Application of Design of Experiments to Expedite Probabilistic Assessment of Reservoir Hydrocarbon Volumes (OOIP)
Design of Experiment (DoE) methodology was used to minimize the number of stochastic earth models that were needed to appropriately evaluate original oil in place (OOIP) uncertainty for a Jurassic-age, Middle East carbonate reservoir. The DoE methodology enables the maximum amount of information to be obtained from the minimum number of experiments (model OOIP) in which multiple parameters (structural uncertainty, facies distribution uncertainty, oil-water contact uncertainty, netto- gross uncertainty, etc.) contribute. The DoE methodology also allows for rapid determination of the magnitude of model parameters to overall OOIP uncertainty. Thus, attention can properly be focused on the few key model parameters that most affect OOIP uncertainty, perhaps to the point of obtaining additional data if cost-justified.
The DoE-based workflow used was as follows: (1) use Plackett-Burman design (one of several DoE methodologies tested) to determine which combinations of model parameters should be evaluated; (2) collect the experimental results (OOIP); (3) analyze the results statistically to determine significant contributors to OOIP uncertainty; (4) use the experimental results to obtain a response “surface” (equation) that describes the relationship between OOIP and the significant contributors to OOIP uncertainty; (5) use the response surface along with appropriate statistical distributions for the significant contributors to OOIP uncertainty in a Monte Carlo-process to obtain P10, P50, and P90 OOIP values. Drained volume uncertainty was also evaluated using the above workflow so that stochastic reservoir models with P10, P50, and P90 drained volumes could be generated using appropriate combinations of geologically reasonable parameters for further sensitivity and optimization studies as well as input to probabilistic economic evaluation.
KeywordsUncertainty Source Bottomhole Pressure Finite Difference Simulation Field Development Optimization Semivariogram Range
Unable to display preview. Download preview PDF.