Implementation of a Scalable Distributed Cloud Computing Platform for Fast Real-Time Reservoir Development

Gutierrez, Edgar (Halliburton) | Wu, Mark Hsu-Hsiang (Halliburton) | Dong, Weixin (Halliburton) | Efimov, Ivan (Halliburton)


Recent developments of new ultra-deep logging-while-drilling (LWD) resistivity tools have increased usage for on-demand computational infrastructure. The tools are capable of providing much deeper determinations on formation geologies than conventional electromagnetic (EM) resistivity tools, allowing more accurate real-time wellbore adjustment and optimization. This technique efficiently explores reservoir insights for maximizing oil production; however, the time to process raw measurements into useful geological information is long owing to the complexity and large amount of data associated with the tools. The conventional computation platforms are not efficient enough for both real-time and post-well formation evaluations based on this tool's measurements. This paper introduces a high-performance computing (HPC) platform which provides flexibility among different deployment architectures and large-scale cloud infrastructure. This enables numerous computational resources to quickly process raw data and provide the information needed to successfully steer a well. The new HPC platform has 50% more efficiency compared to conventional parallelization methods, such as Open Multi-Processing (OpenMP), using same amount of CPUs. Furthermore, faster computation is achievable owing to the scalability of the HPC implementation as well as the flexibility of available assets in the cloud or on-premises environments, which are beneficial for applications with heavily computational requirements and short time constraints.