Explicit Analysis vs Transient Structural Analysis

[EngineerMickeyMouse]

Hi All,

ANSYS documentation gives a recommendation that for the events duration below 1sec (blasts, shocks, etc.), Explicit Analysis is recommended, while for longer times, Transient Structural Analysis is appropriate.
May I ask you, why is it so, and how do you choose whether or not which of those two analysis to use as per your experience? I would appreciate to get more insights since this 1sec red line is not satisfactorily explaining the reasons behind.

Thanks!

 

[JGard1985]

EngineerMickeyMouse

Nice user name!

The biggest difference between transient structural analysis and explicit analysis is the ratio of load duration to your structures natural period.

In the case of a blast load or drops or shocks, your load duration is so much shorter than your structures natural frequency. Only explicit can really capture this in the time domain. In these cases the structure doesn't have time to fully deform because the load is there and then gone. Implicit (transient structural maybe able to capture this but it really needs to break the problem down into very, very, small timesteps. Implicit solvers are typically used for transient structural analysis and use P=Kx and invert stiffness matrix to solve.

In the case of longer duration loads (two or three times the natural period of your structure or longer) your structure will have time to deform according to the load (near quasi-static) and an implicit solver for a transient structural analysis will easily capture this.

In short transient structural analysis is typically a long duration problem that the solving engine "cheats" and breaks up into a series of steady-state implicit problems. It then puts them together to get displacement over time


Hope this helps
Jeff

Jeff
Pipe Stress Analysis Engineer

http://www.xceed-eng.com

 

[EngineerMickeyMouse]

Thanks Jeff. That is a great response!
However, are you sure you have not missed inertia effects to be included in equation P=Kx which transient solver uses?

 

[Dave442]

you are asking what is the difference between implicit and explicit solvers

this question comes up a lot. search the forum:

http://www.eng-tips.com/viewthread.cfm?qid=415684

http://www.eng-tips.com/viewthread.cfm?qid=305063

http://www.eng-tips.com/viewthread.cfm?qid=404576

lots of resources online too if you google implicit/explicit fea or check out the Ansys documentation.

 

[IceBreakerSours]

Dave is correct.

A transient (i.e., time-dependent) analysis can be performed using either an implicit or an explicit numerical scheme. Implicit schemes also solve a system of equations and also commonly referred to as "solvers". In principle, either scheme should give you the exact same solution. Practical considerations end up tipping the balance in favor of one or the other scheme. Either scheme is just a numerical tool to solve for the governing equations of motion (or, any other physics) by discretizing space and time.

Note that I did not mention finite element analysis (or any particular code).

By the way, I would not choose ANSYS if I were running explicit jobs.

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

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

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

 

[EngineerMickeyMouse]

Thanks IceBreakerSours, could you please let me know the reason why you would not have use Ansys for explicit jobs and what software would you use instead?

 

[IceBreakerSours]

I end up having to open up the hood rather often, and sometimes, write my own code or scripts so take these comments with a grain of salt. In other words, none of what is written below is supposed to suggest ANSYS' explicit offering will not solve your needs; it may or may not be a good solution for you.

Let me take the conservative approach and assume you want to simulate some of the more "challenging" problems (mathematically stiff, severely nonlinear, coupled multiphysics, etc.) in small time-frames with some technical support, documentation, a modern graphical user interface, etc.

Brief take-home message:

ANSYS wants the non-FE expert to become an FE user. That is where they see growth. A lot of their product offering makes sense in this context to me. They put a LOT of effort in to making things easy to get set-up and running. I do not need such help. They take too long to deliver what I want, which happen to be advanced capabilities.

If you have more time, here are my sentiments:

1. ANSYS offers add-on products for explicit analyses but ANSYS' bread and butter is implicit. This suggests their investment decisions and support expertise will be relatively tilted in favor of implicit capabilities.

2. ANSYS' focus seems to be growing the market of non-FE expert users, which explains their emphasis on ease of use and some product offerings (e.g., AIM) to appeal to CAD designers.

The technical jargon used by ANSYS is odd to me. Your statement about explicit analysis and transient structural analysis is a case in point. To me, this distinction is ludicrous. To me, deliberate choices in avoiding precise linear algebra terminology by ANSYS suggests that they are focused on the non-FE expert. Another point: ANSYS lets a user choose a unit system. I do not understand why a developer would not put this responsibility on the computational analyst unless, of course, they are trying to appeal to non-FE experts.

Taken together, I feel like if I had to have them develop a new element, a new solution scheme, etc., it is likely they will either be unable to implement it or I will have to jump through many hoops or it will take them years to put it in their development pipeline. Also, if I wanted technical support, it is likely they will end up having me go to the company that develops the add-on product.

I recommend discussing these with ANSYS sales/tech support folks to give them a fair shot at earning your business, if you are interested in using ANSYS.

3. Look for jobs on famous job sites for explicit FEA and see which code keeps popping up more often as a required skill. I'd recommend doing this this exercise for your particular industry because it is likely the software vendor may have developed solutions targeting your industry and has gained sufficient know-how to help you.

4. When I asked ANSYS for an explicit solution for a particularly "challenging" problem, they did not offer ANSYS' "own" product.

I left out technical jargon for a very good reason. As in life, it is (almost) impossible to do a truly fair comparison between products in a competitive marketplace, which is why you must know/define your requirements and budget first. But then, the tricky question is: Do you even know how to define your ideal solution? If you do not know the breadth of problems you want to solve, it might be a good idea to give a tool a test drive for a while (ideally, a super cheap or a free one).

If you think of yourself as a beginner, then user interface, documentation, and technical support are going to be very important in the first year (or five)! What you want in such a case is a large code developer with a support staff that can hold your hand a bit gently. You may not want a small code developer whose offering may be cheap and their staff is likely going to be technically superior but they may not have the bandwidth or the documentation that can help explain the inner workings to you.

I would choose LS-DYNA. It is the cheapest and the best performing implicit (yes, implicit!) solver I have encountered and their explicit capabilities are very well documented. Their developers and staff are good. The quality of their conference publications is more impressive to me than quite a few academic journals. They work with you - if a capability does not exist and you make a case, they will implement it as soon as is humanly possible. All of this is possible due to their business model and the vision set by the founder. Competitors also have equally capable folks working for them but they are typically hard for an end-user to reach; you end up having to go through layers of support staff.

The downside is LS-DYNA is not really geared towards beginners. A beginner will find their user interface overwhelming. In fact, a beginner might get themselves in to trouble because they may end up tweaking a knob they shouldn't have touched in the first place. Their documentation isn't exactly gentle. [ABAQUS sets the gold standard on documentation.] Their technical support folks aren't exactly going to hold your hand for too long.

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

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

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

 

[EngineerMickeyMouse]

I appreciate such an extended presentation point of view of yours. Generally speaking, nowadays ANSYS etc. has became "just" products which are completely differently recognized by their sellers and users. Not mentioning authors, owners and other top managements at each of the ladder's step. That must not be surprising. This one string is pulled by the large amount of hands in the opposite directions. As LS-Dyna being company of different scale some of troubles may not be valid for LS-Dyna at this moment in its product history. Similarly as it was Algor being bought out by Autodesk, now it is going to be discontinued at a very low level of user's satisfaction. These are "standards" in today's world. That is why I am impressed noting how you have properly taken your very good and fair position in this matter.

 

[Dave442]

Not sure how the licensing currently works, but an ANSYS structural/mechanical license used to give you access to Autodyn and Dyna.

http://www.ansys.com/products/structures/ansys-autodyn

http://www.ansys.com/products/structures/ansys-ls-dyna