Construction and Validation of an Integrated 3D Reservoir Characterisation Geomodel for Reservoir Management at Seminole San Andres Unit

The West Texas Seminole San Andres Unit (SSAU) is a mature CO2 flood and one of the ten largest fields in the Permian Basin, with 624 MMBO cumulative production. Recently a full-field reservoir characterisation 3D geomodel was constructed for geologic input into the current CO2 reservoir simulation efforts, and as a tool for integrated reservoir management decisions. Mike Uland (uland@ireservoir.com) of iReservoir.com based in Littleton, Colorado and Ken Kosco (KKosco@hess.com) of Amerada Hess Corporation in Houston, TX present an overview of the project. The article is based on the presentation given to the 9th Annual CO2 Conference, Midland, Texas, 12 December 2003. See end of article for co-authors.

Introduction

This article summarises the reservoir characterisation steps used in the creation of the full-field geomodel. The 27 square mile SSAU geomodel was validated using quality-control checks in the geomodeling workflow, which included flow-simulation calibration within a mini-model area. The input data set for integration into this full-field model consisted of 12,000 ft of core, 630 well logs, 3D seismic, and 60 years' production history. The major geomodel workflow tasks are presented in order as follows:

Geologic Depositional Model

The depositional model and stratigraphic framework tasks provide a basis for constraining the eventual geomodel framework, as well as both geophysics and petrophysics modelling tasks. The core facies description work was key to developing a depositional model, and helped calibrate well log responses. Depositional facies control the reservoir properties with the exception of pore-plugging by late anhydrite. The principal reservoir facies are fusulinid dolowackstone, fusulinid-peloid dolopackstone, and coated-grain dolograinstone. One interesting finding at SSAU is evidence of minor dissolution in some cored intervals that were collected post-CO2 flood.

Sequence Stratigraphic Framework

Stratigraphic framework construction builds off the facies description and depositional modelling. It has a primary impact on all other aspects of reservoir characterisation because the gridded stratigraphic framework constrains all 3D property distributions. The stratigraphic framework was first built from 1D core descriptions consistent with outcrop-based stratigraphy, then tied to 2D log cross-sections and isopach maps, and finally extended with 3D well-log correlation to define the geomodel's internal geometry (Figure 1). Both transgressive (flooding events) and regressive maxima (some were microkarsted) surfaces were correlated. Major transgressive surfaces were the most significant vertical reservoir baffles.

Figure 1: Integration of Core and Log Facies Data into the Geomodel (Click for larger view)

Geologic Facies Predictions

Five facies groups were predicted in uncored wells using fuzzy-logic constrained by available log data and vertical core-facies proportion curves. Facies were predicted at the Seminole field away from core control to provide a quality-control check on the depositional model within the 3D geomodel. The 3D facies distribution was then used to constrain permeability data and to create "rock property regions" by grouped facies to improve the local flow variability in future simulations during the flow simulation history-matching.

Petrophysics for 3D Models

Extensive petrophysical work was done to constrain the extremely broad vintage, quality, and type of log data into a core-calibrated, field-wide data set for distribution in the 3D geomodel. Initial log cleanup and normalisation was done with 1D histograms and 2D cross-sections. The merged porosity logs were ranked by log-quality. All log data were loaded into 3D and gross interval-average porosity maps were used as quality-control checks on the merged porosity logs (Figure 2). After several iterations, 463 of the 630 wells were used as hard-data to constrain porosity distribution within the final 3D geomodel. In some cases, the lower-quality logs were used as soft-data or removed completely to prevent unrealistic porosity data-points in the geomodel.

Figure 2: Well Log Data in the 3D Geomodel which Needs Adjustment after 2D Normalisation (Click for larger view)

Geophysics

The Geophysics modelling task included seismic acoustic impedance (AI) calibrated to rock properties, time-to-depth conversion of the amplitude and inverted volumes, and analysis of coherence volumes. Due to seismic data-quality problems at the San Andres level, the seismic AI is only weakly correlated to the log rock-properties and is not of sufficient quality to be used in the final 3D porosity distribution for the San Andres interval. Seismic time horizons were converted to depth using the geostatistical external-drift method, and added significant control to the structural framework around the geomodel periphery relative to just using well log picks. The seismically constrained depth surfaces also were used for the secondary permeability analysis based on curvature attributes.

3D Geologic Model Construction

A 3D geomodel was used to integrate the available geologic, petrophysics, and geophysics data in the previous tasks in a single shared-earth model. The grid X, Y orientation and vertical fine-layering scheme for the geomodel used the seismic depth horizons and all stratigraphic picks to create the 3D framework. Porosity was distributed using sequential Gaussian simulation and honoured the log data at the wells (Figure 3). Facies was distributed in 3D using indicator-based simulation that had core or log facies-data at 173 wells and was constrained by facies probability trends, which combined horizontal and vertical proportion curves. Permeability was then distributed by facies and used stratigraphic zone-dependent Phi-K cloud transforms. The result was a high-resolution 3D volume of integrated properties from all core, log, seismic, and production data that can now be used as a shared-earth model.

Figure 3: 3D Porosity Distribution in the SSAU Full-Field Geomodel (Click for larger view)

Secondary Permeability and Fracture Integration

Using core observations, semblance lineation maps, and calculated maximum curvature; areally varying "deformation metrics" were developed across the SSAU field. The deformation metrics were expressed as polygons of five distinctly different deformation styles across the field. As an example of how the deformation metric may be used to predict production history, curvature versus production was plotted and compared to an ideal case. Results show that curvature is not the single contributing variable, but that semblance, core descriptions, and production history must also be incorporated in 3D with the deformation features to predict production trends within the field.

Mini-Model "Flow Calibration"

A crucial aspect of building an accurate 3D geomodel is to independently test the framework and data distribution with production data. As part of the quality-control workflow for the SSAU geomodel, a flow simulation "mini-model" was up-scaled and history-matched midway through the full-field workflow to test whether the full-field geomodel effectively captured rock properties and flow units (Figure 4). History-matching through water-flood indicated that the geomodel had accurately captured the local flow-unit response. The mini-model simulation also indicated the possibility for additional facies-based rock-region permeability and permeability anisotropy work for improved flow calibration. This intermediate flow-calibration step independently verified the reservoir geomodel workflow prior to full-field geomodel completion.

Figure 4: Flow-Unit Comparison for Coarse v. Fine Geolayers in the Mini-Model Flow Calibration (Click for larger view)

3D Geomodel: Benefits and Future Uses

The 5,760,000 cell full-field geomodel currently enhances reservoir management by facilitating:

• Remaining reserves estimates.
• Better understanding for CO2 and water conformance.
• Future CO2 pilot ROZ project evaluations.

The operations engineers can use the 3D geomodel to investigate individual well problems that might be related to the local injection pattern geology, such as, pattern re-alignment choices to optimise field operation (Figure 5). The 3D geomodel provides the reservoir engineer with a starting point to initialise a reservoir fluid-flow simulator. The design of tertiary recovery pilots, such as the CO2 injection into ROZ zones, requires the detailed 3D reservoir description provided by this geomodel. Currently (early 2004) simulation work on up-scaled geomodels is ongoing.

Figure 5: Fluid Breakthrough Management using Geomodel High-Permeability Connected Layers (Click for larger view)

Co-authors are Thomas P. Wingate, Jim Bush, Michelle Simon, Scott Pluim and Becky Tupman (Amerada Hess Corporation, Houston, TX ), Laura C. Zahm (Bureau of Economic Geology, Houston, TX ), Chris Zahm (ConocoPhillips Corp., Houston, TX ), Don H Caldwell (Marathon Oil Co., Houston, TX ) and Hai-Zui Meng, K. Lyn Canter and Mark D. Sonnenfeld (iReservoir.com, Littleton CO ).

Disclaimer: The material available on this website is designed to provide general information only. Whilst every effort has been made to ensure that the information provided is accurate, it does not constitute legal or other professional advice.
Please note: The Department of Trade and Industry cannot be held responsible for the contents of any pages referenced by an external link.