Faster seismic history matching in a United Kingdom Continental Shelf reservoir

In seismic history matching, we (the authors) obtained a computerassisted history match to conventional production data but also include the spatio-temporal information offered by time-lapse (4D) seismic to further constrain the model. Good predictive models can only be found with a suitable search of the space of uncertain parameters, such as permeability, porosity, net : gross, and fault transmissibility. We present a method of history matching that speeds up the search process. By using streamlines in place of a finite difference simulator, a factor 10 speed up in simulation can be obtained. The model accuracy is slightly diminished, affecting both the predicted well data along with saturations and pressures used to predict the 4D signal. We show how to reduce any bias effects by calibrating the model error and including it in the misfit calculation. The method has been applied to the United Kingdom Continental Shelf (UKCS) Nelson field using 6 years of production history and two seismic surveys. The field is maintained above bubblepoint and is ideal for streamline simulation. We compared the use of the finite difference simulator with the streamline method in history matching, when we changed net : gross and vertical permeability properties. By addressing the mismatch in seismic data, together with production, we improved our predictive capability. A proper parameter search increases the reliability of predictions made by using simulations and has great benefits for reservoir management. Decisions regarding water and gas handling, well maintenance, and optimization of production can be made with greater confidence and new wells can be planned with reduced risk. Copyright © 2009 Society of Petroleum Engineers.