Even the most powerful supercomputers cannot be productive without suitable operating software and applications. In engineering, finite element analysis (FEA) is used to create 3D digital models of large structures to simulate how they perform under different real-world conditions (stress, vibration, heat, etc.). The challenge of modeling large-scale structures, such as mining equipment, buildings, and oil rigs, is the sheer amount of computation involved. Running these simulations takes many hours on expensive systems, which to engineering firms translates into a lot of time and money.
MIT spinoff Akselos has been working to make the process more efficient, so that it can be used on a wider basis. The Akselos team – which includes chief technology officer David Knezevic, cofounder and former MIT postdoc Phuong Huynh as well as MIT alumnus Thomas Leurent – developed innovative software based on years of research at MIT.
The software relies on precalculated supercomputer data for structural components — like simulated Legos — to significantly reduce simulation times. According to an article on the MIT news site, a simulation that would take hours with traditional FEA software can be carried out in seconds with the Akselos method.
The startup has attracted hundreds of users from the mining, power-generation, and oil and gas industries. An MITx course on structural engineering is introducing the software to new users as well.
The Akselos team is hoping that its technology will make 3D simulations more accessible to researchers around the world. “We’re trying to unlock the value of simulation software, since for many engineers current simulation software is far too slow and labor-intensive, especially for large models,” Knezevic says. “High-fidelity simulation enables more cost-effective designs, better use of energy and materials, and generally an increase in overall efficiency.”
The software runs in tandem with a cloud-based service. A supercomputer precalculates individual components of the model, and this data is pushed to the cloud. The components have adjustable parameters, so engineers can fine-tune variables such as geometry, density, and stiffness. After creating a library of precalculated components, the engineers drag and drop them into an “assembler” platform that links the components. The software then references the precomputed data to create a highly detailed 3D simulation in seconds.
By using the cloud to store and reuse data, algorithms can finish more quickly. Another benefit is that once the data is in place, modifications can be carried out in minutes.
The roots for the project extend back to a novel technique called the reduced basis (RB) component method, co-invented by Anthony Patera, the Ford Professor of Engineering at MIT, and Knezevic and Huynh. This work became the basis for the 2010-era “supercomputing-on-a-smartphone” innovation, before morphing into its current incarnation under the Akselos banner.