We present a case study of hydraulic fracturing treatments for two horizontal wells located in the Horn River Basin, B.C. The wells were completed using a technique that we refer to as Short Interval Re-injection (SIR). For each individual treatment stage, this technique makes use of an initial injection interval using conventional hydraulic fracturing pumping procedures, followed by a “soaking" period that may last from a few hours to about one day in duration, during which the well is temporary shut in. This is followed by a subsequent re-injection interval with a pumping schedule similar to the first interval. Several commercial names are in use to describe this type of approach, which has a desired goal of enhancing the overall effectiveness of the treatment. In this study, we observe a significant increase in the rate of microseismic activity that occurs after the initial soaking period. This type of response has been documented previously and, in some cases, has been empirically related to increased production for wells. We postulate that cohesion of pre-existing fractures is reduced by the initial injection and soaking period, facilitating reactivation of fractures during the second injection. A numerical model has been developed using the software 3DEC in which the cohesion parameter for a discrete fracture network (DFN) is set to zero after the first injection stage. Preliminary results produce a satisfactory match with respect to increased events. Future work will include adjustments to the DFN in order to increase the match with the spatial locations.
The Horn River Basin (HRB) is an important resource play in northeastern British Columbia, Canada. While conventional oil and gas developments have been underway in the HRB for several decades, since 2005 operators have targeted the large shale resources that are in place.
As commodity prices have declined, refracturing has given operators an alternative way to obtain positive returns with lower investment under constrained capital markets. A major operator in the Haynesville shale was interested in determining the optimum method to refracture several laterals located directly offset to each other. In this case, a four-well pad was initially drilled to drain the section. The viability and optimum sequence/design of refracturing this four-well pad was unknown. There were many uncertainties around refracturing this pad including refracturing approach/method, refracture timing, refracture sequence (well order), refracture job size, and number of wells to refracture to obtain the greatest return on investment.
In this paper, an integrated refracturing workflow was created and applied to determine the optimum refracturing strategy for this four-well pad. This comprehensive workflow represents a multidisciplinary approach that integrates complex hydraulic fracture models, geomechanical models, and multiwell production simulation. The unique approach in this workflow was the ability to couple simulated 3D reservoir pressure with a geomechanical finite-element model (FEM) to quantify the changes to the magnitude and azimuth of the in-situ stresses from the depletion. Then, the altered stress field was utilized as the input for modeling the new fracture system created by the refracturing treatment. A separate refracturing workflow was developed to calibrate the proposed four-well refracture study by fracture modeling and production history matching of a previously refractured well a few miles away, and it was also applied to run sensitivities on modeling the proposed refracture treatments on the four-well pad.
This new unique approach to studying pad refracturing was beneficial to understanding the viability of refracturing this four-well pad in the Haynesville shale and the influence of refracturing on the existing fracture networks. The optimized completion strategy and workflow can help operators calibrate expectations and optimize the refracturing process for pad wells to obtain the best return on investment.
The Haynesville shale is a unique dry gas formation located in northeast Texas/northwest Louisiana with high reservoir pressure, gradients of 0.85 to 0.9 psi/ft (Fan et al. 2010; Thompson et al. 2010). The higher reservoir pressure and clay content cause the play to be prone to creep, which causes severe permeability reduction as wells are produced with high drawdowns (Thompson et al. 2010; Okouma Mangha et al. 2011; Indras and Blankenship 2015). By controlling the severe production drawdown, existing fracture networks were maintained for longer periods of time (Baihly et al. 2015).
Wells in unconventional, ultra-low permeability reservoirs have significantly longer duration of transient flow regimes than wells in conventional reservoirs due to different reservoir characteristics and production methodologies. This calls for new methodologies developed specifically for transient flow analysis, such as the pioneering work by Duong (SPE 137748).
This paper presents a critique of the Duong method and proposes a modified and improved version of the method which enhances the accuracy of forecasting production in ultra-low permeability reservoirs.
Our study examined the derivation of the Duong model and highlighted the lack of taking fracture damage into account. This causes inconsistent expressions of flow rate and cumulative production as functions of time. Further, most wells produce with changing pressure drawdowns, and this also creates an inconsistency between the flow and cumulative production equations that Duong assumed in deriving his decline model.
We therefore propose a modified and improved version of the Duong method using basic transient flow equations that account for the presence of fracture skin and changing pressure drawdown. We used simulated production histories to carry out a comparative study of the predictions from the original and modified Duong methods to highlight the effect of fracture skin and varying pressure drawdown on the production forecast results.
Results from our analysis show that Duong method experiences problems in accuracy when early time pressure drawdowns increase and, independently, when fracture damage is present. Our modified method removes the inaccuracies due to fracture damage and can remove inaccuracies due to changing pressure drawdown when pressures are known. For unknown pressure histories, our method can at least minimize the impact of changing pressure drawdown.
This study shows that the Duong method is most accurate under constant pressure drawdown and no fracture damage. With changing bottomhole pressure, and presence of fracture damage, production forecasts using the original Duong model can be quite inaccurate. Our modified Duong method, which uses pressure-normalized rates and accounts for the presence of fracture damage, leads to more accurate production forecasts. The parameters calculated using the modified Duong method also provide meaningful insight into reservoir and fracture properties, whereas the empirical parameters calculated using the original Duong model fail to do so.
Understanding of proppant transport and deposition patterns in a hydraulic fracture is vital for effective and economical production of petroleum hydrocarbons. In this research, a numerical modeling approach, combining Particle Flow Code (PFC) with lattice Boltzmann (LB) simulation, was adopted to advance the understanding of the hydraulic fracture conductivity as a function of proppant concentration under various effective stresses from partial monolayer to multilayer proppant concentrations. PFC was used to simulate effective stress increase and the resultant proppant particle movement and rearrangement during the process of reservoir depletion due to hydrocarbon production. The pore structure of the proppant pack was extracted and used as boundary conditions of the LB simulation to calculate the time-dependent permeability of the proppant pack. We first validated the PFC-LB numerical workflow, and the simulated proppant pack permeabilities as functions of effective stress were in good agreement with laboratory measurements. Furthermore, several proppant packs were generated with various proppant concentrations, ranging from partial monolayer proppant packs to multilayer ones. The fracture conductivities for proppant packs from partial monolayers to multilayers were simulated. A partial monolayer proppant pack with large-diameter proppants may be an alternative to increase the fracture conductivity. Then, three proppant packs with the same average diameter but different diameter distributions were generated. Specifically, we used the coefficient of variation (COV) of diameter, defined as the ratio of standard deviation of diameter to mean diameter, to characterize the heterogeneity of particle size. We obtained proppant pack porosity, permeability, and fracture width reduction (compressed distance) as functions of effective stress. Under the same effective stress, a proppant pack with a higher diameter COV had lower porosity and permeability and larger fracture width reduction. This was because the high diameter COV gave rise to a wider diameter distribution of proppant particles; smaller particles were compressed into the pore space between larger particles with the increasing stress, leading to larger compressed distance and lower porosity and permeability. The transition time distributions for proppant packs with diameter COV5% and COV19% were determined and hydrocarbons transported more difficultly within the COV19% proppant geometry. With identical stress increase, the proppant assembly having a more heterogeneous particle diameter distribution experienced more dramatic changes with respect to pore structure and connectivity, which directly led to a reduced hydrocarbon transport in the hydraulic fracture.
The Vaca Muerta Fm mainly consists of black mudstones, variably calcitic and siliceous, rich or poor in bioclasts and with a variable content of calcitic radiolarians. These mudstones are punctuated by a variable proportion of heterogeneities such as parallel bedded fractures filled with fibrous calcite (the “beef”), calcitic nodules and ash beds. Understanding the genesis of these beef is important because they may represent up to 4% of the facies. These beef will also likely affect the propagation and efficiency of hydraulic fracturing during stimulation.
Several hundred meters of cores from different wells of various maturities were studied at very high resolution, including 1 cm spacing Total Organic Carbon (TOC) measurement by the LIPS (Laser Induced Pyrolysis System) along with detailed sedimentological core descriptions.
At least two main types of beef are observed on cores. The first type is laterally continuous, at the core scale, displaying or not a median line with fibrous calcite growth. The second type, called micro-beef, is thinner (a few mm), discontinuous and in relay, lenticular or sigmoidal. Regarding the continuous beef, no link was evidenced with the surrounding depositional facies of the source rock neither with the mineralogical content (such as calcite). More surprisingly, no clear correlation was found between beef occurrence and TOC values although beef are only found in source rocks. The cumulated thickness of beef increases with maturity and a large proportion of beef develops along or close to heterogeneities such as ash beds (for the most part) or calcitic nodules. This proportion is decreasing with maturity and a phasing in beef formation is evidenced. The clayey ash beds are among the most porous facies (13% porosity in average). Similarly the calcitic nodules were initially also porous facies. The hypothesis which is made is that, first, fluids preferentially circulate and concentrate within these thin high porosity streaks and, due to overpressure created inside these tiny reservoir beds, natural hydraulic fractures are created at their interfaces and subsequently filled by calcite that starts crystallizing from the bed interfaces. In the latter case, a median line, i.e. crystallization up and down from the fracture, is not systematically observed. With increasing maturity and increasing hydrocarbon and aqueous calcitic fluids expulsion and migration, overpressure is generated in the shale matrix near the ash beds and more beef can be formed. Those latter beef develop a median line. With maturity further increasing, beef are generated everywhere else in the series where TOC reaches at least 2%.
Akkutlu, I. Yucel (Texas A&M University) | Baek, Seunghwan (Texas A&M University) | Olorode, Olufemi M. (Texas A&M University) | Wei, Pang (Sinopec Research Institute of Petroleum Eng.) | Tongyi, Zhang (Sinopec Research Institute of Petroleum Eng.) | Shuang, Ai (Sinopec Research Institute of Petroleum Eng.)
Organic-rich shale formations consist of multi-scale pore structure, which includes pores with sizes down to nano-scale, contributing to the storage of hydrocarbons. In this paper, we show that the hydrocarbons in the formation partition into fluids with significantly varying physical properties across the nanopore size distribution of shale. This partitioning is a consequence of multi-component hydrocarbon mixture stored in nanopores showing a significant compositional variation with the pore size. The smaller the pore is, the heavier and the more viscous the hydrocarbon mixture becomes. During the production and pressure depletion, primarily the lighter hydrocarbons of the mixture are released from the nanopores. Hence, the composition of the remaining hydrocarbons inside the pores becomes progressively heavier. The viscosity and apparent molecular weight of the hydrocarbon mixture left behind increase significantly during the depletion. The kinetic mean-free path length of the mixture does not increase, however, as anticipated from the kinetic theory of gases. Further, the length may decrease drastically in small nanopores as an indication of capillary condensation and trapping of the hydrocarbon mixture. These effects significantly limit the release of hydrocarbons from nanopores, in particular those pores with sizes smaller than 10nm.
In the light of these microscopic scale observations, the concept of composition redistribution of the produced fluids is introduced and a new volumetric method is presented honoring the compositional variability in nanopores for an improved accuracy in predicting hydrocarbons in-place in presence of adsorption and nano-confinement effects. The method allows us to differentiate mobile bulk hydrocarbon fluids from the fluids under confinement effects and from the trapped hydrocarbon fluid dissolved in the organic material. Hence, it also reduces the uncertainties in predicting the reserve. The application of the method is presented using produced hydrocarbon fluid composition for dry gas and wet-gas formations and using reservoir flow simulation of production from a multi-stage fractured single horizontal well. We showed that liquids production is mainly due to flow of bulk fluid in large-pore volume.
Measurements of fluid wetting characteristic are made routinely on rock samples. However, there are no published petrophysical models to differentiate between oil-wet and water-wet fractions of a reservoir sequence using commonly available log suites.
This presentation builds on our previous publication that describes the unconventional reservoir petrophysical model we have developed (Holmes 2014). Essentially, we define four porosity components, namely total organic carbon, clay porosity, effective porosity (inorganic), and effective porosity (organic). This last component, which is associated with total organic carbon, is an indirect calculation if the first three components do not sum to total porosity.
Porosity/resistivity plots can be constructed for the total porosity and interpreted in a standard fashion. These will mostly indicate a water-wet system when the effective porosity (inorganic) fraction is examined. A second porosity/resistivity plot compares resistivity with effective porosity (organic), and is interpreted to indicate Archie saturation exponents of much larger than 2 – frequently in excess of 3 – indicating the oil-wet fraction of the reservoir system. Additionally, the plots suggest very low values of the cementation exponent of 1.0.
Examples from the Bakken of North Dakota and the Wolfcamp of Texas are presented showing quantitative distinction of water-wet vs. oil-wet reservoir components.
It is commonly recognized that mixed wetting occurs in unconventional oil reservoir systems – part of the porosity fabric is water-wet and part is oil-wet. Measurements are made on rock samples to define wetting characteristics. However, in addition, there are data available from triple-combo log suites which can be analyzed to define wetting characteristics.
Gherabati, S. Amin (The University of Texas at Austin) | Browning, John (The University of Texas at Austin) | Male, Frank (The University of Texas at Austin) | Hamlin, Scott (The University of Texas at Austin) | Smye, Katie (The University of Texas at Austin) | Walsh, Mark (The University of Texas at Austin) | Ikonnikova, Svetlana A. (The University of Texas at Austin) | McDaid, Guinevere (The University of Texas at Austin) | Lemons , Casee (The University of Texas at Austin)
This paper presents an integrated workflow for hydrocarbon-in-place and recovery factor estimation in Bakken. Evaluating factors that control generation and storage of hydrocarbon such as total organic carbon (TOC), maturity of shale, thickness, porosity and permeability is a challenge in any shale play study. In addition, hybrid nature of the Bakken petroleum system where source and reservoir rock present within short depth interval adds complexity to the production interpretation and outlook of the play. One of the complexities is the contribution from the shaley Upper and Lower Bakken to the production of horizontal wells completed in the Middle Bakken. In addition, presence of local, structural and stratigraphic traps complicates mapping water saturation distribution in Middle Bakken that is mostly affected by hydrocarbon generation in the Upper and Lower Bakken.
We address geological and petrophysical uncertainties and calculate and map hydrocarbon pore volume. For fluid characterization, we use three models in order to accurately cover a range of American Petroleum Institute (API) gravity and gas/oil ratio (GOR). We evaluate the contribution of Upper and Lower Bakken in production by constructing simulation models and used that knowledge to estimate recovery factor of the horizontal wells.
Production depletes the Middle Bakken, creating a pressure difference between the Middle Bakken and the Upper/Lower Bakken, which in turn depletes the Upper/Lower Bakken. Vertical permeability controls production from the Upper and Lower Bakken, and higher vertical permeability increases the contribution of the two shale members. An understanding of maturity and trap mechanism can help to explain the water saturation distribution and understanding these factors is crucial to any future development of the play.
Permeability of fractures is a key uncertainty in the analysis of unconventional oil and gas reservoirs. This uncertainty is greatest at in situ reservoir conditions and when attempting to predict the permeability of fresh fractures created via tensile stress or critical stress shearing. In this study, we measure permeability of shale specimens shear fractured at stressed conditions and combine these measurements with concurrent x-ray tomographic characterization of the fracture systems. The fractures were created using a specialized triaxial system configured for either direct-shear fracturing. Confining stresses up to 30 MPa (4400 psi) were used to simulate moderate reservoir depths. The body of the apparatus is an aluminum coreholder that permits acquisition of simultaneous x-ray radiography or tomography data together with permeability measurements and shear fracture displacement. In both clay-rich Utica and carbonate-rich Marcellus shale, we observe that increasing confining stress during fracturing results in significant decreases in fracture permeability ranging over 3 to 4 orders of magnitude. In a clay-rich direct-shear specimen, negligible permeability development was observed at confining stress of 22 MPa (3200 psi) despite significant deformation. The large variations in permeability as a function of stress at fracture initiation are in contrast to the lesser effect of changes in confining stress on already-formed fractures with a corresponding range of one to two orders of magnitude. These observations of permeability are corroborated by x-ray radiography and tomography that show a reduction in fracture aperture with increasing stress.
Hydrocarbon production from unconventional shale resources depends on a number of factors including the extent and penetration of the hydraulic fractures (HF), distribution of proppant, the hydrocarbon content, matrix properties, pre-existing fractures, etc. A number of studies have concluded that existing, potentially reactivated fracture systems must contribute hydrocarbon to account for observed HF productivity (e.g., Gale et al. 2007; Johri and Zoback 2013; McClure and Horne 2014). However, little is known about “natural” shear fracture permeability and the changes in fracture permeability in response to changing stress or fracture reactivation. This results in significant uncertainty in the potential role of natural fractures in hydrocarbon production. For example, discrete fracture network models of unconventional reservoirs use an array of natural fractures to reproduce production data (Karra et al. 2015).
The accurate placement and positioning of horizontal wells is important in developing an unconventional reservoir. 3D seismic and pilot wells are used to determine an accurate depth and facies model of the subsurface. Carbonate debris flows need to be accurately mapped as they influence horizontal drilling and act as fracture barriers during hydraulic stimulation. Prestack seismic inversion is often used to derive estimates of the rock properties, facies models and help guide the development of the field.
A combination of model based inversion and supervised neural networks is used to develop high resolution rock property volumes from surface seismic data. Several rock property volumes are derived through this process including acoustic impedance, shear impedance, p-wave velocity, s-wave velocity, Young’s modulus, Poisson ratio, brittleness, and critical strain. These volumes have higher frequency and are calibrated to fit the well data. In addition to these rock volumes, the acoustic impedance volume can be used to derive a better seismic volume, a Reflection Coefficient (RC) volume. The RC volume has much higher frequency, better lateral continuity, and ties the well logs better than conventional seismic or frequency enhanced data. By interpreting and mapping with this RC volume, a much more accurate depth model can be built. Reflectors that were previously un-mappable on conventional seismic can be mapped and horizontal wells can be more accurately placed. This paper will explain the inversion work flow followed by an example of a problematic horizontal well drilled using conventional seismic that can be more accurately explained using the RC depth volume.
The development of unconventional reservoirs hinges on the accurate positioning of horizontal wells into brittle organic rich shales. Seismic data has been instrumental in accurately positioning these wells in depth and avoiding large faults, carbonate debris flows or other types of hazards. In addition to steering horizontal wells, rock property volumes can be derived from surface seismic to map out different types of facies.