First Time Utilization of Cloud-Based Technology to Fast Track A 521 Million Cell Gas Condensate Reservoir Dynamic Model: A Case Study from Saudi Arabia

Abstract

Abstract The main objective of this study is to run a high-resolution dynamic simulation on a 521-million cell gas condensate field model for 50 years and capture the effects of gas condensate dropout. Two challenges were encountered on the user and service provider levels. The former is performing such work in a remote location with limited processing hardware resources. The latter is related to resolving memory, CPU allocation, technical support, system resources availability, integration between providers, and simulation needs on user demand. The approach adopted in this field development planning study was to utilize the latest cloud technology to run the 521 million cell simulation on cloud clusters as well as two upscaled versions (5 and 21 million cells). Such procedures can save significant processing time and money. As opposed to direct purchase and installation of clusters that require maintenance, updates, and become outdated over time, with a cloud cluster that is kept updated and maintained by service providers, significant cost overheads (in millions) could be saved. Using such technology allows operators to get global technical support making executing such simulations viable even in the most remote locations. The field under study is a gas condensate field that on its own can present multiple challenges including the gas condensate banking impact and compositional modeling. The main strategy adopted in this study was to utilize the static model with no upscaling, to capture the geological details. With the utilization of cloud technology, all simulations were completed in record time. The 5 million cell model was executed in 23 min, while the 21 million cell model executed in 4 hours, and the 521 million executed in 65 hours. The results of the simulations showed that the gas condensate banking effect was captured clearly after the implementation of local grid refinement (LGR) on the upscaled models. A good match was observed in the production profiles for all key parameters, such as gas rates, oil and condensate rates and their cumulative productions. Using cloud technology saved the operating company over 5 million dollars in cluster hardware direct purchase, support and maintenance costs, making the utilization of the cloud computing technology not only economical, but also bringing about operational efficiencies. This is the first time a cloud-based dynamic simulation is performed on a 521 million cell model in the world and the first time, an on-demand reservoir simulation based on cloud computing technology has been conducted in the Middle East region. This paper will also show that, given the right model parameters, carefully built smaller models can yield results similar to larger models, highlighting the importance of efficiency.