Welcome to Reimagined Mobility podcast series. I'm here with Brett Sheets. Brett's been with AVL for over 20 years. Welcome, Brett, thanks for joining me. Thank you. Simulation we've talked quite a bit already with with some colleagues of yours regarding the different products and services ADL has, But I don't necessarily want to talk about AVL. We want to talk about simulation in general. I think your experience, if it serves me well, my memory here is really in the mechanical simulation space.
So maybe give some of our listeners and viewers an idea what different simulation areas are. And when you say you're more mechanical, so what are the auto guys and what does mechanical mean? Yeah, so for mechanical, it's it started out being static, finite element analysis. So just doing stress analysis. But as as as we get into an engine, say so 25 years ago I started working with engines that got in more into the multi body dynamics.
So the interaction of each finite element model that you were making separately are now connected together inside of a multi by dynamics tools for example like Excite which AVL sells and supports. And there we're able to take the summit through the full simulations and look at how the pressures from the combustion act on the piston. The loads traveled down through the connecting rod produce torque on the crankshaft.
Look at how the, the bearings are supporting these loads and that's looking at say, durability. So looking at the loads, looking at how things are going to wear or fatigue. And as a side product of those simulations we also look at NVH, which is noise, vibration and harshness and that's really what you can feel maybe for an automotive vehicle setting in the in the vehicle. And maybe at the beginning I know my son's car when it starts in the morning, it's trying to warm the engine up.
So it's going through thumps, through things, things that are going to make the engine vibrate, that gets into the chassis and that gets into your seat and under your steering wheel. So looking at how things are vibrating so those are more tactical and then you have the ones that you can hear. So you have the frequency range up to say 3000 hertz where these parts are vibrating. Then that's moving the air that then gets to your eardrum and or could be vibrating other parts.
So that's multi by dynamics, durability, NVH is where it's going really. So I think I answered your question there. It's good. It's very good. So a lot of times when when people talk in the simulation space, they talk about one simulation and talk about the simulation, maybe all two types of simulation. Can you highlight a little bit the difference between what those two or maybe again, additional ones really mean?
Yeah, sorry, only two, but say electric vehicles does it also apply to things as you just mentioned? Hey, you know, basic engines or engines that have been around for 25, 50 plus years. Right. So on the structure side there, we're predominantly people are talking about three dimensional. So we're physically modeling the parts with finite elements, democratizing them up in three dimensions where things are moving in the X, Y and Z direction, but also have rotations.
There is a subset of that where we look at early on in designing engines or not just engines, but also motors and drive lines, which is torsional vibrations, where you can simplify the parts down to just one degree of freedom, which is predominantly just the rotation. So it's a spring damper system. So we simulate the cross-section of a of a part and then we can get the torsional stiffness of that.
The forcing function would either be the electric motor poles as they pulse then causing torques or could be the individual firing of a cylinder, causing those pulses down the system. So those those cause durability issues, those also call it NVH issues. And so you've probably felt that in your own vehicle where you get maybe some shutter or something of this nature where you're surging back and forth.
So generally speaking, we're talking three dimensions, but there are subsets of doing one dimensional simulations. Okay. And in the mechanical simulation, I know there are systems that we have that other ones has as well as it relates to simulation.
We also simulate not just maybe a an E-motor, but to complete, let's say, electric propulsion system, maybe not just a a combustion chamber, but the complete engine is is that something that is is growing is is something that is dependent on if you're a system’s guy versus a specific component and or part guy. More important how is that.
There's the I think on the electric motors side, there is the controls that we take and that we're trying to take into consideration as far as how the direct voltage has to be converted into an alternating current. The the cleanliness or how clean that current is, is if it's not done well, you can get you have to convert something into digital and basically you're making a sine wave ideally, but you can't do that because it's digital.
So you're getting stair steps and these stair steps can then produce torques that aren't perfect and those can get into the system. So that is done on the control side. And but we also have to consider how they could affect downstream on the on the durability side or on the torsional side. I think systems we want to talk about, you know, is is all looking at an internal combustion chamber to simulation for it. Are we are we expanding to the, to the simulation.
I mean to that to the system. Right. I mean the for I could say that for the look when we considering systems on a on an engine, the internal combustion is performing so we have fluid dynamics people that are simulating that that gives us a pressure trace that can come in to our as our forcing function into our multi bio dynamics. On. The valve train is part of the gas exchange.
So we are simulating valve trains and then simultaneously with the crank train and also if there's any gear trains or pulleys or things of this nature, those are solved simultaneously. It's not always necessary to go all out and do all of those calculations. You can model them separately and decouple them. But in certain situations it is important that there can be some some coupling of those systems that can lead to things that you wouldn't see if you decoupled them. Okay, very good.
So you're talking a lot about engines is still very much a field that you're simulation is very important, maybe even more so. Right. Less people are being deployed towards that because the human capital is deployed to more to more evolving EVs and fuel cell based propulsion systems. So into space that we've talked about. And let's stay with engines again, gas and diesel engines.
I've always wondered what what's more important is the processing speed of a simulation tool, more important for your customers, or is it more to the accuracy and a correlation to to the real world of the models that we're using? Or does it depend on what somebody is trying to do? I, I think we're to the point where we can do both.
Actually, I was thinking about this in preparation that 20, 25 years ago when when I started doing this, the we would model things on a pretty coarse approach we were limited to by our computing resources. I almost started back when you had to build tiny elements by hand.
Not nothing automated, but but we had to keep them small because you had limited resources with your computers and with Moore's Law tracking along through the eighties and nineties, things were getting faster every year, every every month, almost. And that allowed us to not just look at what we normally would look at, would say peak cylinder pressure or maximum torque or an overspeed condition where we only looking at individual points throughout the whole load curve.
But we are now able to look at the whole speed range from part load to full load. So we were able to do more. And the idea there is that there could be conditions that you would miss where torsional excitations could get reach or resonance could cause damage. And if you didn't do the so-called speed sweep of multiple loads and so forth, you could miss that. So computers allowed us to, to do that, to be able to submit multiple cases, even of the same model.
We have the luxury of AVL to have unlimited licenses. But you still can submit a lot of a lot of simulations and more recently, the post processing is becoming automated. So you would spend a lot of time just gathering the results and then showing them. Now we have apps and so forth that create these for you automatically and you're looking at things in a consistent manner. And even comparing models, the accuracy part, we've reached that accuracy.
We have correlated models, we know the techniques that we need to do to get the correct stresses in the crankshaft, fill it, say, and we have reached that accuracy. So we're able to do both. I believe at this point. Yeah. Accurate and quick. Good, good. So maybe staying a little bit with this, I think it ties in nicely to my next question. If you were to pick over your last 25 years, your your favorite simulation tool or simulation product, what would it be?
And tell us a little bit, how has that tool evolved? And maybe you're picking a tool that we've had or that you've been using or supporting for ten plus years, whatever it might be.
But pick a tool that really is, again, hopefully one of your favorites and tell us a little bit how has this tool evolved, maybe from, again, from exactly what we just talked about, process, power, post-processing, power modeling, accuracy, the finite element analysis that maybe go into how to maybe even use the user friendliness, all of that stuff. Yeah. So I there's, there's probably a couple of them but I think I've focused on what I've kind of almost become an expert in, I would say.
And that's withinside of EXCITE™ We have the capabilities of not just solving the structures that we're also have the so electrichydrodynamic joint, which solves the fluid film or more particular say the oil in the main bearing or the oil film thickness in the connecting rod big end and this was part of why I enjoyed multibidynamics when I started and I saw this and I one of my first job was to work in this area.
At the very beginning there was it was groundbreaking, I would say, to be able to solve this, but it was very basic. It would solve the Reynolds equation for the fluid film simultaneously. But there was assumptions. It was fully filled. We couldn't model any oil grooves, We couldn't model any oil holes. And then over the years, those were added inside of EXCITE™. So we could see we could put the groove where the oil comes into the bearing.
We could also model any deformations in the bearing due to manufacturing. When you bolt up the main bearing cap to the block. And that was a big improvement to be able to get the boundary conditions right, to get more accurate results in the better beginning, the joint was really to get the nice load distribution from properly from the crankshaft of the block, but it evolved into being able to look at the details of the bearings.
So we're talking very microns of closeness and then that also involved into being able to not just solving the Reynolds equation, but we introduced the ability to look at friction from metal to metal contact, how the parts would if the film thickness got small enough, how the peaks of the parts would touch each other and actually contact and cause additional friction. This can cause heat that has to be worried about it can damage the bearing.
And also we can simulate the wear so we can progressively run a simulator and look at the load. So this is going to remove so much material based on models like Archer's law. And then we can re profile the bearing as if it were happening in real life and then run the simulation a second time. And I find this interesting that we can do this. And then you lay the part on the table and you do an EXCITE™ simulation.
You can see very, very similar wear patterns and I find that very interesting and that's cool. Kind of pretty cool. Yeah. Yeah. So I think my next question has always been I've always had this in my mind. You know, there's this feeling of a simulation tool is super critical, obviously, on how accurate it is, right? So I get as close, closer correlation to real world data.
Once I put to device the engine in the case that we've talked today together versus what I've had to compare it to what I model. Right? Yeah. But I always wondered how important you to your customers, to our customers are in just general to, to simulation tool and product users is to user interface and to user friendliness. So how easy is it for for maybe a layman like me that a tool is easy to use and makes it real simple to, to get to running a model and get some data out of it?
Or is this a different type of application? And some tools are very simple and they're specifically for, let's say, a layman like me. Yeah. And then there's tools that are really for the for the let's call it the really the the nitty gritty engineers that want to go into the details, you know, let's call them the nerds, right. I really every little finite data point. Can you share a little bit a light on that.
Yeah I'm back In the old days everything was command line driven so you would type a command to to to make point A to go to point B and then there's the, there's the computers came in, the graphics, laptops became more popular and so forth. The, the graphics. And I'll admit our tool is the excite tool for examples, I think probably 30 years old plus now.
So the very beginning that we're still using some of those same graphics ideas, but we have recognize this as a company and over the past several years we've started to simultaneously develop and now it's it's part of part of our offering is are similar simulation desktop where that is. More. Progressive. It is what I think new users would expect to see the video gamers, if you will, where you have more flashy graphics. It's still the same solver in the background.
You still getting these the same precise answers that you getting before, but you're getting more visual feedback as you assemble parts, as you put things together. In the past, I think you had to use a little bit more of your imagination because you didn't always have a 3D confirmation that you were putting things together correctly. So this I think there's pros and cons in both kind of old school and I think you should start from the bottom up and not jump in deep in head first.
But it allows you, I think, now to maybe take those bigger leaps and you have more confidence that you're putting the putting things together. So I think as things as things progress, we have more better tools to help us with these these things. And a lot of these also have so-called apps in the background that are automating of the steps that we've had to do manually.
Okay. And this is important because you could get, say, the expertise of AVL will put together a workflow and a new user can follow this workflow. An experienced user could say, Well, I like 99% of what AVL’s doing. Maybe there's something I'd like to do here so he can tweak that as well. So but I think this is an important thing that we're trying to help, is to is to make make more entry level users become quicker experts.
Sure. Sure. And maybe last question with all this, let's say evolution of and progress we have from, again, simulation tools. And you mentioned 25 years ago to now this has moved to UX. How that has changed. How do you stay up on the latest and greatest trends as it relates to simulation technology and maybe even simulation methods? Right? I don't know how much you guys already are playing a game or are playing a big part in the simulation tools. Chat GPT Now is this big?
Allen This big craze right now that we're trying to see where can we implement this? Almost anywhere. How do you stay up on trends? How do you stay up on what maybe makes sense? What doesn't make sense? How do you form your opinion, how you have your professional know how? Yeah, I I'm intrigued by what I hear about this.
AI, I think my only real exposure to this so far has been probably in the last six months or so where part of what we were talking about earlier is when I was thinking about how things have progressed. When we started just doing basic models, there was a point where we could start to do optimization or more dollars because we could do more simulations. I think on the optimization side there, we now have some tools at AVL that have some artificial intelligence built into them.
So by running a few models, this thing learns how the behavior of the system is going to is behaving based on what you've told it. And it more quickly guides you to the optimized direction. Wow. I find that interesting. It's probably a little bit above me. I try to read up on it when I can, but I learned about this, I guess through osmosis by other people are able doing work and following their projects or following the reports that they put out. So okay, very good.
Thanks for your time. And you're welcome. Listeners and viewers tuning in. Thank you. Take care. Bye. Thanks for listening to. Reimagine Mobility Podcast. If you like this episode, please. Subscribe and tell a friend.