Proxy Models as a Substitute for Full Reservoir Simulations (to be applied for SAGD process)

Arash Mirzabozorg
Supervisors: Dr. Long Nghiem, Dr. John Chen


Numerical simulation of complex systems such as SAGD processes has the following problems. Long computational times due to compositional nature and transient temperature behavior of the models used in the solution. Transference of uncertainty from the reservoir and operational parameters to the forecast variables during a SAGD process simulation is almost an impossible and very expensive task.

The proxy model is a good way to solve these problems. Proxy-modeling (also known as surrogate modeling or meta-modeling) is a computationally cheap alternative to full numerical simulation in assisted history matching, production optimization and forecasting. A proxy model is defined as a mathematically or statistically defined function that replicates the simulation model output for selected input parameters. The classical one-parameter-at-a-time approach is simple to understand but only delivers a limited picture on linear relations only. Proxy models, on the other hand, allow deeper investigation of the combined effects of input parameters. Right now the proxy model has already been used in the sensitivity analysis, history matching, optimization and uncertainty assessment.