| Although underground reservoirs which suffered long period and complex geological process are determined,it has complicated structure set and spatial-varied properties, as well as lack of sufficient knowledges and informations about stratigraphy, It's difficult to characterize reservoirs accurately in any circumstances,especially for terrestrial-phase reservoir with strong anisotropy.This makes understands for reservoir uncertain and so called reservoir's property uncertained,i.e. reservoir's randomicity. These cause prediction for reservoir has multiple solutions.It takes hight risk if oil field development strategy established based on only one reservoir-prediction result derived from determined-modeling method.To reduce this risk,stochastic modeling method is introduced and used.Based on the reservoir's multiple datas and geological statistics theory, stochastic modeling could produce more than one optional reservoir-prediction results which have equal probability.These multiple reservoir-prediction results or multiple reservoir property model can be called multiple realizations for stochastic simulation. Studing the difference between these realizations,we could understand the randomicity about properties of reservoir.These realizations will provide more reliable informations for field development strategy deciding.After studied the various datas about Penglaizhen reservoir in Xinchang gas field,We adjusted the stratigraphy set between wells and done the well logs standardization and parameters interpretation.Using sequential Gauss stimulation method, with seismic data as soft constrained condition and well logs as hard condition, multiple reservoir parameter models were built,including porosity , permeability and water saturation.The size about cells in every model are 30m×30m×2m,the sum of cells is 22,804,992.For used by reservoir numerical simulation, parameter models were down-scaled to 75m×75m×11ayer and have been testified their reliability by subsequent...
|
PDF Full Text Request