FEA software ratings/comparison 14

[serpaswj]

My company is looking to purchase several seats of FEA software. We are currently researching what to buy. Does
anyone out there know a good place to look that might compare softwares side-by-side?

Thanks.

 

[DickButkus]

You have to be very careful about the "Checkbox" mentality. Many codes say they can do non-linear, for example, but few can do them well especially when combining simultaneous nonlinearities, such as non-linear materials combined with contact or large deformations. Also, be sure you are comparing apples to apples. The explicit solvers can be extremely powerful, but are only applicable to a small subset of problems.
The most important element in the analysis process is the engineer. A good engineer can make use of even a poor code whereas a poor engineer can take even a top end code such as ABAQUS give incorrect results.

 

[dmech]

I've used Pro/Mechanica, Cosmos/Works, ABAQUS, ANSYS, NASTRAN, LS-DYNA, SDRC I-DEAS, and even ALGOR. In my opinon it depends on what your companies needs are. The experience of the potential users and the price your willing to pay. Most of these companies will drop by and give you a demonstration and leave the leave the software for a test drive. I would highly recommend that you contact several high-end and mid-level companies and try them out. As for a web site check out

http://www.deskeng.com,

which has lots of stuff regarding CAD/CAE/CAM/FEA/CFD, you name it. In particular this link compares most of the popular products:

http://www.deskeng.com/articles/99/Dec/rg_fea/fea.pdf

 

[DickButkus]

I took a look at the posted chart, but it doesn't have ABAQUS listed. It will handle all the procedures listed.

I think that the strengths of the major FEA providers are:

ABAQUS - Robust, general, non-linear, easy to use
ANSYS - Multiphysics (electro-magnetics)
MSC Marc - non-linear, easy to use
Nastran - quick linear dynamics, common CAD integration

For explicit solvers:
ABAQUS - Robust, general, precise
Dyna - Fast, Industry standard

I think ABAQUS and ANSYS cost about the same, Nastran tends to be higher. If you are a new user, most companies will be willing to give you a deal on your first year. It never hurts to ask!

If you know others - feel free to jump in!

 

[oharag]

dmech, what do you mean by "...and even ALGOR." This sounds like Algor is an afterthought. We currently run Algor at our company after evaluating CosmosWorks (coupled with multiple Cosmos products), Nastran 4D (coupled with Visual Nastran) and Algor. The reason for evaluating these solutions was driven by our need to analyze kinematics coupled with dynamic vibration, and that we chose Solidworks as our CAD solution. We wanted tight integration with our CAD, as well as, with multiple modules within the FEA environment.

We were looking for an FEA application that would analysis kinematic motion (in our case centrifugal forces) while also supplying us with dynamic vibration results. The goal was to analyze strength of design, as well as, reduction of vibrations being transmitted to the outside world (we make medical equipment). Other benefits, but not necessary at this time, were rubber, fluid, impact and thermal analysis capabilities. No solution mentioned above besides Algor gave us this capabilities in a economical complete package. With the Cosmos solution we had to purchase multiple products (CosmosWorks for linear, Cosmosmotion for kinematics, DesignStar for dynamic vibrations, and Cosmos/M for nonlinear). The total cost for this solution would be astronomical! The Nastran solution again required multiple products (Natran 4D coupled with VisualNastran). Again, Nastran is expensive when you have to purchase kinematics, vibration and nonlinear. The problem with both these solutions is that they are not tightly integrated with one another as I expected. The Cosmos solution was sweet in that it ran internal to Solidworks on our models, yet you still had to transition between each the models. The Nastran solution I believe was not as tightly integrated as I had hoped (this may change with MSC purchase of Adams). Both of these solutions could not provide us with rubber analysis nor impact analysis. MSC does have Dytran (impact) and Marc (for rubber) analysis capabilities, but no one can touch these products without significant capital outlays. Another note is that the kinematic solutions mentioned above are rigid body analysis only. They do not take into account flexible bodies as Algor does. Algor can also perform rigid body type analysis for quicker solution times if needed. For our budget (and with a little negotiation) we got Algor's MES/NL package coupled with Fluid Flow (Turbulent), Transient Thermal, and Inertial Load transfer! Their solution ties directly to Solidworks (not as nice as Cosmos). I can perform any number of analysis studies all within the same interface. Some analysis results can be shared with one another (ie. transient thermal analysis with an MES study or fluid study). Is it as nice as some of the other solutions out there? Well, there's a package called CosmosFloworks (a repackage of FloWorks) that runs within Solidworks that is sweet! Take your model, setup boundary conditions, attach an intake flow and boom the software does the rest. In Algor a little more work needs to be performed in order to solve a thermal/fluid analysis. But, again buying multiple packages is costly (ie. initial purchase coupled with ongoing training/maintenance costs). The only other solution that I believe I kind of know about is Ansys. Their Multiphysics package is probably awesome. It allows the study of kinematic (with impact) studies, as well as many other analysis types. How well is it tightly integrated with each analysis types? Also, does it tie directly to CAD solutions, or is it an import of a native file format? I believe it's an import. Also, I trained on Ansys 5.7 and the interface was old style with DOS like pull downs. Has this changed? Algor's InCad software is sweet. In fact it kind of looks like CosmosWorks, which is sweet as well. The only negative I see with any FEA solution trying to solve kinematic/dynamic problems is the LONG solution times. This is due to the number of elements needed for accurate solutions, as well as the number of equations to solve. The rigid body solutions (CosmosMotion and Nastran 4D) are very nice in that you can setup kinematic runs quickly, and results seem to be quick as well. Whether they are accurate is another question to be answered.

I called a consultant that was using Algor to make money. He seemed independent and honest with his comments. We spoke about the rigid body applications, and he mentioned that they were nice, and if he had the money he'd buy one of them. He also described an instance where he had worked with a company that was using Ansys at the time. He was trying to evaluate an impulse/impact problem where a charge would be set off against soil, and measurements would be received back to determine whether oil was beneath the ground. He tried to solve it using Ansys, but was not getting the results he had expected. Algor stepped in and solved the problem with little effort. This is part the reason why he uses Algor today (it could also be that you get a lot for minimal capital outlays). Some negatives he mentioned about Algor are that it does not do EMS studies only electrostatics (Ansys is better at this). Also, Algor's fluid flow cannot analyze compressible fluids (above Mach 1.0 - which I believe is supersonic). Again, I believe Ansys can do this well. These two limitations will not impact our ability to analyze the parts we build. One more thing (keep in mind that this is an Algor salesperson talking to me) an Algor salesperson said they just won out an FEA study against Abaqus with one of their customers. Just some additional input I thought was relevant.

To sum it up please do not overlook any FEA solution. I initially had my doubts about Algor. I remember their early interface being terrible (ie DOS like). I also did not realize at the time that they could perform kinematic analysis. An Algor salesperson called me up, and said they could do all of our analysis requirements. Again, for the money and it's capabilities I believe nothing can beat Algor!

PS I do not work for Algor. I'm just impressed with their capabilities, their cost, and their tight integration with our CAD package.

PSS Solidworks RULES as well! If you want another discussion on CAD selections I'll be more passionate with SW than I am with Algor.

Jerry

 

[sancat]

oharag,

Thanks for sharing your thoughts.
How do you work with algor? You do use SD modeling or instead model on SW and use incad to import to algor?
I had come experience with Algor13, and modeling was slow.

I agree with you that there is no reason why a moderatelly priced system will not have the same capabilities compared to others with higher price tags. They get their revenue from a higher sales volume (or so I think, and hope)

sancat

 

[brad]

oharag--
Some comments to your post and dmech's post. I do not intend to start a "best-software" argument, as this is like arguing religion. However, my take on dmech's statement was not a slight on ALGOR, but rather an acknowledgement of him having considered them. ALGOR is often not a consideration within the markets that often use the other codes to which he refers. However, ALGOR (as you have suggested) can be viable for people, and I took his comment to mean that one may want to consider this lower-priced alternative.

Regarding your statements in general, I would add that ABAQUS as a package (basic, as delivered, with no add-ons) can do all that you have suggested, including the kinematics. You seemed to suggest that ALGOR had a distinct technology over the others; I would say ALGOR's advantage is one of price and marketing. They are clearly a good fit for you, but don't imply from your experiences that they would fit for all (as no software does that).

Regards,
Brad

 

[DickButkus]

I would like to state again, watch out for the checkbox mentality. Although a software package says it can do a type of analysis, that doesn't mean it can do them well. I have met many an unhappy user that were surprised when the cheap FEA software they picked could not solve their real production problem. How much does the sofware cost when the engineer has to spend 10 times longer to get the problem to run?

You want to evaluate software based on your specific problem.

Best regards and good luck.

 

[timbo1]

Has anyone had much experience with UG/Scenarios ? Our organisation currently uses NASTRAN, which I am happy with. However, we are considering purchasing UG/Scenarios due to the benefits of integration with the Unigraphics CAD package we use.

Can anyone offer any pros / cons for the package ? In general, our FE applications are not complex, but can cover both static and dynamic analysis.may be static or dynamic.

 

[corus]

timbo's comments about integrating FEA software with CAD packages are relevant in some industries but I've found it to be useful in only a small number of cases and not really a determining factor in choosing a FEA code. I would look at what the programs can't do when they claim they can. All programs are full of bugs and reality tends to be a lot different from the brochure's claims. This can only really be done by listening to other people's experiences about using the software and how good was the software support, how good were they at rectifying bugs, how easy was the program to use, etc. This forum would be a good place to lent vent, but do the libel laws apply here? I use Abaqus and would be pleased to accept a small fee for praising it's virtues, incidentally.

 

[gillesD]

Hi Timbo1 and Corus,

UG/Structures does static and normal modes. It is very tied to the CAD (UG for instance) and gets its own solver (cheap and fast) but also Nastran, Ansys or Abaqus. I saw some very good integration at Delphi: optimization and sysnoise calculations. I would consider it seriously and in addition, with I-deas NX, you will avoid the painful file translations. Give it a try.

Gilles

 

[piedycat]

If you use fortunate to have access to Linux then you can get an Abaqus compatible FEA product for free called Calculix from

http://www.dhondt.de/

It gets good reports.

As for Algor: don't even think about it unless you already have no hair! It's cost of owership is not as low as you may think when you cosider the poor documentation and lack of tutorials. IMHO look for a product where you apply the boundary conditions directly onto the geometry & not the mesh.
I was incorrectly sold Algor, it proved to be incapable for my application. They have dire customer services. They boast 20000+ users worldwide yet the Algor vetted usergroup has only 565 members (some of which are very helpful & knowlegeable)

 

[Berserk]

serpaswj (Industrial) Nov 25, 2002
What FEA software did your company buy?