Using Supercomputers To Reduce Power Plant Emissions
Product News Nov 27, 2019
RJM International, a provider of emissions reduction and combustion improvement solutions, aims to help global power generators and other large combustion plants reduce emissions and improve efficiencies with its first high-performance computing (HPC) cluster. The move also supports the growth of the company’s business in the EU and the UK, particularly in the biomass and Energy from Waste segments of the market. The new HPC environment is designed, integrated and supported by HPC, storage and data analytics integrator, OCF.
RJM International is using the HPC cluster to analyze customers’ existing operational systems and simulate new solutions for their power stations, prior to prototyping. Time taken to complete its analysis has been significantly reduced, whereby a series of Computational Fluid Dynamics (CFD) simulations previously taking around one week, is now complete within only a day. This is, on average, an 86 percent reduction in time. The large memory capacity of the cluster also enables the team to analyze results more efficiently, again saving valuable time and budget.
Anura Perera, Principal CFD Engineer at RJM International, says: “Simulations are becoming larger and more complex for the coal, oil, gas and biomass industries looking to improve new or aging combustion plants. CFD analysis is crucial for the delivery of energy-efficient bespoke solutions for our customers experiencing problems with combustion or high emissions. The new OCF HPC cluster enables us to produce a greater number of simulations at a much faster rate than before and helps support the resolution of complex emissions challenges for our customers.”
The HPC system is comprised of Dell PowerEdge 440 and 740 servers together with Mellanox’s EDR InfiniBand. The solution uses ANSYS® Fluent™ and ANSYS HPC applications and OCF’s HPC management software stack which includes management, monitoring and reporting tools along with an HPC Scheduler to queue up jobs and allow maximum utilization out of the cluster.