To improve the conventional reservoir model optimization techniques

this paper presents a method to introduce experimental design theory to uncertainty reservoir modeling optimization.First

we identified the dominant nondeterministic variables that influence the OOIP calculation and their probability distributions by the actual oilfield data analysis

and made a modeling plan through experimental design.And then

several deterministic models were built to calculate OOIP.Multiple regression was performed on the nondeterministic variables and the OOIP calculated with corresponding models.Finally

the Monte Carlo simulation was run to obtain the OOIP probability curve and P10

P50 and P90 model parameters

which were used to establish the reference P10

P50 and P90 models.This method is not only more objective and efficient

but also capable of identifying the key geologic uncertainties that influence the OOIP calculation and provides a new way to evaluate the uncertainty of OOIP.