Last week I graduated from the University of Strathclyde with an Engineering Doctorate (EngD), which is basically an industrial focused PhD. I owe a great deal of thanks to OnScale for providing this opportunity as my project industrial sponsor. In addition to their support, access to OnScale’s cloud simulation platform enabled me to quickly complete the many simulations required for my thesis corrections.
By: Jeff Dobson, Senior Application Engineer at OnScale
Carbon fiber reinforced polymer (CFRP) components pose formidable challenges to inspections due to their anisotropic material properties and often complex morphology. My research focused on creating a platform to provide simulation of ultrasonic non-destructive evaluation (NDE) inspections for components made of such advanced material. The attributes of CFRP components pose challenges for modelling, where accurate representation of the internal structure can lead to high computational demands. Full numerical simulation has historically not been a viable option for complex simulations due to computational constraints. In addition, for ultrasonic NDE inspections, hundreds of simulations are often required to replicate a full inspection technique such as B-scans and Full Matrix Capture. The combination of OnScale’s capabilities supported by my research, with the ability to access large cloud high-performance computing (HPC) clusters to run simulations in parallel, addressed the major barriers for CFRP simulation. As demonstrated in my thesis, this advanced modelling and simulation capability provided an excellent tool to aid the understanding of the ultrasonic response from CFRP components, as well as the ability to optimize inspection set ups.
It started a little over 6 years ago when I was a summer intern with OnScale (back then, the early product was known as PZFlex, developed and maintained by the solver team at Thornton Tomasetti. In 2017, Thornton Tomasetti spun out that solver team and recruited a group of Silicon Valley entrepreneurs to put the solvers onto the Cloud and create OnScale). I had just finished the fourth year of my Electronic and Electrical Engineering course at the University of Strathclyde, where my project was based on designing a low-profile array transducer for shear wave NDE purposes. During this project I utilized simulations to design the device that I would go on to build, test, and validate for my EngD. At the end of my project I was offered a summer internship with the OnScale team in Glasgow building a MATLAB toolbox – the goal was to allow users to build and run their simulations directly from MATLAB. During the summer, I was also offered the opportunity to undertake an EngD by my academic supervisor as part of the UK Research Center for Non-Destructive Evaluation. Completing a doctorate at the end of my degree was never in my mind. However, after some thought, I decided that it was too good of an opportunity to turn down. Unfortunately, not long after accepting, the opportunity fell through – the industrial company partner in the offer pulled out. Looking back, disappointment was the overriding emotion at the time as I was looking forward to the prospect of undertaking an EngD. However, things were about to change.
I have always believed in the words that my mother often said, ”What’s for you, won’t go by you”. As it turns out, if the initial partner didn’t pull out, I would not have received a phone call offering me the chance to do the EngD with OnScale. Fast forward to today, I am part of an excellent and growing team, getting to work in a wide range of interesting applications with new challenges every day.
Being the industrial sponsor was not the only way OnScale helped me get to graduation last week. After my EngD viva, I had some corrections to do that required some further simulations due to a slight oversight on my part. I am a perfectionist by nature, so this added some further simulations to my task list to ensure I produced the best thesis I possibly could. Most of the simulations in my research replicated ultrasonic B-scan inspections, which is an NDE technique where data is collected along a line of a component to build up an image of the internal structure. Multiple simulations must run to generate all the required data. Typically, each B-scan simulation I performed was around 100 subsimulations that individually took on average up to 10 minutes each. In the early days of my research before accessing the cloud, these had to be simulated sequentially and so each B-scan took hours to complete. During my corrections, OnScale allowed me to simulate an entire B-scan in parallel to reduce the computation time from hours to a matter of minutes. Furthermore, multiple B-scans could be simulated in parallel which allowed me to really enhance my thesis. In total, I added and updated 15 B-scans within a week. Thanks to the simulation platform provided by OnScale, I could quickly run full B-scans in a matter of minutes to quickly update thesis results. If I had access to OnScale at the start of my research, I could have done enough simulations for another thesis – but I won’t be writing a second one anytime soon!
I will forever be grateful to OnScale for the opportunities they have provided me and look forward to being part of the team and driving us forward as industry leaders in engineering simulation software in the cloud.