FLYING WHALES is a French startup that is developing a 60-ton payload cargo airship for the heavy lift and outsize cargo market. The project was born out of France’s ambition to provide efficient, environmentally friendly transportation for collecting wood in remote areas. “We have one of the biggest forested areas in Europe, but these areas are on mountains that are very difficult to access,” says Guillaume Martinat, lead aerodynamics engineer for FLYING WHALES. “This is why we need to create an airship that can load and unload cargo without landing, in hovering flight.”
To design its airship, FLYING WHALES runs complex Computational Fluid Dynamics (CFD), a tool to numerically simulate the flow of any fluid, and structural analysis simulations, which require large amounts of compute capacity. The company cannot perform physical testing because the airship is too large, and testing would prove too expensive and take too much time. Instead, engineers need data to size the airship and define workloads for every flight phase. CFD gives engineers this much-needed data without having to manufacture any parts, enabling a much faster design process. However, each computation requires about 600 cores, and it takes approximately 400 computations to generate one model, requiring significant computational resources.
FLYING WHALES chose to move its HPC environment to the cloud, running its CFD workloads on Amazon Web Services (AWS). “We evaluated several cloud providers, and AWS provided the best performance for us,” says Martinat. Specifically, FLYING WHALES chose to run on Amazon Elastic Compute Cloud (Amazon EC2) C5n.18xlarge instances, which support Elastic Fabric Adapter (EFA) as the Amazon EC2 instance network interface. The C5n instances provide the power and scalability FLYING WHALES needs for its CFD workloads. FLYING WHALES provisions C5n instances using Amazon EC2 Spot Instances. Spot Instances are spare Amazon EC2 capacity available at up to a 90-percent discount. With Spot Instances, FLYING WHALES was able to lower the cost of its HPC clusters by 64 percent.
Additionally, the company uses AWS ParallelCluster to simplify the deployment and management of an HPC cluster to run CFD simulations on AWS. Now, using NICE DCV, FLYING WHALES can securely stream applications while dramatically decreasing data transfer costs, so engineers can inspect solutions without ever having to download them locally.
To learn more about how FLYING WHALES scaled and accelerated performance of their CFD simulations on AWS, read the full case study here.