EFFICIENT HISTORY MATCHING IN NATURALLY FRACTURED RESERVOIRS

For naturally fractured reservoirs, history matching requires that both the matrix and fractures are properly characterized. Log and core data as well other static information can be very useful in building the matrix model. Some elements of the fracture model can be built using 3 D seismic data, well test data and PLT data. However, field experience suggests that it is very difficult to properly quantify fracture conductivity without matching production data. Currently, the procedure involves a trial and error method, where fracture conductivities are adjusted manually to match the production data including water cuts and breakthroughs. One can adjust other parameters to match the data; however, fracture conductivity (especially in the presence of large permeability contrast between matrix and fracture permeabilities) represents first order effect in matching the production data.

This study focuses on automatic history matching of production data by adjusting fracture intensities in a static model. The methodology not only ensures geological consistency in final model, but also includes uncertainties about fracture locations. The calculation of the sensitivity of the production data such as bottom hole flowing pressure, WOR and GOR etc. to reservoir model variables such as fracture permeability, fracture porosity and fracture/matrix coupling factor etc. is made efficient by an efficient direct sparse solver, through building the relations between the fracture intensity and the reservoir model variables, and the gradient of the production data to fracture intensity was estimated by chain rule. Through the transformation of reservoir model variables into the fracture intensity space for history matching, a significant time has been saved because of the reduction in parameter space. The method is validated by conducting history matches for both synthetic and field data sets. The results show that this method is reliable and efficient for naturally fractured reservoir history matching.