While there is no doubt that we all missed meeting up in person at the 2021 NAFEMS World congress, one of the big positives of a virtual event is that we have hundreds and hundreds of hours of content available on demand.
I know you are all busy people and I have picked out some of the event highlights. If you find the selected extracts interesting then I would encourage you to access the full presentations which are available for free using the links in the References section at the bottom of this page.
If you’re looking for a swift, sharp summary of the content at NWC21 I would recommend a viewing of the 5 wrap-up sessions that were organised by the experts on our technical working groups and steering committees. Each of these sessions focussed on a particular theme; Manufacturing Process Simulation [1], Additive Manufacturing [2], Multibody Dynamics [3], Stochastics [4] and Automotive Simulation [5]. The video clip below contains a selection of extracts from the wrap-up sessions - perfect viewing if you want to appear informed when you invite all your simulation friends round for dinner. Highlights include:
NAFEMS Stochastics Working Group member Frank Günther, Director of Virtual Testing and Simulations at Knorr Bremse, gave us a great quote during the Stochastics Wrap-up session:
"I used to have a fairly narrow view of stochastics, probably shared with many others: Using Monte Carlo simulation in the presence of aleatory uncertainty in the input parameters. But really it is much more: it is a frame for describing reality when it is important to understand the uncertainty of our predictions." - Frank Günther
And the award for “Word of the 2021 NAFEMS World Congress” goes to "Reduced Order Modelling".
Reduced Order Models (ROMs) were everywhere and you can see why. We have been hearing for the past half decade about the opportunities provided by Real Time Simulation and the digital twin and the ROM is one of the enablers that is allowing these concepts to be realised.
There were some great industrial applications of this technology in the automotive crash sector from DLR[6][7] and Renault[8].
Who better to turn to than our keynote Karen Willcox [9] of the University of Texas at Austin to guide us through the premise of a projection-based reduced order model.
AI, ML, Data Science - if there weren't so many different abbreviations in this area, we may have had a competitor here for the "word of the congress" award. Everyone with a smartphone or Spotify account can see the value that can be created by providing decisions/guidance/predictions based on data, but understandably, the engineering industry needs to be careful. A computer may identify a strong correlation between a design parameter and a favourable outcome but as engineers we need to understand why this relationship exists before we can exploit it. Correlation is not Causation!
Engineers are right to be wary of black-box algorithms. In her keynote address Karen Willcox[9] told delegates that the way forward must be a fusing of our existing predictive modelling capabilities with data informed approaches. Dr Willcox went onto stress that:
"Unlike many other successful applications of machine learning, we cannot think of machine learning for engineering design in isolation. We may be in the age of big data but when it comes to almost all engineering applications, we will almost never have enough data" - Karen Willcox
The concern with using off-the-shelf algorithms was picked up in the panel session "Artificial Intelligence in Engineering Design"[10] hosted by members of the NAFEMS Engineering Data Science Working Group. In the clip below you can see Dr Mahmood Tabaddor of Underwriters Laboratory talk about his concern with off-the-shelf algorithms. If you are a fan of horror stories of when numerical algorithms go bad, I'd recommend taking a look at Weapons of Math Destruction by Cathy O'Neil.
At NAFEMS, we are seeing lots of interest from organisations starting to augment their design process with a data-based approach to simulation. In the clip below, again taken from the EDSWG panel session, you can see Jeremy Jarrett of Kinetic Vision giving you his view on how you get started in this area.
NWC21 also saw the 5th instalment of the NAFEMS International Conference on Simulation Process & Data Management. This year’s SDM keynote, Mark Norris of theSDMconsultancy used his presentation [11] to walk us through the key milestones that have taken place over the last 20 years.
The benefits of being able to effectively manage your simulation capability are well known but successful implementations are hard to find. It was great to see some strong, end-user examples with one of the highlights being "User-Centric Traceability for Simulation Informed Decisions"[12] by M. Atak et al. from Robert Bosch. This presentation looked to try and avoid many of the pain-points involved in standing up an SDM capability by putting the user at the centre of the process. The work proposed identifying a "department insider" who would act to accelerate the departments uptake.
Multiphysics capabilities have been embedded in the major simulation codes for decades, but the predicted explosion in multiphysics simulation doesn't seem to have happened yet. This assertion, and the reasons why uptake may not have happened as expected at the start of the century, was discussed by members of the NAFEMS Multiphysics working group and NWC21 keynote, Professor Gernot Kurt Bolger of the Zurich University of Applied Sciences, in the panel session "Multiphysics Today & Tomorrow"[13].
The paper that caught the judge’s eye for the "Best Multiphysics paper" was "Magnetic Latches and Fixtures: A Unique Multiphysics Solution"[14] by D. Sarkar of ANSYS. This work was unusual in that it featured structural to electromagnetic coupling in order to better understand the latching behaviour of permanent magnets. The impact behaviour of the latch would have been difficult to capture using a traditional implicit FEA approach, and so a MBD solver was coupled to a CEM tool. As this is a fairly novel application there is no standard interface available to couple the tools together and so two novel workflows are proposed. It is clever stuff, but it also demonstrates the high difficulty tariff involved in tackling multiphysics problems.
The scene in this area was set by keynote Torbyn Syberg of Boeing. Torbyn outlined the progress that a cross-industry working group involving over 60 industry experts from more than 25 organisations had made on the development of a standard that supports airframe structure modelling and simulation for certification. The standard is designed to allow modelling and simulation to play a larger role in the certification process. In the extract below, from his presentation "Recommended Industry Practices to Promote and Support Airframe Structures Modeling & Simulation for Certification"[15], Torbyn walks us through the Credibility Assurance Framework.
It is not hard to see how in static, linear domains, with the correct processes, tools and education in place, we can base some heavy business decisions off the back of simulation results. But how do you feel about trusting simulation results in a highly nonlinear multiphysics environment, when the simulation is being used to support the decision to implant a medical device into your body?
Well, after viewing keynote Dr Mark Palmer, Distinguished Scientist at Medtronic’s presentation [16], I feel both encouraged about the prominent role that simulation is achieving in the healthcare industry and reassured by the level of rigour and understanding that goes into the decision-making process.
In the extract below, Mark compares and contrasts the strengths and weaknesses of four different types of valid scientific evidence that can be used by the regulator for evaluating a medical device. The key point being that modelling and simulation can provide a complimentary source of evidence. Mark also goes onto highlight the key milestones that resulted in us reaching the situation where simulation is viewed as a valuable source of evidence..
It should be mandatory for every NAFEMS World Congress to feature at least one keynote who has the “walked the walk” and developed a simulation capability in an organisation from the ground up. My favourite bit of the video below is the smile on Dr Ivan Krajinovic's face as his Nevera hypercar outpaces the Ferrari SF90 in a sprint. Managing an organisations simulation capability is a fantastic job, and it must be even better when your simulations are the foundation of a product you are passionate about.
In the series of extracts from Ivan's presentation [17] below, he makes the case for using simulation in the automotive industry, walks us through the simulation capability at Rimac that has been built up over a period of 4 years, and goes onto share his plans for expanding that capability in the future. I’ll close out this section with the final line from Ivan's presentation when he was discussing growing the simulation team at Rimac.
"It is easy to buy new equipment, but finding some good people, that is always hard." Ivan Krajinovic
The last word goes to Peter Giddings, Chief Engineer for Energy at National Composites Centre and Chair of the NAFEMS Manufacturing Process Simulation Working Group. Peter closes the show by describing why you should get involved as a member of one of our Technical Working Groups.
Here at NAFEMS we all share his sentiment that we hope to see you in person, in three dimensions, at the 2023 World Congress.
Artifical IntelligenceCertificationData Sciencemulti-physicsNWC21Reduced Order ModellingSDM
All Tags
Stay up to date with our technology updates, events, special offers, news, publications and training
If you want to find out more about NAFEMS and how membership can benefit your organisation, please click below.
Joining NAFEMS© NAFEMS Ltd 2024
Developed By Duo Web Design
