Completions

Understanding Completion Performance: Software-Guided Work Flows and Models

This paper describes a comprehensive field study of eight horizontal wells deployed in the stacked Niobrara and Codell reservoirs in the Wattenberg Field (Denver-Julesburg Basin).

Abstract image of numbers and boxes
Getty Images

This paper describes a comprehensive field study of eight horizontal wells deployed in the stacked Niobrara and Codell reservoirs in the Wattenberg Field (Denver-Julesburg Basin). The overall goal was to understand the geometry of the hydraulic fractures (propped) and the producing volume with respect to completions design, target reservoirs, and well spacing. Through this understanding, the asset can be developed more effectively and economically.

Field Description

The Wattenberg Field is a basin-center petroleum accumulation located northeast of Denver, from which hydrocarbons have been extracted over the last 50 years. Development in the Wattenberg Field began in 1970 from vertical J‑Sandstone wells, with production of the Niobrara following in 1976. In 2009, horizontal development began in the Niobrara formation. The Upper Cretaceous sequence from Codell to Niobrara is the current focus of horizontal development in the Wattenberg Field.

Petrophysical Model

A robust 3D mechanical Earth model (MEM) provides the framework for data integration and physical simulations. In this work flow, the MEM calculates or approximates a number of input parameters. Each of these variables is defined both spatially and stratigraphically and is housed in a 3D geocellular model. Because many of the input calculations are dependent on other variables, a set of variable relationship definitions keeps the model dynamic so that, during the tuning process, all initially established data relationships are honored.

Initially, properties are calculated basinwide on a coarse resolution that is later enhanced locally around data-rich pilot sites for simulation. Regional distributions of mechanical rock properties are calculated from dipole-­sonic logs and compared with core measurements to determine static values of Poisson’s ratio and Young’s modulus. Relationships between the dipole-sonic curves and triple-combination logs are determined to distribute the properties across the basin widely and produce maps of Poisson’s ratio and Young’s modulus, both spatially and stratigraphically. Pore pressure is mapped with interpretations from diagnostic fracture injection tests (DFITs) along with maximum mud weights from horizontal wells. Because of the widespread presence of existing vertical wells, it is important to determine whether the DFIT represents virgin pore-pressure conditions. The maximum-­horizontal-stress orientation is determined from six-arm-­caliper and full-bore formation-microimager logs. The ­vertical-stress magnitude is obtained by integrating a bulk-density log from the surface downward. The minimum horizontal stress is calculated spatially with a poroelastic stress equation and calibrated with DFITs. Use of DFIT interpretations of closure stress along with spatial and stratigraphic grids of Young’s modulus, Poisson’s ratio, and pore pressure make it possible to solve for the tectonic strain variable at the location of each DFIT. In this work flow, the ­horizontal-stress anisotropy remains as a tuning parameter during the geomechanical simulation. Petrophysical analysis of grain density, bulk-density logs, and clay volumes allows for the calculation of a derived effective-porosity log that is mapped regionally and stratigraphically. Porosity/permeability relationships are established from core and used to calculate a derived permeability; however, permeability is used as a significant tuning parameter during the production-history match. Last, a description of the discrete natural-fracture network (DFN) is obtained by use of both formation-­microimager logs and outcrop maps.

Characterization of DFN

Using a horizontal exposure of the Niobrara formation, a 2D map of the ­natural-fracture system was generated. This image was digitized and compared with natural-fracture interpretations from 11 vertical and five horizontal ­formation-microimager logs to confirm that the general orientation (spacing and density) of natural fractures observed at the surface was representative of the reservoir at depth.

After refinement, the 3D MEM is used directly as an input into a hydraulic-­fracture simulator that is capable of modeling the complex interaction between the propagating fractures and the existing natural-fissure network.

This software-guided process integrates petrophysical, geomechanical, ­hydraulic-fracture, and reservoir modeling to gain understanding of the production behavior and the extent of the depletion associated with the completion method and related hydraulic-fracturing treatments.

Hydraulic-Fracture Modeling

Numerous models exist in the industry to simulate hydraulic-fracture propagation. Most of these models assume that hydraulic fractures are planar with symmetrical wings and only model one wing of the hydraulic fracture. Few models simulate both wings of the hydraulic fracture simultaneously and are capable of modeling fracture asymmetry.  In the last few years, a handful of complex hydraulic-fracture models have been introduced. One of these, the unconventional fracture model (UFM), is numerically gridded to simulate not only the asymmetry but also the branching of hydraulic fractures at weak interfaces. UFM uses a rigorous geomodel that takes the rock texture and properties into account while simulating hydraulic-fracture propagation in addition to considering the material balance, fluid and proppant transport, pressure calculations, and interaction among simultaneously growing fractures.

The modeled hydraulic-fracture geo­metry is generally validated by consistently matching observed treatment parameters and microseismic interpretations simultaneously. Actual pump records from each stage were loaded in the software, and the model was adjusted until it reproduced the observed data. Treating-pressure matching is largely a function of simulating friction as a result of the pipe diameter, the fluid/proppant, or the perforations. Instantaneous-­­shut-in-pressure (ISIP) matching is primarily a function of the pressure and stress in the target interval. Complexity and net pressure building through DFN/hydraulic-fracture interactions are used to fine tune the match while honoring DFIT observations of stress and pressure.

The model was calibrated to match treating pressure, ISIP, microseismic geometry, DFIT observations, and ultimately a production-history match. Tuning the horizontal-stress anisotropy is performed by varying the minimum and maximum tectonic-strain variables in the poroelastic-stress equation. This effectively changes the fracture geometry to be more planar with increasing anisotropy. The vertical distribution of minimum horizontal stress is adjusted by altering the minimum and maximum tectonic-strain variables as well as introducing slight alterations of pore pressure within different formations. These stratigraphic variations are the primary driver of calculated height in the model.

Matching the historical production performance of the well with a numerical reservoir simulator involves explicitly gridding the hydraulic fractures in the reservoir model to honor the 3D hydraulic-fracture geometry and proppant distribution together with the MEM. In addition to the reservoir grid, a fluid model, a set of relative permeabilities, stress-­dependent hydraulic-fracture-conductivity profiles, and historical production rates are entered into the reservoir simulator.

Reservoir Modeling

To construct a proper reservoir-­simulation grid that honors the resolution of the complex hydraulic fractures and the 3D MEM, the following gridding parameters were tuned: number of reservoir layers, fracture cell width, vertical layering factor (refinement of the 3D MEM made on the basis of hydraulic-­fracture-results contrast), and zone height. The final grid used during this project included hydraulic-­fracture results from eight wells, had 14 layers, and consisted of approximately 2.5 million cells.

In the process, the software combines production attributes with the MEM and the complex fracture planes to create a nonuniform, locally refined, cellular description of the stimulated reservoir. Cells are divided into regions defined by proppant type. In addition to the properties carried from the MEM, each of these regions can be defined by a unique set of relative permeability and pressure-­dependent permeability curves, as well as fluid and pressure initializations.

Production-History Match

In this project, to properly history match the production from eight wells over approximately 5 months for a pad draining from two reservoirs, multiple physical-behavior considerations were taken into account for the initial state of the model. Different reservoir fluids with different petrophysical properties were included in the simulation. Multiple vertical wells were drilled and produced before the completion of the eight wells; the effect of depleted areas was incorporated into the reservoir-simulation model. Finally, compaction tables for each proppant region within the production grid were used to account for the loss of transmissibility associated with the decreasing pore pressure experienced during production.

History matching the observed well production after a stable period of sustained production is necessary to understand the performance of the ­hydraulic-fracture treatment. Depending on the reservoir and problem, different constraints may be used during the history-matching process. For liquids-rich shale plays such as the Niobrara and the Codell, oil-production rate was used as the control variable and the bottomhole pressure, along with the other two fluid phases (gas and water), was the matching parameter.

A case study involving an eight-well pad in the Wattenberg Field targeting the Niobrara, Codell, and Fort Hays formations is provided in the complete paper.

Conclusions

  • The authors’ field observations and simulation results do not indicate strong indications of uplift in overall stimulated geometry and early well performance by pumping completions larger than 1,600 lbm/ft in the Niobrara.
  • The current well spacing does not appear to be adequate to drain the resource. Pressure-interference and production modeling in the Niobrara suggests that 660-ft well spacing leaves a significant portion of undrained reservoir between wells.
  • The necessity of including depletion considerations to explain well performance suggests that the withdrawal of pore pressure from offset vertical wells is a significant driver on a well’s ability to produce oil. Consideration of existing verticals should influence future pad planning.
  • Hydraulic-fracture-height extension allows initial communication between the Niobrara and Codell reservoirs; however, this connectivity fades during production, likely because of the loss of fracture connectivity vertically.

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 179172, “Understanding Completion Performance in Niobrara-Codell Reservoirs Through the Use of Innovative Software-Guided Work Flows and Models,” by K.J. Wallace, Encana Oil and Gas; P. Reyes Aguirre, Schlumberger; E. Jinks and T.H. Yotter, Encana Oil and Gas; and Raj Malpani and Felipe Silva, Schlumberger, prepared for the 2016 SPE Hydraulic Fracturing Technology Conference, The Woodlands, Texas, USA, 9–11 February. The paper has not been peer reviewed.