Journal of Petroleum Science and Engineering 174 (2019) 235–243
Contents lists available at ScienceDirect
Journal of Petroleum Science and Engineering journal homepage: www.elsevier.com/locate/petrol
Subsurface well spacing optimization in the Permian Basin ∗
T
Baosheng Liang , Meilin Du, Pablo Paez Yanez Chevron North America Exploration & Production, 1400 Smith Street, Houston, TX, 77002, USA
A R T I C LE I N FO
A B S T R A C T
Keywords: Well spacing Completions design Permian basin Reservoir modeling Unconventional reservoir Field analog
Optimum subsurface well spacing is key to developing unconventional reservoirs. From a field application perspective, this paper presents our systematic study on subsurface well spacing in the Permian Basin for unconventional reservoirs which consisted of four main components, i.e., numerical modeling, well interference quantification through simulation and regression, field pilot analogs, and economic evaluation. In this paper, field pilot wells are actual wells in the field to test different well spacing and completion designs. To capture upsized and downsized completions, a wide range of fracture designs in combination with different well spacing scenarios are conducted for both hydraulic fracture modeling and reservoir simulation. At the section level (1 mile by 1 mile, 640 acres), multiple wells are simulated to better capture well interactions and reservoir property variations. A complex fracture network is generated by considering interactions between natural fracture and hydraulic fracture. Well interference, which is determined by estimating ultimate recovery (EUR) difference between a single well and a middle well from multiple wells, is analyzed for general trend regressions and validated through field test results. At the section level, economics is done to evaluate capital efficiency of various scenarios. Results: from both modeling work and pilots indicate that larger hydraulic fracture size without an increase in subsurface well spacing does not necessarily improve section EUR and there is a point of diminishing returns. Larger subsurface well spacing with bigger fracture size is thereby a preferable combination.
1. Introduction Well spacing can be simply defined the space or acreage allocated to a well. It is essentially a measurement of well locations and the number of wells that can be drilled to drain a reservoir economically. In unconventional reservoirs, well spacing is tightly related with well performance, company reserves and economics. Many factors constrain well spacing, including not only commodity price, surface land position, facility capacity, infrastructure, company short- and long-term strategies, economics and recovery factor, but also subsurface reservoir properties and completions design. In this paper, we try to minimize the impacts of surface facilities and operator variations and then refer well spacing to subsurface well spacing only. Due to the nature of depletion process in unconventional reservoir, effective well spacing exhibits three characteristics which should be quantified: spatial distribution, dynamic reservoir changes, and development economics. Spatial distribution means that well spacing is the 3-dimensional problem, with variations of reservoir properties and staggered landings. Dynamic changes refer to time dependent changes in fracture geometries with production. Also, it presents the change of drainage volume while only pressure responses can be acquired and ∗
analyzed. The optimum spacing finally should be determined by economic analysis between well production performance and costs. Smaller spacing may be recommended to compromise economics and well interference. Many authors have studied well spacing and their recent work is summarized in Appendix. In general, one of three methods are used: field trial, data analytics, and simulation. However, there is no systematic while practical study at field development level through integrating complex fracture geometry and well production. 2. Method In this paper, our work leveraged information from different sources: numerical model building and calibration; well interference quantification through simulation and regression; field pilot analogs; and, economic evaluation. Workflow and the associated contents for each step is shown in Fig. 1. Reliable well interference results depend on the assumptions of the accurate numerical simulation and representative field pilots. However, the problem is far more complex. As discussed in the following sections, there are still many uncertainties with respect to properties in 3D earth model, hydraulic fracture
Corresponding author. E-mail address:
[email protected] (B. Liang).
https://doi.org/10.1016/j.petrol.2018.11.010 Received 12 January 2018; Received in revised form 2 November 2018; Accepted 4 November 2018 Available online 07 November 2018 0920-4105/ © 2018 Elsevier B.V. All rights reserved.
Journal of Petroleum Science and Engineering 174 (2019) 235–243
B. Liang et al.
Fig. 1. Workflow and associated contents for each step in this paper.
modeling, reservoir simulation, and field pilot analogs. In this study, wells are assumed to have the same landings. Two well spacing scenarios with a range of proppant intensities were considered in this study, due to the potential large number of simulations, less representative in the field, unavailable completion designs applied to the field, and unavailable field pilots for analogs. Therefore, though the workflow presented in this paper is general, our result with small ranges aimed at the short-term business decisions and searched local optimization instead of the global optimization. First, collected data from log, core, pressure and well production from the pilots and explorations, microseismic events and seismic interpretation were used to build and calibrate numerical geologic, hydraulic and reservoir models in an integrated platform. Non-planar complex hydraulic fracture geometries were modeled by considering fracture propagation at the presence of natural fracture and stress shadow effect within each stage and among neighbor stages. Hydraulic fracture geometries, conductivities, and geomechanical compaction tables are key targets during history matching. Second, multiple simulations at various pumping sizes and well distances were conducted. Both completion design and well spacing are considered simultaneously. Stage length and cluster spacing are further selected based on data analysis of development trends in this study, helping generate three representative completions designs and reduce significantly the required number of simulation runs. Pumping size and well distance are then the focus and the combinations are simulated exhaustively. Those results were further analyzed for general trend regression and simplified the complex process. This step was critical because it helped reveal the inner relationship among the parameters and interference percentage. Such relationship otherwise would not be identified purely through field pilots which are constrained by the limited numbers with large variations of reservoir quality, landing, completions design, well spacing, and operations. Third, more than ten field pilots with different well spacing and completion designs were carefully analyzed and compared with regression results. Data analytics by using public data was tried with little meaningful findings, mainly because of the lack of detailed information
of completions, production history, and the nearby well interference. As real field tests for our simulation work, the pilots mostly are from joint ventures with good data collection and detailed information, covering different locations, completions design, and operators. Overall, analogs showed a good agreement with regression formula. Fourth, economic model was built to analyze profitability and further help the final decisions, followed by discussions of future improvements. 3. Numerical model building and calibration In unconventional reservoir, two key parameters to determine well spacing are completions design and well interference. From modeling perspective, three-dimensional earth model is necessary because it better honors vertical and lateral heterogeneities, spatial variability in reservoir characteristics such as fluid property, reservoir pressure and hydrocarbon in place. More importantly, three-dimensional model is also required to quantify multiple well interference and capture interactions among multi-stages, multi-pads, and multi-landing targets. Therefore, we pursued well spacing issue through well interference quantification by comparing simulation results from single well and multiple wells at the different well spacing and completion designs. Well performance simulation through 3D model is described first, followed by model calibration. There are three integrated steps to simulate well performance in this work: 3D model building, single or multiple well hydraulic fracture modeling, and reservoir simulation of well production with hybrid grids from 3D model and stimulated hydraulic fracture. Building a comprehensive 3D model requires a multi-disciplinary effort, starting with data collection and QC, and integrating structural, stratigraphic, petrophysical and geomechanical models. 3D models were built and calibrated in areas of interest in both Delaware and Midland basins. In the Midland Basin, we leveraged extensive information from existing vertical wells when building the 3D model for horizontal well program. More information about building Midland Basin model could be found in Liang et al. (2017) and Liang et al. 236
Journal of Petroleum Science and Engineering 174 (2019) 235–243
B. Liang et al.
cores with and without proppants were used to construct those compaction tables. After 3D earth models were built, we calibrated them through the following three major steps: (1) initial model QC through analyzing property correlation, histogram, heterogeneity variation, lateral and vertical cross-sectional continuity. (2) hydraulic fracture geometry calibration with results such as microseismic data, pump treatment, pressure communication analysis, and RTA, through adjusting DFN, petrophysical properties and rock mechanical properties. (3) well performance history matching mainly via compaction tables from matrix, different proppants and unpropped area to represent geomechanical effect, initial unpropped conductivity, PVT, and relative permeability curves. Modifications during production history matching is handled by looping back to earth model and then hydraulic fracturing to honor the whole process. Although initial water saturation and water flowback were assessed through adding dilation tables and converting well into water injector to mimic hydraulic fracturing process, we did not match water flowback in this well spacing study and longer water-cut match was fairly good. The following figures are examples from one model in the Delaware Basin. Fig. 2 shows a porosity model. Fig. 3 illustrates the modeled hydraulic fractures along the lateral. Fig. 4 presents match results for one appraisal well. In this well, we constrained oil production rate and matched bottom hole pressure, GOR and water cut. GOR matches in the later time were not as good as the early period because of greater uncertainties in the data after gas lift installation in March 2014. Bottom hole pressure match is quite sufficient during the two periods where downhole gauge data are available. Finally, the longterm water cut matches the field data quite well. The simulation suggests that after the initial flowback period, a large portion of produced water is mobile water from the formation.
Fig. 2. Porosity model in one area of interest in the Delaware Basin.
(2018). While in the Delaware Basin, we utilized more exploration wells in different areas and formations and integrated with datasets from horizontal wells, pilot holes, modern 3D seismic data (Sun et al., 2016; Swenberg et al., 2017) including whole cores, sidewall cores, triple combo logging suite (dipole sonic, nuclear magnetic resonance, and image log) in pilot holes, and triple combo logging suite with image logs for lateral evaluation, microseismic data, surface and downhole pressures, and daily production. Data along the wellbore included advanced cutting analyses, geomechanical tests on cores, fiber optics, tracer tests, DFITs, and PVT tests. Each hydraulic fracture was traditionally treated as bi-wing planar with the absence of complex 3D earth model, natural fractures and interactions among multiple hydraulic fractures. In this work, we used the unconventional fracture model workflow (Wu et al., 2012) and considered hydraulic fracture propagation during the interactions with discrete natural fracture. 2D natural fracture network was assumed the same for all the formations in each development area and it was characterized statistically based on average values and standard deviations of natural fracture orientation, length and density. Such statistical analyses came from openhole logs, cores and seismic interpretation. Because of the large uncertainties and significant impact on hydraulic fracture geometry, 2D natural fracture network was a key parameter for history matching. For each well, perforations, clusters and stages were set up along the lateral wellbore based on completion design. Hydraulic fracture network at each stage was simulated through pumping schedule and interactions between induced fracture and natural fracture. Stress shadow effect among multiple clusters within each stage as well as intra stages were considered during the stimulation. The third step is reservoir simulation of well production with hybrid unstructured grids from 3D model build in the first step and stimulated hydraulic fracture from the second step. One of the disadvantages of using unstructured grid for 3D multi-well reservoir simulation is the significant computational time with possible convergence challenges. By implicitly considering fracture information into the matrix background grid system and avoiding the challenges of complex meshing process, mesh free methods such as multi-segment well (Du et al., 2016) and EDFM (Du et al., 2017), are potential alternatives. However, those methods were in testing stage when this work was studied and unstructured grid was still used here for reservoir simulation. New permeability values were generated from previous hydraulic fracture model and correlations for different regions which were either unpropped or propped by different types of proppants. Note that because heterogeneity was captured from our 3D earth model, hydraulic fracture network exhibited the variation along the horizontal well lateral. In the numerical simulation process, single porosity approach was used. Geomechanical impacts were considered through different compaction tables for different reservoir regions, i.e., matrix, each proppant type, and unpropped region. Conductivity tests measured in the lab from
4. Well interference quantification Well spacing task essentially boils down to well interference quantification given different well distances and completions design. If detailed completions design and well spacing are optimized simultaneously, it will be too complicated to consider all the parameters. As such, in this paper we introduced a sequence concept to first break a complex issue into smaller pieces and then prioritize them to a level that can be practically handled at a given time duration. Fig. 5 shows a general decision sequence for field development in a new unconventional asset, consisting of completions design particularly with key parameters cluster spacing and number of clusters per stage, multi-well decisions mainly including well spacing and proppant intensity, and
Fig. 3. Modeled hydraulic fracture conductivities along well lateral in the Delaware Basin. 237
Journal of Petroleum Science and Engineering 174 (2019) 235–243
B. Liang et al.
Fig. 4. Results from the production history match in one appraisal well in the Delaware Basin. Circle dots are observed data while the solid line denotes history matched result.
• 60 ft cluster spacing with 4 clusters per stage, hybrid, 50/70 and 100
multi-pad decisions such as staggered landing and pad sequence. A basic of completions design, including cluster spacing and number of clusters per stage, is not started from scratch and has been created based on the previous field experiences, data analytics in the basin, and hydraulic fracture modeling results. Field experiences and trend analysis have significantly reduced the searching space of design parameters. It is still highly challenging to obtain detailed information of completions designs in the Permian Basin, even though there are thousands of horizontal wells fractured in the shale formations. After tedious data collection and quality control of our own dataset, Fig. 6 presents the changes of stage length and cluster spacing from about 300 wells in the Permian Basin in the latest 3 years. Each dot represents one horizontal well, colored by proppant intensity from low value (light color) to high value (dark color). The chart indicates that as proppant intensity continues to climb high (moving to the left on the chart), stage length is fairly stable, in the range of 160–240 ft, and cluster spacing is in the range of 20–60 ft. The number of cluster per stage meanwhile is around 4–8. In addition, slickwater or low viscous hybrid fluid and 100 mesh-size proppant are widely used in all the latest pumping designs. Therefore, three representative designs are considered in the study:
mesh proppant.
Next, well distance and proppant intensity are major parameters for decision and will be the focus in this study. Permian basin has multiple target landings and some of them are vertically communicated, leading to the determination of well spacing and completions design in the staggered landing. It has drawn wide attention that depletion from existing vertical/horizontal wells affects fracture geometry and reduces initial reservoir pressure, causing lower production performance in new wells. As a result, pad sequence has to be optimized. Consequently, well spacing and completions design in the depletion scenario will be different with the scenario without depletion. This approach helped us follow the measured decision sequence and clarify the key parameters at each decision level. In practice, the threelevel sequence forms a loop: staggered landing and pad sequence lead to the lookback and review of completions design, well spacing and proppant intensity. Note that although well spacing is a three-dimensional issue, the focus of this paper, for the sake of generality, is lateral spacing impact on EUR of wells landed in both Midland and Delaware Basins and drilled between 6 and 8 wells per section without consideration of offset depletion impact.
• 20 ft cluster spacing with 8 clusters per stage, slickwater, 100 methsize proppant; • 40 ft cluster spacing with 6 clusters per stage, slickwater, 100 meshsize proppant;
Fig. 5. Decision sequence for field development in a new unconventional asset.
238
Journal of Petroleum Science and Engineering 174 (2019) 235–243
B. Liang et al.
Fig. 6. Evolution of stage length and cluster spacing along with proppant intensity. It provides a range of stage length and cluster spacing as proppant intensity rises, helping the determination of three representative completions designs in the study.
different distances and proppant intensities. The results are listed in a table: rows contained proppant intensities; columns consisted of single well EUR, 660 ft middle well EUR, 880 ft middle well EUR, 660 ft interference percentage, 880 ft interference percentage, 660 ft section EUR and 880 ft section EUR. All the results demonstrated that larger spacing (880 ft) with higher proppant intensity (D4) generated highest EUR for the whole section. Fig. 8 plotted some simulation results in the Midland Basin. In addition, we conducted sensitivity analysis of reservoir characterization impact on well interference. Different DFN and matrix permeability were studied but the trend of combining larger spacing with higher proppant intensity was the unchanged.
4.1. Simulation results for different well distances and proppant intensities In this study two well spacing scenarios are considered: 660 ft (8well per section) and 880 ft (6-well per section). The selection of two scenarios of well spacing 660 ft and 880 ft was mainly driven by field operations and business needs. Such decision was the most critical one while there were tests on further closer well spacing. Four designs in a range between 1100 lb/ft and 2200 lb/ft were selected for proppant intensity: D1, D2, D3 and D4. The proppant intensity difference between two adjacent designs is about 300 lb/ft and hence D4 is roughly 1100 lf/ft larger than D1. The parameters of detailed completions design categorized in the first decision sequence in Fig. 5 were remained unchanged while the volumes of fluid and proppant to be pumped are proportionally changed from the basic of designs. Those designs are then checked with the designs applied in the field and found pretty close each other. For each proppant intensity, we simulated three cases: single well with no neighbor well and consequently no well interference; 3-well per pad with 660 ft distance; 3-well per pad with 880 ft distance. We used 3 wells for multiple well interference instead of 6 or 8 wells mainly because of time consumption. It was very time consuming to hydraulically fracture either 5000 ft or 7500 ft lateral horizontal well and to conduct reservoir simulation with millions of unstructured grids. For proppant intensity D4, Fig. 7 presented pressure depletions after 30 years of production in the Midland Basin model for the cases with 660 ft distance on the left and 880 ft distance on the right. Hydraulic fracture geometries in the left well in two cases are the same due to the same completions design. As expected, less interference was observed for the larger distance case. Also, we realized that even though the fracture geometries might be visually identical, production performances were very different due to fracture conductivity variations. 3D models provide a visualization of reservoir heterogeneities and fracture geometry variations along the lateral through different simulations. Because well performance could be dependent upon location, different well locations and different landing depths were modeled to test the sensitivity of single well EUR and multiple well interference and build up confidence with the general trend of well interference at
4.2. Regression between EUR percentage reduction per well with well spacing and proppant intensity Results from models in both basins for EUR percentage reduction for middle well with different well spacing and proppant intensities are shown in Fig. 9. The data are plotted on the basis of 660 ft and 880 ft well spacing and a simple linear relationship was found for a given well spacing, regardless of basins or formations. This means that the same relationship and more importantly same well spacing for all our areas and formations of interest in both Midland and Delaware basins. Referring to our decision sequence in Fig. 2, the results also indicate that well spacing can be identical at the highest level and completions optimizations differentiate reservoir and rock properties at the secondary level. This study focused on an oil system only and there would probably be different well spacing and relationship for gas and condensate gas systems in the Delaware Basin. We also expect that more variations could be evident as more models for different areas are built and calibrated. 5. Field pilot analogs Data from pilots on well spacing and proppant intensity associated with NOJV (Non-Operated Joint Venture) in both Midland and 239
Journal of Petroleum Science and Engineering 174 (2019) 235–243
B. Liang et al.
Fig. 7. For the same completions design, pressure depletions after 30 years of production for the cases of 660 ft distance on the left and 880 ft distance on the right. Black lines denote unstructured grid meshes used in the simulation. Color shows pressure magnitude, from blue (low value) to orange (high value). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
There were no perfect pilot analogs as reservoir heterogeneity along lateral and vertical zones possibly affected well performance. More than 10 pilots were selected for analog analysis by considering single well analog without completion, as well as multiple well cumulative oil productions from different well spacing and different proppant intensities. A minimum six-month production was required in those pilots in order to acquire reasonable production. 5.1. Single well analog for different proppant intensities Five pilots in two areas (one in the Midland Basin and the other in the Delaware Basin) were analyzed for production improvement with the higher proppant intensities. Wells in those pilots were either single well or on a single pad with multiple identical wells developed at the same time. In Fig. 10, solid blue dots are percentage of normalized 30year EUR incremental from simulations in two basins; solid blue line is linear regression of solid blue dots; five red triangles are pilot results based on nine-month oil production. Simulation results are well aligned with actual data in a wide range of proppant intensities. The analysis indicates that larger fracture size leads to higher well performance if there is no well interference. Though nine-month actual production was
Fig. 8. Simulated cumulative oil production from two proppant intensities.
Fig. 9. EUR percentage reduction per middle well with different proppant intensities and well spacing. Blue and green dots denote results from 660 ft to 880 ft spacing, respectively. Solid lines are linear regression of blue and green dots, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
Delaware basins were integrated. It was necessary to understand detailed information including completions design, well spacing, landing, lateral length, possible offset wells, production reliability, production duration, artificial lift, and operator in order to build the viable models.
Fig. 10. Simulated single well performance improvement (blue dots and their regression, i.e., blue line) and actual nine-month production incremental from five wells with different proppant intensities in the field, shown by triangles. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) 240
Journal of Petroleum Science and Engineering 174 (2019) 235–243
B. Liang et al.
Fig. 12. Well production reduction when spacing is down from 880 ft to 660 ft. Horizontal axis shows proppant intensity with the scale 300 lb/ft. Solid line comes from regression result based on simulation results. Solid circles are data points from three pilots based on actual well performance comparisons between 880 ft and 660 ft well spacing.
Fig. 11. Performance reduction percentage comparison between simulation result regression colored blue and actual data colored orange from three pilots. Each pilot has its own single well performance and average well performance from multiple wells which were fractured by the same completions design and put on production at the same time. However, each pilot has its own proppant intensity and well spacing. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
per well, pad cost, and operations cost per well. Based on reservoir modeling and pilot analysis, two scenarios were identified as defining a step-change in field development and well design. These are high-level scenarios that would be further refined in detailed completions optimization and subsequent design and development decisions for specific cases. In scenario A, wells are spaced 660 ft apart and completed with smaller proppant intensity. In scenario B, wells are spaced 880 ft apart and completed with larger proppant intensity. For equal comparison, economic analysis for both scenarios covers the same size of acreage, with 8 wells in scenario A and 6 wells in scenario B. Production profiles are taking from simulation model for each scenario. Per well capital investment is higher for scenario B to account for larger proppant intensity, but overall capital investment is higher for scenario A due to more well count. Fig. 13 shows the outcomes for the economic analysis for the two scenarios in terms of NPV and capital investment. The result shows scenario B delivers higher capital efficiency, with higher NPV and lower overall capital investment. Per well capital investment is higher for scenario B due to larger proppant intensity but this is more than offset by the reduction in well count, and overall capital investment is lower than for scenario A. Under optimal circumstances, the right completion design and well spacing can reduce total capital per section while maintaining or increasing total EUR per section. Even when this is not the case, there is significant potential to improve capital efficiency at the section level.
compared with 30-year simulated EUR incremental, data shows ninemonth production can be a good indicator of well long-term performance. 5.2. Multiple wells for the same proppant intensity and different well spacing For this category, two types of pilots were considered: 1) performance reduction from single well to the well on multiple pads, and 2) given the same proppant intensity, performance reduction from larger to smaller well spacing. It is impossible to find pilots that are exactly comparable. General trend is more important than the exact difference between actual performance and model prediction. Attention was therefore paid to the check if simulations consistently overpredict or under predict the results. Fig. 11 provides the comparisons between our predicted performance reduction from simulation and field pilot performance data for first type of pilots. Fairly good matches of production reduction percentage in these three pilots. Large difference in the third pilot case mainly comes from the short production period, overpressured reservoir, and large uncertainty of EUR. For the second type, we mainly focused on four pilots with both well spacing 660 ft and 880 ft and plotted well production reduction in Fig. 12. The pilot data and regression results from simulation are close, showing that more interference occurs when spacing is down and proppant intensity increases. 6. Economic evaluation and remarks Several clear relationships between well spacing, proppant intensity, and well productivity are illustrated in this work. Higher proppant intensity combined with larger well-spacing results in higher production per well, but these design changes also increase the total capital cost per well. To determine if the production gains are worth the additional capital per well, economic analysis was done at the section level to evaluate the capital efficiency of different well spacing in combination with different completions designs. The economic inputs mainly include oil and gas production profiles, oil and gas prices, surface facility and pipe line costs, drilling cost per well, completion cost
Fig. 13. Economic evaluation for different well spacing in combination with different proppant intensity. 241
Journal of Petroleum Science and Engineering 174 (2019) 235–243
B. Liang et al.
Table 1 Summary of recent studies on well spacing. Approaches Field pilot test and surveillance
Authors
Formation
Fluid Type
Well Spacing Conclusions
Portis et al. (2013) Friedrich and Milliken (2013) Scott et al. (2015) Rucker et al. (2016)
Eagle Ford Wolfcamp in Midland Wolfcamp in Midland Niobrara and Codell
na Oil Oil Oil
Pettegrew et al. (2016) Lascelles et al. (2017) Palisch et al., (2016); Palisch et al. (2017)
Wolfcamp in Delaware Wolfcamp in Midland Bone Springs and Marcellus Bone Springs and Marcellus
Oil Oil na na
na 400 ft na Niobrara: < 200 ft Codell: < 700 ft 1320 ft with 1628 lb/ft na na na
Operator data analytics
Sahai et al. (2012) Schuetter et al. (2015) Mohaghegh et al. (2017)
Marcellus Wolfcamp in Delaware Marcellus
Gas Oil Gas
Variable for Marcellus na na
Numerical and analytical simulation.
Sahai et al. (2012) Lalehrokh and Bouma (2014)
Haynesville Eagle Ford
Gas Black oil; retrograde gas
Yu and Sepehrnoori (2014) Siddiqui and Kumar (2016) Belyadi et al. (2016) Li et al. (2017) Ramanathan et al. (2015) Suarez and Pichon (2016)
Bakken Eagle Ford Utica Niobrara Duvernay Vaca Muerta
Black oil Retrograde gas Gas Gas Retrograde gas Oil and gas
1056 ft for Haynesville 330–400 ft for blackoil 440–450 ft for retrograde 880 ft 400 ft for single landing 1200–1300 ft 2000 ft 200 m is not optimum na
7. Conclusions
In this study, we demonstrated that given the proppant intensity designs within 2200 lb/ft, there was a linear relationship between proppant intensity and EUR percentage reduction due to well interference. Hence, larger well spacing is beneficial for avoiding interference between wells and brings about better economic returns. Note that only two scenarios of well spacing, i.e., 660 ft (8 wells per section) and 880 ft (6 wells per section), were selected mainly due to the following three reasons. First, it was primarily driven by field operations and business needs. The two selected scenarios were widely used in the Permian basin. The business decision was the most critical and representative while there were quite some tests on further closer well spacing (10 or 12 wells per section for example) one year ago and some tests on 4 or 5 wells per section recently. Second, well spacing is tightly related with completion design. Completion design itself has been aggressively changing with many parameters involved, such as the number of clusters per stage, fluid intensity, and cluster spacing. Lots of those designs were not available in the field during this study. Therefore, we focused on the actual range of practices in the Permian basin and did not intend to stretch too much through only increasing or decreasing proppant intensity. Third, our work relied on actual performance analysis from field pilot. Cases such as giant proppant intensity and large well spacing were not available in the field when the work was conducted. On the other hand, it is known that there is a physical limit for fluid and proppant transport away from wellbore. If both proppant intensity and well spacing further increase, the linear relationship found in this paper is expected to be curved at a certain point and the benefit of increasing proppant intensity will be diminishing in terms of both EUR and economics. In other words, an optimum well spacing and proppant intensity are expected from total EUR or NPV per section. Our recommendation of 6 well per section is not the global optimization. The full optimum point can be determined in future work through the workflow illustrated in this paper: the combination of numerical modeling and actual field pilot tests.
In this paper, we integrated fracture-reservoir modeling with more realistic interference among wells and wide range of completions designs, conducting economic evaluation and performance analysis of field pilots. A range of proppant intensities and two well spacing distances (660 ft and 880 ft) were studied in Midland and Delaware basins. Results from both modeling work and pilots in two different basins aligned well.
• For a given well spacing (660 ft or 880 ft), cluster spacing and
• •
number of clusters per stage, simulation results showed a linear relationship between proppant intensity and EUR percentage reduction due to well interference, regardless basins and formations. This indicated that the same relationship and well spacing could be used for areas and formations of interest in both Midland and Delaware basins. More than 10 field pilots which were carefully collected and analyzed provided invaluable analogs to calibrate and support simulation results. At the section level, modeling work and pilots have demonstrated that larger hydraulic fracture size without an increase in subsurface well spacing does not necessarily improve section EUR and economics. Economic evaluation has showed that larger well spacing with bigger fracture size is a preferable combination. However, the economics will be diminished at certain point if both well spacing and fracture size increase. There is expected an optimum combination of well spacing and completion design.
Acknowledgements The paper was significantly revised from URTeC 2671346, originally presented at the Unconventional Resources Technology Conference held in Austin, Texas, USA, 24–26 July 2017. The authors would like to thank Chevron North America E&P management for permission to publish this paper.
Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.petrol.2018.11.010.
242
Journal of Petroleum Science and Engineering 174 (2019) 235–243
B. Liang et al.
Appendix. Summary of Recent Studies on Well Spacing Many authors have studied well spacing and recent work is summarized in Table 1 with three main approaches. (1) field test on different well spacing and surveillance such as microseismic, pressure and tracer, DNA sequencing, and proppant detection to infer drainage volume and interwell communication (Portis et al., 2013; Friedrich and Milliken, 2013; Scott et al., 2015; Rucker et al., 2016; Pettegrew et al., 2016; Lascelles et al., 2017; Palisch et al. 2016, 2017). Those methods are traditional trial-and-error, and technically easy to implement, but difficult to find a solution because there are too many possible combinations to test. More importantly, none of those approaches directly measure drainage volume, nevertheless its dynamic change with time. (2) Operator data analytics to develop the potential correlation between rock and performance and to identify the trend of well spacing (Sahai et al., 2012; Schuetter et al., 2015; Mohaghegh et al., 2017). Such methods have gained popularity under the current environment of big data and artificial intelligence. The main bottleneck for the application to the unconventional development lies in the lack of large amount of accurate data to cover a sufficient range and discover a good correlation. In most scenarios, data is either missing or inaccurate. As such, adding information from more wells does not necessarily improve the analysis because those wells probably introduce higher dimensions of uncertainties with more parameters. Additionally, it is challenging for most artificial intelligence algorithms to extrapolate and prediction. (3) Numerical and analytical simulation. Most efforts were focused on either well spacing without consideration of completions design, or evaluation of infill drilling and depletion impact on existing and new wells. Fracture geometry or Stimulated Reservoir Volume (SRV) is either directly assumed (i.e., planar and symmetric fracture with half length, conductivity, and height are given or in a range) (Sahai et al., 2012; Lalehrokh and Bouma, 2014; Yu and Sepehrnoori, 2014; Siddiqui and Kumar 2016), or derived from Rate Transient Analysis (RTA), microseismic data and production log (Belyadi et al., 2016; Li et al., 2017). Recently, complex fracture geometry through interactions between hydraulic fractures and pre-set natural fracture network has been linked with flow simulation (Ramanathan et al., 2015; Li et al., 2017; Suarez and Pichon, 2016).
workflow to describe the relationship between well spacing and EUR. In: Paper URTEC 2464916 Presented at the Unconventional Resources Technology Conference, 1-3 August, San Antonio, Texas. Portis, D., Bello, H., Murray, M., Barzola, G., Clarke, P., 2013. Searching for the optimal well spacing in the eagle ford shale: a practical tool-kit. In: Paper URTEC 1581750 Presented at the Unconventional Resources Technology Conference, 12-14 August, Denver, Colorado. Ramanathan, V., Boskovic, D., Zhmodik, A., Li, Q., Ansarizadeh, M., Perez Michi, O., Garcia, G., 2015. A simulation approach to modelling and understanding fracture geometry with respect to well spacing in multi well pads in the duvernay – a case study. In: Paper SPE 175928 Presented at the SPE/CSUR Unconventional Resources Conference, 20-22 October, Calgary, Alberta. Rucker, W., Bobich, J., Wallace, K., 2016. Low cost field application of pressure transient communication for rapid determination of the upper limit of horizontal well spacing. In: Paper URTEC 2460806 Presented at the Unconventional Resources Technology Conference, 1-3 August, San Antonio, Texas. Sahai, V., Jackson, G., Rai, R., Coble, L., 2012. Optimal well spacing configurations for unconventional gas reservoirs. In: Paper SPE 155751 Presented at the SPE Americas Unconventional Resources Conference, 5-7 June, Pittsburgh, Pennsylvania. Schuetter, J., Mishra, S., Zhong, M., LaFollette, R., 2015. Data analytics for production optimization in unconventional reservoirs. In: Paper URTEC 2167005 Presented at the Unconventional Resources Technology Conference, 20-22 July, San Antonio, Texas. Scott, K., Chu, W., Flumerfelt, R., 2015. Application of real-time bottom-hole pressure to improve field development strategies in the Midland Basin wolfcamp shale. In: Paper URTEC 2154675 Presented at the Unconventional Resources Technology Conference, 20-22 July, San Antonio, Texas. Swenberg, M., Schwartz, K., Merino, M., Taha, P., Sherlock, D., Best, J., 2017. An integrated study of geophysical, petrophysical, and geochemical data to define optimal reservoir development of the avalon shale in the salado draw field, Delaware basin, lea county, New Mexico. In: Paper URTEC 2668789 to be Presented at the Unconventional Resources Technology Conference, 24-26 July, Austin, Texas. Siddiqui, S., Kumar, A., 2016. Well interference effects for multiwell configurations in unconventional reservoirs. In: Paper SPE 183064 Presented at the Abu Dhabi International Petroleum Exhibition & Conference, 7-10 November, Abu Dhabi. Suarez, M., Pichon, S., 2016. Horizontal wells in pad development in the vaca muerta shale. In: Paper SPE 180956 Presented at the SPE Argentina Exploration and Production of Unconventional Resources Symposium, 1-3 June, Buenos Aires. Sun, H., Zhou, D., Chawathe, A., Du, M., 2016. Quantifying shale oil production mechanisms by integrating a Delaware basin well data from fracturing to production. In: Paper URTEC 2425721 Presented at the Unconventional Resources Technology Conference, 1-3 August, San Antonio, Texas. Wu, R., Kresse, O., Weng, X., et al., 2012. Modeling of interaction of hydraulic fractures in complex fracture networks. In: Presented at the SPE Hydraulic Fracturing Technology Conference, The Woodlands, Texas, USA, 6–8 February, SPE-152052-MS. Yu, W., Sepehrnoori, K., 2014. Optimization of well spacing for bakken tight oil reservoirs. In: Paper URTEC 1922108 Presented at the Unconventional Resources Technology Conference, 25-27 August, Denver, Colorado.
References Belyadi, H., Yuyi, J., Ahmad, M., Wyatt, J., 2016. Deep dry utica well spacing analysis with case study. In: Paper SPE 184045 Presented at the SPE Eastern Regional Meeting, 13-15 September, Canton, Ohio. Du, S., Youshida, N., Liang, B., Chen, J., 2016. Application of multi-segment well Approach: dynamic modeling of hydraulic fractures. J. Nat. Gas Sci. Eng. 34, 886–897. Du, S., Liang, B., Lin, Y., 2017. Field study: embedded discrete fracture modeling with artificial intelligence in Permian basin for shale formation. In: Paper SPE 187202 Presented at the SPE Annual Technical Conference and Exhibition, 9011 October, San Antonio, Texas. Friedrich, M., Milliken, M., 2013. Determining the contributing reservoir volume from hydraulically fractured horizontal wells in the wolfcamp formation in the Midland Basin. In: Paper SPE 168839 Presented at the Unconventional Resources Technology Conference, 12-14 August, Denver, Colorado. Lalehrokh, F., Bouma, J., 2014. Well spacing optimization in eagle ford. In: Paper SPE 171640 Presented at the SPE/CSUR Unconventional Resources Conference – Canada, 30 September–2 October, Calgary, Alberta. Lascelles, P., Wan, J., Robinson, L., Allmon, R., Evans, G., Ursell, L., Scott, N., Chase, J., Jablanovic, J., Karimi, M., Rao, V., 2017. Applying subsurface DNA sequencing in wolfcamp shales, Midland Basin. In: Paper SPE 184869 Presented at the SPE Hydraulic Fracturing Technology Conference and Exhibition, 24–26 January, the Woodlands, Texas. Li, N., Lolon, E., Mayerhofer, M., Cordts, Y., White, R., Childers, A., 2017. Optimizing well spacing and well performance in the piceance basin niobrara formation. In: Paper SPE 184848 Presented at the SPE Hydraulic Fracturing Technology Conference and Exhibition, 24–26 January, the Woodlands, Texas. Liang, B., Khan, S., Puspita, S., 2018. An integrated modeling work flow with hydraulic fracturing, reservoir simulation, and uncertainty analysis for unconventional-reservoir development. SPE Reservoir Eval. Eng. 21 (2), 462–475 SPE-187963-PA. Liang, B., Khan, S., Tang, Y., 2017. Fracture hit monitoring and its mitigation through integrated 3D modeling in the wolfcamp stacked pay in the Midland Basin. In: Paper URTEC 267336 Presented at the Unconventional Resources Technology Conference, 24-26 July, Austin, Texas. Mohaghegh, S., Gaskari, R., Maysami, M., 2017. Shale analytics: making production and operational decisions based on facts: a case study in marcellus shale. In: Paper SPE 184822 Presented at the SPE Hydraulic Fracturing Technology Conference and Exhibition, 24–26 January, the Woodlands, Texas. Palisch, T., Al-Tailji, W., Bartel, L., Cannan, C., Czapski, M., Lynch, K., 2016. Recent advancements in far-field proppant detection. In: Paper SPE 179161 Presented at the SPE Hydraulic Fracturing Technology Conference, 9-11 February, the Woodlands, Texas. Palisch, T., Al-Tailji, W., Bartel, L., Cannan, C., Zhang, J., Czapski, M., Lynch, K., 2017. Far-field proppant detection using electromagnetic methods - latest field results. In: Paper SPE 184880 Presented at the SPE Hydraulic Fracturing Technology Conference and Exhibition, 24–26 January, the Woodlands, Texas. Pettegrew, J., Qiu, J., Zhan, L., 2016. Understanding wolfcamp well performance – a
243