Related Posts

Watch Simulation World Now!

Running Simulations on HPC Cloud Improves FLSmidth’s Scalability and Workflow Runtime

Running Simulations on HPC Cloud Improves FLSmidth’s Scalability and Workflow Runtime

Engineers in a variety of industries are challenged to produce bigger, more complex simulations to bring optimized products to market faster. Cloud-based high-performance computing (HPC) is a great tool to speed up simulation workflows during shrinking development cycles.

FLSmidth, a supplier of engineering equipment and service solutions in the mining and cement industries, needed HPC cloud services to keep up with market demand.

To achieve this, FLSmidth partnered with UberCloud. The team built a fully automated self-service cluster architecture in the cloud that was able to compute jobs of various sizes. The architecture could be expanded to other FLSmidth sites around the world.

With improved HPC capabilities, FLSmidth could simulate its JETFLEX® kiln burner and optimize the product’s performance with a range of fuels (from wood chips to sewage sludge), while maintaining optimal flame shape and low nitrogen oxide (NOx) emissions.

HPC Cloud Can Improve Workflow Runtimes

FLSmidth uses ANSYS CFX to optimize the JETFLEX. In addition, FLSmidth uses other simulation software to optimize its other products. These include:

The JETFLEX simulation shows a multiphase Lagrangian CFD calculation, including solid fuel combustion.

All the software FLSmidth uses needs to run on the HPC cloud system. UberCloud helped FLSmidth create the application software, container registries, license servers, resource management systems, command line interface (CLI) and benchmarks to run the software in the cloud.

UberCloud and FLSmidth used Microsoft Azure™’s CycleCloud to set up this cloud HPC system. Using CycleCloud, the team gained access to self-service cluster provisioning and management capabilities without adding any overhead, hardware or IT requirements.

Benchmarks showed that the HPC Cloud system improved FLSmidth’s workflow runtimes compared to their on-premise Intel Haswell-based HPC cluster with 512 cores. The cloud’s compute nodes showed good scalability.

For more on the current and future HPC Cloud infrastructure at FLSmidth as well as the ANSYS managed software-as-a-service (SaaS) offering named ANSYS Cloud, watch the webinar: Moving Digital Manufacturing to the Cloud.