Click
to view figures in PDF format
Predictive Modeling of Sedimentology and Early Diagenesis in an Icehouse Isolated Carbonate Platform: Salt Creek Field, West Texas
Gareth D. Jones1, Fiona F. Whitaker2, Roger Barnaby1, Michele Thomas1, Hsin-Yi Tseng1 and Yitian
Xiao1
1ExxonMobil Upstream Research Company, P.O. Box 2189, Houston, TX 77027,
USA
[email protected]
2Department of Earth Sciences, University of Bristol, Bristol BS8 1RJ, UK
Introduction
A fundamental challenge in carbonate reservoir characterization is predicting the spatial distribution of early diagenesis, which is often a critical control on porosity and permeability heterogeneity. Conceptual models that qualitatively relate diagenetic processes and products in a sequence stratigraphic framework (Moore, 2001) predict the general styles of diagenesis but lack the quantitative spatial data necessary to populate reservoir flow models. Diagenetic overprinting in response to high amplitude sealevel fluctuations can be discordant to sequence stratigraphic surfaces, which introduces additional uncertainty. Furthermore, in data-poor environments a high-resolution sequence stratigraphic framework is generally not available.
Numerical process-based forward models present an alternative approach for quantitative spatial predictions of carbonate rock properties that can assist reservoir assessment. These models have primarily focused on simulating the stratigraphic evolution of carbonate platforms in response to specified rates of sediment production, erosion, redeposition, and accommodation space change (e.g. Bosence et al., 1990). An exception is the forward model developed by Whitaker et al. (1997, 1999) which couples the early diagenetic processes of stabilization, dissolution, cementation and dolomitization to the sedimentological routines described by Bosence et al. (1990). In the Whitaker et al. model, hydrological zones (hydrozones) act as a template to define the 2D spatial distribution of sediments affected by different diagenetic processes. Relative sealevel, platform width, and ratio of recharge to platform-scale permeability define the distribution of the four hydrozones: vadose, meteoric, mixed and marine. Depositional mineralogy and porosity are user-defined and changes are tracked in time and space. Porosity modification by early diagenesis is calculated as the sum of the product of residence time within each of the hydrozones and the associated diagenetic rate. Here we report on the utility of the Whitaker et al. (1997) forward model to predict early diagenesis in a reservoir example: the Northwest Extension, which is part of the Salt Creek Field in West Texas.
Geologic Setting and Approach
The Northwest Extension of Salt Creek is a small (c. 25 km2) Pennsylvanian isolated carbonate platform that is located on the NE flank of the Horseshoe Atoll in the Midland Basin. The Canyon Formation reservoir interval consists of cyclically stacked wackestones, packstones and oolitic-skeletal grainstones. Early diagenesis has significantly modified reservoir quality by: 1) the precipitation of minor isopachous marine cements, 2) major pore filling meteoric cements and 3) major fabric-selective dissolution of ooids and phylloid algae to generate abundant moldic porosity. This diagenetic sequence is consistent with high amplitude sealevel fluctuations and aragonite mineralogy that dominated the Pennsylvanian global icehouse system.
To investigate model application in a data-poor environment, we simulate a representative 2D section of the NW Extension using an initial surface (the top of the Strawn Formation), the external platform geometry, a regional scale burial history and literature constraints on Pennsylvanian sealevel and climate. We first decompacted the platform, which resulted in a 40 % increase relative to present-day subsurface thickness. The workflow was divided into simulating first the sedimentology and then the diagenesis.
Sedimentological Modeling
The external platform geometry of the NW Extension can be adequately simulated in the estimated 5 My interval of platform growth with 1) a sealevel curve that is dominated by icehouse high amplitude (70 m) asymmetric cycles with a frequency of 400 ky, that also includes lower and higher order components of smaller amplitude, 2) relatively low rates of depth-dependent carbonate production (maximum 0.3 m/ky) and 3) a subsidence rate of 0.027 m/ky. Sensitivity analyses demonstrated that platform evolution is critically sensitive to the amplitude and frequency of the composite sealevel curve and also to the phase
shift
between cycles of different orders. The majority of specified configurations in sensible parameter space result in platforms that are grossly different to the external geometry observed.
Diagenetic Modeling
The Midland Basin in the Pennsylvanian experienced a semiarid climate and we thus specify groundwater recharge as 0.1 m/ky, surface dissolution as 0.01 m/ky ,and an isotropic platform-scale permeability of 60 Darcies. A relative sealevel fall that exposes the platform results in the development of a freshwater lens, the geometry of which defined the platform space occupied by each of the four hydrozones. The cumulative residence time of sediments in different hydrozones was tracked (Figure 1A,B). At the time of platform demise, after 5 My, the maximum residence times recorded in the vadose, meteoric and mixed hydrozones were 250 – 300 ky. Thus, the platform spent the vast majority of time in the marine hydrozone. However, it is the residence time in the non-marine hydrozones that is critical for porosity modification because the associated diagenetic rates are relatively fast. The distribution of residence times reflects the hydrozone geometry. Meteoric residence times are greater in the platform interior where the freshwater lens is thickest and decrease toward the coast as the lens thins (Figure 1A). The mixed hydrozone is thickest at the coast and thins inland (Figure 1B). Our simulated meteoric and mixed residence times also exhibit distinct vertical trends in response to lower order sealevel cycles (Figure 1A,B). Sensitivity analyses demonstrated that the specified platform-scale permeability is a critical control on the size of the freshwater lens. For the range tested (12–1200 Darcies) 12 D yielded a maximum residence time of 545 ky and 1200 D yielded 145 ky.
We investigated early diagenetic porosity modification by specifying a homogenous depositional porosity of 40 %, a depositional mineralogy of 75 % aragonite and 25 % calcite, and semiarid diagenetic rates from Whitaker et al. (1999, p. 344) for 1) aragonite stabilization to calcite, 2) aragonite dissolution, 3) calcite dissolution and 4) calcite precipitation. Early dolomite is not significant in Salt Creek or other reservoirs on the Horseshoe Atoll so dolomitization was not simulated. At the time of platform demise, the simulated post-early diagenesis porosity distribution is highly heterogeneous ranging from < 10 % to 58 % (Figure 1C). The largest simulated increase in porosity is up to 30 % and associated with dissolution in the meteoric hydrozone. Porosity occlusion, up to 35 %, is dominated by cementation in the vadose and the meteoric hydrozones, primarily due to the reprecipitation of calcite derived from surface dissolution but also due to the stabilization of aragonite to calcite.
Comparison of Predictions to Reservoir Porosity and Permeability
There are a number of known limitations in the model applied, for example diagenetic rates in each hydrozone are uniformly specified and a compaction function to predict “burial” porosity is not incorporated. Therefore a refined “early” porosity distribution was generated using proprietary diagenetic rate data from reactive transport simulations and our hydrozone residence times. To predict reservoir porosity our “early” porosity was compacted based on sediment texture and reservoir depth (c. 2km). We compared our predicted reservoir porosity to ooid grainstone intervals in seven cored wells in the Northwest Extension (Figure 2A). We report a relatively good match between predicted and measured porosity, however we overestimate low porosity (<10 % class) and the highest porosity (30-40 % class).
The correlation between porosity and permeability for ooid grainstones is known to be poor in the Salt Creek reservoir due to the varying connectivity of different pore types. Rules were developed to predict the evolution of pore type in response to early diagenesis, based on quantitative petrography. Three pore types were tracked: 1) between particle, 2) touching molds and 3) isolated molds. Ooid grainstone matrix scale reservoir permeability was then predicted from the proportion of different pore types and our predicted reservoir porosity (Figure 2B). We successfully predict the reservoir permeability distribution, although we underestimate the 10-100 mD class and overestimate low permeability (<1 md). Considering the uncertainty in many of the processes described in our forward model approach we are encouraged by these preliminary predictions of reservoir porosity and permeability in ooid grainstones.
Summary
Carbonate forward models, with refined sedimentological, diagenetic and compaction algorithms have the potential to be a powerful tool for quantitative carbonate reservoir characterization, particularly in data-poor environments.
References
Bosence, D.J.W. and Waltham, D.A., 1990, Computer modelling the internal architecture of carbonate platforms: Geology v. 18, p. 26-30
Moore, C.H., 2001, Carbonate Reservoirs: porosity evolution and diagenesis in a sequence stratigraphic framework: Developments in Sedimentology 55, Elsevier, 444 p.
Whitaker, F.F., Smart, P.L., Hague, Y., Waltham, D.A. and Bosence, D.J.W., 1997, A coupled two-dimensional sedimentological and diagenetic model for carbonate platform evolution: Geology v. 25, p. 175-178
Whitaker, F.F., Hague, Y., Smart, P.L., Waltham, D.A. and Bosence, D.J.W., 1999, Coupled modeling of carbonate diagenesis and sedimentology: structure and function of a coupled 2-dimensional diagenetic and sedimentological model of carbonate platform evolution: SEPM Special Publication 62, p. 69-84
Acknowledgements
We thank ExxonMobil for permission to publish the results of this study. G. D. Jones thanks his ExxonMobil colleagues in the Salt Creek Production Team (Linda Price, Dave Smith and Pak Wong) and Peter Holterhoff. F. F. Whitaker is grateful to Peter Smart, Dave Waltham and Yvette Hague for collaboration in model development and the Leverhulme Trust for financial support (grant F182AE).