Need assistance finding the right FEA/CAE package 2

[TehMightyEngineer]

I'm sure this has been asked before but I've yet to come up with a truly accurate answer to my question(s) so I'll just ask here.

The company I work for does an extremely wide variety of engineering consulting. The relevant ones are things like pressure vessel design (ASME), tank and silo design (API and similar), equipment framework (steel), complex steel structural analysis (bent/damaged beams, odd custom shapes, complex designs requiring contact elements, etc.), below-the-hook lifting devices, cranes and monorails, shafts and drives, pressure piping, steel structures, pipe supports, and on and on. The ability to do reinforced concrete design would be a big plus but is not necessary.

Currently we're scrapping by with hand calculations and basic linear elastic FEA analysis using Staad.Pro but this is clearly not efficient on our end and is limiting the work we can do. We're at the point where we can't avoid buying a true FEA package capable of non-linear analysis (both geometric and material), post-buckling analysis, fatigue analysis, residual stresses, mode shapes, contact elements, and so on. As we're a small company our current budget is around $14,000.

We started out looking at CAD based FEA software such as Solidworks combined with Solidworks Simulaton Premium. However, after I've dug into this a little more this might not be the best way to go. Solidworks still seems to be the best bet for CAD but it appears that Solidworks can act as a pre/post processor for most of the major FEA players out there. However, just from past experience it seems that using the same software developer for both the FEA package and the CAD side of the program would make the most sense. It does not appear that Solidworks Simulation is the best FEA package that is in our budget when considering our applications.

Thus, my question what is the best FEA software that meets our budget and also provides the functionality we need?

Solidworks + Abacus is too expensive and Solidworks + Algor appeared to not play as nicely together as one would like. Solidworks + Nastran seemed good but it's really hard to tell. And, yes, once we get closer were going to have all the options demoed and select the most appropriate. This is just so we can help narrow down which FEA packages we should consider demoing.

In addition, we're looking to improve our ability to create complex meshes with ease. Solidworks appears to be the best solution but is there a better one? Is a CAD program even recommended for our applications?

Any pointers in the right direction would be much appreciated.

Maine Professional and Structural Engineer.
(Just passed the 16-hour SE exam, woohoo!)

 

[Naffoussi]

Hello,

Femap + NX Nastran is for me the best combination to do analysis you mentionned.

Why Femap ??

http://www.plm.automation.siemens.com/en_us/products/velocity/femap/why_femap/index.shtml

Some case studies:

Link

Seif Eddine Naffoussi, Stress Engineer

http://www.Innovamech.com

33650 Martillac û France

 

[TehMightyEngineer]

Hmmmm, haven't heard of Femap before. I'll definitely check it out.

Maine Professional and Structural Engineer.
(Just passed the 16-hour SE exam, woohoo!)

 

[Brad805]

We too have been looking at this and have some of the same constraints budget wise. I think your budget precludes many of the NLFEA packages mentioned. We have quotes for basic NLFEA options (ABAQUS/ANSYS/Diana) and they all exceed your budget by one magnitude plus they have a $10 - $15k annual servicing cost. Our needs are a bit different analysis wise, but we already have a modeling package to start with. What you require is one or two steps beyond our needs. We are in Canada, so the packages may be cheaper in the US.

As the engineer we tend to be all about the analysis. I would take a step back and look at the overall picture. I suspect right now the workflow you use entails CAD drafting, dxf/dwg exports to the analysis package, and then hours modeling. What you are contemplating can vastly improve this situation if the model can be transferred directly into the analysis package, and then detailed from the final model. Given your budget I would be tempted to start with the modeling software and then plan to add in the future. If that is the case I would be looking at which is the best package for your workflow and others down the line if you have thoughts of any detailing. We started with Solidworks and have been using simulation for a while now. It has its uses and it has been quite an eye opener to figuring out the tricks to create a solid model for analysis. I understand Autodesk has a simulation package as well, so Inventor might be another consideration.

Personally, I would try to use one supplier for both modeling and analysis. I have been working between multiple packages for years and I think too many software companies lie when it comes to interoperability. This also helps to avoid future problems associated with mergers. Maybe it is different in the NLFEA field, but I am reluctant to believe so.

Another consideration I think is the practicality of this for day to day use. Many on this forum seem to be in the manufacturing field. The NLFEA tools make great sense for them. They can spend endless hours going over the smallest details because they are going to build the product for years. I suspect you do a lot of designs for components that are built once. We also work on a few of the things you mentioned. I doubt the NLFEA will be practical on a day to day basis for some of the one off designs you are faced with. The more I have looked into this the more I find the most common use in the structural field for this type of software is University research and manufacturing.

Also keep in mind the time and cost to implement this. There is a huge learning curve to modeling and NLFEA appears to be even worse once you dig into this. We have been working in 3D modeling for years, and adding SW by itself was a significant undertaking to learn. We underestimated this aspect.

 

[IceBreakerSours]

This topic has been covered extensively over the years in this forum. Please search the archives (or google and look for links on this site).

Are you new to this forum? If so, please read these FAQ:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083

 

[TehMightyEngineer]

Brad, exactly the kind of advice I was looking for. Much appreciated.

It does sound like we have very similar situations. My boss and I were also blown away at the annual servicing costs. Solidworks + Simulation Premium cost about 50% of the total cost for the original license. At that cost it's much more worth while for us to not buy the annual maintenance and just save our pennies and buy another license in two years. I just don't understand their pricing model. At $1-2 thousand we can easily just write that off but when you talk about the cost of hiring an intern for 6 months at a small company it just doesn't make sense.

As the engineer we tend to be all about the analysis. I would take a step back and look at the overall picture. I suspect right now the workflow you use entails CAD drafting, dxf/dwg exports to the analysis package, and then hours modeling. What you are contemplating can vastly improve this situation if the model can be transferred directly into the analysis package, and then detailed from the final model.

Our workflow right now exactly matches what you suspected. Model in 3D in autocad, generate a mesh, copy mesh to dxf file, import into Staad, spend MANY hours revising mesh and finalizing the model, spend MANY hour running and revising the model further for the limitations of staad very basic FEA capabilities, use results combined with hand verification to come up with a workable design. Very tedious and inefficient way to design, to say nothing about being limited to linear elastic FEA.

One advantage of solidworks is both the markets we work in often use it and it strikes me as the most "polished" CAD/CAE program for our needs. It also provides a great marketing tool as solidworks output models are very impressive for the non-technical clients.

Personally, I would try to use one supplier for both modeling and analysis. I have been working between multiple packages for years and I think too many software companies lie when it comes to interoperability. This also helps to avoid future problems associated with mergers. Maybe it is different in the NLFEA field, but I am reluctant to believe so.

I also agree on using one supplier for both pre/post processor and the actual analysis software. To me FEA is just way too feisty and unless everything plays together really nicely then it can definitely be frustrating. I also would have big concerns of an update from one supplier causing the other software to become incompatible or have other issues. I'm the companies IT guy as well and often run into this problem on other software. Thus, just keeping the pre/post processor and the analysis software under the same roof is a big plus for me.

Another consideration I think is the practicality of this for day to day use. Many on this forum seem to be in the manufacturing field. The NLFEA tools make great sense for them. They can spend endless hours going over the smallest details because they are going to build the product for years. I suspect you do a lot of designs for components that are built once. We also work on a few of the things you mentioned. I doubt the NLFEA will be practical on a day to day basis for some of the one off designs you are faced with. The more I have looked into this the more I find the most common use in the structural field for this type of software is University research and manufacturing.

You hit the nail on the head so hard with this one. This is exactly the problem I'm having finding information. Everything is geared towards the manufacturing field (and rightly so, it's the obvious client for NLFEA). Thus, it's really hard to find the right NLFEA package for someone who needs to be able to efficiently make a mesh, run it as linear elastic, revise it, then run it as non-linear, and then take that design and provide drawings/models or calculations/reports to the client. I don't need to revise a part 2,000 times just to tweak exactly the performance I need. I do need to check whether that dented crane beam can still function adequately because it costs $100,000/hour to shut down the crane.

And, yes, you are correct that NLFEA isn't our only consideration, obviously most of these designs will be linear-elastic only. However, the ability to offer NLFEA would really expand our companies potential in addition to being almost a requirement for some of our potential upcoming work.

Also keep in mind the time and cost to implement this. There is a huge learning curve to modeling and NLFEA appears to be even worse once you dig into this. We have been working in 3D modeling for years, and adding SW by itself was a significant undertaking to learn. We underestimated this aspect.

Agreed. However, we have a senior engineer (not me) who is very well versed in NLFEA analysis and being a child of the internet I'm very quick at picking up new software. But, yes, even with that we have budgeted a serious amount of time and lost productivity towards learning any new software packages.

Maine Professional and Structural Engineer.
(Just passed the 16-hour SE exam, woohoo!)

 

[TehMightyEngineer]

This topic has been covered extensively over the years in this forum. Please search the archives (or google and look for links on this site).

Oh, I'm sure it has. I stated as much in the first paragraph of my original post.

However, after about 10 hours of searching eng-tips.com, various FEA developer websites, FEA articles, case studies, and general google searches I was not any closer to finding an answer.

My issue is exactly what Brad805 above mentioned. We don't do manufacturing which is what 90% of NLFEA seems to be used for. Thus, almost every single topic, article, discussion, etc. I found wasn't applicable. It was honestly a little frustrating.

But, that said, you're entirely right. I'm sure exactly what I'm looking for has been discussed, I just can't find it through all the wealth of (non-relevant) information.

Maine Professional and Structural Engineer.
(Just passed the 16-hour SE exam, woohoo!)

 

[structSU10]

One thing we have dealt with is our managers tend to ask for too little fee for projects we have done using such packages. The amount of time spent modeling, tweaking, refining mesh, applying loads, etc. is so great, and the cost of the software is so large, normal rates for normal structural stuff just doesn't apply well.

We currently have alibre (3D systems now) and Algor (now autodesk simulation). In the beginning of both programs, they worked well together, with a direct link. Now the new versions do not work well together, and are almost impossible to smoothly transfer from CAD to FEA program. Its a major headache. All these companies buy each other up, the cut the direct links, which is very tough to deal with.

 

[TehMightyEngineer]

StructSU10: I agree, my boss does seem to lowball some of the work we do right now with our somewhat limited FEA capabilities so I think we're going to make a push to charge more when we use these newer capabilities. Additionally, we're hoping that moving to a more modern CAD and FEA software such as solidworks will help us be more productive in the long run.

Good to know that my fear of software becoming less integrated is at least somewhat valid.

Maine Professional and Structural Engineer.
(Just passed the 16-hour SE exam, woohoo!)

 

[Brad805]

I was floored by the prices as well. The pricing arrangements and options seemed too confusing until I watched the video of the Abaqus stent analysis. After that I started to appreciate the scope of this software.

Your crane example is a perfect application for this type of design. He will pay heaps of money to figure out how to fix that problem. Have you looked for examples at Grabcad? There are many analysis models posted by guys all over the world on that site. The Imechanica forum has some structural information as well. I think the savings in modeling time will pay for itself by streamlining the process, but that is hard to track. The other benefit I see to this type of analysis is the illustrations you can provide. It is far easier to explain to a design to a client with a rendered image that does not have all the elements and assumptions that tend to confuse the non-engineer. I have done quite a few video captures and it gives the client a greater confidence. The question is how to monetize this in a consulting practice.

I found one supplier in the UK that leases Abaqus, but I have not had time to inquire if that is an option in NA. I tried that years ago when STAAD offered something similar, but I found it was not very practical. We are waiting for a client to decide about their future project before going further.

You are miles ahead of me if your boss has the NLFEA experience. I purchased the Practitioner's Guide to Finite Element Analysis of Concrete and have been working my way thru that to decide how practical it is for us. Right now it seems very daunting and fraught with considerations that seem to require a method to calibrate.

Jared on the solidworks forum is quite helpful if you have questions about the limitations of SW simulation.

It is interesting to hear struc's comments about the links. It would be interesting to know if he has hear how well Autodesk has setup their simulation? When companies buyout others there is a tendency to make the software work rather than re-programing to work make it work properly.

I feel struc's pain too. We added ETABs to link with our Tekla license this summer and Tekla was not too honest about their link for concrete parts not working flawlessly. That caused endless hours of grief on a project a few months ago. Another frustration is being in the middle of IT guys claiming it is the other guys problem.

Sometimes it is nice to simply talk to others. Nobody I know is interested in my gibber jabber, so that is why we come here.

 

[adfergusson]

If your budget really is only 14K the you're options are going to be very limited based on the products/pricing I'm aware of.

I'm also going to provide the caveat that my company is an Autodesk Clean Tech partner so we get free Autodesk software. We're not paid by them, don't resell their software, etc... but thought it worth disclosing. However, we did pay full whack for NEi Nastran before NEi became part of Autodesk, and we continue to use NEi/Autodesk Nastran.

One of the few things that might be in this price bracket, if you have Solidworks, is Nastran-In-CAD. I'm not sure about Solidworks simulation pricing; a while ago (a good few years now) I saw some pricing for this over the basic version and think it might not work for you, but this might have changed in order to counteract the pricing and capabilities of Nastran-in-CAD . I think Siemens have something similar in the form of a stripped down version of FEMAP and NX for SolidEdge, but its only for SolidEdge. I don't use the likes of Solidworks Simulation or Nastran-in-CAD but from what I know Nastran-in-CAD is a more advanced product as it basically uses the full fat NEi/Autodesk Nastran implicit solvers. We have this for Inventor (I assume its functionality is the same in Solidworks) and its a great package for what it is.I don't use in-CAD, but our main experimental testing engineer uses it for stress analysis when it comes to test fixtures, load cells, mechatronic/CNC devices, etc... and he thinks its great; in his words 'this is so much faster than Femap'.

However, don't buy Nastran-in-CAD (or similar) thinking it will provide you with the capabilities you'd get from something like stand-alone Autodesk Nastran with FEMAP; it won't. We also have licenses for Femap and Autodesk Nastran (a lot more expensive, assuming what we paid for NEi solvers is about the same now that they're part of Autodesk) as well as Abaqus (basically the most expensive FEA package around) and use them because we need some very specific functionality that they offer. I'd suggest putting together some scenarios/jobs for where you'd be looking at using such software and then benchmarking them in terms of the additional jobs/work they'd allow you to do and how much work it takes to do so for each package.

Brad805 has been kind enough to provide you a bunch of good info and he makes a lot of good points but with the caveat that everyone's workflow is different and in regards to:

Personally, I would try to use one supplier for both modelling and analysis. I have been working between multiple packages for years and I think too many software companies lie when it comes to interoperability. This also helps to avoid future problems associated with mergers. Maybe it is different in the NLFEA field, but I am reluctant to believe so.

There are certainly niggles when using products from different suppliers but sometimes the capabilities obtained by going to different suppliers are worth it. We bought NEi Nastran and Femap years ago having tested some other options, including Femap bundled with NX Nastran. We also use Abaqus but if I need to clean up some geometry imported from CAD for an Abaqus simulation then my first port of call would be Femap rather than Abaqus CAE (CAE is pre/post environment for Abaqus).


Anyway, you can download a free trial of Nastran-in-CAD so I think it would be worth a look.

 

[TehMightyEngineer]

Jared on the solidworks forum is quite helpful if you have questions about the limitations of SW simulation.

Thanks. We've arranged with solidworks to provide a demo of their software and discuss options with us so I should be able to have that side of my questions covered well enough. I'll definitely make a post there and browse those forums more if I have any very technical questions regarding solidworks and simulation.

If your budget really is only 14K the you're options are going to be very limited based on the products/pricing I'm aware of.

Completely seeing that now. However, for what we need I'm expecting this to be adequate.

I've of course read a lot about Nastran and will definitely look into Nastran-In-CAD some more. It does seem like one of the best fits.

There are certainly niggles when using products from different suppliers but sometimes the capabilities obtained by going to different suppliers are worth it. We bought NEi Nastran and Femap years ago having tested some other options, including Femap bundled with NX Nastran. We also use Abaqus but if I need to clean up some geometry imported from CAD for an Abaqus simulation then my first port of call would be Femap rather than Abaqus CAE (CAE is pre/post environment for Abaqus).

Agreed on the advantages of mixing and matching capabilities. That's one of the reasons I'm looking at Nastran as it does appear to work well enough with Solidworks and appears to offer much more features than Simulation. I take a look at the Nastran-in-CAD demo and we'll benchmark it against the Solidworks people when they provide their demonstration.

Thanks for the help.

Maine Professional and Structural Engineer.
(Just passed the 16-hour SE exam, woohoo!)

 

[IRstuff]

Not my field at all, but I work with a bunch of guys that do FEA, and while we're in a completely different industry, we also do mostly one-off designs, and we use the same tools as everyone else.

I would think that operationally, there is little difference between doing a so-called "production" design vs. a one-off design, insofar as you still need a solid model, you still need to mesh, and you still need to analyze, and you probably still need to iterate. Whatever niceties there might be for "production" design, they probably pale in comparison to the preponderance of functionality that's needed to do any design at all.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[Brad805]

Good luck. Be sure to post what you decide. Our decision into SW was a bit haphazard as it started with wanting to create decent models for certain projects. In retrospect I wish we would have contemplated what I might do with the software in the future and looked at some other products

 

[TehMightyEngineer]

IRStuff: Probably true, however I find that often the published comparisons and listed features of various CAD + FEA packages tend to focus on the manufacturing clients to the point that it's hard to judge how applicable their software is to someone not in the "production" field. But perhaps you're correct and certainly there will be design similarities. Where I suspect our methods will differ will be in we would value efficiency over accuracy and features. If we could have a cheap FEA software that allowed us to do basic non-linear analysis and check convergence but had great CAD and mesh editing tools with it then that would be much more valuable to us than a knock-your-socks-off high end FEA program that took us 4 months to learn and was clunky and hard to use.

Though in the end I guess my problem is this; you say you use the tools as "everyone else", but what are these tools? What do people use in my industry of industrial consulting and structural design? I'm having a hard time finding out even that.

Brad: Sure, I'll probably update this when I find out more. I'm sure someone else can find it or at the very least it can be added to the (probably large) pile of people asking questions on which FEA program is best.

We also came to the same consensus in our office to not pick a software package on a whim and effectively evaluate which ones are best. I was tasked with finding the options to investigate and demo as well as research the pros and cons of each and present it to my boss. Thus my dilemma of spending way too much non-billable time trying to find out which FEA programs to rule out and which to include in options for us to demo that meet our rough budget.

Maine Professional and Structural Engineer.
(Just passed the 16-hour SE exam, woohoo!)

 

[ThomasH]

Hi

I think you could fit Autodesk Nastran (NEiNastran) with the budget constraints. The problem is that you will probably not get any pre/post to work within the same budget.

And the discussion between once-off designs and production. I don't thing the difference is so very significant. In both cases you need a reliable model otherwise the analysis is useless. But, of course, a high degree of repetition allows you to improve the model. Since I work with structures (buildings etc) most of the models are unique. But that does not mean that you can't improve the modelling over time.

A suggestion, do an evaluation that includes testing the software for real projects. If your budget constraints means that you have to buy the wrong software ,perhaps you should discuss either the budget or the requirements.

Good luck

Thomas

 

[Brad805]

Thomas, I agree that from an analysis stand point there is no difference for a one off design vs. a production shop. The difference is the fee structure. If it costs $10k in fees for a one off design and the same for a production shop where I could end up making 10's/100's or more, there will be a difference in how a client reacts to the idea of this kind of design.

 

[TehMightyEngineer]

Thomas:

Yes, from discussing this here I believe that Nastran In-CAD to be the ideal alternative to Solidworks Simulation. It appears to offer more features and be a better FEA program at the downside of being slightly outside our current budget. It would also have the issue of not having a truly native CAD program unlike Solidworks CAD paired with Solidworks simulation. Though, being owned by Autodesk, if we went with Nastran it would make sense using an Autodesk CAD product and we currently use an older version of Autocad in our office.

As you said, it will come down to testing the products. If Solidworks Simulation meets our needs sufficiently then combined with the lower price it may make the most sense. If it does not meet our needs then I agree that finding a way to increase the budget is the better alternative to buying an expensive piece of software that doesn't do all we want it to do.

Brad:

I believe you're saying more or less what I was getting at. In our line of work our clients often value rapid quality design over precision and project cost savings. If we said we can cut the cost by 30% but it will take us a week it might not be worth it to our clients. They want rapid, accurate answers yesterday. This is why we need to switch to a non-linear, more advanced FEA program because we're spending too much time trying to make a limited linear-elastic analysis work for our projects.

Thus, our design and analysis is quite similar to production-style design, however our ultimate goal in why we market FEA cababilities to our client is vastly different. We're not trying to help them shave the most weight off a part but rather trying to help them fix an issue when they don't have time to do it right and need accurate results for complex problems right away.

Maine Professional and Structural Engineer.
(Just passed the 16-hour SE exam, woohoo!)

 

[Brad805]

I have an idea of you are trying to do from your crane example and we both know it will work incredibly well for both yourself and the client. The key problem I have had with the decision is determining if the investment will pay off. If you foresee weekly clients that will pay for this level of service, the decision is easy, but if this is for random jobs the decision is much harder to make. We are are in the random application category where these projects seem to come in groups, and then disappear for long periods of time. I really would like to expand our license, but I am having a hard time justifying it to my business partner.

 

[TehMightyEngineer]

Similar situation here Brad. We could buy this and then not use it for 6 months. That's why we've put buying this software until now, we just couldn't justify the expense. However, three things have happened that have pushed us toward this:

1) Many of our clients use solidworks and having the ability to trade solidworks files (or similar) back and forth would be a big advantage. In addition, we have the potential to get some fairly large clients this year and really need to provide that "wow" factor so they feel confident in our abilities. Our company has always tried to be at the forefront of engineering technology (we were doing full 3D CAD models similar to BIM back in the 90's) and we want to keep on doing this.

2) We've had many jobs where we made little profit due to the amount of time we had to spend working with a limited FEA program. Our primary engineer who does FEA analysis has been crying for a true FEA program for years now and with my relatively limited FEA experience I agree. I also expect that over the course of a few years I would be shocked if we didn't pay back the software in terms of manhours saved on completing projects.

3) The tipping point was we did a job involving a dented pressure vessel as a fit-for-service analysis. Due to some of the job constraints we came very close to not being able to complete the job due to our lack of non-linear FEA analysis. We conservatively concluded a repair was in order but we almost got backed into a corner with being able to justify this conclusion using linear-elastic analysis alone. This would have really hurt our company. Our boss decided that we have enough work of this nature that it would be a liability concern to continue without getting the capability to do some non-linear FEA analysis.

One thing that makes sense to me is, if you do get an expanded license, that you only offer one price. Don't offer a "cheaper" FEA analysis because it provides the illusion of a "better deal". Every time my boss offers one price and then says "but for this fee we can give you this full 3D cad model and all these other things" it gets turned down. Now we just simply say "with our fee you get all the bells and whistles we can offer". This seems to work much better and allowed us to utilize any advanced tools we have more often.

Maine Professional and Structural Engineer.
(Just passed the 16-hour SE exam, woohoo!)