Modelling fine-scale heterogeneity for optimal history matching from aeolian gas reservoirs, Neptune Field, UKCS

Abstract

Improved matching of reservoir simulation model results to production data in gas fields allows improved reservoir management and optimization of hydrocarbon production. In high net:gross aeolian sandstones this requires the full range of scales of permeability heterogeneity to be represented in a model. This necessitates the modelling and appropriate upscaling of rock properties, including the millimetre-scale lamina effects within the aeolian dune slipface facies, the contrast between dune apron and slipface facies and the impact of first- and second-order bounding surfaces within the dune complexes and the super-bounding surfaces that separate them. The Neptune Gas Field lies on the northwest margin of the Southern North Sea gas basin and in Block 47/4b and extends into Blocks 47/5a and 42/29. Gas is reservoired in the Lower Leman Formation which in the Neptune Field, comprises a succession of stacked dry aeolian sandstones and thin, wet interdune deposits. Permeability varies significantly at the lamina and bedform scale and these variations impact the performance of production wells in the field. A deterministic reservoir simulation model was constructed prior to development drilling, with permeability determined from porosity–permeability transforms based on core analysis data. The collection of continuous bottom-hole pressure (BHP) data from three of the development wells showed that this model failed to match accurately the rate of pressure decline whilst flowing, and over-predicted the build-up pressure when the wells were shut in. Review of the sedimentological information revealed that heterogeneity in the aeolian sands was being lost in the modelling of permeability and the coarse upscaling. To test whether the poor dynamic history match was due to the deterministic model not incorporating the permeability heterogeneity, a fine-scale geocellular model was built to characterize the expected geological heterogeneity and then upscaled for flow simulation. To accommodate the range in scales of heterogeneity present, the modelling was undertaken in two principal steps. First, the lamina-scale effects were modelled to determine average flow properties perpendicular and parallel to dune migration directions. The results of the lamina-scale heterogeneity modelling were then incorporated into petrophysical models of dune bodies that were characterized by low permeability dune apron facies and high permeability base of dune slipface facies that grade upwards into moderate permeability slipface facies. Upscaling of the stochastic permeability grid captured the impact of the fine-scale heterogenetity and gave BHP responses at the development wells in better agreement with the production data than any deterministic permeability distribution modelled. The stochastic rock property distributions have, thus, been adopted as the base case in all future history match and reservoir performance prediction studies.