Is Algor the best FEA package for me?

[sbrad]

I've decided to take the leap and buy an FEA package for in-house analysis. We currently sub-contract and are regularly taken to the cleaners by our supplier. I've been looking at a few packages and Algor and NEiNastran stand out for me. NEiNastran gave a limited node demo version which is good for getting to grips with the software. Algor won't give me a demo , other than one of their on-line sessions. This was good, but obviously it's better to get time on your own with the software. Despite this, I am leaning towards Algor. I'm involved with the design of high temperature pressure vessel / heat exchanger internals. Our sub-contractor used Abaqus / Patran but this seems to be for hard core fea types. The 2 packages I've looked at seem to be learnable...although time will tell. Any advice out there?

Have I missed any packages? I've looked at Strand7 and Cosmos, but haven't looked at these in detail.

I am an experienced Mechanical engineer and have worked for close to 20 years in pressure systems type work. I've done a lot of axi-symmetric analysis using the FAST 1 package and some fundemental FEA...but this was only as part of a design by analysis course.

Advice would be much appreciated.


sbb

 

[GBor]

I'm a strong proponent of Algor with about 10 years of experience using it. I've also used NEiNastran off and on for the last few years. Both are very good packages for the price. They produce similar results. NEiNastran will tell you a great deal about how much more accurate they are than anything else on the market, but FEA is only about +/- 10% considering the inaccuracies of the modeling, the inputs, the boundary conditions, etc. Algor will generally work with you much more than they are apparently showing and they've been around since the early 80's. I think you will be happy with either package. I do think Algor is easier to learn, but that may be because of my long-term use of the product.

Since you are doing pressure vessels and heat exchangers, you may want to look for Paulin Research's software (

http://www.paulin.com).

They do a lot of work in the pressure vessel area and I think their software is probably a little less expensive than some of these general FEA packages. If Pat LaLonde is still working for Paulin, tell him "Garland" suggested them...he may still remember me.

If you want a demo of Algor, contact Ed Simmons at 800-482-5467 and tell him that you really need a 60-day license or you may have to purchase NENastran. Algor and Noran Engineering were looking at possibly merging some aspects of their software at one time. The deal went poorly and now they are BITTER rivals. You can generally get a violent reaction out of one by mentioning the other...not sure how vendictive you are...

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com

 

[sbrad]

Garland

Thanks very much for that. Sounds like I'm on the right track.

NEi Nastran is marketed in the uk by epsilon structural analysis and they are offering (what seems like) a very good deal. To be fair, Algor seem quite keen to do business and they are helpful, but epsilon are throwing in the 4 day training package free. Which means the L1-5 package + training for less than £10k. (approx $17k) I'm sure this may be negotiated down further.

I don't know about being vindictive....but a little healthy competition never hurt anyone...especially when you're paying!

After playing around a little with NEiNastran I could start to generate some very basic models fairly easily, but I did find it a little long winded in the way you have to go about it. This would obviously get better with practice, but perhaps the Algor package will be better for me. I may well press them for a proper demo before I decide.







Thanks again

Sbb

 

[GBor]

In the UK, you should probably talk to Julian Holt. Not sure exactly where in the UK he is located (old age is limiting my memory), perhaps Leicestershire? He is the brains behind Fatigue Wizard (

http://www.fatiguewizard.com)

and has been using Algor for 15 years or so. If he is the one with whom you are dealing, tell him I said that he needs to do some real work for a living!

Algor has a habit of tossing in some training in the US. Not sure about their international operation. If you are looking at L1-L5 of NENastran, you should definitely be interesting to most of the major FEA dealers...I would think they would be interested in helping you decide in their favor.

Cosmos (you mentioned earlier) tried to stay with Algor and NENastran and it is a very good package, but I don't think the support is too readily available, however, since you mention you are in the UK...Cosmos is owned by Dessault Systems (spelling may not be correct) located in France. The support may be better in your part of the world.

Ultimately, it should probably come down to support. If you can't get the answers to your questions when you need them, the software either ties up time that you could be making money, or is set aside until you can get back to it with the answers. As much as I love Algor, if they aren't supporting you, take your business to NEiNastran. The deal that they are offering looks pretty good. Normally, that package will run $20k+ US (maybe even close the $30k US). Both companies usually cut good deals to small firms (perhaps as much as a 30% discount), but your deal looks even better from my perspective.

You must be doing something more than just pressure vessels if you are looking at L1-L5?

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com

 

[sbrad]

Thanks again Garland...much appreciated.

These particular applications are rather unique as our design temperature is up to 1030deg C (1886 F) and we have large displacements etc. Also need to consider transient conditions and the lag effect due to internal refractory / insulation systems.

We also have frictional contact issues with a sliding seal system so there is a lot going on.

It's difficult to summarise the application but it ticks most of the boxes with respect to problems!

I'll keep you posted


sbb

 

[GBor]

I love analysis! Your particular application sounds most interesting! Assuming your motion and frictional contact is over a long period of time (relatively speaking) so that it is non-linear static, I still think either package will do. If it is short duration so that it requires consideration of impact-type loads, I definitely think you need a demo and some analysis that you can compare to some known data. Don't take the software package's word that it can do what you need it to do.

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com

 

[gohj]

I have been using Algor for the past year struggling to do meshes and analysis of thin walled structures with it.

When you talk to Algor tech support they almost always end up recommending you build whatever model you want to analyze from scratch using their superdraw program, when all you want to do is import you solidworks model.

Have attended one Algor training session and it was a complete waste of time.

 

[GBor]

I use SolidWorks and Algor fairly extensively. What kind of problems are you having?

I'm going to guess that you've modeled the thin cross-section to great accuracy in SolidWorks. Depending on the complexity of the structure, a mid-plane mesh is complicated. An alternative is to build your SW model as a solid (no hollow interior - before you shell the model) and surface mesh it.

Sorry to 'hear' your Algor training session was a waste of time. I've had diverse experiences in their training sessions, but I usually walk away with something useful.

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com

 

[ttoole]

I agree with gohj with regard to meshing thin walled structures in Algor. We have had a lot of difficulty with importing surface geometry and meshing it. We are looking at some other products for this purpose now including NEiWorks which seems to be a very attractive alternative to COSMOSWorks. What attracted us was that this was truly embedded in SolidWorks. I do not think besides COSMOS there is another truly embedded product. If anyone knows of one please let me know so we can compare. We have some end of year money to spend.

I personally did not take any of the Algor training but others here have and did not come back any better. The main complaint was the class was focused on a fixed curriculum. I have joined this site hoping to get some better support.

 

[GBor]

I have meshed plenty of thin-walled structures, but the trick is not necessarily in the Algor mesh engine...it's often in the CAD modeling. I've had Algor integrated with SolidWorks for about the last 9 months to year and have had great success. I'd be happy to help where I can including, depending on where you are located, providing some specific training, which I know Algor will provide as well...although I'm not certain of the cost.

I do like SolidWorks and COSMOS is integrated well for certain types of analyses. If you are operating within those analytical types, it may very well be a viable alternative, but don't lose your Algor capability...just add to it with COSMOS.

NEiWorks has a very good product, but I haven't had any better success with meshing in it over anything else.

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com

 

[ttoole]

Thanks Garland,

Is there any special steps to handling thin-walled structures? Are you saying export surface geometry over solid geometry?

What version of NEiWorks do you have? They claim to have added a lot of new features to their latest release.

 

[GBor]

I do not have the latest of NEiWorks. I look forward to "playing" with it soon. My effort with NENastran is less frequent than Algor. I have access to it through another company with whom I work closely.

As for thin-walled structures, if you model the structure with a wall thickness and the geometry is complicated, there are issues with corners. If, however, you model the article as a solid, you can surface mesh more easily. The surface mesh would then be the bounding surface of the solid model. This mesh can be manipulated through a variety of ways either in the CAD software depending on the type of export you are using, or through the CAD import settings of your FEA package. I would suggest you model a box and play around with the CAD export settings and with the FEA import/mesh settings.


Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.

http://www.borowskiengineering.com