Advantages of Automated Rock Type Adjusting History Matching Workflow Driven by Adjoint Approach with Sensitivity Analysis of Different Model Parameters and Saturation Functions

Abstract

During history matching, physical inconsistencies can be introduced and remain unrevealed. A rich parameter space can mimic the historical reservoir behaviour, possibly by sacrificing geological constraints. Improved geological consistency is proposed in this work using rock-type adjustments. Sensitivity analysis of the parameters which characterise different rock types is essential. This paper demonstrates the advantages and sensitivities of the rock-typing workflow when coupled with an adjoint-based approach through various history-matching cases. In order to analyse the effect of individual model parameters on the rock-type driven history matching workflow, three different realisations of a quarter of a five-spot model were created with nine simulation cases. The analysed and modified properties are the porosity, absolute permeability, relative permeability and capillary pressure functions. The rock-type adjusting workflow is based on a Mahalanobis distance calculation; the results are compared to the standard adjoint-based workflow. The history matching process is split into stages with corresponding objective functions, e.g. rate or pressure focus, where the switch is automatic based on preliminarily defined criteria. The rock typing workflow shows more favourable results than the standard approach. The success of rock-type adjustments (validation and correction) lies in preserving the correlations between different petrophysical properties. The automated rock typing workflow can indirectly adjust the relative permeability and capillary pressure through the rock type correction. The extended workflow's performance concerning the parameters characterising different rock types is analysed. The sensitivity analysis provides a comprehensive comparison of the significance of each model parameter on the rock type adjusting history matching workflow. The results subsequently conclude the necessity of rock type adjustments for achieving a better match in the dynamic results while better maintaining the geological constraints of the static model. Conclusively a "smart", well-constrained automation is beneficial for achieving maximum efficiency and quality in history matching. This paper reveals a comparison between the influence of each model parameter on the rock type adjusting history matching workflow through two different automated setup combinations. While monitoring the objective functions of the dynamic results, two geological consistency indicators were introduced. These are the Valid Rock Types (VRT) and the Valid Mahalanobis Distance (VMD), which objectively quantify the validity of the final model regarding the static model assumptions. The automated rock-typing workflow improves the quality and reliability of the history-matching procedure, honouring geological consistency. Moreover, it also enhances the convergence rate and accuracy of the match. It includes automation between the different history matching sequences, which improves finding the right set of model parameters. The results show the necessity of applying the automated rock-typing feature with complex models and demonstrate the degree of inconsistency produced by applying the standard workflow.