Reservoir model calibration against dynamic response data is often constrained by a prior conceptual modelof geologic scenario that specifies the expected types of spatial variability and features in the solution.However,geologists have significant uncertainty in developing a conceptual model,e.g.,due to limited data,process-based modeling assumptions,and subjectivity.Therefore,it is prudent to consider the uncertaintyin the geologic scenario when solving the model calibration problem as it will provide an opportunity toutilize the response data in supporting or rejecting the proposed scenarios.We present a new approach forgeologic scenario identification based on dynamic response data by combining gradient-based inversionfor feature extraction and a convolutional neural network for feature recognition.To compactly representeach scenario while ensuring data sensitivity,the approach relies on extremely low-rank parameterizationof individual geologic scenarios based on the Principal Components Analysis(PCA).The PCA basiselements of each scenario are then combined to capture the salient features in any of the scenarios,ortheir possible combinations.An iterative least-squares formulation is then formulated to detect scenariosthat are supported by the observed data.The inversion results in an approximate(smooth)spatial solutionthat contains the dominant spatial patterns.A pre-trained convolutional neural network(CNN)is thenused to identify the relevant geologic scenarios based on the reconstructed spatial solution.Two mainadvantages of the workflow include:(i)the ability to combine different scenarios if supported by data,instead of evaluating individual scenarios,and(ii)efficient gradient-based implementation that does notrequire extensive forward simulation runs.In addition,the training of CNN is implemented using onlygeologic realizations without requiring additional reservoir simulation.The performance of the workflowis evaluated using tomographic inversion and model calibration in fluvial reservoirs.
展开▼
