Where are we going? A simulation view
The good news: we’re no longer evangelizing about the benefits of simulation. Most people and companies understand that CAE can help create better products, more quickly. The bad news: many still don’t get the fact that simulating one facet of a product isn’t sufficient — to really examine a use case, the simulation has to merge as many aspects of operation as possible. When I drop my phone, I “crash” the structure (the case) into the ground; I may crack the screen, and jiggle loose the antenna or other electrical connections. To truly simulate the function of the phone after the drop, I need to look at all aspects of how I want to use it and how one aspect may affect the next. Companies often have experts in structures, emag, optics and so on, whatever is relevant to what they make; these are the people who make simulations “sing” and come close to representing reality. They are incredibly smart people but not, usually, generalists. To make better modern, complex products, we need to involve generalists in simulation, since they often have a different view of the problems simulation is trying to solve.
An example: when I was at MIT, I had to take a class in welding — not learning how to weld, but about the effect welding has on steel plates. Not knowing anything about welding, steel plates or blow torches, I tried to understand what problem we were trying to solve. Are the plates weaker at the point of weld? Do we see cracks forming? Are welds too hot/cool/slow/fast/wrong? What? I’m an engineer, a problem solver; the professor and teaching assistant were materials specialists. To them, the only possible investigation involved the material properties of steel and how they change as a result of welding. A broader view might not be as educational, but could solve a problem more practically.**
My point: engineers have to be more involved in simulation, and start simulating across physics domains — calling on specialists when needed but using their generalist view of find-problem-solve-problem to truly make a difference.
This means many things. CAE tools have to be accessible to engineers; from a business perspective, the costs have to align to the benefits and usage patterns. Technology has to adapt, too — software needs to be easier to use, targeted at more casual use of very sophisticated solvers, pre- and post- tools. The software has to help the user, by pointing out omissions or errors and highlighting next steps. It needs to remember that these users are experts in their area, but not necessarily experts in the use of that software tool. Where appropriate, it has to support automation so that a simulation subject matter expert can create templates for best practices that the average engineer may not know about. At the very least, the simulation solutions need to not get in the way — they need to be a lot less “software-y” and a lot more “engineering-y”.
User companies have to change their practices, too. Engineers need to be trained in simulation and encouraged to use it — this might mean changing project timelines to increase the time allocated for early noodling on design alternatives, making it up at the end in fewer physical prototypes. It might mean changing the power structure on a project. it might mean buying HPC clusters or pondering a cloud simulation strategy. None of this is easy …
Then, of course, there’s the holy grail that’s increasingly capturing attention. An engineer today says, “Here’s my design; how does it work?” when she really wants to know “Here’s my use case; what’s the best design?” We’re approaching the point when software will be able to find design alternatives we haven’t even thought of. Today’s design optimization software is an important step along that path and something to explore, if you’re not currently using it.
Later on this week I’ll be at ASSESS 2016, the (first) congress on Analysis, Simulation, and Systems Engineering Software Strategies. I’ll be co-chairing a session on the democratization of analysis, simulation, and systems engineering technologies. I actually hate the term “democratization”, but get the meaning: to make CAE available to more users. I’m interested in learning how other attendees (and you, here, if you’re not going to be at ASSESS) view this: who should be using CAE, why, and how? What would your organization’s first step be in “democratizing” its use? Are you thinking about this at all?
** BTW: Not knocking my MIT education. It taught me how to think and question, research, and draw conclusions. It didn’t teach much that was practical, however, and I still don’t know how to weld anything. A fact that gave my colleagues at Bath Iron Works a great deal of amusement. I have come to be at peace with that omission in my education.
Image courtesy of NASA.