Reservoir simulation

New Modeling and Simulation Techniques Optimize Completion Design and Well Spacing

Proper lateral and vertical well spacing is critical for efficient development of unconventional reservoirs. Much research has focused on lateral well spacing but little on vertical spacing, which is challenging for stacked-bench plays such as the Permian Basin.

jpt-2020-04-194367f1andhero.jpg

Proper lateral and vertical well spacing is critical for efficient development of unconventional reservoirs. Much research has focused on lateral well spacing but little on vertical spacing, which is challenging for stacked-bench plays such as the Permian Basin. Following a previous, successful single-well study in paper SPE 189855, the authors have performed a seven-well case study in which the latest complex fracture modeling and reservoir-simulation technologies have been applied. This synopsis will concentrate on the methodology behind the study; the reader is encouraged to view the complete paper for specific comparisons of completion designs.

Introduction

Complex-fracture-modeling tools are used frequently to study well spacing. Most research has focused on lateral well spacing rather than on vertical spacing, though the industry has seen many fracturing hits and hydraulic communications between wellbores placed vertically in stacked plays.

Field pilot tests have been used extensively to test and optimize lateral and vertical well spacings and to optimize well-completion designs. These pilot tests, however, take considerable time to implement and are very expensive.

In this multiple-well study, the authors used an established work flow to study the fracture interaction between wellbores and lateral and vertical well spacings. A calibrated model was then used to optimize well-completion designs for the Wolfcamp formation.

Work Flow

The work flow to build and calibrate the complex-fracture-network and reservoir-performance-simulation models, and then use the models to conduct sensitivity analysis on the fracture networks, resulted from different completion designs and corresponding well performance.

Because some critical data sets are spread throughout a wide area—including core data, comprehensive well-logging data, and geomechanical properties—the authors first built a regional geological model, then sliced the sector models from the regional model. This enabled use of available data that were spread sparsely across the area. On the basis of the well locations, a sector model was built by cutting a section from the regional geological model. In the process, the properties were repopulated with a finer grid size.

Previous studies discussed using the seismic data to build a discrete fracture network (DFN). For the authors of this paper, however, the DFN represents not only the natural fracture networks but also includes geological bedding or layering information. Those beddings or layerings may be mechanical-weakness connections, which influence hydraulic-fracturing propagations. The detailed well-completion information was then plugged in.

Seven-Well Case-History-Modeling Setup

Seven wells drilled and completed in four different zones within the Wolfcamp formation were studied. A sector model was cut from the full-field (regional) model on the basis of the well locations, considering the possible fracture-propagation-modeling needs. The sector model covers a length of 12,600 ft, a width of 4,200 ft, and a thickness of 3,000 ft. In the process, a much finer grid size was used (refined from 300×300 ft in the regional model to 33×33 ft in the sector model) with consideration of perforation cluster spacings. Then, all formation properties were repopulated to the smaller grid cells with the same property models used in the regional model. The wellbores and perforations then were input into the sector model.

Complex-Fracture-Modeling Results

To calibrate the fracturing model, the wellhead treatment pressure of all 190 stages pumped into those seven wells stage by stage were history matched. The authors focused particularly on matching the instantaneous shut-in pressure (ISIP) and the following shut-in pressure of each stage. To find the best matches, rock mechanical properties, in-situ horizontal stresses, and other reservoir properties were sensitized. Two rounds of pumping pressure matching were completed. The first was to check overall matching results, while the opportunity existed to modify systematically the geomechanical properties globally within the whole sector model. In the following round, the geomechanical properties were modified and calibrated locally along the whole wellbore/geological zone if necessary. In general, the geomechanical properties were calibrated by scaling factors based on geological facies or true vertical depth. The primary matching criterion was to reach 10 to 15% ISIPs difference between simulation and real measurements; another criterion was to match the shut-in pressure trends. Complex-fracture-modeling results and analysis are presented in detail in the complete paper.

Reservoir Simulation Modeling

The complex fracture networks were converted into an unstructured-grid well-performance model to accommodate the complex fracture geometries resulting from the fracturing pumping history match. The model grid size is significant. The model in this study contains seven wells not only spreading laterally a couple of thousand feet apart but also vertically more than 1,000 ft apart. Vertically stacked wellbores with complex fracture networks resulted in excessively partialized unstructured gridblocks. Several attempts were made to optimize the model gridding to minimize the grid size while trying to maintain representation of the complex fracture networks. The resulting unstructured simulation grid is shown in Fig. 1. The total number of the grid cells was close to 10,000,000, a size that presents a huge challenge for simulation. To reduce the grid cell size, two wells were left out of the simulation model.

jpt-2020-04-194367f1andhero.jpg
Fig. 1—Complex unstructured reservoir-simulation grid.

 

To save computation time, the simulation grids were reduced upon the principle of the depletion drainage being limited to the area adjacent to the fracture networks. Cells far away from fracture networks and wellbores were deactivated. The final grid size was close to 3,000,000 cells for the reservoir-simulation model. The history-­matching process involved calibrating a series of parameters, including, but not limited to, permeability distributions, relative permeability curves, and compaction tables in different regions.

Optimizing Completion Design With the Calibrated Model

When the geological model and the geomechanical model with the completion pumping history and production history had been calibrated, the focus of the study moved to optimization of well-completion designs. Well spacing is, more or less, a business decision. However, once the well-spacing decision is made, the corresponding completion design ought to be optimized to reach higher production rate and higher recovery. Thus, the goal of the completion design is to create

  • Greater fracture surface area, because a larger fracture surface area will bring a higher production rate
  • More-uniform fracture length and height

To illustrate the process of well-­completion-design optimization, one of the seven wells from the model was used to compare completion-­design results. The compete paper presents a breakdown of different completion designs and how the work flow allowed optimization of the design-selection process.

Discussion

Computation time was the primary challenge of the study. The processes of regional geological modeling, ­complex-fracture modeling, and reservoir-performance modeling were extremely time-consuming. Significant computation time limited the ability to perform more sensitivity studies. At the same time, needs and efforts had to be balanced by clear definition of the objectives and goals of each step, and of the study as a whole.

Like any modeling tool, the one used by the authors has constraints. Some fracturing-modeling results had to be evaluated carefully before they were accepted or rejected. The authors remained in communication with the tool developer for assumption clarification and to provide feedback for further development.

Conclusions

  • Completion designs in unconventional reservoirs can be optimized by complex-fracturing modeling with a calibrated geological model, which is cheaper and faster than the field pilot tests.
  • Optimizing well-completion designs could be a better way to minimize fracturing hits or hydraulic communication instead of setting the wellbores farther apart.
  • Building a robust geological model and calibrating the hydraulic-fracturing-propagation model may be time-consuming, but it might speed up the entire process by focusing on the clearly defined goals and objectives of a specific study.
  • Fewer perforation clusters per stage may improve cluster efficiency.
  • Effective cluster spacing seems to be the key to improving completion effectiveness. For the Wolfcamp formation, tighter cluster spacing with fewer perforation clusters per stage may create a larger fracture surface area, with high fluid and proppant intensity.

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 194367, “Optimize Completion Design and Well Spacing With the Latest Complex Fracture Modeling and Reservoir-Simulation Technologies—A Permian Basin Case Study With Seven Wells,” by Hongjie Xiong, SPE, and Songxia Liu, University Lands, and Feng Feng, SPE, Texas A&M University, et al., prepared for the 2019 SPE Hydraulic Fracturing Technology Conference, The Woodlands, Texas, 5–7 February. The paper has not been peer reviewed.