Rescale

Bombardier Aerospace, a global leader in aviation product design, manufacturing, and support, was seeking to demonstrate the viability and superiority of industrial-scale simulation in the cloud for both large and small companies across most engineering and scientific applications. The project involved the setup and simulation of unsteady airflow around aircraft landing gear, a complex task requiring significant computing capacity and specialized software. The challenge was to provide a platform that could handle the computational demands of the project, while also being user-friendly and accessible. The project was conducted in conjunction with the HPC Experiment organized by Ubercloud, using open-source data and simulation tools exclusively.

Eco-Runner Team Delft, a student engineering team at TU Delft, has been designing and building the most efficient hydrogen-powered vehicles since 2005. They participate in the Shell Eco-marathon competition annually, where the goal is to complete a 16-kilometer course using the least amount of fuel. The challenge lies in designing a vehicle with a fuel-efficient propulsion system and a body with minimal resistance or weight. The weight of the vehicle is particularly crucial to fuel efficiency, especially in the 2017 Eco-marathon course that featured a hill. The team's design efforts were focused on making every component of the car as light as possible to reduce the energy requirement to climb the hill. However, they faced the challenge of exploring and simulating many weight-saving design options while maintaining structural integrity, all within a constrained project schedule and a student budget.

Optisys, a small radiofrequency (RF) antenna design company based in Utah, faced several challenges in its operations. The company specializes in 3D printing custom RF antenna solutions for military and commercial applications, a field that requires intensive simulations. However, the company's IT infrastructure was not up to the task. Optisys' team members were spread across different regions of the US, making it difficult to collaborate effectively. The company was also limited by the need to set up electromagnetics licenses and mechanical package licenses on local workstations, which meant only one person could use them at a time. Furthermore, the company did not want to invest in setting up a local intranet and server to check out licenses, as this was outside of their core competency. Another challenge was the need to purchase new high-performance computers for simulations every two to three years, a costly and inefficient process.

Adaptive Corporation, a leading Digital to Physical Product Lifecycle Company, was facing significant challenges with its High-Performance Computing (HPC) system. The company's workload varied greatly throughout the year, leading to either underutilized computing capacity or insufficient capacity when multiple projects were underway simultaneously. This resulted in bottlenecks when multiple jobs needed to be solved at the same time, causing projects to run over schedule. Additionally, the on-premise hardware was becoming outdated within 2-3 years, necessitating frequent upgrades and a dedicated IT support team to maintain the servers and licenses. These issues were undermining Adaptive Corporation's mission to provide tailored solutions that help customers shorten development and production cycles.

American Axle & Manufacturing, a leading global Tier 1 automotive supplier, was facing challenges with their simulation needs. They use computational fluid dynamics to simulate the lubrication process, aiming to optimize the way oil goes onto bearings or gears inside their products. These simulations are complex and require significant computing resources. Their models required anywhere between 32 to 150 cores to run, dictating the size of the server needed. Before using Rescale, they had a single server with 64 cores. The main challenge was the scalability of hardware requirement. To optimize a specific axle, they needed to conduct several runs, which could be different angles of inclination of a moving car or an SUV driving at different speeds. Depending on how soon they needed an answer, they might want to run a number of these jobs simultaneously, instead of waiting to run them one by one.

Boom Supersonic, a Denver-based startup, is aiming to reinvent air travel by designing a new supersonic passenger jet. The challenge lies in the fact that the world's first commercial supersonic aircraft, Concorde, was not commercially successful due to high operating costs and the inability to fill 100 seats at the $20,000 round-trip ticket price. The Concorde was retired in 2003 after Airbus could no longer source parts, marking a rare backward step in technology progress. Fourteen years later, Boom is reigniting the dream of supersonic travel and showcasing its updated XB-1 supersonic demonstrator at the Paris Air Show. However, the design phase this time is very different from the 1960s, requiring advanced computational capabilities and efficient data management workflows.

Brother Industries, a global business with operations in over 40 countries, was seeking ways to increase growth and improve cost competitiveness in the highly competitive global market. As part of this strategy, the company decided to transition its internal IT infrastructure to the cloud. The first step in this transition was the introduction of Rescale, a high-performance computing (HPC) solution built for the cloud, to their printer development team. However, the company faced a challenge when the maintenance contract for the hardware built into their in-house developed analysis automation system, ATLAS, expired in 2019. The company was forced to decide whether to renew the contract or seek an alternative solution. Additionally, some of the analysis tools built into ATLAS required high-spec CPUs, which put pressure on the company's budget.

LSIS, a Korean enterprise, was keen on building a universal high-performance computing (HPC) cloud-based CAE platform. The company had established an organization called C2 in 2015, with the mission to develop an effective and reliable system to support CAE and CAD. The objective was to create a system that their in-house engineers could use anytime, anywhere without any limitations. However, the cost of building a private HPC platform was over 5 million USD, which was prohibitively expensive. The high cost of constructing an HPC data center made it more sensible to transition to the public cloud. LSIS also faced the challenge of finding a cloud platform that could effectively run CAE software like ANSYS, STAR-CCM+, and Abaqus without requiring significant time and money to install the software on the infrastructure.

Manor Racing, a small but ambitious Formula 1 team, was seeking a technological edge to enhance their performance during the 2016 FIA Formula 1 World Championship challenge. The team needed a solution that could handle the complexity of Formula 1 racing, where every thousandth of a second counts. They required a platform that could work in tandem with their existing race strategy simulation software, enabling their engineers to evaluate thousands of simulations and strategies. The goal was to be at the cutting edge of innovative decision-making during a Grand Prix weekend, all from a laptop web browser.

Modine Manufacturing Company, a global provider of heat transfer solutions, was facing challenges with their computational fluid dynamics (CFD) simulations. Their workstation computing systems in North America were relatively large but lacked an in-house grid computing environment. These workstations, with 10 cores and 64 gigabytes of memory, were constrained by memory and computing bandwidth, limiting the amount of parallelization they could use in their simulations. Modine Germany had a grid updated every five years, with 32 processors and plans for expansion. However, the company wanted to eliminate assumptions in their models, which added complexity and increased the model size. They also wanted to do more transient simulations, which are complex models that take a long time to run. Running a transient simulation on ten cores took them about three months, which was practically impossible. They realized they didn’t have the computing power to quickly investigate methodologies without purchasing their own computing equipment, which would have significantly delayed their development.

Resolved Analytics, a boutique computational engineering and optimization consultant firm, faced a significant challenge in managing their fluctuating simulation needs. The number of customers and the computational demands of their problems varied from month to month, making it difficult to manage with their internal capability of 16 processors. Owning a standalone software license capable of handling the highest workloads was not economically viable, especially during months when only 8 or 16 processors were needed. This resulted in underutilized resources and inefficiencies. Furthermore, the firm found that 16 processors were insufficient for complex multi-physics problems, which required 64 or 128 processors for efficient execution.

Eric Lee, a graduate student at the University of Kansas, was conducting research on nanofluidics, specifically nanowetting problems. His work involved the use of molecular dynamics codes, such as LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator), for computational analyses. However, to fully leverage the scalability of LAMMPS and expedite his analyses, he required access to a large computing cluster that could work seamlessly with LAMMPS. He also needed a platform that could support his specific version of LAMMPS and any custom scripts he wanted to upload. The challenge was to find a solution that could provide the necessary computational power and software compatibility, while also being user-friendly and responsive to his specific needs.

A³, with its project Vahana, aimed to revolutionize personal air transportation by introducing an autonomous, pilotless aircraft. The challenge lay in the design phase of this vertical takeoff and landing (VTOL) aircraft. The project had a vast design space to explore, with designers seeking sci-fi inspired shapes, structural engineers focusing on weight, and aerodynamicists aiming for sleek, high-performance flight surfaces. The task was to create a large number of virtual prototypes to cover the design space and satisfy all stakeholders. However, achieving this with a minimal in-house IT footprint was a significant challenge.