Stochastic history matching to time-lapse seismic of a CO2-EOR project sector model. (November 2016)

• Record Type: Journal Article
• Authors: Leeuwenburgh, Olwijn; Meekes, Sjef; Vandeweijer, Vincent; Brouwer, Jan
• Is Part Of: International journal of greenhouse gas control. Volume 54:Part 2(2016:Nov.)
• Page Start: 441
• Page End: 453
• Publication Date: 2016-11
• Subjects: Seismic -- History matching -- Ensemble Kalman Smoother -- CO2-EOR -- Weyburn; Greenhouse gases -- Environmental aspects -- Periodicals; Air -- Purification -- Technological innovations -- Periodicals; Gaz à effet de serre -- Périodiques; Gaz à effet de serre -- Réduction -- Périodiques; Air -- Purification -- Technological innovations; Greenhouse gases -- Environmental aspects; Periodicals; 363.73874605
• DOI: 10.1016/j.ijggc.2016.05.027
• Languages: English
• ISSNs: 1750-5836

Abstract

Highlights: An Ensemble Kalman Smoother algorithm was applied to history matching of seismic and production data from the Weyburn field. A novel distance parameterization was used to quantify differences between seismic maps and simulation results. The history matching methodology was found to be effective in obtaining improved models in a synthetic data experiments. Application of the workflow to the real data led to a reduction in root-mean-square data mismatch of up to 80%. It is concluded that the assumption that only grid cell permeability and porosity are uncertain should be reconsidered. Abstract: A stochastic seismic history matching workflow has been applied to a single well pattern sector from the Phase 1A development area of the Weyburn CO2 -EOR project. Several experiments were conducted aimed at updating permeability and porosity at all grid cells in the model such that mismatches with time-lapse seismic data interpretations of CO2 spreading over the period 1999–2005 are minimized. The applied history matching methodology consists of the combination of an iterative Ensemble Kalman Smoother and a novel seismic data parameterization based on front positions. Synthetic data experiments were first conducted to verify the suitability of the methodology. These synthetic experiments showed that the method is able to produce multiple models that simulate CO2 front positions that are consistent with those determined from time-lapse seismic data. The algorithm was subsequently applied to inversion of the real data. A number of indicators such as low signal to noise ratio, non-monotonic CO2 spreading, differences in published interpretations, and inconsistency with simulations, suggest that the quality of the real seismic data is rather low. As a result, use of the real data set produced results of a more diverse nature. It was found that the 2002 time-lapse seismic map could be reproduced more accurately after history matching than before history matching and more accurately than the available reference model. The 2001 and 2004 maps could not be reproduced with the same accuracy. A combined match to the 2002 seismic and production data showed a 78% reduction of the total data mismatch. These results could be achieved at relatively low computational cost and thus demonstrate the potential for use of assisted history matching workflows in the management of CO2 -EOR projects. The results also point towards limitations in both the seismic data and model quality for this particular case. The validity of some of the assumptions used in constructing the sector model was investigated in more detail. It was concluded that they cannot fully explain the remaining mismatch. Low grid resolution and possibly the absence of a fracture network in the model are identified as possible candidates for further investigation.