Upscaling of Multimillion Cell Geological Model into a Practical Simulation Model of a Major Carbonate Oil Reservoir in Offshore Abu Dhabi

fax 01-972-952-9435. AbstractIn reservoir characterization studies, static reservoir rock types, SRRTs, are utilized to guide the modeling of reservoir petrophysical properties in static geological models. However, since the reservoir dynamic properties are not considered in the static rock typing schemes, the SRRTs should not be usedas being practiced now -to guide the distribution of the reservoir multiphase dynamic properties of capillary pressure, Pc, and relative permeability, Kr, curves to the cells of dynamic simulation models. Considering the rock attributes that goes into any SRRTs scheme and assuming identical interfacial tension between the different reservoir fluids, this paper suggests that the dynamic reservoir rock types, DRRTs, would only be a function of wettability for any SRRT. Thus by imposing wettability on the SRRTs, reservoir engineers should be able to generate the DRRTs, the proper guide for generating and distributing Pc and Kr curves for simulation models. This technique is applied for building the simulation model of a major carbonate oil reservoir. The gradual change of reservoir wettability from oil wet to water wet dictated the generation of many DRRTs for any SRRT, each defines that SRRT at a different reservoir level between the top and the upper limit of the transition zone. Available experimental Pc and Kr data, acquired for only few SRRTs, were utilized to develop correlations to determine the saturation and Kr endpoints of the Kr curves. The connate water saturation, Swc, of any DRRT is picked from the Pc curve of the parent SRRT at that DRRT reservoir level. Swc correlated with the residual oil saturation, Sorw. Sorw is in turn used to establish the maximum water Kr point of Krw max . Oil Kr maximum point, Kro max , is assumed to be 1. Furthermore, the shape of the different Kr curves would be devised by picking the corresponding Corey exponents as they are correlated with wettability for any SRRT. The consistently derived sets of Kr curves helped build a very well initialized and history matched simulation model.

show abstract

Reliable Simulation Model Permeability Arrays By Integration of Core, Test, and Log-Derived Permeability Data

Ghedan¹,

Boyd²,

Thiebot³

2003

SPE Annual Technical Conference and Exhibition

View full text Add to dashboard Cite

The generation of permeability models that best describe the reservoir heterogeneity and its flow dynamics is a difficult objective to achieve. This paper addresses the multi-step procedure followed and the various analytical techniques utilized to integrate different types of permeability data to generate multiple realizations of permeability, which were fed and tested in a dynamic reservoir simulation model for initialization and history matching. Due to the limited availability of core permeability data and lack of vertical resolution in test permeability data, most modelers rely on statistically log derived permeability predictors to generate permeability models. This paper discusses the design, merits and drawbacks associated with three permeability predictors included in the static geological model of a major carbonate reservoir in offshore Abu Dhabi, along with the rock-typing scheme utilized. Four analysis techniques are utilized to screen and choose the right permeability predictor for the right sub reservoir or layer of the field. These are merged to generate the first realization of permeability for the simulation model after upscaling. Test KH conditioned to core KH is taken as a reference in the screening and testing process. The conditioning process is performed at the wells levels to add the core permeability vertical resolution to test permeability data and produce a new log for that matter. This new log is utilized to produce two additional permeability realizations for the simulation model by either conditioning at the wells level followed by modeling and upscaling, or conditioning at the geological model after modeling followed by upscaling. The initialization of the simulation model is completed with each of the three permeability and rock type realizations. The fluids in place are very close and model is equilibrated. The simulation model performance of the initial history matching runs was found to be very close to the field observed performance for a very complicated and heterogeneous carbonate reservoir. In our case, a combination of the conditioned permeability models are found to be best for achieving a quality history matched simulation model. Background Essential to the building of robust reservoir simulation models is an accurate representation of permeability variability and quantity. Common sources of permeability are well test and core. Available on many of the wells in this offshore reservoir, well tests lack the fine detailed permeability profiles of core and are restricted to the completed reservoir layer. Core coverage is scarce however, comprising only small percentage of the vertical footage drilled. To supplement these information gaps, numerous approaches to relating core permeability to conventional log response were explored over the past eighteen years. None were satisfactory. Abu Dhabi's carbonate rocks have highly complex pore geometries. Variations in facies, cementation, diagenesis, pore throat size, grain size etc. nearly obliterates the link between porosity and permeability. For the most difficult cases, lithology remains constant, porosity and water saturation fluctuate little, while permeability varies three magnitudes. The most enduring permeability estimate1 was arrived at using a program called ESTIME*. Core permeabilities were related to accompanying well log values via a multidimentionsonal look-up table. This approach presented an unrealistic homogeneous picture of the reservoir permeability variability.

show abstract

Upscaling of Multimillion Cell Geological Model into a Practical Simulation Model of a Major Carbonate Oil Reservoir in Offshore Abu Dhabi

Cited by 5 publications

References 6 publications

Quad-Tree decomposition method for areal upscaling of heterogeneous reservoirs: Application to arbitrary shaped reservoirs

Quad-Tree decomposition method for areal upscaling of heterogeneous reservoirs: Application to arbitrary shaped reservoirs

Dynamic Rock Types for Generating Reliable and Consistent Saturation Functions for Simulation Models

Reliable Simulation Model Permeability Arrays By Integration of Core, Test, and Log-Derived Permeability Data

Contact Info

Product

Resources

About