A lot of us believe in doing what we can to help fight climate change, and for many, that includes recycling. Of particular urgency is the recycling and management of plastics, one of the most impactful and pressing challenges in the sustainability sphere. Encina is an innovator at the highest level in this realm, breaking new ground with partner CPFD Software. With the help of Microsoft Azure high-performance computing (HPC), the company is designing the most powerful and promising plastic recycling solution on the horizon: circular products for a circular economy. To advance the urgent and vital work of designing next-generation recycling facilities, Encina and CPFD ran groundbreaking simulations using CPFD’s Barracuda Virtual Reactor software. In one of the largest real-world Virtual Reactor simulations ever run on any public cloud platform, they took full advantage of an advanced system from Azure and full stack accelerated computing from NVIDIA. Azure is enabling Encina and an expanding and varied pool of companies to provide innovative sustainability products and services to customers, generating potential solutions for critical global problems.
“Having the ability to develop these models using Azure HPC allows us to take leaps in sustainability initiatives instead of slowly creeping forward with preexisting technology. You can model innovative ideas with great confidence.”
Sheida Sahandy, Chief Sustainability Officer, Encina
Encina’s circular vision for the future
Consider a rather grim statistic: only about 10 percent of plastics are currently recycled. The hundreds of millions of tons that are landfilled, incinerated, or end up in oceans has become a pressing global issue. Encina’s proprietary process represents a potential paradigm shift, targeting the other 90 percent of plastics that are currently lost, which includes some surprising materials. “That 90 percent is a combination of many things we interact with as consumers that typically are mismanaged or might not be classified as single use, such as packaging, electronics, car parts, and durables,” explains Carlo Badiola, Senior Vice President of Engineering and Technology at Encina.
The challenge for Encina has two parts: the work needed to convert the plastics, and then taking the resulting material and reinserting it back into the manufacturing stream. The company’s recycling process produces valuable circular chemicals, primarily something called BTX/P, by economically and sustainably breaking down end-of-life plastics into their constituent molecules. Encina’s customers can seamlessly incorporate the BTX/P into their manufacturing, thereby increasing the recycled, circular content of their consumer products, enabling them to easily respond to growing market and customer demands for sustainable products. The same material can be used to make plastics again and again without degradation to create products like electronics, furniture, building construction materials, paint, detergent, and adhesives.
“These difficult-to-recycle plastics are our focus, trying to get them into a circular system,” explains Sheida Sahandy, Chief Sustainability Officer at Encina. “We’re giving these molecules a kind of infinite life—they can come in as waste and go back out to become almost anything.” The potential benefits for society of Encina’s work are twofold. First is that a market is created, deriving value from items that might otherwise be considered waste. And second, market demand is driven toward companies and customers that are focusing conscious attention on what they’re doing with these materials.
CPFD helps point the way
Making the circular system possible requires building a physical facility with large reactors to accomplish the enormously complex chemical process of breaking down plastic waste. Encina already has a working smaller-scale unit online in Texas, and the company is presently constructing its first commercial-scale plant in Pennsylvania.
Designing this unique hardware requires massive computational calculations in the design process, and the reactors are enormously expensive, which is where Azure and CPFD come in. CPFD’s Barracuda Virtual Reactor is an industry-standard tool for researching new sustainability technologies, simulating the complex behaviors of fluid-particle systems in 3D. CPFD’s clients use simulations to effectively communicate technological capabilities and data to partners, customers, and investors so they can deploy and scale with confidence.
GPU revolution: Azure and NVIDIA help Encina break new ground with faster, less expensive solution
CPFD was an early adopter of NVIDIA GPUs for accelerated computing and began parallelizing its code with CUDA in 2012. In 2021, CPFD added multiple-GPU parallelization capabilities to its offering, enabling Barracuda Virtual Reactor to efficiently scale and run on HPC systems, such as NVIDIA DGX hardware systems and Azure GPU-accelerated virtual machines (VMs) with multiple GPUs. “We were early adopters of parallelization on NVIDIA technology,” says Peter Blaser, Vice President of Operations at CPFD Software. “We knew it would give our customers more speed but couldn’t then imagine the true possibilities.”
“For this project with Encina, we used an ND A100 v4-series VM, which has eight NVIDIA A100 Tensor Core GPUs available,” adds Sam Clark, Barracuda Virtual Reactor Product Manager at CPFD Software. “The NVIDIA Ampere-generation of GPU cards have great performance for running Virtual Reactor simulations, and in this case, they were about three times faster than previous generations of V100 GPUs. If a company wanted to buy a comparable system, the machine would likely cost more than $100,000. That was an attractive draw of Azure—Encina didn’t need to spend that money on a hardware purchase, and always being able to access the most performant technology on the Azure cloud platform made all of this possible.”
Running on a four-GPU VM, simulations were completed 400 times faster than on an on-premises, CPU-based workstation. To demonstrate the power of GPU speedups, CPFD revisited a 2012 simulation running on CPUs that modeled 400,000 cells and 4.6 million computational particles, which would have taken three years to complete and was never finished. Today, a comparable model running on Azure HPC can be completed in 13 hours. In addition to exponentially faster model completion, these speedups bring the possibility of running larger domain models—models with finer spatial resolution, more detailed physics, and more complex chemical reactions. These additional insights into system dynamics can inform design and operational decisions, and with such advanced simulations at their fingertips, operators can improve reliability and performance to previously unattainable standards. These improvements have a positive impact on refinery economics, uptime, customer satisfaction, and provide an overall competitive advantage.
Using advanced Azure VMs for both capacity and speed to run one particular model exemplifies Encina’s groundbreaking scale. “This simulation was trying to resolve what’s happening in industrial units that could be 30 feet in diameter with 1016 particles inside them. The math of those numbers isn’t billions or trillions—it’s bigger than that,” explains Blaser. Additionally, CPFD and Azure achieved a 506X speed boost compared to the CPU version of the results. “This was the first time we’ve ever seen anybody break more than 500 times faster than running on CPUs—all on a single Azure VM!”
Paradoxically, the blazing speed didn’t come at a financial cost. In fact, the opposite was true. “Azure tested a couple of different GPU systems because the higher-end ND A100 v4-series VM is much more expensive per hour,” says Blaser. “But on Azure, the fastest possible time to solution using the ND A100 v4-series was actually less expensive because it ran so fast that the simulation needed fewer hours. That was really counterintuitive. It’s not just time to market and getting value—it was literally about half the cost.”
Startups measure up with Azure solutions for a complex world
The Encina project illustrates an important trend in sustainability: innovative solutions are coming from startups, and Azure can help them do more with less. “We’re seeing startup sustainability companies with great ideas, but they don’t have the infrastructure,” says Blaser. “So, to instantly have access to powerful HPC and AI capabilities on Azure is really exciting, and they’re not locked in for years.” Adds Clark, “The ability to use massive computational power in a very flexible manner is really attractive for our customers. They want top-of-the-line hardware, and having it available on Azure is a great option for a smaller company.”
Using the power of Azure and Barracuda Virtual Reactor effectively eliminates technical constraints for Encina. In the past, the company had to contend with less refined, blurrier data. “Now, we don’t need to simplify anything,” says Song Wang, Technology Team Lead at Encina. “After we run a simulation, we learn how to optimize the reactor design and can mitigate risks. We just think about a question and then go to the Azure platform.” Adds Sahandy, “Having the ability to develop these models using Azure HPC allows us to take leaps in sustainability initiatives instead of slowly creeping forward with preexisting technology. You can model innovative ideas with great confidence.”
Find out more about Encina on LinkedIn. #AzureHPCAI #MakeAIYourReality
“The ability to use massive computational power in a very flexible manner is really attractive for our customers. They want top-of-the-line hardware, and having it available on Azure is a great option for a smaller company.”
Sam Clark, Barracuda Virtual Reactor Product Manager, CPFD Software
Follow Microsoft