AAPG Middle East Region GTW, Maximizing Asset Value: Integrating Geoscience with Reservoir Management & Facilities Optimization

Datapages, Inc.Print this page

Reservoir Modeling in a Complex Giant Carbonate Reservoir: An Integrated Approach to Constrain Permeability for Better History Match and Waterflood Development


The mid-Cretaceous Mauddud Formation is the principle carbonate reservoir in the giant Sabriyah field in North Kuwait. Oil production from the Mauddud in Sabriyah field commenced in the late 1950s. Initially the reservoir was produced under natural depletion. However, there is little aquifer support and full field waterflood development was introduced to arrest pressure decline, boost offtake rates and enhance recovery factors. The Mauddud Formation in North Kuwait consists of a diverse suite of carbonate facies deposited in a shallow marine ramp setting. Reservoir heterogeneity, both laterally and vertically, is high and reservoir quality varies significantly across the field. These variations are attributed to a combination of depositional controls and the complex diagenetic overprint. Natural fractures are observed in core and BHI but fracture density is low and contribution to flow is minor. Fracturing can be more intense near faults although the impact of faulting and fractures on the waterflood is thought to be only secondary. High permeability layers, where K is enhanced by more than two orders of magnitude compared to the matrix, provide a particular challenge for the waterflood scheme with regard to early water breakthrough and conformance management. Typically, the high permeability streaks are associated with rudist dominated facies and/or high energy inner ramp skeletal packstones/grainstones with solution enhanced porosity. The facies setting and diagenetic overprint (mainly leaching) favoring the development of high permeability layers are sequence stratigraphically controlled. High K streaks are typically present at the top of shallowing upward trends. However, the lateral extent of high K layers is less well constrained and carries a higher degree of uncertainty. Correlation distances range from hundreds of meters to several kilometers for individual high K streaks. The fact that the high K layers do not show a consistent characteristic log signature hampers their identification and further complicates mapping of high K streaks. A recent static/dynamic model update was undertaken featuring an integrated reservoir modelling approach in order to realistically represent the high K layers. Close iterative feedback between static modeling and learnings from the dynamic simulation was a key success factor. A series of static subsurface realizations was tested in the simulator and the results used to improve the static model. The distribution of high K layers in the model was driven by PLT data within a sequence stratigraphic framework derived from detailed facies analysis. The resulting model achieved a significantly improved history match at both field and well by well levels. Water breakthrough and flowing BHP match were improved which is an important result in this waterflooded reservoir. In addition, PORV multipliers used in previous models for better average reservoir pressure match were not required anymore and could be discarded. Also, predicted recovery improved as a result of better sweep compared to previous model realizations.