Summary The water recovered from hydraulic-fracturing operations (i.e., flowback water) is highly saline, and can be analyzed for reservoir characterization. Past studies measured ion-concentration data during imbibition experiments to explain the production of saline flowback water. However, the reported laboratory data of ion concentration are approximately three orders of magnitude lower than those reported in the field. It has been hypothesized that the significant surface area created by hydraulic-fracturing operations is one of the primary reasons for the highly saline flowback water. In this study, we investigate shale/water interactions by measuring the mass of total ion produced (TIP) during water-imbibition experiments. We conduct two sets of imbibition experiments at low-temperature/low-pressure (LT/LP) and high-temperature and high-pressure (HT/HP) conditions. We study the effects of rock surface area (As), temperature, and pressure on TIP during imbibition experiments. Laboratory results indicate that pressure does not have a significant effect on TIP, whereas increasing As and temperature both increase TIP. We use the flowback-chemical data and the laboratory data of ion concentration to estimate the fracture surface area (Af) for two wells completed in the Horn River Basin (HRB), Canada. For both wells, the estimated Af values from LT/LP and HT/HP test results have similar orders of magnitude (approximately 5.0×106 m2) compared with those calculated from production and flowback rate-transient analysis (RTA) (approximately 106 m2). The proposed scaleup procedure can be used as an alternative approach for a quick estimation of Af using early-flowback chemical data.

Abstract Continuous monitoring of wellhead pressures, captured from passive wellbores on a multi-well pad during fracturing operation, has been used as a cost efficient diagnostic tool to better characterize stimulated fracture network and reservoir drainage in Horn River Shale Basin. This paper aims to generate pressure hit catalogues and to quantify their effectiveness to deliver real-time on-site improvement in completion treatments, parent/child interactions, and re-fracturing designs. Pressure data of a multi-well pad was initially analyzed to establish measurable attributes for identification of hydro-mechanic responses from direct hydraulic communications. A fully-coupled hydro-mechanical code, with explicit inclusion of discontinuities, explored the interaction mechanisms across natural and induced fractures by reproducing observed pressures of passive wells. Employing multivariate experimental design on a subset of pressure data, pressure hit catalogues were generated with notion to the sensitivity of hydro-mechanical properties of fracture/intact rock, geometrical/statistical properties of fracture network, in-situ stresses, and completion design. Using the unexploited subset of pressure data, the practicality of pressure catalogues was verified against coupled simulations. The pre-processing of pressure data, to assign attributes to pressure hits, was found essential for analysis of interference mechanisms during fracture treatments prior to incorporating the data into a coupled simulation. Explicit modeling of discrete fractures allowed to evaluate how key sensitive parameters, mainly fracture/intact rock properties, and altered stress environment due to continuous multi-stage fracturing operation, could affect passive pressure signature. Using a subset of pressure hit data obtained from a multi-well pad in Horn River Shale Basin, the calibrated coupled simulations helped to constrain the statistical complexity of fracture network realizations, which ultimately observed to closely align with available microseismic data. The efforts went into the calibration of pressure/stress shadows, using wellhead pressure data at subsequent stages, were summarized. Given the model size and dealing with too large data sets, history-matching of individual pressure hits proved to be computationally intensive. The steps to generate and employ pressure hit catalogues were also illustrated. The results obtained from application of pressure catalogues on the unexploited subset of pressure data indicated close agreement with coupled numerical simulation. A procedure in potential deployment of pressure hit catalogues for real-time modification of multi-well completion and re-fracturing design was lastly provided. Because of the inherited uncertainty associated with locating the subsurface origin of passive pressure responses, yet measured at wellhead, more robust interpretation techniques are required for a better on-site assessment of fracturing operations. The possibility of employing pressure hit catalogues for evaluation of well interference, parent-child interactions, complex structure of stimulated fractures, and reservoir drainage, is an effort in that direction, aiming to monitor, and if required, to remediate well spacing, completion treatments, and re-fracturing designs.

Abstract A design of hydraulic fracturing in variably-stressed zones is one of key components for an effective multi-zone, multi-horizontal well pad treatment. In the recent literature, optimum completion strategies catering for stimulation-induced in-situ stress changes are discussed, however, only few of these focus on vertical stress changes and its impact on multi-zone fracture geometries. In this paper, we present an approach to design contained hydraulic fractures in a high stress layers by studying the role of vertical stress shadowing on actual field data. In modeling hydraulic fractures with pseudo-3D models, if fracture simulations are initiated in high stress zones, "artificially" unbounded height growth results in very limited lateral propagation. On the other hand, 3D hydraulic fracturing models are too computationally expensive to optimize large design jobs, for example, in multi-horizontal well pads. In this paper, we employ a Stacked Height Growth Model, whereby fractures are also discretized vertically yet retain the numerical formulation pseudo-3D models. Coupling with finite element stress solvers then allows to identify vertical stress changes in the vicinity of induced hydraulic fractures and to understand the interference between hydraulic fracture sequences and their respective microseismic signatures. Considering a potential combination of fracturing sequences, it was revealed that stress perturbations from the neighboring well hydraulic fractures initiating from low stress layers can be used to increase stress within the same zone and also potentially reduce stresses in higher-stress layers above and below. By modeling and calibrating an actual multi-zone, multi-horizontal stimulation job, we elaborate on the benefits of increasing stress barriers before fracturing in higher-stress layer to avoid the chances of re-fracturing from high stress zones. Regarding hydraulic fracture geometries, we explain our results by analyzing actual microseismic observations with respect to simulated stress patterns after stimulation. We explore the notion of deliberately ordering hydraulic fracture to manage vertical interference and create more contained fractures in a multi-zone horizontal well pad. Fracturing in a higher-stress zone will naturally divert the energy into low stress, potentially unproductive zones. In an effort to manage this phenomenon, this paper presents one of the few data-rich case studies on multi-zone, multi-well engineered stimulation design. The approach shown in this paper can be a helpful reference to understand fracture height growth in the presence of both vertical and horizontal stress shadowing.

Abstract To determine the degree of connectivity and complexity of a stimulated fracture network, a prescriptive completion program was undertaken in the Horn River Shale Basin which enabled continuous monitoring of pressure interactions among horizontal wells during multi-stage hydraulic fracturing. This paper introduces a novel approach to characterize the stimulated fracture network, and consequently, to optimize the stimulation, wellbore placement, and re-fracturing designs, by integrating the pressure hits captured from passive wellbores on a pad during fracturing operations. If effective, it may also provide a cost-effective alternative to microseismic monitoring. The workflow initially considers a rigorous data analysis on available pressure hits at each frac-stage in time and space, including the location of pressure events, time of flights to offsetting stimulation, and the magnitude and intensity of pressure hits/falloffs. Streamline simulation, assisted with a hydraulic fracturing module, is then used to match the pressure hits/falloffs in the passive wells. This ultimately provides a dynamic probabilistic 3D map of the fracture network growth, reservoir complexity and inter-well connectivity. The fundamental mechanisms of hydraulic fracturing and the interactions across natural and induced fractures (fracture initiation/propagation/growth) are implemented by means of an advanced coupled hydro-mechanical code, based on distinct element method. Results from initial data analyses were fed into a hydro-mechanical model, which incorporated the physics of the hydraulic fracturing process, in order to reproduce the pressure hit signatures. An assisted streamline-based technique was used to simulate various scenarios of pressure hit responses to construct a database of standard pressure hit/falloff patterns. This database, compiled into a dynamic 3D map, facilitated a probabilistic approach to calculate a robust estimate range of stimulated fracture network of the pad area. This database can be subsequently used in future stimulation and re-fracturing designs. The backbone of the highly complex fracture network, extracted from pressure hit/falloff data, was found to closely align with high-resolution microseismic data. Calibration of the hydro-mechanical model using the pressure hit data provides increased confidence in the use of the model to optimize well placement and hydraulic fracturing designs. In the absence of microseismic data, this unique workflow has the potential to deliver real-time on-site monitoring of fracturing operation at a reduced cost and acceptable accuracy, to provide additional statistics on complexity of stimulated reservoir volume, and to offer a better assessment of the likely range of the induced fracture network among horizontal wells.

Abstract Imbibition of water into the shale matrix is known as the primary reason for inefficient water recovery after hydraulic fracturing treatments. The hydration of clay minerals may induce microfractures in clay-rich shale samples. The increased porosity and permeability due to induced microfractures has been considered to be partly responsible for 1) excessive water uptake of gas shales, and 2) increase in hydrocarbon production rate after prolonged shut-in periods. To test this hypothesis, it is necessary to measure imbibition-induced strain and stress under representative laboratory conditions. In this study, we conduct laboratory tests to 1) measure the strain and stress induced by water imbibition in gas shales and 2) investigate the effect of confining load on the rate of water imbibition. We conduct a three-phase study on rock samples from the Horn River Basin (HRB) and the Duvernay (DUV) Formation, located in the Western Canadian Sedimentary Basin. In the first phase, we measure baselines for water and kerosene imbibition into the rock samples by conducting spontaneous imbibition tests. In the second phase, we measure expansion of the rock samples during imbibition of water and kerosene, in separate tests, using a linear variable differential transformer (LVDT). In the third phase, we measure imbibition-induced tensile stress during water imbibition into the samples. The results show that both HRB and DUV shale samples imbibe more water than kerosene, due to water adsorption by clay minerals. Imbibition of water increases the porosity of the HRB and the DUV samples by up to 0.94 and 0.25 percentage points, respectively. Expansion of all samples is anisotropic, with higher expansion perpendicular to the depositional lamination. Water imbibition into the samples induces an expansive stress as high as 17 psi. Moreover, applying confining stress reduces the imbibition of water by up to 18.1% and 33.7% in the HRB and DUV samples, respectively.

Abstract As observed in many shale-gas plays, the produced flowback water is highly saline and the salt concentration increases with time. Several past studies investigated water-rock interactions to interpret flowback chemical data, evaluate reservoir performance, and investigate the environmental impacts of fracturing operations. In this study, we measure the total ion produced (TIP) during flowback process for two wells completed in the Horn River Basin. We also conduct two sets of imbibition experiments to investigate the effects of water-rock surface area (As) and rock volume (Vs) on the TIP in laboratory. Furthermore, we compare the experimental correlations between As - TIP and Vs - TIP with the TIP measured in the field flowback water to estimate fracture surface area (AFrac) and invaded reservoir volume (IRV). In order to investigate the effect of As on the TIP, we conduct a series of imbibition experiments using shale samples of different As but similar Vs at constant temperature. The experiments are performed at T = 23, 45, and 65°C to investigate the temperature effect on the TIP. The experimental correlation between TIP and As at constant temperature is applied to estimate AFrac using field data of TIP. We further utilize AFrac - T correlation to extrapolate AFrac at reservoir temperature. In order to evaluate the estimated AFrac values we also calculate AFrac by rate-transient-analysis (RTA). In order to investigate the effect of Vs on the TIP, we conduct a series of imbibition experiments using shale samples of different Vs but similar As at constant temperature. Experimental results indicate that the TIP increases with both As and temperature. The calculated AFrac value at reservoir temperature is approximately 106m2 for both target wells. These results are in agreement with RTA calculation of AFrac values for both target wells (≈ 106m2). Our estimated values of AFrac are also in agreement with the field data of water recovery. The well with higher estimated value of AFrac has lower water recovery in the field as opposed to the well with lower estimated value of AFrac and higher water recovery in the field. Additionally, the estimated IRV is approximately 105 - 106m3 for both target wells. Our estimated values of IRV are also in agreement with the field data of water recovery and experimental results of water uptake. The well with higher estimated value of IRV has higher water uptake during imbibition experiments and also higher leak-off rate in the field. In contrast, the well with lower estimated value of IRV has lower water uptake during imbibition experiments and also lower leak-off rate in the field.

Summary After hydraulic fracturing, only 10 to 50% of the fracturing fluids is typically recovered. This paper investigates how the remaining fracturing fluids are imbibed by shale as a function of time, and it investigates the influence of various parameters on the imbibition process that include lithology, reservoir characteristics, and fluid properties. In addition, on the basis of experimental results, a numerical model has been developed to estimate the volume and rate of spontaneous imbibition over the entire fracture face. The rock samples are from the Horn River formation onshore Canada. The fracturing fluids used in the experiments included 2% KCl, 0.07% friction reducer, and 2% KCl substitute. In the experimental control group, distilled water was used. Through spontaneous-imbibition experiments, the relationship between imbibed fluid volume and time indicated that clay content was the most important factor that affected the total imbibed amount. Shale matrix with high clay content could imbibe more fracturing fluids than its measured porous space because of the clay's strong ability to expand and hold water. According to contact-angle-test results, the strongly water-wet shale samples had a faster imbibed rate. Total organic carbon (TOC) and porosity had no influence on imbibed volume and rate. These experimental findings can contribute to an improved fracturing-fluid design for different shale-formation conditions to reduce fluid loss. The experiment showed that 2% KCl and 2% KCl substitute fracturing fluids were imbibed from 10 to 40% less than 0.07% friction reducer in the shale formation with high clay content, whereas in the shale formation with low clay content, the opposite occurred. In the low-clay-content shale, 0.07%-friction-reducer test fluid was imbibed from 10 to 30% less than 2% KCl fluid, but had an imbibed amount similar to that of 2% KCl substitute fluid. The numerical-model result was matched with the experimental result to estimate a relative permeability in the model that could represent the rock properties. This model could be used to estimate the total imbibed volume along fracture faces through spontaneous imbibition.

Flowback water is usually highly saline and the salt concentration varies by time and well location. Understanding the origin of the flowback salts is essential for evaluating fracturing and flowback processes. In this study, laboratory and field analyses are performed to investigate the origin of the flowback salts. The field data includes the total salt concentration (salinity), individual ion concentration, pH, and dissolved oxygen measured during the flowback process for three wells completed in the Horn River Basin. The rock mineralogy is determined using XRD. The cation exchange capacity (CEC) of shale samples are measured using ammonium acetate method. Water and oil imbibition experiments are conducted for shale samples of different surface-to-volume ratio. The individual ion concentration is measured during the water imbibition experiments using ICP-MS and IC. EDXS analysis is used to investigate the surface of natural fractures.Noticeable amount of barium found on the surface of natural fractures suggests that the barium in the flowback water primarily originates from the natural fractures. Furthermore, the samples with higher clay content have higher CEC. During the water imbibition process, these samples have higher and faster ion transfer from shale -to -water; suggesting the mobilization of the exchangeable ions from the clays. During the water imbibition experiment, the Na/CI and K/CI ratios are initially high and decrease at the later times. Leaching of the exchangeable sodium and potassium ions from the clay minerals is a possible reason for the initial high Na/CI and K/CI molar ratios. The dissolution of chloride -bearing components increases the chloride concentration, which decreases the Na/CI and K/CI molar ratios at later times. The measured pH is slightly above 8 for all of the flowback water samples. The presence of natural buffer systems such as calcite and dolomite may explain the neutral pH range of the flowback water. (C) 2016 Elsevier Ltd. All rights reserved.

Hydraulic fracturing technology has been proven to significantly increase production from shale gas and oil formations. However, during a hydraulic fracturing treatment a large percentage of the fracturing fluid usually remains unrecovered. Therefore, the reasons for this low fracturing fluid recovery have become the focus of many studies. Imbibition of fracturing fluid in the shale is believed to be one of the explanations for the low amount of the fracture fluid recovery. The fluid is imbibed by the shale matrix and trapped inside the rock.Capillarity has generally been the primary mechanism considered during imbibition in conventional formations, such as sandstone and carbonate formations. In shale formations osmosis diffusion also exists and cannot be ignored because the clay in the shale rock functions similarly to that of a membrane. It is believed that both capillarity and osmosis diffusion work together to result in imbibition during hydraulic fracturing in shale formations. This paper investigates the effects of both capillarity and osmosis diffusion as the key mechanisms in fluid imbibition through simultaneous imbibition experiments. The results of these tests illustrate that the imbibition process is dominated by both capillarity and osmosis diffusion. This domination is based on the change of water saturation in shale rocks. In addition, the capillary and osmotic pressures, which influence the imbibed rate, can be qualitatively determined by the contact angle and salinity, respectively. Higher capillary and osmotic pressures correlate to faster rates of imbibition.This study, which examines the mechanisms of imbibition and their influences, can improve the understanding of fluid behavior when imbibition occurs during hydraulic fracturing in shale formations. The understanding of this behavior is useful for further simulation research. (C) 2016 Elsevier B.V. All rights reserved.

Summary As observed in many shale-gas operations, salt concentration of flowback water increases with time. Usually, the shape of salt-concentration/load-recovery plots is different from one well to another. We hypothesize that the shape of the salinity profile during the flowback process provides useful information about the complexity of the fracture network. In this study, we propose a model to describe the relationship between salinity and cumulative water production. We also compare the model results and flowback-salinity data to characterize the fracture network. Flowback-salinity data are collected from three multifractured horizontal wells completed in the three shale members [Muskwa (Mu), Otter-Park (OP), and Evie (Ev)] of the Horn River Basin. The salinity profiles for the Mu and OP wells initially increase and finally reach a plateau, whereas the salinity profile for the Ev well shows a continuous increase and does not show a plateau. We hypothesize that the early water with lower salt concentration at the onset of the flowback process is mainly produced from the primary fractures with larger aperture size. Also, we believe that the fractures with smaller aperture size become more important as the flowback process progresses, and therefore, the high-salinity water produced at later times is mainly produced from secondary fractures. We also propose a model to describe the salinity-profile behaviors. The model presents the aperture-size distribution (ASD) of the fracture network. A comparative analysis of the model results and the flowback-salinity data indicates that the Ev well with a steady increase in its salinity profile has a wider ASD compared with the Mu and OP wells with a plateau in their salinity profiles. This suggests that the fracture network is more complex in Ev compared with those in Mu and OP. More-complex fracture network in Ev is also in agreement with its higher gas and lower water recovery during the flowback process as opposed to the lower gas and higher water recovery in Mu and OP. The presented model for describing the behavior of the salinity profile during the flowback process and its meaningful relationship to the fracture-network complexity provide an alternative approach for reservoir characterization. This study encourages the industry to manage the flowback operations carefully and to monitor the water chemistry.

Abstract Recently, flowback chemical analysis has been considered as a complementary approach for evaluating fracturing operations and characterizing reservoir properties. Understanding the source of flowback salts and the mechanisms controlling the water chemistry is essential but also challenging due to the complexity of shale-water interactions. In this study, samples of flowback water and downhole shales are analyzed to investigate the mechanisms controlling the chemistry of flowback water. The water samples at different flowback times and the shale samples are collected from three wells completed in the Muskwa, Otter-Park, and Evie members of the Horn River Basin. The water samples consist of aqueous solution and precipitated salts. The water samples are digested in nitric acid to dissolve the precipitated salts, and are analyzed at both intact and acid-digested conditions using ICP-MS. The flowback salts are weighted and analyzed using XRD and SEM-EDXS. A sequential ion-extraction is performed on the shale samples; and the extracted ions are categorized into three tiers of loosely-, moderately-, and strongly-attached ions. The concentration of monovalent cations in both intact and acid-digested samples is higher than that of divalent cations. Also, the concentration of all cations is higher in the acid-digested samples compared with that in the intact samples. The ratio of divalent cations concentration in the acid-digested samples to that in the intact samples is higher than that for the monovalent cations. This ratio increases for the divalent cations over time, while it remains constant for the monovalent cations. Additionally, for the acid-digested samples the monovalent cations concentration has an initial sharp increase followed by a slower increase at later flowback stages; while the divalent cations concentration increases continuously over time. These results suggest that the majority of the ions in the early flowback water are loosely-attached monovalent ions. These ions can be originated from the mixing with in-situ formation brine, dissolution of soluble precipitated salts, or leaching of exchangeable cations from the clay minerals. Similarly, the role of relatively slow water-rock interactions (such as leaching of divalent exchangeable cations, e.g. Ca2+,) increases at the later flowback stages. XRD and SEM-EDXS analyzes of the flowback salts indicate that sodium chloride, potassium chloride, and calcium carbonate are the major salts. The sequential ion-extraction reveals that the majority of the monovalent cations are in the loosely-attached tier. However, majority of the divalent cations are moderately- /strongly-attached to the rock. The strongly-attached portion of the ions is determined by acid digestion of the rock sample at the final stage of sequential extraction process. These strongly-attached ions cannot be easily released by hydraulic fracturing and therefore, has small effect on the flowback water chemistry.

Abstract Several past studies have focused on the saline flowback water to evaluate the hydraulic fracturing operations. The origin of the salts in the flowback water is important for the assessment of the flowback process. In this study, laboratory and field analyses are performed to provide a better understanding about the origin of the flowback salts. The field study analyzes the total salt concentration (salinity) and ion concentration data measured during the flowback process for the Muskwa (Mu), Otter-Park (OP), and Evie (Ev) formations. The concentration profiles of both the barium and chloride during the flowback process, whereas the iron concentration declines after experiencing an initial increase. The laboratory study encompasses contact angle, XRD, imbibition, individual ion concentration, surface element, and adsorption isotherm experiments for samples from the OP and Ev formations. To investigate the effects of fluid-rock interface area on the liquid uptake and diffusion rate of individual ions, a series of imbibition experiments are carried out for different values of surface to volume ratios (specific surface or "Asp"). The electrical conductivity and individual ion concentrations are measured during the imbibition process. XRD data is analyzed to determine the mineralogy of the samples. SEM-EDX analysis is performed to determine the distribution of the elements on fresh break and natural fracture surfaces of the samples (in addition to a sample from the Lower Keg (LK) formation). Finally, since both ion transfer and water adsorption processes occur during the imbibition experiment, an adsorption isotherm experiment is carried out to prevent the ion transfer into/out of the rock in order to solely study the water adsorption process. The laboratory results show that barium is mainly concentrated in the natural fractures; and therefore the shape of the barium concentration profile in the flowback water maybe an indication of the complexity of the fracture network.

Abstract During some hydraulic fracturing processes, it is found that only 10% to 50% of fracturing fluids is recovered. This paper investigates how much of the remaining fracturing fluids are imbibed by shale rocks as a function of time, and also investigates the influence of various parameters on the imbibition process including: lithology, reservoir characteristics, and fluid properties. In addition, based on the experimental results, a numerical model is developed to estimate the amount and rate of spontaneous imbibition during fracturing over the entire fracture face. The rock samples are from the Horn River formation. The fracturing fluids used in the experiments include 2% KCL, 0.07% friction reducer and 2% KCL substitute. Distilled water is also used in experiments as control groups. Through spontaneous imbibition experiments, the relationships between imbibed weight of fluid and time show that the content of clay is the most important factor which affects the total amount imbibed. Shale matrix with high clay content can imbibe volume of fluid greater than its measured porous space because of the clay's strong ability to expand and hold water. Small porosity with less total organic carbon (TOC) results in the highest imbibed rate. Contact angle results show the stronger water wet shale samples have a faster imbibed rate. Temperature also influence amount of imbibition. The total imbibed volume decreases as the environment temperature rises. From this paper, it can help optimally design fracturing fluid for different conditions of shale formation to reduce fluid loss. We find that 2% KCL and 2% KCL substitute fracturing fluid are imbibed 10% to 40% less than 0.07% friction reducer in shale formation with high clay content; while in shale formation with low clay content, the opposite occurs. 0.07% friction reducer is imbibed 10% to 30% less than 2% KCL, but has similar imbibed amount with 2% KCL substitute. The numerical model results are matched with the experiment results in order to estimate relative permeability and initial water saturation in the model which can represent the properties of rocks. This model can use to estimate the total imbibed volume along fracture faces through spontaneous imbibition.

Abstract Understanding water uptake of gas shales is critical for designing fracturing and treatment fluids. Previous imbibition experiments on unconfined gas shales have led to several key observations. The water uptake of dry shales is higher than their oil uptake. Furthermore, water imbibition results in sample expansion and microfracture induction. This study provides additional experimental data to understand the effects of rock fabric, complex pore network, and clay swelling on imbibition behavior. We systematically measure the imbibition rates of fresh water, brine and oil into the confined and unconfined rock samples and crushed packs from different shale members of the Horn River Basin. We also measure the ion diffusion rate from shale into water during imbibition experiments. The results show that confining the shale samples decreases the water imbibition rate of samples tested parallel to the bedding. However, it has a negligible effect on water uptake of samples tested perpendicular to the bedding and on ion diffusion rates. The comparative study suggests that, for both confined and unconfined samples, water uptake is higher than oil uptake. The liquid imbibition and ion diffusion rates along the bedding are higher than those against the bedding. Surprisingly, the crushed samples show a completely different behavior. The oil uptake of crushed packs is higher than their water uptake. The data suggest that the connected pore network of the intact samples is water wet while the majority of rock including poorly connected pores is oil wet. This argument is backed by complete spreading of oil on fresh break surfaces of the rock.

Abstract Field studies demonstrate low flowback efficiency determined by volumetric analysis of injected and recovered fracturing water. However, the reasons for inefficient water recovery, and its impact on short-term and long term production are poorly understood. Furthermore, volumetric water analysis is not sufficient for determining the source of recovered water and the true load recovery. This paper aims at understanding how flowback efficiency is related to the imbibition process, presence or absence of natural fractures, and the complexity of induced fracture network. We interpret the flowback rate and salt concentration, measured from several multi-fractured horizontal wells recently completed in different members of the Horn River basin. We also measure and analyze water imbibition and salt diffusion rate in actual cores drilled from the same shale members. The wells are classified into those with 1) low water and high gas production, 2) high water and low gas production. This classification is explained by lab imbibition data and possible fracture patterns. Furthermore, the flowback salt concentration change is explained by the diffusion data measured in the laboratory. This systematic study provides a practical database for understanding the factors impacting the water recovery that will potentially help the operators to optimize the flowback operations, and obtain useful information about the induced fracture network.