Automated approach to reservoir zonation: A case study from the Upper Permian Dalan (Khuff) carbonate ramp, Persian Gulf

Abstract Reservoir zonation is one of the fundamental processes conducted on permeability-porosity core data to classify a reservoir thickness based on its productivity potential. In all the methods currently used, human decision plays a prominent role at every stage, from basic reservoir rock clustering to final flow zone determination. This study aims to reduce the often biased human decision-making role at each stage of the reservoir zonation by utilizing simple mathematical algorithms, and enhance the reservoir tock clustering approach. To do so, we propose the application of the elbow algorithm as a straightforward yet effective method to optimize the cluster numbers, and the Pruned Exact Linear Time (PELT) algorithm to effectively determine the zones with identical fluid flow potential (in this paper regarded as Hydraulic Flow Unit – HFU). Both the elbow and PELT algorithms are employed to effectively classify reservoir thickness based on the commonly used flow zone indicator (FZI), reservoir quality index (RQI), and flow unit speed (FUS). The efficiency of the suggested mathematical algorithms was tested against the results of the sedimentological and petrophysical study of the Upper Dalan (Khuff) gas reservoir, the Persian Gulf. Three lithofacies associations (sabkha and tidal flat, lagoon, and shoal) with 11 distinctive lithofacies plus crystalline dolomite have been identified and the major diagenetic processes influencing their petrophysical characteristics have been assessed. Based on the automated approach, the studied Upper Dalan succession can be subdivided into five HFUs with consistent poroperm values and distinct lithofacies and diagenetic modifications: barrier unit (HFU 1; permeability (k)  100 mD). The study suggests that the automated approach may be an effective alternative to a conventional reservoir zonation of similar mixed carbonate-evaporite reservoirs.