Abaqus suitable for non PhD?

[cnuk]

I need to begin to venture into analysis of elastomers. I have only used (hate to admit it) Mechanica to date along with some experimentation using Ansys. Abaqus keeps coming up when I search for elastomer FEA. Problem is, I am not a PhD...only a B.SC ME. I have what I consider a good background in strength of materials which should be a starting point, but I've purged the advanced calculus shortly after leaving university. We are getting loads of DMTA data characterizing our elastomers so I think I have good material property data. Our "typical" problem is not large strain, includes some temperature loads and contact analysis with a steel part. The ultimate would be to predict some operating temperature that incorporates hysteretic heating, and time dependent pressure loads.

I know it's a general question with little in the way of specifics, but do you think Abaqus is an option for me? If not, any suggestions on a more suitable package?

Thank You

 

[SWComposites]

You don't need a PhD to run any FE code, including Abaqus. Howvever, you do definitely need to have a good knowledge of practical FE principles and techniques (understand boundary conditions, load application, element selection, meshing issues, etc. - none of which you need much calculus for). In your case you also need to have a good understanding of the material behaviour, and know how to model the behavior with the appropriate meterial model in the FE code (you should study up on elatomeric material response). Abaqus is a very good non-linear FE code, both for contact and non-linear material models. Abaqus has a variety of very good training classes, from basics thru very specialized topics (see their web site) and I strongly suggest attending one or more of the classes related to the nonlinear modelling that you will be doing. For instance, I know someone who took the Contact Modelling class, and the instructor stated that a lot of the tricks and techniques for sucessful contact modelling covered in the class are not documented anywhere (there are a large number of options related to contact modelling in Abaqus and it is not at all clear which to use when). I also suggest that you start with very simple models with the non-linear material and show good correlation with your test data before moving on to more complex models. Non-linear material behavior can also be tricky.

 

[gwolf]

I agree with SWComposites, you don't need a PhD to use an FE code but you definitely need a formal education in structural mechanics to about B.Sc level. More importantly, you need to have really understood what you were taught. Don't worry about the calculus, you probably won't need it in this application. In general I find that the more highly qualified people become, the less able they are to actually do anything!!

On your particular job, ABAQUS is definitely the tool for you. It has fantastic contact, medium-high quality elastomer models, elastomeric test data correlation facilities, hysteretic heating - the whole shooting match.

Although ABAQUS is well thought out and straightforward to use, what you are proposing is right at the top end of analysis complexity (for any FE package). You will need help and training to use do this properly. As a tool for the future I highly recomend it but don't expect miracles in 3 weeks, you will have to put in a significant effort to learn how to use it. I strongly agree with SWComposites about the 'tricks of the trade' issues in contact modelling.

 

[cnuk]

Thanks for your comments. I know it will be a lot of effort to figure it out but I'm ready for a new challenge in my job. In an unfortunate turn of events, it seems that our company already owns two seats of MSC Marc so I may be forced into using that rather than Abaqus. I am certain Marc is more than capable but from what little I know, it sure looks like a dated, cryptic program interface.

Thanks Again

 

[gwolf]

MARC: It is a dated, cryptic interface but the capabilities are quite good for rubbery things in contact. Many of HKS who wrote ABAQUS are ex-MARC employees. Good luck, it could have been a lot worse if you had been burdened with NASTRAN or ANSYS.

 

[ekline]

Your company may have the Mentat interface for Marc (which is dated and cryptic). You could propose purchasing a new pre/post-processor like Patran, which would be cheaper than getting a new FEA package, but would still be thousands of dollars.

 

[cnuk]

I think we do own Patran as well as Mentat. From what you're saying, Patran is a step up from Mentat. From the screenshots I can see I'd agree. Does it offer all the capabilities of Mentat though? If so, I wonder why Mentat still exists. I'll have to check into Patran some more and report back.

 

[ekline]

cnuk, I think Mentat still exists as a legacy product for people who are used to it. I haven't used Mentat, Marc, or Patran since I was in grad school 7 years ago, but I doubt MSC has spent any money on Mentat since aquiring it. Patran was a very capable mesher and had lots of options, and I thought it was pretty intuitive. I don't know if it's improved or changed any since I used it. MSC also purchased Patran since I last used it.

 

[cnuk]

Thanks for the information. I'm trying to get an answer from MSC as to what the differences are and why choose one over the other.