Kasper Steen Andersen, a CAE manager for Danish exhaust and emissions system manufacturer Dinex, talks with us about how his group uses Rescale’s ScaleX big compute platform for CFD and FEA. Keep reading to hear how ScaleX has accelerated their design process, shifted their staffing needs, and improved their competitiveness.
Rescale: Tell us about Dinex.
Kasper Steen Andersen: Dinex is an exhaust manufacturer and developer. We produce standard mufflers and full exhaust aftertreatment systems for trucks, buses, and off-road equipment. Our technologies include pipe-bending and flexible sections installation, and then the whole canning and stuffing of substrates. We also do both coating and manufacturing of emission technology substrates.
Rescale: What is your role in the organization?
Kasper Steen Andersen: I’m a computer-aided engineering (CAE) manager for a team that designs and simulates exhaust aftertreatment systems. I am responsible for the team that does computational fluid dynamics (CFD), finite element analysis (FEA), and acoustic simulations, as well as the computer-aided design (CAD) of these exhaust aftertreatment systems.
Rescale: Describe how you use simulation at Dinex.
Kasper Steen Andersen: The type of simulation we’re doing is CFD. We do single phase and multi-phase simulations such as urea spray simulations (to lower NOx emissions in diesel engine exhaust systems). We do both CFD for conjugate heat transfer, in which we map fluid to structure; FEA to see thermal stresses on thermal parts; and FEA for structural dynamic and static simulations. We use ANSYS Mechanical for FEA and AVL FIRE for CFD.
Rescale: Can you describe what you were using for your simulation IT before you started using ScaleX?
Kasper Steen Andersen: Before we tried ScaleX from Rescale, we used our own small in-house server for AVL FIRE simulations. But since we were generating quite large models, we needed to reduce the runtime. That’s why we turned to Rescale—because we could reduce the simulation runtime from about three weeks to about three days.
Rescale: Can you tell us more about your on-premise server?
Kasper Steen Andersen: We have an in-house server of 32 cores. But it needs to support both FEA and CFD, so it’s split up. So CFD has 16 cores, and FEA has 16 cores with the GPU. We only have 16 cores in-house for CFD exclusively.
Rescale: Was there a specific project or deadline that made you want to move your simulation to the cloud?
Kasper Steen Andersen: Yes, we had a specific project that required a large computer model and had a tight deadline. We needed to do the development within weeks. To meet these deadlines, we needed to reduce the turnaround time for our simulations.
Rescale: Can you tell us a little bit about that project?
Kasper Steen Andersen: Dinex develops exhaust aftertreatment systems for the US, China, India, and Russia. We were developing a new SCR (selective catalytic reduction) exhaust system for a large Chinese truck manufacturer. The new systems are designed to meet the China 5, which are the emission standards that new Chinese trucks need to meet.
“Our higher productivity will enable us to handle more projects. The Rescale platform makes us more efficient and more attractive to our customers, which then brings us more business and more development projects.”
– Kasper Steen Andersen, CAE Manager, Dinex
Rescale: Were you using AVL FIRE for that project?
Kasper Steen Andersen: Yes.
Rescale: What were the IT options that you considered you accelerate your simulations? Did you consider buying more on-premise servers or did you consider moving your simulations to the cloud or maybe just bursting to the cloud?
Kasper Steen Andersen: We thought about investing in more on-premise servers, but it would have simply taken too much time to get a new server up and running. It would also have been too expensive, and we were not sure if it would meet our requirements in the future. The Rescale platform was a more flexible solution for us.
Rescale: Did you consider other cloud providers?
Kasper Steen Andersen: We previously looked into Microsoft Azure, but that was simply too complicated. In contrast, Rescale’s ScaleX platform was easy because we could rent both the software and the hardware and ScaleX has an easy process workflow. All in all, we chose Rescale over Microsoft Azure because it gave us both ease-of-use and scalability.
Rescale: Let’s talk about how you use the Rescale big compute platform. What software do you use and what types of simulations do you run on ScaleX?
Kasper Steen Andersen: We use AVL FIRE for CFD. When we’re running a job, we’re using about 100 cores on average. Our models are about four to five million cells. They are transient multiphase CFD simulations, which typically have a long runtime because we need to capture the fluid and droplet interactions.
Rescale: Can you give us a specific example of a simulation that you ran on ScaleX and how ScaleX impacted the process or the results?
Kasper Steen Andersen: In the project for the Chinese truck manufacturer, we ran six design optimizations of multiphase simulations. Each optimization included two simulations, so in total that was 12 runs. In between each optimization, there were design changes to improve the ammonia uniformity and reduce the risk of deposits. Those are the main targets when doing urea spray simulations. The Rescale platform helped us to reduce our turnaround time for each design optimization to one week—each optimization includes making design changes, generating the mesh, setting up the model on ScaleX, running it, viewing results, and making decisions. We do all of that within a week now.
Rescale: How long did it take before you began using the Rescale platform?
Kasper Steen Andersen: Four weeks. We reduced our simulation turnaround time from four weeks to one week with Rescale.
Rescale: What impact does that have on your business?
Kasper Steen Andersen: In this case, it had the impact that we could meet the customer expectations and deadlines. Since it was an important project, it had a huge business impact.
Rescale: Speaking more generally, what has ScaleX enabled you to do that you couldn’t do before?
Kasper Steen Andersen: It has enabled us to reduce our overall development time quite significantly because we often need to make these types of design optimizations. In the case of the Chinese truck manufacturer, we ran six design optimizations. Each one normally took four weeks, but now with ScaleX they take only one week. So, we are reducing our development time by 18 weeks for the project. That’s quite significant considering development takes about 1.5 years.
Rescale: That’s a 75% reduction in turnaround time and almost a 25% reduction in time to market—that’s wonderful. How has Rescale’s ScaleX helped you operate better?
Kasper Steen Andersen: Our staff are more productive now. Faster simulation actually requires more work from the design and simulation engineers because you get quicker feedback when you have a quicker runtime. It puts some pressure on those engineers. As a result, they can work faster and do quicker design changes since now the simulation is no longer the bottleneck.
Rescale: Do you think you’ll hire more simulation engineers as a result?
Kasper Steen Andersen: Yes, I think so because our higher productivity will enable us to handle more projects. The Rescale platform makes us more efficient and more attractive to our customers, which then brings us more business and more development projects. So yes, we are considering hiring more simulation engineers.
Rescale: Has the Rescale platform impacted your business in any other way?
Kasper Steen Andersen: Since we use Rescale’s cloud solution, which is where we see the software and simulation market going, we consider ourselves to be in the forefront. This is good marketing as it is attractive to customers that we use the latest technologies and are up to speed on simulation trends.
Rescale: Do you have any other thoughts or comments that you want to add?
Kasper Steen Andersen: Rescale has enabled us to be quite flexible and fast in our development time. ScaleX’s ease of use is a clear advantage, and the ability to set up a model directly in ScaleX and view your results is another huge advantage.
This article was written by Mika Pegors.