Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations waross on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Future Advancements in FEA 4

Status
Not open for further replies.

Scobie33

Mechanical
Jan 10, 2020
3
Hi there,

I've been reading quite a few papers on the Nurbs-Enhanced Finite Element Method (NEFEM) and I was just wondering what you guys would consider as its greatest advantages/disadvantages?
From what I can see, it appears to be far more accurate in the vast majority of cases. Also, why are no FEA packages (such as Ansys) using this?

Finally, what do you consider to be some of the upcoming advancements in FEA technology in the next 10/20years in terms of accuracy of solution and computing time?

Cheers, any insight appreciated :)
 
Replies continue below

Recommended for you

I think the best opportunity for advancement is in reducing/solving PEBKAC issues.

another day in paradise, or is paradise one day closer ?
 
Traditional "large" (software) companies like ANSYS/Dassault/.. try to avoid cost and risk which are always going to be a part of investing in anything fundamentally novel, which is why they prefer to grow by acquisitions.

Isogeometric analysis and phase-field modeling are a couple of things I am looking forward to playing with in LS-DYNA (now part of ANSYS). Both technologies hold a fair bit of promise. I am also interested in some ideas crossing over from Nitsche-based methods like Cut-FEM to the more traditional methods.


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

 
rb1957 said:
I think the best opportunity for advancement is in reducing/solving PEBKAC issues.

I had to look that up, but having done so, I agree.

How we go about it is a difficult question though. I doubt that it is high on the priority list of most software suppliers.

Doug Jenkins
Interactive Design Services
 
I'd prefer a mallet, or electro-shock, for every time people over-constrain models.

another day in paradise, or is paradise one day closer ?
 
replace FEA with random number generator, or free style artist (colouring the elements). We'd get something like what the Germans used in WW1.

another day in paradise, or is paradise one day closer ?
 
Amusingly, this subject never gets old! We had discussions about PEBKAC back in 1979, when MSINC was still a serious competitor to SPICE, although we didn't have a cute PEBKAC acronym.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
the only thing older than PEBKAC would be keyboards. Before keyboards the only people using computers were smart enough to figure things out for themselves.

I guess before PEBKAC it would've been PEBPAC (P for pen, pencil, or S for stylus or stick ?)

another day in paradise, or is paradise one day closer ?
 
Before keyboards the only people using computers were smart enough to figure things out for themselves.

I don't think that's strictly true in the way you meant. The word "computer" prior to mid-last century was used to refer to human beings whose sole job was to crank calculations, essentially as piece-part producers. So, the "people" using "computers" were probably smart enough, but also smart enough to hire someone to do the grunt work, but errors in correctly specifying the problem, initial conditions, and exact equations probably still existed.

As for NEFEM, it's unclear whether it offers that much improvement in accuracy, and whether the complexity of having to segregate boundary elements vs. bulk elements and the concomitant processing complexity is worth the effort, since the NEFEM appears to be closer to convention FEM than to exact solutions: We've had occasion to run FEM on deflections and the end results weren't close to reality, although they might have been good enough to compare relative merits of different design solutions. The net effect, I suspect, is that the extra accuracy of NEFEN doesn't really exist, in general, and the added complexity and processing time is not cost-beneficial, and you're still better off building and testing. I think that Boeing found that to be the case in the design of the 787.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
I am aware of "human calculators" … poor b@st@ards, usually women (sexist pigs we were), given a spreadsheet to fill in. I once worked in an office where there were still two remaining (even though there were real live (?) computers on every desk) … I guess the company didn't have the heart to lay them off.

I thought P-elements was supposed to be the next big thing (or has that already happened ?).

another day in paradise, or is paradise one day closer ?
 
I was actually referring to before even ENIAC, or even mechanical calculators. Star positions, navy almanacs, etc., had been published since the 17th century with human "computers." Now, those, were "men." ;-)

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
so a slide rule is a "computer" ? how about a nomogram ?

another day in paradise, or is paradise one day closer ?
 
Is it more energy efficient to use NEFEM instead of FEM for a given level of accuracy?
Would NURBS be implying a more coarse mesh, less sparse matrix structures, and thus requiring less computing power ?
 
I was looking forward to answers to the intended question - aaaaaanyyyyyy day, now.

Fair enough.

I know next to nothing about NEFEM, but an online paper on the subject from 2008:

states that:
"NEFEM is at least one order of magnitude more precise than the corresponding
isoparametric FE in every numerical example shown. This is the case for
both continuous and discontinuous Galerkin formulations. Moreover, for a
desired precision NEFEM is also more computational efficient, as shown in
the numerical examples."

Frankly I'm sceptical, but I would be interested in the comments of others.

I should probably read the full paper as well.

Doug Jenkins
Interactive Design Services
 
rotw said:
Is it more energy efficient to use NEFEM instead of FEM for a given level of accuracy?
Would NURBS be implying a more coarse mesh, less sparse matrix structures, and thus requiring less computing power?

Tom Hughes is an excellent communicator and an inventor of the isogeometric analysis. I recommend watching a lecture or two by him. He gave a 4-hr long one at Aachen; its up on YouTube. There are other shorter ones by him as well. I am no mathematical wizard so I can't judge how much of what he claims is useful and how much of it is of academic interest. However, Dyna developers invested in it heavily and have been steadily showing its value. Until I have it in my toolkit, its in the 'promising tech' space for me.

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

 
Status
Not open for further replies.

Part and Inventory Search

Sponsor