Capillary Pressure Effect on Injected Water Movement and Upscaled Relative Permeability in a Heterogeneous Carbonate Reservoir

Abstract This paper presents the effect of capillary pressure on injected water movement in a fine grid numerical simulation model and demonstrates the necessity of correctly upscaling relative permeability data for large scale simulation modelling of a heterogeneous carbonate reservoir. Premature water breakthrough was observed in a giant oil field under five spot pattern water injection offshore Abu Dhabi. Open hole log resistivity anomalies and time lapse thermal decay time logs have suggested that thin high permeability streaks conduit injected water to producers. In this study, a simple and effective approach was used to generate oil wet capillary pressure data from pore throat size distributions determined from mercury injection capillary pressure data classified by lithology and permeability range. Average oil-water relative permeability curves were generated based on the same classification. Once the lithology and permeability are specified, the appropriate capillary pressure and oil-water relative permeability curves are assigned. Results from fine grid 2D simulation models demonstrate the difference between water wet and oil wet system performances. Upscaled relative permeability for use in large scale simulation grids were derived from these results using Kyte and Berry's "dynamic pseudo functions". Water wet capillary pressure, usually used for model initialisation, is not suitable to simulate injected water performance in heterogeneous oil wet reservoirs. Oil wet capillary pressure data derived from pore throat size distributions appear to correctly model injected water movement in the reservoir. In a heterogeneous oil wet reservoir, correctly upscaled relative permeability honouring vertical heterogeneity combined with oil wet capillary pressure data is required for large scale simulation modelling. 1. Introduction Premature water breakthrough was observed in a giant oil field under five spot pattern water injection offshore Abu Dhabi. Open hole log resistivity anomalies and time lapse pulsed neutron logs have suggested that thin high permeability streaks conduit injected water to producers. Namba and Hiraoka1 pointed out the limitation of the conventional simulation approaches which utilise initialisation capillary pressure data to simulate dynamic conditions, and have suggested that oil wet capillary pressure could prevent water slumping without any reduction of vertical permeability. Unfortunately, for this field, the number of negative capillary pressure measurements are limited. Faced with the same problem, previous authors employed a "J-function" to apply measured capillary pressure to different porosity and permeability rocks. In this paper, the authors introduce a simple and effective approach to generate oil wet capillary pressure data from pore throat size distributions, which are derived from mercury injection capillary pressure data. In a fine scale simulation model, the appropriate capillary pressure and oil-water relative permeability curves are assigned based on the lithology and permeability range.