Accelerating Innovation Through HPC-Enabled Simulations

HPC Ferrari RacesMany engineers are using powerful simulation software but are still not deploying HPC to the full extent. Case in point, I presume most of you have heard about the 24 Hours of Le Mans race. There is one starting June 17. I find it very exciting, not least because teams of three drivers per car compete to complete the most laps around the 13.629-km Circuit de la Sarthe in 24 hours! The race cars reach more than 320 km/h on the straightaway, spending most of the 24 hours at full throttle.

Imagine the roar of the engine drowning out the cheers of the crowds as you speed smoothly around the track in a finely tuned (thanks to simulation) race car. Now imagine the track is a country road or dirt road, not so smooth or speedy now, is it?

It doesn’t make sense to enter the race with the same car you use to fight rush-hour traffic or open up a race car tuned for Le Mans on a country lane. The same holds true for simulation and computing technologies — access to more computing power is better than ever before. Workstations are becoming more powerful — with 8-12 CPU cores standard — and both on-site clusters and cloud-based resources are getting easier to deploy.

Use the Right HPC Hardware and Software

If you think that doesn’t happen anymore, given all the progress in making HPC accessible and economical, you’d be surprised. As I travel to user conferences, I hear many engineers lament about their increased workloads, increasing product complexity and shorter deadlines. However, many aren’t taking advantage of the “supercomputing” power right at their fingertips that could alleviate many of the bottlenecks in their workflows.

hpc speed and simulationAs illustrated in the Speed Simulation and Innovation digital issue we produced with the help of Digital Engineering magazine, companies that are taking advantage of HPC – on the cloud or locally – are seeing amazing benefits. Just to name a few:

  • Aavid Thermalloy uses cloud-based access to ANSYS so that its 60 simulation experts around the globe can access the software, data, and HPC resources they need from any location, any time of day or night. As a result, the simulation team’s output has increased by 200 to 300%.
  • Rolls-Royce used an HPC cloud approach, coupling ANSYS Fluent with an in-house structural solver to provide heat flux predictions for the component walls of jet engines. Running the simulation on an HPC cluster in the cloud provided a five-times increase in speed.
  • KeelWit Technology used ANSYS Fluent on a cloud-based cluster from Gompute in the design of a massive vertical wind tunnel. The company performed simulations of over 50 different operating conditions and reduced development times by a factor of four.

For example, engineers can use (centralized) HPC to improve collaboration with colleagues, work more efficiently remotely, store and manage large datasets, and ensure against disaster recovery. Once companies see how many ways HPC can transform their businesses, they are quick to embrace it for many enterprise applications.

Of course, when it comes to engineering, the highest level of benefits stem from combining the right hardware and simulation software. Our developers have made considerable strides in scaling ANSYS software to take advantage of parallel processing in workstations, clusters and the cloud. For example, they have developed a completely new, frequency-based domain decomposition approach in ANSYS Mechanical, which can easily speed up harmonic analyses and cyclic symmetry modal analyses by a factor of three compared with the traditional mesh-based approach. Our CFD developers set a record last year by scaling ANSYS Fluent software to 172,032 CPU cores on the Cray XC40 supercomputer. You can find all the details in the Speed Simulation and Innovation digital issue.

Unfortunately, I won’t be able to attend Le Mans this year, but speed will be on my mind that week. I’ll be at Science & Innovation 2017 in London on June 13th, giving a presentation entitled “Accelerating Innovation Through HPC-Enabled Simulations”. I’ll explain how to cope with product design complexity with pervasive engineering simulation, the importance of a cloud-enabled platform for scaling the deployment of simulation, and what innovative companies have accomplished via HPC-enabled simulations, on premise and in the cloud. Drive on over if you’re in the neighborhood, but stick to the posted speeds!

  1. How to do constrained optimization in Adjoint solver of ANSYS FLUENT? I am working on minimizing the coefficient of drag on an airfoil section