Uncertainty Analysis and Assisted History Matching Workflow in Shale Oil Reservoirs

For newly developed shale oil reservoirs, it is a challenging task to arrive at reasonable long-term production forecasts due to both large uncertainties associated with reservoir parameters and short production history. Assisted history matching plays an important role in integrating key uncertainties in order to arrive at a calibrated production prediction.

In this paper, we present two workflows to utilize a stochastic history matching method to a multi-fracture horizontal well in Eagle Ford shale oil reservoir. First, we discuss the impact of reservoir properties, hydraulic fractures, microfracs, phase behavior and rock characteristics on production behavior using sensitivity analysis. Next, we use the key uncertainties to calibrate the model against historical data using genetic algorithms. Three different geo-models were considered in all cases. However, in one workflow, they were evolved separately while in another one, they were evolved as a group. Production forecasting based on updated models from both workflows were categorized into several groups using cluster analysis. Then, the suggested workflows were compared according to their advantages and limitations. The results indicated that for workflow I, inaccuracy in uncertainty ranges could results in an incomplete set of updated models during evolution. For workflow II, reasonable probability must be provided; otherwise good model for certain geo-models may be ignored because the results could be constrained by less-probable geo-models. For unconventional reservoirs with very short limited static and dynamic data, our proposed workflows provide a flexible framework for capturing key uncertainties. Thus, they can be applied flexibly for long-term production forecasting or for identifying key areas for further data acquisition.