback water chemistry; however, there were examples of wells from the same site with similar completion depths with different rates of increase for the major ions over the flowback period.  This variability is due to the influence of additional variables on the flowback water chemistry, including the length of the shut-in period and the fracture complexity.  A challenge in comparing flowback water between wells is that the length of the flowback period varies between wells; therefore, it is not possible to simply compare the median or maximum ion concentrations.  A trend analysis using linear regression and linear mixed effects models was presented as a method to compare the flowback water chemistry between wells.  This method allowed the sites to be grouped together using the slope 191  coefficients from the regression of major ion concentration as a function of time.  It was found that nearby sites often have similar slope coefficients, likely due to similarities in the formation water chemistry and the properties of the formation.  The linear regression analysis completed for the volume of flowback water produced as a function of time did not show any obvious correlations with the flowback water chemistry.   An additional complication with evaluating the flowback water chemistry from different wells is that there are a plethora of variables that can impact the chemistry.  Several of these variables were considered in a multiple regression analysis.  This analysis identified shut-in time as a dominant variable.  Longer shut-in times were consistently correlated to higher major ion concentrations in both the early flowback water and the flowback water after 1,000 m3 of fluid had been produced from a well.  However, the relative importance of the shut-in time decreases later in the flowback period.  The initial high ion concentrations associated with longer shut-in times are likely related to a prolonged time for mixing within the reservoir between the hydraulic fracturing fluid and the formation water accessed in the fracture network prior to the initiation of flowback from a well.  In addition, a lengthier shut-in period would allow for a greater contribution of formation water produced by countercurrent imbibition, as well as more diffusion and/or osmosis (Balashov et al., 2015; Zolfaghari et al., 2016; Wang et al., 2016; 2017).  The impact of the length of the shut-in period on flowback water chemistry would have implications for water treatment and re-use, as longer shut-in periods would produce flowback water with higher TDS.   Other notable variables identified in the multiple regression analysis are the hydraulic fracturing fluid chemistry and the formation water chemistry, although the influence of these variables was not consistent between the different ions.  In the early flowback water, the 192  hydraulic fracturing fluid chemistry was found to be important for Cl, Ca, Mg, and Sr concentrations, while in the flowback water for samples collected later in the flowback after 1,000m3 of flowback water had been produced, the formation water chemistry is an important variable for TDS, Cl, Ca, and Mg concentrations.  These variables were anticipated to be identified as dominant variables impacting the flowback water chemistry.  The complexity of the system and the interrelationships between many variables may obscure the importance of these variables in the multiple regression analysis.  Furthermore, the relative importance of the formation water chemistry may be impacted by the low variability in this parameter between sites as well as the uncertainty in the data since this water was not directly sampled. The minor ions considered in the study are Ba, B, and Li.  The concentrations of these ions exhibited more variability in the flowback water across the study area relative to the major ions. The Ba concentrations as well as the B/Li ratio generally had a decreasing trend from the northwest sites to the southeast site. Mixing between the hydraulic fracturing fluid and the formation water is likely the dominant geochemical process impacting the B and Li concentrations in flowback water, while the Ba concentrations are influenced by barite dissolution and precipitation, in addition to mixing between the injected fluid and formation water.  The relative importance of these geochemical processes on Ba concentrations was found to vary between the different wells.  The SO4 concentrations are another control on Ba in flowback water.  As proposed by others, the SO4 concentrations in flowback water are impacted by pyrite oxidation (Wilke et al., 2015; Harrison et al., 2017) and bacterial SO4 reduction (Engle and Rowan, 2014), which can impact the saturation index of barite and determine if this mineral will dissolve or precipitate in the well.  The examination of these additional ions shows that there is variability in the Montney Formation flowback water which may have implications for both 193  water treatment and for determining the processes occurring in the formation at the different sites. The summary of the flowback water chemistry from multiple wells completed in the Montney Formation provides a basis for future flowback water research.  The use of multiple wells in the study illustrates the complexity of the system and demonstrates that there are numerous interacting variables influencing the flowback water chemistry.  These factors must be considered when conducting research in this field of study.  By improving on the understanding of the variability and controls on flowback water chemistry, it will be possible to use the changing chemistry to further investigate reservoir processes and to develop predictions for the flowback water chemistry to assist with water management at future hydraulic fracturing operations.   6.2 Recommendations Based on this thesis, additional areas of study that can provide further insight into the Montney Formation flowback water chemistry have been identified and include:  Further sampling of flowback water from wells completed in the Montney Formation in the northwest and southeast regions of the study area.  Potential regional trends in Sr and Ba concentrations and the B/Li ratios were identified in our studies; however, only two wells in the northwest region and one well in the southeast region were sampled for the present series of studies.  Additional sampling in these regions may provide support for the observed trends which will provide information on the variability in the formation water chemistry and the geochemical processes taking place in the formation.    Further investigation into the variables that are impacting the flowback water chemistry at different sites.  Potential variables were listed in Table 3.1 (Chapter 3).  Identification 194  and quantification of other relevant variables should be completed through more detailed investigations at well sites.  For example, rock sampling along the horizontal portion of the well to better understand the controls of mineralogy on flowback water chemistry, continued sampling of the produced water from the well until the chemistry stabilizes to obtain a more accurate formation water chemistry, and microseismic analysis to determine the fracture complexity at the site.  Laboratory studies could be conducted in support of field investigations and include studies on the fluid chemistry of rock samples exposed to either deionized water or hydraulic fracturing fluid.  Completing laboratory and field studies on rock and fluid samples from the same site would help with validating the laboratory experiments and provide a more complete understanding of the site.   A more detailed study on the mineralogy and organic matter in the horizontal portion of hydraulically fractured wells would assist with determining if the B and Li concentrations are impacted by the amount of illite and the organic matter present in the formation.  The site-specific mineralogy may help with the interpretation of other ions in the flowback water.    Additional geochemical studies should be completed to look at trace elements (e.g., As, Cu, and Zn), minor elements (e.g., Fe, Mn, and Si), and organic compounds in the Montney Formation flowback water.  Understanding the occurrence and the variability of these additional elements and compounds is important as they may impact water treatment and recycling and have the potential to provide a more complete understanding of the processes taking place in the reservoir during hydraulic fracturing.   195  References  Abualfaraj, N., Gurian, P.L. and Olson, M.S. 2014. Characterization of Marcellus Shale Flowback Water: Environmental Engineering Science, v. 31(9), p. 514-524, doi: 10.1089/ees.2014.0001.  Aitchison, J. 1982. The Statistical Analysis of Compositional Data: Journal of the Royal Statistical Society. Series B (Methodological), v. 44(2), p. 139-177.  Akob, D.M., Cozzarelli, I.M., Dunlap, D.S., Rowan, E.L. and Lorah, M.M. 2015. Organic and inorganic composition and microbiology of produced waters from Pennsylvania shale gas wells: Applied Chemistry, v. 60, p. 116-125, doi: 10.1016/j.apgeochem.2015.04.011.  Alessi, D.S., Zolfaghari, A., Kletke, S., Gehman, J., Allen, D.M. and Goss, G.G. 2017. Comparative analysis of hydraulic fracturing wastewater practices in unconventional shale development: Water sourcing, treatment and disposal practices: Canadian Water Resources Journal/Revue canadienne des ressources hydriques, v. 42(2), p. 105-121, doi: 10.1080/07011784.2016.1238782  Anderson, T.F. and Chai, B.H. 1974. Oxygen isotope exchange between calcite and water under hydrothermal conditions, in A.W. Hofmann, B.J. Giletti, H.S. Yoder, Jr. and R.A. Yund, eds., Geochemical Transport and Kinetics, Carnegie Institution of Washington, p. 219-227.  Appelo, C.A.J. 1994. Cation and proton exchange, pH variations, and carbonate reactions in a freshening aquifer: Water Resources Research, v. 30 (10), p. 2793-2805, doi: 10.1029/94wr01048.    Bagheri, R., Nadri, A., Raeisi, E., Eggenkamp, H.G.M., Kazemi, G.A. and Montaseri, A. 2014. Hydrochemical and isotopic (δ18O, δ2H, 87Sr/86Sr, δ37Cl and δ81Br) evidence for the origin of saline formation water in a gas reservoir: Chemical Geology, v. 384, p. 62-75, doi: 10.1016/j.chemgeo.2014.06.017.  Balashov, V.N., Engelder, T., Gu, X., Fantle, M.S. and Brantley, S.L. 2015. A model describing flowback chemistry changes with time after Marcellus Shale hydraulic fracturing: AAPG Bulletin, v. 99(1), p. 143-154, doi: 10.1306/06041413119.  Barbot E., Vidic N.S., Gregory K.B. and Vidic R.D. 2013. Spatial and Temporal Correlation of Water Quality Parameters of Produced Waters from Devonian-Age Shale following Hydraulic Fracturing: Environmental Science & Technology, v. 47, p. 2562-2569, doi: 10.1021/es304638h.  Bates, D.M. 2010. lme4: Mixed-effects modeling with R. Madison, WI, USA: Springer. 196  Bates, D., Maechler, M., Bolker, B. and Walker, S. 2015. Fitting Linear Mixed-Effects Models Using lme4: Journal of Statistical Software, v. 67(1), p. 1-48, doi: 10.18637/jss.v067.i01.  BC Oil and Gas Commission (BC OGC). 2012. Montney Formation Play Atlas NEBC. Retrieved from https://www.bcogc.ca/montney-formation-play-atlas-nebc.  Bearinger, D. 2013. Message in a Bottle: SPE 168891/URTeC 1618676, Prepared for presentation at the 2013 Unconventional Resources Technology Conference, August 12-14, 2013, Denver, Colorado, USA, doi: 10.1190/urtec2013-148.  Bird, T.D., Barclay, J.E., Campbell, R.I. and Lee, P.J. 1994. Part I: Geological Play Analysis and Resource Assessment In Triassic Gas Resources of the Western Canada Sedimentary Basin, Interior Plains: Geological Survey of Canada Bulletin 483, p. 1-66, doi: 10.4095/194156.  Blauch M.E., Myers R.R., Moore T.R., Lipinski B.A. and Houston N.A. 2009. Marcellus Shale Post-Frac Flowback Waters – Where is All the Salt Coming From and What are the Implications?: SPE 125740, Prepared for presentation at the 2009 SPE Eastern Regional Meeting, September 23-25, 2009, Charleston, West Virginia, USA, doi: 10.2118/125740-ms.  Blount, C.W. 1977. Barite solubilities and thermodynamic quantities up to 300°C and 1400 bars: American Mineralogist, v. 62, p. 942-957.   Capo, R.C., Stewart, B.W., Rowan, E.L., Kolesar Kohl, C.A., Wall, A.J., Chapman, E.C., Hammack, R.W. and Schroeder, K.T. 2014. The strontium isotopic evolution of Marcellus Formation produced waters, southwestern Pennsylvania: International Journal of Coal Geology, v. 126, p. 57-63, doi: 10.1016/j.coal.2013.12.010.  Carpenter, A.B. 1978. Origin and chemical evolution of brines in sedimentary basins: SPE 7504, Prepared for presentation at the 53rd Annual Fall Technical Conference and Exhibition of the Society of Petroleum Engineers, October 1-3, 1978, Houston, Texas, USA, doi: 10.2523/7504-ms.  Chalmers, G.R.L. and Bustin, R.M. 2012. Geological evaluation of Halfway-Doig-Montney hybrid gas shale-tight gas reservoir, northeastern British Columbia: Marine and Petroleum Geology, v. 38, p. 53-72, doi: 10.1016/j.marpetgeo.2012.08.004.  Chan, L.-H., Starinsky, A. and Katz, A. 2002. The behavior of lithium and its isotopes in oilfield brines: Evidence from the Heletz–Kokhav field, Israel: Geochimica et Cosmochimica Acta, v. 66(4), p. 615-623, doi: 10.1016/s0016-7037(01)00800-6.    197  Chapman, E.C., Capo, R.C., Stewart, B.W., Kirby, C.S., Hammack, R.W., Schroeder, K.T. and Edenborn, H.M. 2012. Geochemical and strontium isotope characterization of produced water from Marcellus Shale natural gas extraction: Environmental Science & Technology, v. 46, p. 3545-3553, doi: 10.1021/es204005g.  Chaudhuri, S. and Clauer, N. 1993. Strontium isotopic compositions and potassium and rubidium contents of formation waters in sedimentary basins: Clues to the origin of the solutes: Geochimica et Cosmochimica Acta, v. 57(2), p. 429-437, doi: 10.1016/0016-7037(93)90441-x.  Chayes, F. 1960. On correlation between variables of constant sum: Journal of Geophysical Research, v. 65(12), p. 4185-4193, doi: 10.1029/jz065i012p04185.  Clarkson, C.R. 2013. Production data analysis of unconventional gas wells: Review of theory and best practices: International Journal of Coal Geology, v. 109-110, p. 101-146, doi: 10.1016/j.coal.2013.01.002.  Clauer, N., Williams, L.B. and Fallick, A.E. 2014. Genesis of nanometric illite crystals elucidated by light-element (hydrogen, lithium, boron and oxygen) isotope tracing, and K–Ar and Rb–Sr dating: Chemical Geology, v. 383, p. 26-50, doi: 10.1016/j.chemgeo.2014.05.025.  Connolly, C.A., Walter, L.M., Baadsgaard, H. and Longstaffe, F.J. 1990. Origin and evolution of formation waters, Alberta Basin, Western Canada Sedimentary Basin. 1. Chemistry: Applied Geochemistry, v. 5, p. 375-395, doi: 10.1016/0883-2927(90)90016-X.  Connor, J.A., Molofsky, L.J., Richardson, S.D. and Bianchi-Mosquera, G.C. 2015. Environmental issues and answers related to shale gas development: SPE-174164-MS, Prepared for presentation at the SPE Latin American and Caribbean Health, Safety, Environment and Sustainability Conference, July 7-8, 2015, Bogotá, Columbia, doi: 10.2118/174164-ms.  Cooley, H. and Donnelly, K. 2012. Hydraulic Fracturing and Water Resources: Separating the Frack from the Fiction. Oakland, California: Pacific Institute. Retrieved from http://pacinst.org/publication/  Davies, G.R., Moslow, T.F. and Sherwin, M.D. 1997. The Lower Triassic Montney Formation, west-central Alberta: Bulletin of Canadian Petroleum Geology, v. 45 (4), p. 474-505.   Davies, G.R. and Hume, D. 2016. Lowstand/slope onlap wedges in the Montney: Stratigraphic and sequence framework, reservoir significance: Canadian Society of Petroleum Geologists, GeoConvention 2016: Optimizing Resources, March 7-11, 2016, Calgary, Alberta, Canada, Abstract.   198  Dehghanpour, H., Lan, Q., Saeed, Y., Fei, H. and Qi, Z. 2013. Spontaneous Imbibition of Brine and Oil in Gas Shales: Effect of Water Adsorption and Resulting Microfractures: Energy Fuels, v. 27, p. 3039-3049, doi: 10.1021/ef4002814.  Desrocher, S., Hutcheon, I., Kirste, D. and Henderson, C.M. 2004. Constraints on the generation of H2S and CO2 in the subsurface Triassic, Alberta Basin, Canada: Chemical Geology, v. 204, p. 237-254, doi: 10.1016/j.chemgeo.2003.11.012.   Dieterich, M., Kutchko, B. and Goodman, A. 2016. Characterization of Marcellus Shale and Huntersville Chert before and after exposure to hydraulic fracturing fluid via feature relocation using field-emission scanning electron microscopy: Fuel, v. 182, p. 227-235, doi: 10.1016/j.fuel.2016.05.061.   Dixon J. 2000. Regional lithostratigraphic units in the Triassic Montney Formation of western Canada: Bulletin of Canadian Petroleum Geology, v. 48 (1), p. 80-83, doi: 10.2113/48.1.80.  Dresel, P.E. 1985. The Geochemistry of Oilfield Brines from Western Pennsylvania. Pennsylvania State University, M.Sc. Thesis.   Dresel, P.E. and Rose, A.W. 2010. Chemistry and Origin of Oil and Gas Well Brines in Western Pennsylvania: Pennsylvania Geological Survey, 4th ser., Open-File Report OFOG 10-01.0, 48 p., doi: 10.3996/052013-JFWM-033.S11.   Droge, S. and Goss, K.-U. 2012. Effect of Sodium and Calcium Cations on the Ion-Exchange Affinity of Organic Cations for Soil Organic Matter: Environmental Science & Technology, v. 46, p. 5894-5901, doi: 10.1021/es204449r.  Edwards D.E., Barclay J.E., Gibson D.W., Kvill G.E. and Halton E. 1994. Triassic Strata of Western Canada Sedimentary Basin In Geological Atlas of the Western Canada Sedimentary Basin, G.D. Mossop and I. Shetsen (comp.), Canadian Society of Petroleum Geologists and Alberta Research Council. Retrieved from http://www.ags.gov.ab.ca/publications/wcsb_atlas/atlas.html.  Engelder, T., Cathles, L.M. and Bryndzia, L.T. 2014. The fate of residual treatment water in gas shale: Journal of Unconventional Oil and Gas Resources, v. 7, p. 33-48, doi: 10.1016/j.juogr.2014.03.002.  Engle, M.A., Reyes, F.R., Varonka, M.S., Orem, W.H., Ma, L., Ianno, A.J., Schell, T.M., Xu, P. and Carroll, K.C. 2016. Geochemistry of formation waters from the Wolfcamp and “Cline” shales: Insights into brine origin, reservoir connectivity, and fluid flow in the Permian Basin, USA: Chemical Geology, v. 425, p. 76-92, doi: 10.1016/j.chemgeo.2016.01.025.  199  Engle, M. and Rowan, E.L. 2014. Geochemical evolution of produced waters from hydraulic fracturing of the Marcellus Shale, northern Appalachian Basin: A multivariate compositional data analysis approach: International Journal of Coal Geology, v. 126, p. 45-56, doi: 10.1016/j.coal.2013.11.010.  Esmaeilirad, N., White, S., Terry, C., Prior, A. and Carlson, K. 2016. Influence of inorganic ions in recycled produced water on gel-based hydraulic fracturing fluid viscosity: Journal of Petroleum Science and Engineering, v. 139, p. 104-111, doi: 10.1016/j.petrol.2015.12.021.  Fontenelle, L., Weston, M., Lord, P. and Haggstrom, J. 2013. Recycling water: Case studies in designing fracturing fluids using flowback, produced, and nontraditional water sources: SPE 165641, Prepared for presentation at SPE Latin American and Caribbean Health, Safety, Environment and Sustainability Conference, June 26-27, 2013, Lima, Peru, doi: 10.2118/165641-ms.  FracFocus. 2016. What Chemicals Are Used: FracFocus Chemical Disclosure Registry (Canada). Web. Date of access 15 February 2017. Retrieved from http://fracfocus.ca/chemical-use/what-chemicals-are-used.  Freeman, J.T. 2007. The use of bromide and chloride mass ratios to differentiate salt-dissolution and formation brines in shallow groundwaters of the Western Canadian Sedimentary Basin: Hydrogeology Journal, v. 15, p. 1377-1385, doi: 10.1007/s10040-007-0201-1.  Ghanbari, E., Abbasi, M.A., Dehghanpour, H. and Bearinger, D. 2013. Flowback volumetric and chemical analysis for evaluating load recovery and its impact on early-time production: SPE 167165-MS, Prepared for presentation at the SPE Unconventional Resources Conference – Canada, November 5-7, 2013, Calgary, Alberta, Canada, doi: 10.2118/167165-ms.   Goss, G., Alessi, D., Allen, D., Gehman, J. Brisbois, J., Kletke, S., Zolfaghari Sharak, A., Notte, C., Thompson, D.Y., Hong, K., Covalski Junes, V.R., Guedes de Araujo Neto, W.B. and Prosser, C. 2015. Unconventional Wastewater Management: A Comparative Review and Analysis of Hydraulic Fracturing Wastewater Management Practices Across Four North American Basins. Canadian Water Network. Retrieved from http://www.cwn-rce.ca/reports/  Gregory, K.B., Vidic, R.D. and Dzombak, D.A. 2011. Water Management Challenges Associated with the Production of Shale Gas by Hydraulic Fracturing: Elements, v.7, p. 181-186, doi: 10.2113/gselements.7.3.181.  Grömping, U. 2006. Relative Importance for Linear Regression in R: The Package relaimpo: Journal of Statistical Software, v. 17(1), p. 1-27. Version 2.2-2, doi: 10.18637/jss.v017.i01.  200  Gupta, I., Wilson, A.M. and Rostron, B.J. 2012. Cl/Br compositions as indicators of the origin of brines: Hydrogeologic simulations of the Alberta Basin, Canada: GSA Bulletin, v. 124 (1&2), p. 200-212, doi: 10.1130/B30252.1.  Haluszczak L.O., Rose A.W. and Kump L.R. 2013. Geochemical evaluation of flowback brine from Marcellus gas wells in Pennsylvania, USA: Applied Geochemistry, v. 28, p. 55-61, doi: 10.1016/j.apgeochem.2012.10.002.  Harrison A.L., Jew, A.D., Dustin, M.K., Thomas, D.L., Joe-Wong, C.M., Bargar, J.R., Johnson, N., Brown Jr., G.E. and Maher, K. 2017. Element release and reaction-induced porosity alteration during shale-hydraulic fracturing fluid interactions: Applied Geochemistry, v. 82, p. 47-62, doi: 10.1016/j.apgeochem.2017.05.001.  He, C., Li., M., Liu, W., Barbot, E. and Vidic, R.D. Kinetics and Equilibrium of Barium and Strontium Sulfate Formation in Marcellus Shale Flowback Water: Journal of Environmental Engineering, v. 140(5), p. B4014001-1-9, doi: 10.1061/(ASCE)EE.1943-7870.0000807.  Johnson, J.D., Graney, J.R., Capo, R.C. and Stewart, B.W. 2015. Identification and quantification of regional brine and road salt sources in watersheds along the New York/Pennsylvania border, USA: Applied Geochemistry, v. 60, p. 37-50, doi: 10.1016/j.apgeochem.2014.08.002.  Kanfar, M.S. and Clarkson, C.R. 2016. Reconciling flowback and production data: A novel history matching approach for liquid rich shale wells: Journal of Natural Gas Science and Engineering, v. 33, p. 1134-1148, doi: 10.1016/j.jngse.2016.04.042.    Kirste, D., Desrocher, S., Spence, B., Hoyne, B., Tsang, B. and Hutcheon, I. 1997. Fluid flow, water chemistry, gas chemistry and diagenesis in the subsurface Triassic in Alberta and British Columbia: Bulletin of Canadian Petroleum Geology, v. 45 (4), p. 742-764.  Kolesar Kohl C.A., Capo R.C., Stewart B.W., Wall A.J., Schroeder K.T., Hammack R.W. and Guthrie G.D. 2014. Strontium Isotopes Test Long-Term Zonal Isolation of Injected and Marcellus Formation Water after Hydraulic Fracturing: Environmental Science & Technology, v. 48, p. 9867-9873, doi: 10.1021/es501099k.  Kondash A.J., Albright, E. and Vengosh, A. 2017. Quantity of flowback and produced water from unconventional oil and gas exploration: Science of the Total Environment, v. 574, p. 314-321, doi: 10.1016/j.scitotenv.2016.09.069.  Lan, Q., Ghanbari, E., Dehghanpour, H. and Hawkes, R. 2014. Water loss versus soaking time: Spontaneous imbibition in tight rocks: Energy Technology, v. 2, p. 1033-1039, doi: 10.1002/ente.201402039.  201  Lester, Y., Ferrer, I., Thurman, E.M., Sitterley, K.A., Korak, J.A., Aiken, G. and Linden, K.G. 2015. Characterization of hydraulic fracturing flowback water in Colorado: Implications for water treatment: Science of the Total Environment, v. 512-513, p. 637-644, doi: 10.1016/j.scitotenv.2015.01.043.  Lindeman, R.H., Merenda, P.F. and Gold, R.Z. 1980. Introduction to Bivariate and Multivariate Analysis. Glenview, Illinois: Scott, Foresman.  Lu, J., Mickler, P.J., Nicot, J.-P., Choi, W., Esch, W.L. and Darvari, R. 2017. Geochemical interactions of shale and brine in autoclave experiments – Understanding mineral reactions during hydraulic fracturing of Marcellus and Eagle Ford Shales: AAPG Bulletin, v. 101(10), p. 1567-1597, doi: 10.1306/11101616026.  Machel, H.G. 2001. Bacterial and thermochemical sulfate reduction in diagentic settings – old and new insights: Sedimentary Geology, v. 140, p. 143-175, doi: 10.1016/s0037-0738(00)00176-7.   Macpherson, G.L., Capo, R.C., Stewart, B.W., Phan, T.T., Schroeder, K. and Hammack, R.W. 2014. Temperature-dependent Li isotope ratios in Appalachian Plateau and Gulf Coast Sedimentary Basin saline water: Geofluids, v. 14, p. 419-429, doi: 10.1111/gfl.12084.  Macpherson, G.L. 2015. Lithium in fluids from Paleozoic-aged reservoirs, Appalachian Plateau region, USA: Applied Geochemistry, v. 60, p. 72-77, doi: 10.1016/j.apgeochem.2015.04.013.  Makhanov, K., Habibi, A., Dehghanpour, H. and Kuru, E. 2014. Liquid uptake of gas shales: A workflow to estimate water loss during shut-in periods after fracturing operations: Journal of Unconventional Oil and Gas Resources, v. 7, p. 22-32, doi: 10.1016/j.juogr.2014.04.001.  Marcon, V., Joseph, C., Carter, K.E., Hedges, S.W., Lopano, C.L., Guthrie, G.D. and Hakala, J.A. 2017. Experimental insights into geochemical changes in hydraulically fractured Marcellus Shale: Applied Geochemistry, v. 76, p. 36-50, doi: 10.1016/j.apgeochem.2016.11.005.   McCaffrey, M.A., Lazar, B. and Holland, H.D. 1987. The evaporation path of seawater and the coprecipitation of Br- and K+ with halite: Journal of Sedimentary Petrology, v. 57(5), p. 928-937, doi: 10.1306/212f8cab-2b24-11d7-8648000102c1865d.  McLaughlin, M.C., Borch, T. and Blotevogel, J. 2016. Spills of Hydraulic Fracturing Chemicals on Agricultural Topsoil: Biodegradation, Sorption, and Co-contaminant Interactions: Environmental Science & Technology, v. 50, p. 6071-6078, doi: 10.1021/acs.est.6b00240.  202  Michael, K., Machel, H.G. and Bachu, S. 2003. New insights into the origin and migration of brines in deep Devonian aquifers, Alberta, Canada: Journal of Geochemical Exploration, v. 80, p. 193-219, doi: 10.1016/S0375-6742(03)00191-2.   Munson, E.O., Chalmers, G.R.L., Bustin, R.M. and Li, K. 2016. Utilizing smear mounts for X-ray diffraction as a fully quantitative approach in rapidly characterizing the mineralogy of shale gas reservoirs: Journal of Unconventional Oil and Gas Resources, v. 14, p. 22-31, doi: 10.1016/j.juogr.2016.01.001.  Nelson, A.W., May, D., Knight, A.W., Eitrheim, E.S., Mehrhoff, M., Shannon, R., Litman, R. and Schultz, M.K. 2014. Matrix Complications in the Determination of Radium Levels in Hydraulic Fracturing Flowback Water from Marcellus Shale: Environmental Science & Technology – Letters, v. 1, p. 204-208, doi: 10.1021/ez5000379.  Nelson, A.W., Eitrheim, E.S., Knight, A.W., May, D., Mehrhoff, M.A., Shannon, R., Litman, R., Burnett, W.C., Forbes, T.Z. and Schultz, M.W. 2015. Understanding the Radioactive Ingrowth and Decay of Naturally Occurring Radioactive Materials in the Environment: An Analysis of Produced Fluids from the Marcellus Shale: Environmental Health Perspectives, doi: 10.1289/ehp.1408855.  Nelson, A.W., Johns, A.J., Eitrheim, E.S., Knight, A.W., Basile, M., Bettis III, E.A., Schultz, M.K. and Forbes, T.Z. 2016. Partitioning of naturally-occurring radionuclides (NORM) in Marcellus Shale produced fluids influenced by chemical matrix: The Royal Society of Chemistry – Environmental Science: Processes & Impacts, v. 18, p. 456-463, doi: 10.1039/C5EM00540J.  Notte, C., Allen, D.M., Gehman, J., Alessi, D.S. and Goss, G.G. 2016. Comparative analysis of hydraulic fracturing wastewater practices in unconventional shale developments: Regulatory regimes: Canadian Water Resources Journal/Revue canadienne des ressources hydriques, doi: 10.1080/07011784.2016.1218795.  Olsson, O., Weichgrebe, D. and Rosenwinkel, K.-H. 2013. Hydraulic fracturing wastewater in Germany: composition, treatment, concerns: Environmental Earth Sciences, v. 70, p. 3895-3906, doi: 10.1007/s12665-013-2535-4.  O'Neil, J.R. and Kharaka, Y.K. 1976. Hydrogen and oxygen isotope exchange reactions between clay minerals and water: Geochimica et Cosmochimica Acta, v. 40, p. 241-246, doi: 10.1016/0016-7037(76)90181-2.  Parkhurst, D.L. and Appelo, C.A.J. 2013. Description of input and examples for PHREEQC version 3 – A computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations: U.S. Geological Survey Techniques and Methods, book 6, chap. A43, 497 p., available at https://pubs.usgs.gov/tm/06/a43/.  203  Paukert Vankeuren, A.N., Hakala, J.A., Jarvis, K. and Moore, J.E. 2017. Mineral Reactions in Shale Gas Reservoirs: Barite Scale Formation from Reusing Produced Water as Hydraulic Fracturing Fluid: Environmental Science & Technology, v. 51, p. 9391-9402, doi: 10.1021/acs.est.7b01979.  Pawlowsky-Glahn, V. and Egozcue, J.J. 2006. Compositional data and their analysis: an introduction. In Buccianti, A., Mateu-Figueras, G. and Pawlowsky-Glahn, V. (Eds.): Compositional Data Analysis in the Geosciences: From Theory to Practice. Geological Society, London, Special Publications, v. 264, p. 1-10. Retrieved from http://sp.lyellcollection.org/     Phan, T.T., Capo, R.C., Stewart, B.W., Graney, J.R., Johnson, J.D., Sharma, S. and Toro, J. 2015. Trace metal distribution and mobility in drill cuttings and produced waters from Marcellus Shale gas extraction: Uranium, arsenic, barium: Applied Geochemistry, v. 60, p. 89-103, doi: 10.1016/j.apgeochem.2015.01.013.  Phan, T.T., Capo, R.C., Stewart, B.W., Macpherson, G.L., Rowan, E.L. and Hammack, R.W. 2016. Factors controlling Li concentration and isotopic composition in formation waters and host rocks of Marcellus Shale, Appalachian Basin: Chemical Geology, v. 420, p. 162-179, doi: 10.1016/j.chemgeo.2015.11.003.  R Core Team. 2017. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/.  Reible, D.D., Honarparvar, S., Chen, C.-C., Illangasekare, T.H. and MacDonell, M. 2016. Chapter 6 – Environmental Impacts of Hydraulic Fracturing. In S. Orszulik (Ed.), Environmental Technology in the Oil Industry, p. 199-219, Switzerland: Springer International Publishing, doi: 978-3-319-24334-4_6.   Renock, D., Landis, J.D. and Sharma, M. 2016. Reductive weathering of black shale and release of barium during hydraulic fracturing: Applied Geochemistry, v. 65, p. 73-86, doi: 10.1016/j.apgeochem.2015.11.001.  Rietveld, H.M. 1967. Line profiles of neutron powder-diffraction peaks for structure refinement. Acta Crystallographica, v. 22(1), p. 151-152, doi: 10.1107/S0365110X67000234.  Rietveld, H.M. 1969. A profile refinement method for nuclear and magnetic structures. Journal of Applied Crystallography, v. 2(2), p. 65-71, doi: 10.1107/s0021889869006558.  Rivard, C., Lavoie, D., Lefebvre, R., Séjourné, S., Lamontagne, C. and Duchesne, M. 2014. An overview of Canadian shale gas production and environmental concerns: International Journal of Coal Geology, v. 126, p. 64-76, doi: 10.1016/j.coal.2013.12.004.  204  Rokhforouz, M.R. and Amiri, H.A.A. 2017. Phase-field simulation of counter-current spontaneous imbibition in a fractured heterogeneous porous medium: Physics of Fluids, v. 29, p. 062104-1-9, doi: 10.1063/1.4985290.  Rosenblum, J., Nelson, A.W., Ruyle, B., Schultz, M.K., Ryan, J.N. and Linden, K.G. 2017. Temporal characterization of flowback and produced water quality from a hydraulically fractured oil and gas well: Science of the Total Environment, v. 596-597, p. 369-377, doi: 10.1016/j.scitotenv.2017.03.294.  Rostron, B. and Arkadakskiy, S. 2014. Fingerprinting "stray" formation fluids associated with hydrocarbon exploration and production: Elements, v. 10, p. 285-290, doi: 10.2113/gselements.10.4.285.  Rostron, B.J. and Holmden, C. 2000. Fingerprinting formation-waters using stable isotopes, Midale Area, Williston Basin, Canada: Journal of Geochemical Exploration, v. 69-70, p. 219-223, doi: 10.1016/s0375-6742(00)00024-8.  Rowan, E.L., Engle, M.A., Kraemer, T.F., Schroeder, K.T., Hammack, R.W. and Doughten, M.W. 2015. Geochemical and isotopic evolution of water produced from Middle Devonian Marcellus shale gas wells, Appalachian basin, Pennsylvania: AAPG Bulletin, v. 99 (2), p. 181-206, doi: 10.1306/07071413146.  Roychaudhuri, B., Tsotsis, T.T. & Jessen, K. 2013. An experimental investigation of spontaneous imbibition in gas shales. Journal of Petroleum Science and Engineering, v. 111, p. 87-97, doi: 10.1016/j.petrol.2013.10.002.  Salman, A. and Kurtoglu, B. 2014. Analysis of Chemical Tracer Flowback in Unconventional Reservoirs: SPE-171656-MW, Prepared for presentation at the SPE/CSUR Unconventional Resources Conference – Canada, September 30-October 2, 2014, Calgary, Alberta, Canada, doi: 10.2118/171656-ms.  Seales, M.B., Dilmore, R., Ertekin, T. and Wang, J.Y. 2016. Numerical analysis of the source of excessive Na+ and Cl- species in flowback water from hydraulically fractured shale formations, SPE 180911: 2016 SPE Journal, p. 1-14, doi: 10.2118/180911-pa.  Shipman, S., McConnell, D., McCutchan M.P., and Seth, K. 2013. Maximizing Flowback Reuse and Reducing Freshwater Demand: Case Studies from the Challenging Marcellus Shale: SPE-165693-MS, Prepared for presentation at the SPE Eastern Regional Meeting, August 20-22, 2013, Pittsburgh, Pennsylvania, USA.  Sharma, S., Mulder, M.L., Sack, A., Schroeder, K. and Hammack, R. 2014. Isotope approach to assess hydrologic connections during Marcellus Shale drilling: Groundwater, v. 52 (3), p. 424-433, doi: 10.1111/gwat.12083.  205  Stewart, B.W., Chapman, E.C., Capo, R.C., Johnson, J.D., Graney, J.R., Kirby, C.S. and Schroeder, K.T. 2015. Origin of brines, salts and carbonate from shales of the Marcellus Formation: Evidence from geochemical and Sr isotope study of sequentially extracted fluids: Applied Geochemistry, v. 60, p. 78-88, doi: 10.1016/j.apgeochem.2015.01.004.  Strong, L.C., Gould, T., Kasinkas, L., Sadowsky, M.J., Aksan, A. and Wackett, L.P. 2014. Biodegradation in waters from hydraulic fracturing: Chemistry, Microbiology, and Engineering: Journal of Environmental Engineering, v. 140, p. 1-8, doi: 10.1061/(ASCE)EE.1943-7870.0000792.  Stueber, A.M and Walter, L.M. 1991. Origin and chemical evolution of formation waters from Silurian-Devonian strata in the Illinois basin, USA: Geochimica et Cosmochimica Acta, v. 55, p. 309-325, doi: 10.1016/0016-7037(91)90420-A.  Sullivan, R., Woodroof, R., Steinberger-Glaser, A., Fielder, R. and Asadi, M. 2004. Optimizing fracturing fluid cleanup in the Bossier sand using chemical frac tracers and aggressive gel breaker deployment: SPE 90030, Prepared for presentation at the SPE Annual Technical Conference and Exhibition, September 26-29, 2004, Houston, Texas, USA, doi: 10.2523/90030-ms.  Vengosh, A., Warner, N., Jackson, R. and Darrah, T. 2013. The effects of shale gas exploration and hydraulic fracturing on the quality of water resources in the United States: Procedia Earth and Planetary Science, v. 7, p. 863-866, doi: 10.1016/j.proeps.2013.03.213.  Vengosh, A., Warner, N.R., Kondash, A., Harkness, J.S., Lauer, N., Millot, R., Kloppman, W. and Darrah, T.H. 2015. Isotopic fingerprints for delineating the environmental effects of hydraulic fracturing fluids: Procedia Earth and Planetary Science, v. 13, p. 244-247, doi: 10.1016/j.proeps.2015.07.057.  Vengosh, A., Kondash, A., Harkness, J., Lauer, N., Warner, N. and Darrah, T.H. 2017. The geochemistry of hydraulic fracturing fluids: Procedia Earth and Planetary Science, v. 17, p. 21-24, doi: 10.1016/j.proeps.2016.12.011.  Vidic R.D., Brantley S.L., Vandenbossche J.M., Yoxtheimer D. and Abad J.D. 2013. Impact of Shale Gas Development on Regional Water Quality: Science, v. 340, p. 826-835, doi: 10.1126/science.1235009.  Wang, F., Pan, Z. and Zhang, S. 2016. Modeling fracturing-fluid flowback behavior in hydraulically fractured shale gas under chemical potential dominated conditions: Applied Geochemistry, v. 74, p. 194-202, doi: 10.1016/j.apgeochem.2016.10.008.  Wang, F., Pan, Z. and Zhang, S. 2017. Impact of chemical osmosis on water leakoff and flowback behavior from hydraulically fractured gas shale: Journal of Petroleum Science and Engineering, v. 151, p. 264-274, doi: 10.1016/j.petrol.2017.01.018.  206  Ward, H., Eykelbosh, A. and Nicol, A.-M. 2016. Addressing uncertainty in public health risks due to hydraulic fracturing: Environmental Health Review, v. 59(2), p. 57-61, doi: 10.5864/d2016-013.  Warner, N.R., Darrah, T.H., Jackson, R.B., Millot, R., Kloppmann, W. and Vengosh, A. 2014. New Tracers Identify Hydraulic Fracturing Fluids and Accidental Releases from Oil and Gas Operations: Environmental Science & Technology, v. 48, p. 12552-12560, doi: 10.1021/es5032135.  Wei, W., Clinch, S., Shafer, J.L. and Lasswell, P. 2010. Improving connate water salinity analysis on preserved clay rich cores: SCA2010-29, Prepared for presentation at the International Symposium of the Society of Core Analysts, October 4-7, 2010, Halifax, Nova Scotia, Canada.  Wilke, F.D.H., Vieth-Hillebrand, A., Naumann, R., Erzinger, J. and Horsfield, B. 2015. Induced mobility of inorganic and organic solutes from black shales using water extraction: Implications for shale gas exploitation: Applied Geochemistry, v. 63, p. 158-168, doi: 10.1016/j.apgeochem.2015.07.008.  Williams, L.B., Hervig, R.L., Wieser, M.E. and Hutcheon, I. 2001. The influence of organic matter on the boron isotope geochemistry of the gulf coast sedimentary basin, USA: Chemical Geology, v. 174, p. 445-461, doi: 10.1016/s0009-2541(00)00289-8.  Williams, L.B. and Hervig, R.L. 2002. Exploring intra-crystalline B-isotope variations in mixed-layer illite-smectite: American Mineralogist, v. 87, p. 1564-1570, doi: 10.2138/am-2002-11-1206.  Williams, L.B. and Hervig, R.L. 2005. Lithium and boron isotopes in illite-smectite: The importance of crystal size: Geochimica et Cosmochimica Acta, v. 69(24), p. 5705-5716, doi: 10.1016/j.gca.2005.08.005.  Williams-Kovacs, J.D. and Clarkson, C.R. 2016. A modified approach for modeling two-phase flowback from multifractured horizontal shale gas wells: Journal of Natural Gas Science and Engineering, v. 30, p. 127-147, doi: 10.1016/j.jngse.2016.02.003.  Wood, J.M. 2015. Crushed-rock vs. full-diameter core samples for water-saturation determination in a tight-gas siltstone play: SPE Reservoir Evaluation & Engineering, v. 18 (3), p. 407-416, doi: 10.2118/174548-pa.  Zheng, Z., Zhang, H., Chen, Z., Li, X., Zhu, P. and Cui, X. 2017. Hydrogeochemical and Isotopic Indicators of Hydraulic Fracturing Flowback Fluids in Shallow Groundwater and Stream Water, derived from Dameigou Shale Gas Extraction in the Northern Qaidam Basin: Environmental Science & Technology, v. 51, p. 5889-5898, doi: 10.1021/acs.est.6b04269.  207  Zhou, Q., Dilmore, R., Kleit, A. and Wang, J.Y. 2015. Evaluating fracturing fluid flowback in Marcellus using data mining technologies: SPE-173364-MS, Prepared for presentation at the SPE Hydraulic Fracturing Technology Conference, February 3-5, 2015, The Woodlands, Texas, USA, doi: 10.2118/173364-ms.  Ziemkiewicz, P.F. and He, Y.T. 2015. Evolution of water chemistry during Marcellus Shale gas development: A case study in West Virginia: Chemosphere, v. 134, p. 224-231, doi: 10.1016/j.chemosphere.2015.04.040.  Zolfaghari, A., Noel, M., Dehghanpour, H. and Bearinger, D. 2014. Understanding the origin of flowback salts: A laboratory and field study: SPE-171647-MS, Prepared for presentation at the SPE/CSUR Unconventional Resources Conference – Canada, September 30-October 2, 2014, Calgary, Alberta, Canada, doi: 10.2118/171647-ms.  Zolfaghari, A., Dehghanpour, H., Ghanbari, E. and Bearinger, D. 2015a. Fracture characterization using flowback salt-concentration transient: 2015 SPE Journal, p. 1-12, originally prepared for presentation at the SPE Hydraulic Fracturing Technology Conference, February 4-6, 2014, The Woodlands, Texas, USA.   Zolfaghari, A., Tang, Y., Holyk, J., Binazadeh, M., Dehghanpour, H., and Bearinger, D. 2015b. Chemical Analysis of Flowback Water and Downhole Gas Shale Samples: SPE-175925-MS, Prepared for presentation at the SPE/CSUR Unconventional Resources Conference, October 20-22, 2015, Calgary, Alberta, Canada, doi: 10.2118/175925-ms.  Zolfaghari, A., Dehghanpour, H., Noel, M. and Bearinger, D. 2016. Laboratory and field analysis of flowback water from gas shales: Journal of Unconventional Oil and Gas Resources, v. 14, p. 113-127, doi: 10.1016/j.juogr.2016.03.004.  Zonneveld, J.-P., Gingras, M.K. and Beatty, T.W. 2010. Diverse ichnofossil assemblages following the P-T mass extinction, Lower Triassic, Alberta and British Columbia, Canada: Evidence for shallow marine refugia on the northwestern coast of Pangaea: Palaios, v. 25, p. 368-392, doi: 10.2110/palo.2009.p09-135r.  Zonneveld, J.-P., Golding, M., Moslow, T.F., Orchard, M.J., Playter, T. and Wilson, N. 2011. Depositional framework of the Lower Triassic Montney Formation, west-central Alberta and Northeastern British Columbia: Recovery - 2011 CSPG CSEG CWLS Convention.  Zonneveld, J.-P. and Moslow, T.F. 2014. Perennial river deltas of the Montney Formation: Alberta and British Columbia subcrop edge: GeoConvention 2014 – Focus.   208  Appendices  Appendix A  Detailed sample preparation method  A summary of the sample preparation method for the dissolved metals analysis was provided in the methods sections of chapters 2, 3, 4, and 5.  This section of the appendix provides a detailed step-by-step breakdown of the method used to prepare the samples for inductively coupled plasma-optical emission spectrometry (ICP-OES).   All glassware was acid washed in a 5% hydrochloric acid (HCl) bath for a minimum of 4 hours and transferred to a 10% nitric acid (HNO3) bath overnight.  All acids used are trace metal grade.  The test tubes used in sampling were new and were acid washed in the 10% HNO3 acid bath for a minimum of 4 hours.  Glassware and test tubes were rinsed with deionized (DI) water prior to soaking in the acid baths and again when removed from the acid baths.  Acid washed containers were air-dried and stored in sealed plastic bags.     The procedure for the hydraulic fracturing fluids, flowback water, and produced water samples followed the subsequent steps: 1) All samples were stored at 4°C prior to analysis; 2) If an oil layer was visible in the sample, the sample was put into a separatory funnel and the water portion was collected as a sub-sample (flowback water and produced water only); 3) The water sample was gently mixed by turning the container upside down two to three times and then poured into an acid-washed 120 mL Teflon® container.  The sample was heated to reservoir temperature (75-80°C) in a hot water bath for approximately 24 hours.  209  This step was included to return the fluid to reservoir temperature and re-dissolve any precipitate that had formed during sample shipment and storage (flowback water and produced water only); 4) The heated sample was filtered with a 0.45 µm filter and divided for 1) pH, conductivity, and alkalinity (20 mL of sample); 2) anions (40-45 mL of sample); and 3) dissolved metals (40 mL of sample).  The dissolved metals sample was preserved to pH < 2 with HNO3; 5) A 50 mL glass beaker was weighed then 5 mL of sample was added to the beaker and re-weighed.  For the first step of the acid-digestion, 0.5 mL of HCl and 1mL of HNO3 were added to the sample in the fumehood.  The beaker was placed on a hot plate in the fumehood and covered with a ribbed watch glass, which minimizes contamination to the sample in the open beaker while allowing the sample to evaporate.  The sample was heated until almost dry.  The second step of the acid-digestion was adding an additional 1 mL of HNO3 and re-evaporating the solution until almost dry.  The remaining precipitate was re-dissolved using a 1% HNO3 solution and transferred to a weighed 15 mL test tube.  The test tube was filled to 10 mL (2x dilution) and re-weighed. Further dilutions were required due to the high concentrations in the samples – 100x and 1000x dilutions were made for each sample.  All dilutions were calculated by weight; 6) The samples were run on the ICP-OES in two batches, the first run analyzed the 1000x dilutions for Na and Ca.  For some samples, (e.g., for the hydraulic fracturing fluids), a lower dilution was used for the Na and Ca analysis if the concentrations of these elements were below the detection limit in the 1000x dilution.  The second run analyzed the 100x and 2x dilutions for the remaining cations of interest.  An internal standard of 10 ppm 210  indium (In) was used and was added to all samples as the sample was injected into the ICP-OES.   211  Appendix B  Example of code used for PHREEQC geochemical mixing model The following PHREEQC script is provided as an example of the geochemical mixing model for mixing between the injected hydraulic fracturing fluid and the formation water.  The data used are an example from site I, well 1.  The first model includes mixing only; the second model includes mixing and secondary mineral precipitation; the third model includes mixing, secondary mineral precipitation, and ion exchange; and the fourth model includes mixing, secondary mineral precipitation, carbonate dissolution, and ion exchange.  SOLUTION 1 # formation water – site I     temp      75     pH       5.9     pe        -4     redox     pe     units     mg/l     density   1.09     Ca        10000     Cl        89168     K         2040     Mg        1140     Na        54000     S(6)      0.05     Sr        1620     Alkalinity 63     -water    1 # kg  SOLUTION 2 # hydraulic fracturing fluid – well I-1     temp      25     pH        7.7     pe        4     redox     pe     units     mg/l     density   1     Ca        74     Cl        1204     K         8     Mg        16     Na        66     S(6)      0.05     Sr        0.5     Alkalinity 242     -water    1 # kg     212  MIX 1 # All models 1 0   # proportion of solution 1 (formation water) [ran model for 0, 0.05, 0.1, 0.15,… , 0.95, 1]  2 1   # proportion of solution 2 (hydraulic fracturing fluid) [ran model for 1, 0.95, 0.9, 0.85,… , 0.05, 0]   EQUILIBRIUM_PHASES 1 # mineral phases can precipitate but not dissolve; Models 2 and 3      Calcite   0 0     Dolomite  0 0     Huntite   0 0     Magnesite 0 0     Celestite 0 0   EXCHANGE 1 # Models 3 and 4     X       0.5     -equilibrate with solution 1 # formation water     -pitzer_exchange_gammas true   EQUILIBRIUM_PHASES 1 # calcite and dolomite can dissolve; Model 4     Calcite   0 1     Dolomite  0 1     Huntite   0 0     Magnesite 0 0     Celestite 0 0    213  Appendix C  Percentages of formation water for all analyzed samples using δ2H isotopes, δ18O isotopes, and Cl concentrations as conservative tracers Well Sample Percentage of Formation Water (%)   Well Sample Percentage of Formation Water (%) Based on δ2H Based on δ18O Based on Cl concentrations   Based on δ2H Based on δ18O Based on Cl concentrations A-1 1 79.2 91.8 54.1   B-2 1 2.8 -16.2 9.9 A-1 3 85.3 106.6 58.3   B-2 2 n.a. n.a. 25.8 A-2 1 31.0 17.9 25.3   B-2 3 36.5 101.4 32.4 A-2 3 42.8 10.6 38.4   B-2 4 n.a. n.a. 33.3 A-2 5 n.a. n.a. 48.0   B-2 6 33.4 147.4 39.5 A-2 7 43.8 43.8 57.0   B-2 8 n.a. n.a. 46.8 A-3 1 29.3 26.5 42.4   B-2 9 29.2 91.0 25.4 A-3 3 n.a. n.a. 47.2   B-2 10 n.a. n.a. 60.6 A-3 5 36.5 41.7 48.4   B-2 12 31.3 112.7 59.3 A-4 1 33.2 33.7 26.9   B-2 15 35.4 124.8 55.9 A-4 3 32.0 44.0 34.5   B-2 17 67.3 224.2 53.6 A-4 5 28.0 35.2 37.6   B-2 19 n.a. n.a. 57.9 A-5 1 46.5 76.9 41.1   B-2 21 56.2 206.8 63.4 A-5 2 53.1 68.7 51.2   D-1 1 23.6 30.5 45.8 A-6 1 55.9 57.4 59.3   D-1 2 n.a. n.a. 46.5 A-6 4 78.8 103.7 68.8   D-1 3 26.5 38.6 38.6 A-7 1 29.2 61.5 27.7   D-1 4 n.a. n.a. 69.7 A-7 2 34.5 56.2 28.8   D-1 5 68.8 121.0 72.3 A-7 4 38.5 53.4 37.2   D-1 6 23.9 28.6 31.9 A-8 1 60.2 68.6 42.9   D-1 7 32.0 58.8 67.2 A-8 2 n.a. n.a. 31.1   D-1 8 n.a. n.a. 83.5 A-8 5 43.1 52.3 35.2   D-1 9 46.8 86.9 84.6 B-1 1 1.4 65.7 18.4   D-2 1 20.2 18.8 47.0 B-1 2 n.a. n.a. 23.5   D-2 2 n.a. n.a. 44.9 B-1 3 4.0 74.5 35.2   D-2 3 n.a. n.a. 50.6 B-1 4 n.a. n.a. 38.3   D-2 4 23.5 40.3 62.0 B-1 5 n.a. n.a. 40.3   D-2 5 n.a. n.a. 64.5 B-1 6 11.8 76.5 44.8   D-2 6 23.8 46.1 65.0 B-1 7 n.a. n.a. 39.4   D-2 7 n.a. n.a. 81.1 B-1 8 23.9 101.3 31.3   D-2 8 29.5 53.6 81.1 B-1 9 n.a. n.a. 33.9   D-2 9 n.a. n.a. 58.9 B-1 10 n.a. n.a. 42.0   D-2 10 23.8 36.4 63.3 B-1 11 n.a. n.a. 38.4   D-3 1 16.9 47.8 4.9 B-1 12 n.a. n.a. 41.0   D-3 2 n.a. n.a. 6.2 B-1 13 41.1 84.5 42.3   D-3 3 16.1 17.3 10.4 B-1 14 n.a. n.a. 39.3   D-3 4 n.a. n.a. 11.4 B-1 16 46.9 75.3 100.9   D-3 5 n.a. n.a. 16.5 B-1 18 n.a. n.a. 102.3   D-3 6 18.3 22.8 15.2 B-1 20 40.0 67.7 128.7   D-3 7 n.a. n.a. 24.6 214  Well Sample Percentage of Formation Water (%)   Well Sample Percentage of Formation Water (%) Based on δ2H Based on δ18O Based on Cl concentrations   Based on δ2H Based on δ18O Based on Cl concentrations D-3 8 31.3 36.1 33.6   F-1 8 34.4 17.4 n.a. D-4 1 -3.3 7.6 -0.3   F-1 9 34.5 12.0 19.0 D-4 2 0.7 0.4 4.4   F-1 10 32.7 19.5 16.2 D-4 3 8.0 12.9 18.0   F-2 1 34.9 33.0 5.4 D-4 4 n.a. n.a. 16.9   F-2 2 34.9 29.3 8.9 D-4 5 9.9 12.4 16.3   F-2 3 49.1 72.4 7.1 D-4 6 n.a. n.a. 18.0   F-2 4 37.9 34.7 n.a. D-4 7 22.9 37.8 38.4   F-2 5 32.1 23.8 11.6 D-4 8 n.a. n.a. 32.5   F-2 6 41.5 47.8 n.a. D-4 9 17.3 23.0 28.8   F-2 7 31.3 21.5 n.a. D-4 10 n.a. n.a. 42.4   F-2 8 32.2 17.1 n.a. D-4 11 27.3 24.9 44.1   F-2 9 35.2 18.2 22.5 D-4 12 n.a. n.a. 48.0   F-2 10 34.9 20.3 26.6 D-4 13 n.a. n.a. 55.9   G-1 1 1.5 13.8 16.3 D-4 14 29.3 23.3 49.4   G-1 3 n.a. n.a. 32.1 E-1 1 1.9 30.2 3.3   G-1 5 17.1 21.1 32.4 E-1 3 2.2 6.0 4.7   G-1 7 26.8 32.0 46.9 E-1 5 4.3 28.9 7.2   G-1 8 n.a. n.a. 70.9 E-1 7 n.a. n.a. 8.9   G-1 9 43.3 53.1 85.8 E-1 8 19.1 32.6 15.0   G-1 11 n.a. n.a. 95.3 E-1 10 n.a. n.a. 17.6   G-1 13 57.6 63.7 102.5 E-1 12 24.0 27.3 28.2   H-1 1 8.4 8.4 15.8 E-2 1 16.3 31.2 7.3   H-1 2 7.6 9.6 19.3 E-2 3 n.a. n.a. 10.1   H-1 3 20.9 20.4 24.4 E-2 5 18.0 42.2 11.2   H-1 4 n.a. n.a. 19.3 E-2 7 n.a. n.a. 13.6   H-1 5 18.5 23.8 26.9 E-2 9 22.3 44.5 13.6   H-1 6 n.a. n.a. 31.2 E-2 11 n.a. n.a. 20.3   H-1 7 26.2 38.0 45.7 E-2 14 31.2 68.6 23.6   H-1 8 n.a. n.a. 55.7 E-3 1 15.3 27.9 15.7   H-1 9 33.1 14.5 60.6 E-3 3 5.7 -30.3 19.6   H-1 10 n.a. n.a. 65.5 E-3 6 24.0 45.6 20.0   H-1 11 n.a. n.a. 74.3 E-3 10 n.a. n.a. 25.8   H-1 12 31.1 25.0 69.2 E-3 11 38.9 98.0 26.1   H-1 13 n.a. n.a. 77.6 E-4 1 1.1 11.5 3.5   H-1 14 n.a. n.a. 77.5 E-4 3 2.0 8.0 8.2   H-1 15 25.5 51.4 81.8 E-4 5 n.a. n.a. 3.5   H-1 16 n.a. n.a. 79.9 E-4 7 7.9 10.4 12.8   H-1 17 n.a. n.a. 82.9 E-4 9 n.a. n.a. 15.5   H-1 18 30.7 50.8 86.2 E-4 11 4.6 -40.2 21.9   I-1 1 7.5 7.3 1.2 F-1 2 38.8 46.1 n.a.   I-1 3 9.9 10.6 3.3 F-1 4 32.1 22.9 14.2   I-1 4 n.a. n.a. 4.2 F-1 5 43.2 49.6 n.a.   I-1 7 11.9 13.6 6.5 F-1 6 31.8 16.1 13.1   I-1 9 n.a. n.a. 7.3 F-1 7 33.7 21.3 13.5  I-1 11 16.1 17.1 8.6 215  Well Sample Percentage of Formation Water (%)       Based on δ2H Based on δ18O Based on Cl concentrations       I-1 13 20.7 17.2 12.3       I-1 16 23.2 18.7 14.0       I-1 18 n.a. n.a. 18.4       I-1 20 28.5 21.2 19.1       I-1 23 n.a. n.a. 23.2       I-1 26 28.8 30.6 29.4       I-1 29 n.a. n.a. 32.2       I-1 32 31.1 32.2 33.0       I-2 1 13.5 15.7 4.6       I-2 3 21.7 21.2 8.8       I-2 5 n.a. n.a. 8.9       I-2 7 23.9 24.0 11.9       I-2 9 n.a. n.a. 15.6       I-2 11 25.9 25.9 20.1       I-2 13 n.a. n.a. 26.6       I-2 14 45.4 43.1 30.8       I-2 18 n.a. n.a. 39.0       I-2 21 51.8 41.5 43.5       I-2 23 n.a. n.a. 46.0       I-2 26 55.2 43.2 50.8       I-2 29 n.a. n.a. 54.4       I-2 32 59.1 39.5 56.9        Note: n.a. indicates Not Analyzed 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0362574/manifest

Comment

Related Items