Leaders in CFD Mesh Generation and Simulation

Broadcast in Software

Call in to speak with the host

h:254863
s:3809831
archived
Tecplot Radio

Tecplot Radio

×  

Follow This Show

If you liked this show, you should follow Tecplot Radio.

Listen in as friends from two sides of the CFD business discuss -

  • The biggest challenges facing CFD & CAD in the coming years
  • The democratization of CFD
  • Advice for young people entering the field

Our guests today are John Chawner, President of Pointwise, leaders in CFD mesh generation and Mike Peery, Chairman of Tecplot, Inc.

This will definitely be an entertaining show.

Transcript

0:15 Steve St. Clair

Welcome back to another episode of Tecplot Radio everyone. My name is Steve St. Clair. Our guests today, and that's plural, this is almost the first for us. Our guests today are John Chawner, President of Pointwise. They are leaders in CFD Mesh Generation and our other guest is Mike Peery, Chairman of Tecplot. Between them, they have over 60 years of experience in helping engineers make better decisions about their designs, so I can't think of two men more qualified to ask a few of the important questions facing engineers who use CFD. Welcome John. Welcome Mike. How is it going, John?

0:49 John Chawner

Its going great, thanks for having me. Anyone who is willing to ask me questions and record the results deserves whatever they get.

0:59 Steve St. Clair

And Mike, how have you been?

1:00 Mike Peery

It's so great having you. Oh, I have been fine. I am good. I am happy to be here, too and I am particularly happy to have the witty and famous John Chawner on the radio with you and that's fantastic.

1:13 Steve St. Clair

I should point out that I do have a mute button, so... I am only kidding.

1:20 Steve St. Clair

No, there is no delay, sorry.

1:20 John Chawner

Are we able to delete or is this on a delayed broadcast...

1:26 John Chawner

...something that some of the words that comes out. Forget about it well. Be careful, man.

1:30 Steve St. Clair

All the flops come out.

1:31 John Chawner

Alright.

1:32 Steve St. Clair

So I want to jump in because we are on a bit of a tight timeline here. But I do appreciate you guys both joining us and I know how difficult that is in your schedules. First, I just like to point out as a side note that today in Fort Worth, Texas where John is from, it's 88 degrees, the skies are crystal clear and it's 49% humidity. Mike is in Seattle and it's a different story, it's 55 degrees with clouds and 77% humidity. On that note, we will jump in. The first area I would like to get your thoughts on is regarding the challenges facing the CFD and CAD business in the coming years. First, Moore's law is still working. The computing power is still allowing engineers to create more and more simulations and that's the good news. The bad news is they are creating far more data than they can every fully understand at this point at anyway. Mike, Tecplot has coined a phrase and invented a software that helps with all of this data, right?

2:27 Mike Peery

Yes, we sure have. It's called simulation analytics. And with that I assume you want me to describe that.

2:36 Steve St. Clair

Sure. Jump in.

2:38 Mike Peery

Okay. Well -- I mean, just to go back to your first thought is what are the big challenges in simulations today particularly, on computational fluid dynamics and certainly one of them is the increasing number and size of cases that engineers are running. Obviously, everyone knows that's listening to this broadcast that cases today are orders and orders of magnitude larger than they were just a few years ago. Certainly, when I started this business in 19... No. Yeah, in CFD that would be in the early or late 70s, a big case on those days was all about a thousands cells on a two dimensional grip, that was huge. Of course today or you know a billion-cell model is not unusual to run for us. So that's the big thing. And simulation analytics is really Tecplot's offering as a solution to engineers that are faced with processing, analyzing, giving information out of large sets of simulation results. They used to be -- you can work on one at the time. Great, we have good postprocessing tools from many vendors for doing that, but there is virtually nothing except for now our offering to really process a large set of simulation results. So there you go.

4:11 Steve St. Clair

And that offering is called Tecplot course.

4:14 Mike Peery

There you go.

4:15 Steve St. Clair

You missed your chance at the plot there. So I had to turn in for you.

4:18 Mike Peery

Well, I was trusting that you are going to do that.

4:21 Steve St. Clair

That's right. John, what about Mesh Generation? How has all these changed the function of a Mesh generation in all of this?

4:32 John Chawner

Well, one of the truisms I think of CFD or scientific computing in general is that work expands to consume all available resources. That's, you know, discs, CPU and sometimes budget. So as the capacity of modern computers has gotten that much larger, our customers continue to push the envelope in terms of the size of simulations they are running in the number of cells and the degree of geometric complexity. For example, we are very fortunate to have Lockheed Martin as one of our customers and there are people there simulating the F-35 Joint Strike Fighter in its landing configuration with all sort of interesting doors and louvers open and operating in the flow field and the geometry is extremely complex. So from a measuring standpoint, you know that the need to model that and to do that in high fidelity where you would require, you know, as Mike said hundreds of millions or billions of cells with length scales varying over 7, 8, 9 orders of magnitude, that becomes quite a challenge for the Mesh Generation. There is a difference between the pre-processing side and the post-processing side which is somehow I always tease Mike about. He has the benefit of a product like Tecplot course where you can bring in a thousand of the customers daily sets all at once and see it all right there when someone sits down in front of a Mesh Generator, you are confronted with the deadly black screen. We can fulfill that problem up from there. So its a different bit of an environment and because of that unlike this other guys, its not an computed intensive thing, its more of an interactive automation type of environment.

6:20 Steve St. Clair

To achieve that you have got a new version of your software Pointwise available. Can you tell us a little more about that?

6:28 John Chawner

Well, Pointwise has been a development offered over the past several years. We just recently released what we call version 17. The number isn't really all that significant, but what it indicates is as our next generation product, it encapsulates all the core measuring functionality in our classic products, the cornerstone product which is called Gridgen which has been around since 1984. But it does it in a new framework that streamlines a lot of that interaction that I was talking about, but also does it in a way that has a lot more capability than the old code ever had. It was a case of a 10-pound and a 5-pound box and you know we took a multiyear effort to kind of bring this new software provision and now we are set, you know for another 25 to 30 years of Mesh Generation development.

7:22 Steve St. Clair

That's great. I have been watching some of your notes go back and forth, yours and Mike's and one thing that you have been talking about -- a related issue of what we were talking about before and that is this computing power and other things have allowed more and more non-CFD experts to weigh it into the process. Can we talk about why that's happening and potential concerns you guys might have about it and opportunities with that?

7:51 Mike Peery

Sure. I'll jump in on that one.

7:53 Steve St. Clair

Yeah. Mike, jump in.

7:56 Mike Peery

Of course like you said, there are a lot of trans in CFD going on now. One being the larger data sets and many larger cases, larger sets of cases and so forth and actually there is a lot of challenges and I don't want to discuss, we may not be able to mention all of them. There are a lot of other big things, right? I thought I'll just give you essentially the things that are on my mind and I think are on most of our customer's minds that trouble them about doing CFD simulation. Probably the top of the list is uncertain dequantification is that they make these runs and they have -- they are not sure how accurate it is. They were trying to design something calculated to our angular force moment or something on a vehicle or a park and there is really no dependable way that will give the engineer of that pliancy of the measure of the uncertainty and the results. That's I think the number one thing. Then there are secondary things on, for us developing models for the physics, from turbulence, multidiscipline simulations, multiphase, all those things. I am not gonna talk about any of those today, but those are big issues particularly when you are computing things that are very sensitive to full dynamics like drag, __9:18__, transitions and operation and then there is the one more thing that is on our customer's minds or now that CFD is so prevalent throughout the industry as far as use and design that there is a challenge on how to appropriately use it in design and so there has been a number of papers and books and stuff on best practices in computer aid and engineering simulation process and it is really a new region or new area for study that will hopefully improve max.

10:00 Mike Peery

And I think one of the areas that will improve is the one that you really started out with, that is as the CFD or simulation in general is being more prevalent in the industry and design more and more nonexperts are applying CFD and that is a big problem that lacks because of the things that I just mentioned there is not an easy way for that nonexpert to know how accurate these results are -- and so they represent to stay a huge problem and there was a conference recently I think John, you got to remind me which one that was, but one of the issues was okay as more and more non-users start using it, there is one campus that says, "how great that is". It is more power to the people, but the other camp which I think or the camps would journalize on that fall and it was like, "Whoa, better be careful". CFD is not a push button analysis and you really have to know what you're doing as far setting up the problem, the battery conditions that that they extend at the domain in next generation and on top of that, you really have to know full dynamics not just the computer analysis and so that word that they use is called democratization and to me, all the stuff right there that John comment on that first batch and then we can hammer on that word of democratization which is -- there he goes, one problem is hard to pronounce.

11:34 Steve St. Clair

It's got too many syllables.

11:36 John Chawner

Well, if you are gonna serve me up software like that I wanna hit that guy. That is just a terrible word. I do not like that word at all. It is a made up word that really doesn't mean anything. If to me, at least, I think it was one of the PLM think tanks that had an event where the whole idea of the expanded role simulation where it is going to take place and how it's gonna to be implemented and you are right, there are two camps that you know, it is a good thing to -- it is a good thing to make this technique in the software in CFD. Put a fine point on it available earlier in the design process where everyone says it needs to be done to get the most bang for the book, but it also means they are gonna be putting it to the hands of probably people who are not dedicated, analysts people are more here you know if you're an engineer or you're a designer, so the question is how do you make sure that they're producing the accurate results, the results they need -- and coincidentally, I was just literally yesterday talking about this very topic on the phone with Jill Wolfram in __12:50__ - I do not know if you know him. He and I were of the same mind regarding this issue. It's that yes, it is a good thing. It is not a bad thing. It is not that we need to put the brakes down, but we all need to go under it with our eyes open and make sure that in partnership between companies like ours, the software companies and the end-users that were delivering the tools, that really make all that possible in a way that is accurate and effective. It is not, the burden does not need to be born by you know just one. It is just not our problem. It is just not the end-user problem, it is kind of a collaborative thing and I have got two examples of that if you want me to go into it.

13:32 Steve St. Clair

Yeah that would be great.

13:34 John Chawner

One of the examples that kind of touches on what Mike said is the issue of mesh quality and how do you make sure that a non-expert analyst knows that the mesh they just generated is sufficient for the class of problem that they are running for the solve where they are using for the results that they expect. Yeah that's hard enough for the expert users. There is a quote that I love from a professor friend of mine from UBC in Vancouver who says that we now embarrassing a little about the effect of the mesh on the solution and that sounds like a pretty downing indictment, but it gives you an idea of how you know the steps were needed to take to quantify. What makes a good mesh from a standpoint of using this for an aircraft. Is it an automotive under hood. Is it rotating machinery. Is it a stent in a blood vessel which solver am I using because the solver has numerics effect of that. Is it an in house tool. Is it open foam or do I affluent or star CC and plus license, that cut those numerics come in at play and then from our side is the sense with a measuring side. It's not only a matter of saying, okay if we know that solvers in this applications like cells with you know certain aspect ratios or certain other geometric qualities are priority, then it becomes you know we are very good at saying well this mesh does not meet those requirements the next step would be how do we go ahead and proactively ensure that the mesh meets those requirements so that is one of the things that opens up when we kind of deliver this functionality to non-expert users you know by definition they are probably not too interested in the mesh really.

15:29 Mike Peery

Really? Not too many people are with especially not them and the more that burn that we can assume and automate for them the better and that is one of the caveats who opening up kind of CFD to kind of a wider audience to help them ensure that the thing that what you know what are giving in.

15:48 Steve St. Clair

At the end of the day, if this -- sorry go ahead.

15:51 John Chawner

Oh no, you have a question.

15:53 Steve St. Clair

Well at the end of the day, if these companies go out and build products and we are talking about the impact of it on their design experiences something that can be turned into a product that goes in someone's veins like a stent or something it has potentially huge implications, right?

16:11 Mike Peery

Yeah which is -- go ahead.

16:12 John Chawner

It is very important that we have a process to do simulation within a larger design process that would result in safe and efficient products and today about doing a way to do that for sure is they still have a role for physical testing of the product before you actually put it in to your customer's hands. And that's true for I think most products, but there are a few that is probably not true, although it is not like the space shuttle. Pretty sure that they could not test that in full scale still with their signals and remark numbers and so forth of the full scale model until they actually pull it so a lot of that -- it was designed based on computer modeling. Most things are cars, commercial airplanes and software figure out the very intensive testing period and so really over the years, you can look at the role of CFD in particular, all this is true for all simulation. It has changed somewhat from. Originally, it was an interesting way to study some candidates in a design process maybe the biggest benefit is instead having to test a thousand different variations of candidates, CFD had reduced that down to maybe just a hundred of interest to test and that was 20 years ago, but now CFD is used from very early preliminary design through a final design and tweaking of clients details and testing instead of just being - instead of used to be a process or giving data back for various conditions. Now, it is really a test. For the most part, a way to verify that the computer models are working, and of course the final test that the product itself needs and its objectives.

18:19 Steve St. Clair

John, how do you go about insuring that this process exists for those who weren't experts. To see if they work. If you guys talked about methods of doing that as is there are organization mainly to get formed or how do people get educated.

18:36 John Chawner

Right. There are a couple of things. Remember I talked to you before about the shared burden. The sort of vendors, of course they are all doing the verification and validation so that when you start as a user, you start, adopt a tool, and you decide that is the one you are going to use, where you have access to this body of material that says, look -- company access test that they are solver on this would be the cases and here is the accuracy, the results and things look good. The users' portion is to say, "Okay, I'm going to use solver X. So I'm going to do my own verification studies on the type of geometries that I am going to be doing everyday. My bread and butter widget, I'm going to do my own validation and make sure that when I apply the tool to my type of problem, the results I get match the other results we have in our intellectual property involving the accumulated data of experimenting in the field of measurements of how your thing works." In that shared as well, and then the other part is the other tools that you are using as part of the process simply need to be flexible enough to support that, and both in terms of the capabilities that are offered and in terms of the companies of billing, willingness and ability to work with you. Now you ask about organizations, the one organization that I can talk to this addressing part of this problem and that would be a technical committee of the American Institute of Aeronautics and Astronautics (AIAA) that is involved in measuring visualization and computational environments and last year, they had a conference specifically on the effect of good quality on solution results, kind of it is a kickoff effort to say to say here is the current state of the art where the known problems and what we needed to doing to continue to work on that.

20:47 John Chawner

That brings together all the constituent folks that I have talked about before. Mesh generation people like us that flows over the people and representatives of the unuser community. So we are kind of all on the same page with respect to that. So that the start that's moving in that direction that kind of make the -- you know, the software much more amenable and overall the kind of avoid as many pitfalls as we can.

21:12 Steve St. Clair

Mike, any thoughts from that? How were your thoughts?

21:17 Mike Peery

What I think you are....say it again?

21:20 Steve St. Clair

No, go ahead.

21:21 Mike Peery

Oh. Well, I was just going to get back to your question and I thought you were moving towards the new engineers?

21:33 Steve St. Clair

Yes.

21:34 Mike Peery

Yeah. So if the question, since we are running out of time, I think the question up maybe important to answer or address is what does John and I suggest or give advice to engineers entering these fields.

21:52 Steve St. Clair

Yes.

21:53 Mike Peery

Liquid assimilation, computerated engineering, design process, that are done all of the above or specializing in grid solvers, post processing, whatever, or design and the first, to answer that question. The first thing that comes out of my mind which I give advised to anyone as they're going to a school and preparing to enter the work world in the rest of their life is to find something that you are passionate about, because if you don't like going to work everyday or let's day, most days. Its going to be hard for the rest of your life and so for me, its a lot about learning and new challenges that excite me which usually have to do with computer modeling and math and science and things like that. The CFD was just like the sweet spot that brought all those good things together, so that for me was what I really like to do and I can remember driving down to Oregon on a family vacation and I'm driving on this four-hour drive and my mind is I'm trying to forget the new kind of bandwidth edition for something, because that's just what I like to do. And so for me that wasn't really work. It was just what I like to do. So that would be my number one recommendation for anyone is to, you know, could have tried to figure out what you want to do and you have your whole life and wants to kind of graduate through it, so you don't have to be in too much of a rush, but that's what you want to do and then find a job that you can do. Follow your passion and, so that, you can, you also provide for yourself and your family and so that's really the trick is the passionate job that can really sustain the lifestyle that you'd like and third thing on the list in way maybe, way down is don't worry too much about how much money that you're going to make just make sure you make enough and have a lot of fun in your life.

23:53 Steve St. Clair

So what will you say John? John what advice would you give to the folks entering the field?

23:59 John Chawner

You know, I got one thing that I got to go around and talk to a lot of college students from time to time, you know engineers who are on the verge of entering the profession. For example, I've been fortunate enough to be on the advisory board of my alma mater Syracuse University in Mechanical and Aerospace Engineering department for many, many years now. And one thing that I try to stress when I get the chance to talk to students is the idea of striving for excellence in their everyday work and that is partly because, you know within engineering, you don't design a space shuttle everyday, you know lot of the work is which you would call Monday and then boring and believe me having work in a Mesh Generation field for almost 30 years now I know what unglamorous looks like. So you need that focus on excellence to push you through the things that aren't necessarily exciting about engineering and just to make that part of your normal workday product, there is a management guy named Tom Peters, whose he almost made a career out of excellence and organizational excellence and leadership excellence and he has a very good quote about what excellence is, he said "its not an aspiration, it's the next five minutes" and if you look at it in that regard, it really manifests itself in that idea of always striving for excellence, manifest itself in a very tangible way. And I try to lead students with that, you know regardless of what career path they take.

25:46 Steve St. Clair

Very nicely said and I want to believe the group with this and you just, you mentioned that the Mesh might be the more unglamorous side of the business, are you saying that Mike is glamorous?

25:58 John Chawner

Absolutely. Just leave it at that, just assuming absolutely saying that. So its times up.

26:05 Steve St. Clair

You. I got a click... Gentlemen, I really appreciate this, it's been a lot of fun. We will do this again sometime as time allows. Very good information here, people can learn more by following John's blog called Another Fine Mesh at pointwise.com. They can learn more of that simulation analytics in Mike Perry's work and the rest of the team at Tecplot at tecplot.com. Both are on Facebook and Twitter and have a lot of fun with social media. Gentlemen, thank you very much.

26:46 John Chawner

You're welcome.

26:46 Mike Peery

Thank you very much.

26:49 Steve St. Clair

My pleasure. (Crosstalk)

26:49 John Chawner

Let's do it again very soon.

26:49 John Chawner

We could have a debate on this. During the presidential debates, we can do that.

26:58 Steve St. Clair

Can be the glamorous question, can we finally get an answer, you know the definite answer on that.

27:05 Mike Peery

As long as you don't mind the fact that no else would be listening. I don't know.

27:14 Steve St. Clair

Alright gentlemen, thank you very much have a good one.

27:16 Mike Peery

Alright, bye bye.

Comments

 comments