Society of Petroleum Engineers - Abu Dhabi International Petroleum Exhibition and Conference, ADIPEC 2015
Reservoir modeling involves the characterization of the internal gridded petrophysical properties distribution and the simulation of fluid production. However, a common problem associated with reservoir modeling is the highly non-linear relationship between the distribution of the petrophysical parameters (frequently with a non-stationary character) and the fluid production. To tackle this problem, this paper presents a new methodology for integrated reservoir modeling by addressing a multiscale optimization approach applied for non-stationary geostatistical history matching of complex connectivity hydrocarbon reservoirs. The methodology comprises a two staged procedure for multiscale optimization, which includes a global optimization stage followed by a local refining stage. The former consists of optimizing the geological spatial anisotropy trend by coupling the stochastic optimization over the anisotropy multiparameter space with an image generation algorithm - direct sequential simulation with local anisotropy correction. The latter aims at refining the small scale heterogeneity by performing the local optimization based on a regional image perturbation method. The local refining optimization is achieved by taking into account the best individual well production matches. A complex deltaic reservoir case study is presented to illustrate the applicability of the proposed methodology. An ensemble of multiple optimized history matched models is obtained respecting both the production data and the geological settings. The results show that the deltaic complex channelized pattern is well reproduced and also that the multiscale optimization improves the match between the simulated fluid profiles and the observed production data. As such, optimization is performed at multiple scales: the anisotropy trend model accounts for larger scale variability of the structures, while consecutive local refining improves convergence to dynamic response around individual wells. The contribution of this work to petroleum technology is the implementation of a novel methodology for the reproduction of the fluids production profiles through the perturbation of the subsurface petro-physical models while honoring complex geological constrains. Copyright 2015, Society of Petroleum Engineers.
Year of publication: 2015