Nonlinear inversion for estimating reservoir parameters from time-lapse seismic data: Fid-Bau Portal

Nonlinear inversion for estimating reservoir parameters from time-lapse seismic data

Mohsen Dadashpour / Martin Landrø / Jon Kleppe

Saturation and pore pressure changes within a reservoir can be estimated by a history matching process based on production data. If time-lapse seismic data are available, the same parameters might be estimated directly from the seismic data as well. There are several ways to combine these data sources for estimating these reservoir parameters. In this work, we formulate a nonlinear inversion scheme to estimate pressure and saturation changes from time-lapse seismic data. We believe that such a formulation will enable us to include seismic data in the reservoir simulator in an efficient way, by including a second term in the least- squares objective function. A nonlinear Gauss–Newton inversion method is tested on a 2D synthetic dataset inspired by a field offshore from Norway. A conventional reservoir simulator has been used to produce saturation and pore pressure changes as a function of production time. A rock physics model converts these data into synthetic time-lapse seismic data. Finally, the synthetic time-lapse data are used to test the derived inversion algorithm. We find that the inversion results are strongly dependent on the input model, and this is expected since we are dealing with an ill-posed inversion problem. Since we estimate pressure and saturation change for each grid cell in the reservoir model, the number of model parameters is high, and therefore the problem is undetermined. From testing, using this particular dataset, we assume neither pressure nor saturation changes for the initial model. Although uncertainties associated with the proposed method are high, we think this might be a useful tool, since there are ways to reduce the number of model parameters and constrain the objective function by including production data and reservoir simulation data into this algorithm.