Use of Cluster Analysis to Improve Representative Model Selection: A Case Study

Abstract

Abstract An accurate representation of the range of possible subsurface outcomes is critical for effective appraisal and development of petroleum resources. The infinite number of possible subsurface realisations is typically represented by a small number of representative reservoir models that are then used for more detailed analysis. This study investigated the applicability of cluster analysis to deterministic model selection for an undeveloped gas reservoir, using the results of response surface models. Two cluster analysis techniques were compared against standard model selection methods. Cluster analysis was found to perform better than traditional methods of model selection in generating robust, representative and varied models for scenario analysis and development planning. Generalised sensitivity analysis, GSA, was performed to complement the model selection process.