FEA Meshing Programs used in different industries.

[VN1981]

Hello,
So far, I (we) have been using MSC Patran and/or Femap (depending on the client) for meshing our CAD models. A small group use Altair Hypermesh and increasingly we are using Hypermesh in lieu of Patran as a pre-processor. So I receive bdf files generated in Hypermesh and we import the same in to Patran for further analysis.

I've been trying to explore other opportunities and I've been told that knowledge of Hypermesh varies from essential to optional depending on the industry. It is rather difficult for me to learn Hypermesh at work currently. I have to attend some sort of training courses to get access to the software and its usage.

I just wanted to gather opinions from Stress Engineers, especially working in Aerospace & automotive domain, about the different programs that are used for pre-processing.

 

[aerostress82]

Soon MSC Apex will take over all other pre-processors.
I myself am proficient in 6 pre-processors, but due to the fact that MSC Patran and MSC Apex will be leading the market in the next 10 years with the new capabilities added up, I would advice you to learn these 2 at full capability. You can learn other pre-processors like ANSA, Hypermesh, Solidworks & etc, but if you want to be competitive in the aerospace and automotive domain (I worked on both as well), these 2 will keep you going for a long time. Just train your stress team members to become perfectly proficient with these 2 and you'll have a good competition. FEMAP has also been used recently, but I don't
See it better than MSC Apex in any way. MSC Apex to FEMAP is like Hypermesh to Patran 2005..

Spaceship!!
Aerospace Engineer, M.Sc. / Aircraft Stress Engineer

 

[VN1981]

I have used MSC Apex trial version and although in terms of preprocessing, its a huge improvement over Patran, I feel it has a long way to go. I also realize it is very much in active development.

I find it that it is not very easy to refine the mesh the way you want it yet in Apex. I wish there was more control on how surfaces/geometry are split (I still don't know how to draw straight lines on surfaces...if I press ctrl or shift during drawing line command, the line doesn't get orthogonal like in Word and other CAD/FEA programs). But I agree with you that in next few years, as the program matures, it will perhaps become the defacto pre-processor.

Thanks aerostress82 for your input.

Edit: My time frame is 6 months to a year. So, would still like inputs about most used programs in Aerospace & automotive industries.

Thanks

 

[Mustaine3]

> Soon MSC Apex will take over all other pre-processors.
>I myself am proficient in 6 pre-processors, but due to the fact that MSC Patran and MSC Apex will be leading the market in the next 10 years with the new capabilities added up [...]

Yeah, dream on...

 

[aerostress82]

Mustaine3
From your posts it looks like you are an Abaqus guy. So, you are probably living in a different domain with advanced nonlinear problems-which is not exact aim of Nastran in the first place. So you might solve things with Abaqus for advanced problems, but when you have to deal with linear or crash domains, you'll most probably end in those software I mentioned.

VN1981,
In addition to MSC Apex and Patran, I would like to add Hypermesh, ANSA & Solidworks. But especially for aerospace and automotive domain, you may remove Solidworks as it is more efficient where design+production goes hand in hand (space industry). Of course you know FEMAP as well.

I've used Ansys for only 3 months but have been in the indistry for around 9 years now, so I would advise you not to get involved with Ansys too much as my feeling is that whichever company uses Ansys, they beasically do their Ansys work themselves-not outsource it. Whatever they would be outsourcing as part of an Ansys project, they could get it via Hypermesh FEA models and convert them to Ansys.

I've given quotes for some projects in some of my positions too, so I had to make extensive research on the capabilities of all above software and used 6 of them (each between 1-7 years) in projects multiple times.

I don't think there could be any additions to these for automotive and aerospace at all for pre-processors..

Spaceship!!
Aerospace Engineer, M.Sc. / Aircraft Stress Engineer

 

[GregLocock]

So far as I know we are 100% Hypermesh at an automotive OEM. I used to think it was horrible until I tried Patran.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Burner2k]

Patran is horrible as a pre-processor. I have heard from friends that Hypermesh is a very powerful pre-processing unit although many hate the UI.

 

[aerostress82]

You guys just don't realize the power of Patran. Things I couldn't do with Hypermesh, I could do all with Patran. Even checked it with Altair, the answer I got was that Hypermesh wasn't designed for automation tasks. But yeah, it is good with "trimmed" and composite trimmed surface handling.

Greglocock, if you are working in automotive domain, I would advise you all guys to switch to ANSA. Have used Patran for 7 years, Hypermesh for 6 years. Ansa for 1 year. ANSA is by far the best pre-processor until Apex will take over. Things I suffered in Hypermesh were in 2 clicks in ANSA. I've become to be one of the fastest meshers with Hypermesh, but ANSA is a whole different animal. Just give it a try for 6 months, and you won't want to leave it if you are aboe to work in black and white screens

Spaceship!!
Aerospace Engineer, M.Sc. / Aircraft Stress Engineer

 

[VN1981]

Thanks folks. It looks like I better start exploring options on how to familiarize myself with Hypermesh program.

Abaqus as a program has intrigued me. Back when I had a temp gig with a university, I had to play around with Abaqus Student Edition, cause we could end up using that program for an upcoming project. The project never arrived and I did not follow up beyond basic usage of Abaqus. Abaqus is the only FEA program which contains a CAD like sketcher part (AFAIK), which kinda makes it easier to create geometry easily. Of course, the workflow is to have CAD part imported in to FEA program and take it forward.

I know Nastran was never intended for major non-linear applications and MSC Marc is the preferred program, but I am curious to find out about the capabilities of Abaqus, especially in the non-linear domain.

Any open source material which captures the above?

Aerostress82, I think increase preference of Hypermesh over Patran can also be due to cost reasons. For general pre-processing, especially meshing & geometry clean up, Hypermesh seems a lot more capable than Patran. We still use Patran to define our Material, loading & other cards (mainly because it we like to post-process in Patran although I hear good words about Hyperview as well).

 

[ThomasH]

Hi

aerostress82

I can't help but wonder what you base your claims regarding Patrans superiority on. A few years ago a colleague and myself compared Patran and Femap. The reason was that we (the company) have both and somebody wanted to kick out Femap, primarily because he sold Patran. "Patran is obviously superior since it is more expensive".

So we tested and the result was that they were fairly comparable. There were things that you could do in Patran that was easier in Femap and vise-versa. But we did not find anything that was impossible. Now my colleague had 30+ years of Patran use (he is now retired) and I have 20+ years of Femap use. So I would say we are fairly familiar with the interfaces.

Since the question was aerospace and automotive and as far as I know Patran is big in that domain it is probably good advise the learn Patran. But that does not necessarily mean that it is "the best" from a technical standpoint.

Thomas

 

[aerostress82]

Thomas,
Patran is very stable and structured. You can't believe the level of automation you can manage with Patran from pre-processing to post processing. I have signed off projects where 5-10 people would actually need to work on - whereas I completed on my own or with another stress guy so many times!!! The deal with Patran isn't meshing, it is programming and automating. Now that Apex will be meshing very soon, that's why I'm saying the above. You cannot say you know Patran without knowing how to automate it or even program it. A good stress guy who knows coding can do a lot in Patran compared to the othet team of 4-5 stress guys. And especially, once you do this over and over and start customizing & programming Patran utilities, you are looking at a software which you can develop however you want!! I have worked with so many 20-30 years experienced stress chiefs and leads who were amazed to see the capabilities and automation we could integrate into projects with Patran. I also showed them how Hypermesh and Patran completed each other at the time. But right now as Apex is around the corner, that will change is my argument.

You might think you have seen a lot but I just want to remind you that I've surprised so many senior stress people and boosted the team's speed at so many aspects at all times with the help of Patran. So keep an open mind to what you can automate with Patran if you haven't done anything like this yet. And please do some research to see what this automation and customization really is.

VN1981,
Hyperview is neat but gets stuck with big *.op2 files so many times. I would advise Meta if you are dealing with very large output results. But Hyperview is neat looking too with filter and other functions in it. Once you spend a week figuring it out carefully, you could do a lot really. But remember to be innovative with it while playing around. I've seen guys using it not efficient at all. It is perfectly efficient actually if your *.op2 sizes are good enough.

Spaceship!!
Aerospace Engineer, M.Sc. / Aircraft Stress Engineer

 

[ThomasH]

aerostress82

So you are into programming and that is the big thing? My colleague worked a lot with programming and automation and it can be a wonderful tool. I have myself build a toolbox for my specific needs in Femap. I have actually seen a guy build an entire module building from a few parameters in an Excel sheet and several lines of code behind it, in Femap.

I did not claim to know Patran, my collegue did. But I do know Femap fairly well. You can program Femap also, there is no difference. There may be some difference in the details if you have very specific needs.

The problem with my work (I work with structures like buildings) is that each one is usually unique. So fully automated modelling does not really make sense. But there are things that can be automated.

Interesting discussion but it did not tell me anything I didn't already know.

Thomas

 

[ThomasH]

I can't say for sure but isn't most of the "modern" pre/post processors possible to costumize and program? In Femap I know of three ways to automate tasks. You can't do everything three different ways but so far I have not seen anything that proved to be impossible in all of the options. They all have pros and cons.

I like it when people believe in "their" software, but when one software should be considered more powerful that another you have to know both i detail. That was why we did our comparison the way we did. My colleague knew Patran and I know Femap. He had always considered Femap to be more limited but he revised his opimion after our test. I would not say that we were into every detail but for our needs it was enough.

I have over the years more than once heard somebode clam that one software i better than another. In my experience the conclusion is often based on that the user knows one software better than another. So, the software the user knows better is better.

Edit: I could make a small addition. Recently a colleague tried to import solid geometries into Patran, and he failed miserably. When I tried the same files in Femap it worked with no problems. After some adjustments Partan also made it but I think the key is not what software you use. It' is to know how to use it.

Regards

Thomas

 

[aerostress82]

I completely agree with all you say. Knowing a software at full capacity is what makes it more advantegous. My argument was regarding the Aerospace domain-and that Patran is customizable. But any new software is. But companies tend to use the robust solver+pre/post-processor. This was again another point regarding Patran.

Thanks for the discussion, you've become to be the most knowledgeable stress person for programming I've come to know so far. Sorry if I sounded a little cocky, I was trying to emphasize what Patran could do, but turns out you already knew

Spaceship!!
Aerospace Engineer, M.Sc. / Aircraft Stress Engineer

 

[aerostress82]

Reading your post again about not knowing everything in detail, I just want to add the below workflow to point out what is possible in Patran at full automation level. 2 different tasks:

First task
I have compiled some built-in custom utility functions in Patran that were added to a new Utility Drop-Down Menu in Patran UI. This came in handy at so many levels and aerospace companies are also having these compiled at all times when they want a time-taking process automated. This is also available in some other new software like FEMAP & etc. But not sure if the below "Second task" would be available in FEMAP. I have tried it in Hypermesh, and turned out it is not possible (which I had mentioned above in a previous post).
Second task
[ol 1]
[li]The workflow of this second task was as follows:[/li]
[li]First I meshed a whole assembly in Hypermesh to get the mesh out fastener in 30-40% of the time I would spend in Patran,[/li]
[li]After having the mesh I automated the below process in order for 20 Patran databases[/li]
[li]Delete the coarse grid FEM representing the new detailed FEM[/li]
[li]Import detailed FEM into the coarse grid FEM[/li]
[li]Perform mesh transitioning between coarse grid FEM & detailed FEM[/li]
[li]Update properties/loadcases/boundary conditions for loadcase matrices of 15-40 loads/boundary condition cards[/li]
[li]Equivalence nodes[/li]
[li]Run Nastran analysis[/li]
[li]Post-process results via Patran reporting/snapshots/*.cvs file extractions[/li]
[li]Perform SOL105 analysis on the DFEM local area of the whole FEM via Patran Pre-processing of the *.cvs files above[/li]
[li]Perform loadcase eliminations on a domain of 10000 loadcases[/li]
[li]Perform freebody loadcase extractions from a domain of 1000 cross-sections and 500 loadcases for further hand calculations.[/li]
[/ol]



I want to add a third one which will probably indicate the automation level at full:
On Excel sheet, input geometry/load/bc parameters/material/mesh size-type/"design goal" parameters on a couple of cells
Click "Run" button (which I added to activate a macro as usual),
Pre-processing in Patran>Nastran analysis>Post-processing in Patran>checking the results from Patran and interpolating the inputs to converge to the "design goal">Re-do the Patran pre-processing>Nastran>Patran post-processing steps again until the solution converges to the design goal. (which is also a SOL200 thing - but in my case the design goals were fancier)


Hope this makes the automation & programming more clear regarding Patran/Nastran together for aerospace tasks.
I used to use Hypermesh+Patran/Nastran previously when I had to deal with Detailed FEM instead of Coarse Grid FEM, but it looks like Apex has a strong possibility to take over Hypermesh's spot where Hypermesh+Patran combinations were used (a lot of big aerospace companies do this as I've seen & heard - please correct me if you have any other information regarding those big aerospace companies' software combinations choice)

Thanks All

Spaceship!!
Aerospace Engineer, M.Sc. / Aircraft Stress Engineer

 

[ThomasH]

aerostress82

Like I said, I don't really know Patran or Aerospace but I have programmed Femap. Actually, I have built a toolbox for my own needs. It's nothing fancy but it has developed over something like 10-15 years. Some functions I have suggested to Siemens and they have implemented in Femap, for others I'm still waiting of haven't suggested them at all. Some of them is just because I found something to be boring so I programmed it instead.

A colleague of mine used to work (as consultant)for a company that built living quarters on offshore platforms. They were built from modules and he had an Excel sheet where he defined "everything". He started with defining a base layout, number of floors and the loading, all in Excel and that he automatically modelled it in Femap. Then he did the analysis with Nastran and after that he imported the forces from Femap to Excel to do the code based analysis for US codes och Eurocode or whatever.

And I think the same could also be done in Patran, but not by the same guy .

My own programming has, so far, not been so much in automated modelling and meshing. I work primarily with structures like buildings and the are usually so "unique" that it is not meaningful to automate modelling on a large scale. The tools already available in Femap usually meet my needs. And when they don't I try to do sometning about it. But I don't switch software. That, in my experience, only gives me other headaches.

As for MSC Apex I can't have an opinion. Only that with 20+ years of experience I have heard it before, this or that will "take over". It has not happended yet and I don't think it will, ever.

Thomas