Automatic seismic and production history matching in Nelson using different updating schemes

Timelapse (4D) seismic is becoming a commonly used tool and helps reservoir engineers history match field data and make forecasts. Automated history matching processes are often restricted by large numbers of parameters that control changes to models. In this work we introduce and apply different parameter search strategies in the Nelson field. We reduce the volume of the parameter space heuristically by considering changes that are local within the reservoir and compare to those that are applied everywhere simultaneously. We find that we get similar results but local changes are applied more quickly if only one locality is modified at a time. We also investigate the improvements that are obtained through using time-laps seismic and report that seismic and production data are not always complimentary. In this study, production data is best used to modify reservoir properties near the well while seismic is more useful in the inter-well regions. Overall, improved seismic and production predictions were obtained and forecasts were also improved.