Strand7 FEA software review 4

[ALK205]

Hello, I am currently in the process of evaluating FEA software packages for use in analyzing various structural problems including welded/bolted connections, stresses in steel vessels, and plate/shell structures. Our office currently uses RISA-3D for structural frame analysis (beams, columns, etc.) as well as some plate structures. We also use SAP2000 which has cable elements and staged construction analysis capabilities. We would like to add a software package for analyzing more complicated FEA problems which require contact elements and nonlinear analysis capabilities. At this point we do not want to invest in a package such as ANSYS or Abacus since we perform this type of analysis on a limited number of projects. So far I have seriously looked at SolidWorks Simulation (which combines SolidWorks modelling capabilities with the Cosmos FEA solver), SAP2000, and Strand7. SolidWorks has great modelling features, however, I am not convinced that the FEA capabilities are rigorous enough for the type of problems we wish to solve. SAP2000 has FEA capabilities including shell elements, solid elements, meshing tools, link elements, and nonlinear material properties; however, I have found the user interface to be extremely cumbersome. Strand7 seems like a good option, however, this program is not as well known, and does not seem to be very widely used. I am very interested to hear from current Strand7 users to find out what type of problems they use Strand7 to solve, and what they think of the user interface as well as the versatility of the software.

 

[IceBreakerSours]

This particular subject has been discussed - at length! - on several threads on this forum alone:

[link ]

http://www.eng-tips.com/viewthread.cfm?qid=326048

[/url]
[link ]

http://www.eng-tips.com/viewthread.cfm?qid=127019

[/url]
[link ]

http://www.eng-tips.com/viewthread.cfm?qid=120567

[/url]
[link ]

http://www.eng-tips.com/viewthread.cfm?qid=74650

[/url]
[link ]

http://www.eng-tips.com/viewthread.cfm?qid=105128

[/url]
[link ]

http://www.eng-tips.com/viewthread.cfm?qid=37632

[/url]
[link ]

http://www.eng-tips.com/viewthread.cfm?qid=67468

[/url]

I don't know about Strand7 but the links above *will* definitely help you make up your mind or confuse the heck out of you!

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

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

 

[ALK205]

IceBreakerSours, thank you for your reply. Before I started this thread, I did take the time to read the threads that you mentioned. As you alluded to, there are plenty of opinions and trains of thought regarding FEA software. While Strand7 is mentioned in several of these posts, it is not discussed at length like some of the other, better known FEA software options. The threads that specifically discuss Strand7 are limited to specific user questions. I am trying to get a broader feel for the software, and get feedback from current Strand7 users. I have spent some time researching this topic, however, I have had a hard time finding specific information on Strand7, other than what is provided by those who are marketing the software. I would appreciate feedback from anyone who has used this software. Thanks.

 

[IceBreakerSours]

In that case, I would simply email researchers who are using Strand7. I found quite a few publications by people using Strand7 via. Google Scholar.

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

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

 

[ALK205]

IceBreakerSours, as I mentioned previously, my goal in starting this thread is to obtain useful feedback from current Strand7 users. I have spent considerable time researching this software on my own, including using a trial version, and I am interested in the opinions/experience of other industry professionals who have used this software.

 

[IDS]

I don't know about Strand7 but the links above *will* definitely help you make up your mind or confuse the heck out of you!

As far as I can see those threads contain precisely zero information about Strand7 (other than a link to the Strand7 site). Is there some problem with asking for comments on a specific FEA application in the FEA section of an engineering forum?

I have been a regular user of Strand7 since 1994 (when it was Strand6) and I can recommend it as a comprehensive package at a reasonable price. I find the user interface friendly, and pre and post processing features good. I now make a lot of use of the API which allows complex analyses to be automated via Excel (or the language of your choice). I have found technical support to be excellent, but I'm based in Australia (where it originated), and I can't speak for elsewhere. My main applications are bridge analysis and 2D soil-structure interaction problems. It now has a good staged analysis capability, which makes the latter problems much easier to handle. I have done limited work in 3D solids analysis and dynamic analysis and found it quite capable in those areas.

Please ask if you have any questions on specific aspects of the package.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[IceBreakerSours]

Is there some problem with asking for comments on a specific FEA application in the FEA section of an engineering forum?

No but since when has a little humorous tone become something worth frowning upon? Did I not try and help as I felt was needed? I made the effort of looking up the threads that seemed to fit the bill, did I not? Could I not simply have misunderstood the context of the question?

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

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

 

[jhardy1]

Strand7 originated in Australia, where it has a substantial and very loyal following. It’s less well known outside Australia, in my experience.

I suspect it is the No. 1 general-purpose FEA package used by general structural / mechanical engineering practitioners in Australia (based on number of users). I think it is the best "bang for your buck" FEA available. It will handle pretty well all of your basic and mid-level FEA requirements, and will even do some surprisingly "high-end" stuff as well.

The user interface is very easy and intuitive to learn, and it has a good complement of solvers and post-processing tools, etc.

Probably the two main issues which mean I have to occasionally look elsewhere for specific applications:

1. While it will import and auto-mesh most of the standard 3D CAD formats (ACIS, IGES, STEP, etc), it has minimal ability to create and edit geometry, apart from some basic tools to "de-feature" slivers, stitch small gaps, etc. That is, if you are given a CAD model to analyse, and you need to modify the basic geometry (such as adding or moving a stiffener), Strand7 does not have the tools to manipulate the imported geometry before meshing, so you need to go back to the generating CAD application, and do your geometry changes there, then re-export the geometry to Strand7.

2. It does not have a tool for "mid-plane extraction" (the ability to take a solid model of a thin-shell structure, and automatically simply to a properly-connected single-surface representation of each plate). This makes creating good quality plate-shell meshes from imported solid geometry of welded steel platework etc a very intensive manual task. (It’s by no means impossible, but it’s a LOT easier and faster if you have access to a tool for automatic mid-plane extraction.)

Even with these limitations, it is highly recommended. Why not download the free demo from the Strand7 website to see if it will do what you want?

http://julianh72.blogspot.com

 

[ALK205]

IDS and jhardy1, thanks for your responses. I downloaded the free demo several months ago, and did a few verification models, however, I only had so much time to "kick the tires" so to speak. I have gotten a lot of helpful feedback from the Strand7 rep that I am working with, however, I wanted to also get feedback from Strand7 users who are not trying to sell the software.

One specific question I have is whether or not it is worth spending a little extra to get the sparse solver, or if the skyline solver included with the basic package will be sufficient. Any thoughts? Thanks.

 

[IDS]

On the sparse solver question, it depends what you are using it for, but to give some indication I have run a couple of my typical (for me) models with the sparse solver and the skyline solver:

Staged non-linear analysis, 13 stages with 20-50 iterations per stage, say 300 iterations in all, 2D model with 1000 nodes, 65 beams (contact elements) and 300 8-noded plates:
Sparse solver 23 seconds; Skyline solver 43 seconds

Linear analysis with one load case. 3D model with 26,000 nodes, 4,000 beams, 2,000 plates, 3000 20-noded bricks:
Sparse solver 60 seconds; Skyline solver indicated 97% completion after 7 minutes, then seemed to get stuck. Task Manager is currently indicating 1-10% CPU allocation to the St7Solve32.exe process, and about 1.9 GB memory allocation, so it may be it has started using hard disk memory to complete the task.

The second model was actually taken from a staged non-linear analysis with 35 stages and about 20 iterations per stage, so it really needs the sparse solver to be practicable.

In summary, for small to medium linear 2D analyses it wouldn't be worth worrying about, for staged non-linear analyses of even quite small models it gives a reasonable saving, and for 3D non-linear work with brick elements (other than small models) it's pretty well essential.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[Denial]

I agree with Doug. Strand7's sparse solver can make a HUGE difference to the solution times, and is well worth its additional cost.

 

[jhardy1]

+1 for the Sparse Solver:

If your models are all "small" (hundreds of nodes to a few thousand nodes, say), and your runs are linear, or involve only a small number of time steps / load increments, then the Skyline Solver works fine. Indeed, my own results suggest that for such problems, using the Skyline Solver with a well-sorted matrix can actually be faster than the Sparse Solver - but we are comparing run-times of just a few seconds either way, so it doesn't really mean much. I guess the simple "brute force" approach on a small tightly-banded matrix may actually be more efficient than the "computational overhead" required for the Sparse Solver?

However, for "medium" to "large" models (tens of thousands of nodes or more), and / or if you are doing intensive non-linear and / or transient dynamic analysis, the Sparse Solver will pay for itself many times over. On such models, elapsed times can be five times (or more) faster.

When planning an analysis, I tend to break problems into a few "time scale" brackets:

"Instantaneous" - results are available in less than a minute, so it doesn't really matter whether one solver takes 10 seconds and the other takes 15 seconds - you will just watch the progress bar approach 100% either way!

"Coffee break" - the run will take a few minutes to run; long enough to grab a coffee, visit the toilet, etc, and your solution is ready when you get back to your desk.

"Lunch break" - now you are talking about an hour or two, so you might trigger a run before going out for lunch, or else let it run in the background while you perform some other task for an hour or so. With this time scale, it really starts to pay to check your runs carefully before hitting the "Solve" button - there is nothing more frustrating than initiating the Solver, going away for an hour or two, and returning to find that your loads / restraints are incorrectly specified!

"Overnight" - these runs take several to many hours to perform, so the comments above about checking your data REALLY apply. Typically, launch the Solver when you leave work, and hope that you have meaningful results when you come back to work tomorrow.

For "instantaneous" problems, it really doesn't matter much which Solver you use. Where the Sparse Solver comes into its own is that it may turn a "Lunch beak" problem into a "Coffee break" problem, or an "Overnight" problem into a "Lunch break" problem.

If money is tight, just buy the regular Skyline Solver initially - you can always add the optional Sparse Solver later if necessary.

http://julianh72.blogspot.com

 

[Hansmeister]

I use Strand7 quite a bit (but not full time), and love it. I have all the options, but don't have time to explore all of them. Ann Delvoux is the USA Tech Rep, and is very helpful. In one of my models I had about 120,000 bricks, lots of plates, etc. and it took about 10 minutes.

 

[blihpandgeorge]

for what it is worth, i have used a number of different FEA packages from ROBOT / SOFISTIK / SAP 2000 / ANSYS and Strand 7. I would personally place Strand in the same league as SAP 2000. From what i can recall, it is pretty cheap for the amount of punch you get.

As mentioned earlier, it is Australian developed and is probably the main general purpose FEA package used in Au for 'general' FEA. I have and continue to use it for large complex building project such as tall buildings and long span roofs to stadia. I also know it is routinely used by our mining guys doing materials handling. Like all packages it is has its little nuances on how it wants to do things, but nothing strange there.

A few features i like is that is readily allows you to do batch solving, so from one model you can set off lots of different types of long running analysis to be run as 'batch' when you leave work at the end of the day. It will then run a series of analysis so when come in the morning it has a heap of results for you.

Notwithstanding the above, the long and short of it is you need to have a good play with it to see if it does what you want.