Approximately 123 years ago, Thomas Alva Edison established the Edison General Electric Company. For well more than a century, our company has not strayed from an empirical Edisonian culture where trial and error experimentation is encouraged and failure is not seen as an impediment, but rather one step closer to a solution.
However, we are now beginning to greatly benefit from the ability to test and experiment virtually by leveraging state of the art simulation and modeling of complex scenarios and scientific problems. We still remain steadfast to our Edisonian heritage but are merely using a new means of experimentation, testing and validation.
Here at GE Research, investments in highly scalable computing hardware and software, including the 2011 purchase of a Cray supercomputer, are opening doors that allow us to do work that was not previously possible internally. We are now able to push forward the state-of-the-art.
We are deploying computational modeling across nearly every GE business area. It is reducing cycle times and engineering costs. Physical tests for an aircraft engine can take months to conduct, cost millions of dollars in custom hardware and then several more months to analyze the data. We can run numerical simulations in a fraction of the time, testing dozens of configurations at a tremendous reduction in the cost.
For our turbomachinery businesses, the use of computational models is critical to the business unit remaining competitive. Modeling and simulation is allowing us to understand the physics and multiple interactions of physical systems at scales that were never possible in the past. Some phenomena significant to design decisions are difficult or impossible to observe in a physical experiment, but we can make such measurements in high-fidelity computer models. Critical to this process is the validation and verification of the models to accurately reflect the data collected from physical experiments.
We are constantly creating new alloys and composites that can withstand challenging operating environments. We are building capabilities to simulate new materials in the conditions and temperatures inside an engine to discover a design and process for a material that would be capable of performing its engineering tasks. Prior to modeling, alloy and composite creation was much like cooking, except this experimentation could take months and tens of thousands of dollars for each candidate material. There is no doubt that software and computational modeling is not the same as the trial and error that scientists have relied on for decades. But, it is still grounded in the same principle. Empirical tests are still critical to validating products that require high reliability. We just now have the ability to test more at reduced cost, increasing speed-to-market and reducing innovation times by factors of x.
Rick Arthur was the first Computer Engineer graduate from Clarkson University. He also holds Masters in Engineering (Software Systems Engineering) from the Rensselaer Polytechnic Institute and an MBA from The University at Albany. Arthur, a 20+ year GE veteran, is a member of the U.S. Council on Competiveness’ High Performance Computing Advisory Council. You can follow Rick on Twitter at @arthurrge.