Seismic assisted history matching using binary maps

Abstract

This paper addresses the challenge of integrating 4D seismic data with production data in a quantitative manner in order to improve the forecasting ability of a reservoir model and reduce the associated uncertainty. It presents a history matching workflow that has been applied to production data and time lapse seismic data. In this procedure, the production data objective function is calculated by the conventional least squares misfit between the historical data and simulation predictions, while the seismic objective function uses the Current measurement metric between a binary image of saturation change. This approach is implemented on a real field data from the United Kingdom Continental Shelf (UKCS), where uncertain reservoir parameters which consist of global and local parameters are initially assessed. These parameters include flow based multipliers (permeability, transmissibility), volume based multipliers (net-to-gross, pore volume), as well as the end points of the relative permeability curves (critical saturation points). After the initial screening, sensitive parameters are selected based on the sensitivity analysis. An initial ensemble of fluid flow simulation models is created where the full range of uncertain parameters are acknowledged using experimental design methods, and an evolutionary algorithm is used for optimization in the history matching process. It is found that the primary control parameters for the binary seismic gas match are the permeability and critical gas saturation, while the volumetric parameters are important for the binary seismic water match in this particular reservoir. This approach is compared to seismic history matching using full seismic modelling, preserving all amplitudes. The results demonstrate that the binary approach gives a good match to gas saturation distribution and water saturation distribution, and the reservoir parameters converge towards a solution. The conventional approach does not capture some signals of hardening and softening in the seismic data, and hence in summary, the binary approach seems more suitable as a quick-look reservoir management tool. A unique feature of this study is the application of the binary approach using Current measurement metric for seismic data history matching analysis, as this circumvents the use of the uncertain petroelastic model. This approach is easy to implement, and also helps achieve an effective global history match.