Topic 2b "Leveraging the Uintah Computational Framework for Commercial Simulation of Industrial Flares"

Advanced modeling and simulation software, such as the Uintah Computational Framework (Uintah), has been developed in the US, through significant funding by the Department of Energy Office of Advanced Scientific Computing, but has been underutilized by US industry. We will leverage Uintah for commercial simulation of industrial flares on publicly available HPC facilities, and we will further develop a graphical user interface incorporating machine learning and surrogate model development, making this software available to operators and designers to improve combustion efficiency and reduce emissions. The small business will collaborate with developers of Uintah as well as flare manufacturers, testing companies, and operators to improve the accuracy of flare simulations. The Uintah software will be hardened for efficient simulations of industrial flares. The small business will make the hardened Uintah Computational Framework available to industrial flare operators through a web based graphical user interface hosted by commercial servers accessing commercially available high performance computing facilities in the US. In phase I and phase II, the necessary components of Uintah were installed at a commercial high performance computing provider. The small business, in collaboration with the research institution, carried out a parametric evaluation of two industrial flares, and developed a hand-off strategy for simulation of multipoint ground flares. In addition, the small business developed a web-based interface to ease the setup, simulation, and post processing steps. In Phase IIB, the small business will continue work with the research institution, and a petrochemicals producer to carry out Uintah simulations of their elevated flare system including surrogate model development using machine learning. Enhanced capability involving concurrent near-field and far-field simulations in Uintah will be developed to improve capability for evaluating multipoint ground flares. The expanded capabilities will be integrated into the existing web interface. The improved flare simulation capability will provide more accuracy related to combustion efficiency and emissions from flares. The software-as- a-service model for industrial flare simulations will be used to make high performance computing available to designers and users of flares to improve performance and reduce emissions in a more economical fashion than can be achieved through legacy simulation and experimentation alone.