The paper presents an application of an innovative workflow for multiple history matching. Initially streamline-based simulations are used to rank multiple geological images and to efficiently identify representative models. Afterwards, the selected realizations are fully simulated and multiple history matching is achieved through the application of genetic algorithms.
The workflow has been implemented on a North African field whose production started in 1996. The main hydrocarbon reservoir is a Triassic sedimentary series developed in a fluvial environment characterized by high heterogeneity and uncertainty. Such uncertainty has been captured by generating more than a thousand alternative reservoir models by varying key geological parameters. For each realization, streamline simulation was performed and the heterogeneity has been measured based on flow capacity-storage capacity plot. Streamlines simulation requires only few minutes to solve the entire set of models. Cluster analysis was used to select five models with different heterogeneity. History matching calibration has been set up on the reduced number of realizations using genetic algorithm to speed up and automate the process.
The final output of the workflow is a set of alternative models properly matched that represent the ideal suite to estimate the uncertainty in forecast and to evaluate infilling opportunities.
Well placement optimization has been then implemented all over the different geological scenarios.
The application demonstrated that the innovative workflow allowed managing more than a thousand reservoirs models in a very short time, to efficiently select representative realizations and to speed up the history matching process. Moreover, dealing with multiple history matched models put the basis for a robust estimation of future production together with a more reliable evaluation of new infilling opportunities.
As a result of the study, two over three approved wells have been already drilled and put on stream with positive results, in line with the predictions.