| dc.description.abstract | Integration of time-lapse seismic data into the reservoir model offers great potential in understanding reservoir flow patterns as well as reservoir properties. However, it also requires the solution of an inverse problem, which poses challenges in terms of dynamic reservoir modeling and seismic history matching to infer reservoir characterization.
In this dissertation, we first present a method for assessing the inversion results in underdetermined problems, resulting in a multi scale data integration, Then, we introduce a novel history matching approach to integrate frequent seismic surveys (4D) using onset times.
In the first part, an analysis of spatial resolution is incorporated into an efficient history matching approach, in order to indicate the reliability of the estimated solution. By examining the spatial resolution in seismic data integration, as a function of derivation type, we evaluate quantitatively the contribution of pressure and saturation changes on the calibrated permeability field.
Next, we present a novel and efficient approach to integrate frequent time lapse (4D) seismic data into high resolution reservoir models based on seismic onset times. Our approach reduces multiple time-lapse seismic survey data into a single map of onset times, leading to substantial data reduction for history matching while capturing all relevant information regarding fluid flow in the reservoir. We demonstrate the practical feasibility of our proposed approach through the heavy oil reservoir at Pad 31 in the Peace River Field (Alberta, Canada) with daily time lapse seismic surveys recorded by a permanently buried seismic monitoring system.
Finally, we quantitatively investigate the effectiveness of the onset time and the amplitude inversion to solve the inverse problem associated with integrating 4D seismic data into the reservoir model.
The results of the study demonstrate the effectiveness of the onset time approach for integrating a large number of seismic surveys by compressing them into a single map. Also, the onset times appear to be relatively insensitive to the petro elastic model but sensitive to the steam/fluid propagation, making it a robust method for history matching of time lapse surveys. | en |
