Battolyser Systems Is Going Green Faster with New High-Performance Compute

It might seem that the human heart and renewable energy have little in common, but that couldn’t be further from the truth.

I studied mechanical engineering with focus on fluid dynamics, how liquids and gases move in the physical world. My research revolved around blood flow in the heart, a biomechanical process that works the same way as a motorized pump and other machines. I developed an interest in renewable energy, specifically using hydrogen—a naturally occurring gas.

I came to Battolyser Systems, a clean-tech scale-up, as the company’s 12th employee. Now, I have more than 100 colleagues. Battolyser® is a hybrid technology: it combines a  electrolyser which converts water into oxygen and hydrogen to power vehicles and industrial applications, hence the name, with a built in battery that stores electricity from renewable sources. 

It’s a complex system using inexpensive, recyclable components to provide clean energy. The electrochemical processes that generate hydrogen happen at the microscopic level, so we build computer models and run simulations to understand what’s happening. 

Multiphysics and Industry Growth

I head up Battolyser Systems’s multiphysics team. We are ten researchers specializing in computational physics, electrochemistry, battery and fluid dynamics. We model the electrochemical processes that power our technology for our engineers and product designers. No matter how small, every change in our product design can alter the oxygen-hydrogen mix and energy output, so solving one engineering challenge might create another. Hydrogen is also highly flammable, and we need to ensure our designs are leakproof to mitigate risk.

Our motto is, “Choose green, go fast, stay safe.” Research into clean energy is advancing quickly, and we have to move faster than our competitors to stay ahead of the industry curve. Computational modeling is one way to accelerate product development.

Unfortunately, we were constrained by our resources. We were running simulations on a small, 40-core high-performance computer (HPC), and simulating a complex model took an extraordinarily long time. I once modeled a fluid flow behavior in a component before I went on vacation, and it was still running when I returned three weeks later. Sometimes, a simulation would take up to two months.

Those are extreme examples, but even a less complex model took five days to run, and we had to reduce the number of variables—and consequently, the accuracy of the results—to run the needed simulations. We had to queue our simulations because we had to wait for one model to run before proceeding with the next one. The lengthy process sometimes prevented people from trying new things.

In an industry where change happens in days instead of years, we couldn’t iterate or innovate fast enough, and we risked falling behind.

A High-Powered Collaboration

We needed to scale our computing power to speed up our models and run more experiments. Based on recommendations from our peers and the company’s stellar reputation, we started talking to HPE.

When we told them about our mission to bring cleantech and green hydrogen to the world, they wanted to support us. Articulating our message was important because the decision wasn’t only about technology; we wanted to work with a vendor that shared our values, offering cutting-edge, sustainable solutions and outstanding support. Partnering with a company that understood and valued our goals was essential, and HPE was the perfect match.

A clear advantage of working with HPE is that the company was attentive to our needs. We explained that we were solving partial differential equations and simulating fluid dynamics and electrochemistry problems using Ansys, a popular open engineering simulation platform. The team is primarily physicists, not IT professionals, so we put HPE and Ansys together and told them what we wanted to accomplish.

HPE and Ansys collaborated to build a solution with sufficient CPU cores and memory nodes. You can have 1,000 CPU cores, but you’re wasting money if your solver only uses 10. We had a good idea of how much storage and compute we needed, but Ansys made some recommendations, and HPE used their input to optimize our solution.

Accelerating the Competitive Advantage

HPE delivered a right-sized solution that vastly improved our processing speed and the flexibility to scale. Our initial HPE high-power computing deployment consisted of a 

HPE ProLiant Gen11 server with 192 CPU cores and 3 x 10TB storage nodes.

Our new HPE hardware is 20X faster than our previous HPC, and we have a lot more storage—and this is just a starting point! We can use more variables in our simulations, improving the accuracy of our results. We can also queue more models in less time, as it’s faster to run one simulation after another using as many cores as possible compared to running multiple workloads with fewer cores assigned to each.

We have 10 physicists competing for computer time, and a faster server allows us to do a lot more in the same amount of time. 

A Fast Payoff

We began implementation in fall 2023, and we got to work assigning cores and working with Ansys to solve equations, multi-scale and multiphysics models. We had to optimize data flows in our simulations to take advantage of our new hardware.

Investing in HPE technology paid off quickly. Within three months of deploying our new system, we ran a multidimensional simulation with multiple variables that had been too complex for our old system. It was a thorny problem that held back the development of our proprietary technology, and the team had been the bottleneck. After the physicists ran the simulations on our new HPE ProLiant Gen11 server, our engineers and designers could finally move forward.

Our solver divides the computational domain into millions of cells, which makes the calculations extremely time consuming. The more processors we have, the faster the simulation runs. With enough power, we could even simulate multi-scales problems ranging from micrometer to meter. 

More Power Means More Savings

Our HPE ProLiant Gen11 server gives us room to add GPUs, which are better suited to running millions of tiny calculations than CPUs. We have the flexibility to add processing power tailored to the specific simulations we’re running, and that’s incredibly valuable to our team.

Running more simulations in less time means Battolyser Systems’s engineers have to perform fewer physical experiments, which are incredibly costly. Although we can model anything, we have to run real-world hardware tests to validate our simulations. The more we can simulate, the lower our costs will be, and the faster we can move to prototyping and commercializing our products, which will impact our bottom line. 

Using a cost-efficient, performant HPC to run simulations faster aligns us with industry practices. Our competitors are using computational models to develop their hydrogen solutions, and we can’t allow ourselves to fall behind. If we’re six months faster to market, our investment in HPE technology will yield far more than the money we spent.

The Future Is Green

Our HPE ProLiant Gen11 server has transformed our organizational culture. We run many more simulations and can test far more ideas than previously. Our new equipment allows us to tackle complex problems, and I expect we’ll see even tougher challenges in the coming years.

Investing in green technology is investing in the future. We may be a startup, but if our product takes off, it’ll mean clean hydrogen for everyone, bringing humanity closer to solving the climate crisis. Working at Battolyser Systems, I see we’re not just building a clean energy solution. We’re inspiring others to create a better world.