Characterization of A Heterogeneous Reservoir in West Virginia

摘要

The production performance of a reservoir is invariablyrninfluenced by the degree of heterogeneity present. Therefore,rnproduction performance of a heterogeneous reservoir cannotrnbe realistically predicted without accurate reservoirrncharacterization. In this study, a systematic and synergisticrnapproach has been employed to integrate and interpret variousrngeological and engineering data that are obtained at differentrnscales to characterize a complex oil reservoir in West Virginia.rnThe lack of detailed data significantly hampered the reservoirrncharacterization efforts. However, innovative approaches thatrncombined statistical and artificial intelligence techniques werernsuccessfully developed and utilized to predict the missingrninformation. These approaches help reconcile stratigraphy,rnstructural, and petrographic data with well log data, corernanalysis, and production data. The results were utilized torndevelop a model of the reservoir and to identify the flow units.rnThe attributes of the flow units were quantified from well logrndata with help of Artificial Neural Networks (ANN). Anrninnovative approach for training and testing of the ANN wasrndeveloped which provided consistent and reliable predictions.rnPrimary and secondary production data have been utilized tornevaluate and modify the model. The flow unit modelrnsubstantially improved the simulation of the secondaryrnrecovery performance. The simulation results confirmed thernpresence of heterogeneities which have had profound impactrnon the performance of the reservoir. The methodologyrnpresented in this paper can serve as a guideline for therncharacterization of reservoirs with limited data.