My question is about which FEA software, with concrete plate design ability, is the best for the US. It seems that STAAD is the most popular and has the majority of the market. What are your opinions of STAAD overall and would you recommend buying it compared with competitors? I have reviewed STAAD, Robot, SAP200, IES VisualAnalysis, RISA 3D(No Concrete until 2002), and STRAP, but it is very hard to get a good grasp of the programs with demo versions and limited experience. From past threads it seems that RISA is the best, with STAAD offering many problems. Thanks for your input.
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations cowski on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
STAAD vs. competitiors 11
- Thread starter Mark10
- Start date
-
1
- #41
i agree with PTSIStructural - SCS FLOOR is indeed a very good design and analysis tool for RC and PT floors, and mat foundations, based on FEA.
one of the better aspects of FLOOR is that the authors and support team are structural engineers who have actually designed (and constructed) PT and RC floors...not your "closet programmers" producing black box solutions etc..without understanding real design and construction aspects.
my 2 cents worth...
one of the better aspects of FLOOR is that the authors and support team are structural engineers who have actually designed (and constructed) PT and RC floors...not your "closet programmers" producing black box solutions etc..without understanding real design and construction aspects.
my 2 cents worth...
Some years ago I researched (then) FEA CAE software
and wrote a treatise for Architectural & Engr'g mag
on my results. Here's what I discovered:
In reading between the lines of software developers' warrantees, and university peer comments, the accuracy
of any CAE software, whether it's STAAD, SAP, GTSTRUDL,
ANSYS, COSMOS, et al, is no better than ±35%, about the
margin of safety (in steel) between yield and failure!
Then using various solved models from pioneer
FEA-analysis texts, I also found wide disparity
between the CAE results, and a mathematical solution.
Believe me, it's very disconcerting to build a simple
2D truss model, load it into your favorite CAE, then
find the printouts don't match the solved problem!!
Yet contacting CAE vendors with this quandary yielded
no direct answers, only "we are the best" gibberish,
or "we have the most installed seats" baloney, and
a bunch of "well, then there outta be a certification agency" hooey. Face it, they're selling a commodity.
And from a purely practical viewpoint, how often have
engineers gone out to the site after their design is
complete, and seen the client has added another story,
or another piece of A/C equipment, or changed the rebar
configuration, that the original CAE model was based on?
There are more common errors than CAE accuracy. I've seen structural analyses which showed inconsistent stress results between load combinations, due to the engineer's simple failure to give individual loads the correct sign ±, yet the both engineer and peer plan checkers missed that.
Writing your own CAE software is even more of a liability.
They'll kill you in a CAE shootout with your own gun, and
then beat you to death with the Code check!
What about uncontrollable variables? That 100-year record windstorm you can't prove because the nearest recorder is miles away, or the 100-year snow that melted before you could make your field inspection, or a roof drain that got plugged and now is buried under 100T's of collapsed steel.
Bottom line, figure whatever CAE software you choose:
1) you get what you pay for, 2) always test it with solved problems, 3) printout, archive and peer review everything,
and 4) carry good E&O coverage.
Cause it's the little stuff that's gonna get you....
and wrote a treatise for Architectural & Engr'g mag
on my results. Here's what I discovered:
In reading between the lines of software developers' warrantees, and university peer comments, the accuracy
of any CAE software, whether it's STAAD, SAP, GTSTRUDL,
ANSYS, COSMOS, et al, is no better than ±35%, about the
margin of safety (in steel) between yield and failure!
Then using various solved models from pioneer
FEA-analysis texts, I also found wide disparity
between the CAE results, and a mathematical solution.
Believe me, it's very disconcerting to build a simple
2D truss model, load it into your favorite CAE, then
find the printouts don't match the solved problem!!
Yet contacting CAE vendors with this quandary yielded
no direct answers, only "we are the best" gibberish,
or "we have the most installed seats" baloney, and
a bunch of "well, then there outta be a certification agency" hooey. Face it, they're selling a commodity.
And from a purely practical viewpoint, how often have
engineers gone out to the site after their design is
complete, and seen the client has added another story,
or another piece of A/C equipment, or changed the rebar
configuration, that the original CAE model was based on?
There are more common errors than CAE accuracy. I've seen structural analyses which showed inconsistent stress results between load combinations, due to the engineer's simple failure to give individual loads the correct sign ±, yet the both engineer and peer plan checkers missed that.
Writing your own CAE software is even more of a liability.
They'll kill you in a CAE shootout with your own gun, and
then beat you to death with the Code check!
What about uncontrollable variables? That 100-year record windstorm you can't prove because the nearest recorder is miles away, or the 100-year snow that melted before you could make your field inspection, or a roof drain that got plugged and now is buried under 100T's of collapsed steel.
Bottom line, figure whatever CAE software you choose:
1) you get what you pay for, 2) always test it with solved problems, 3) printout, archive and peer review everything,
and 4) carry good E&O coverage.
Cause it's the little stuff that's gonna get you....
Hi, anthae.
That must be close to the most depressing (and surprising)thing I have read for ages. Do you really mean that none of the software you looked at got closer than 35% to your mathematical solutions?
Whenever I have checked my aged DOS software (microsafe) against text book solutions (eg for plate bending) I have got pretty close correlation. In my world that means within less than 5% discrepancy.
Just what do you mean by your 35%? Does that apply to the major design forces in a truss, or just the worst percentage resulting from small errors in very small forces?
That must be close to the most depressing (and surprising)thing I have read for ages. Do you really mean that none of the software you looked at got closer than 35% to your mathematical solutions?
Whenever I have checked my aged DOS software (microsafe) against text book solutions (eg for plate bending) I have got pretty close correlation. In my world that means within less than 5% discrepancy.
Just what do you mean by your 35%? Does that apply to the major design forces in a truss, or just the worst percentage resulting from small errors in very small forces?
-
1
- #45
I would highly question anthae's findings.I have worked in the field for awhile, and done numerous comparisons between codes. I have never seen this level of error due to FEA code issues.
There is currently a relevant discussion which came out of anthae's comments going on in the Finite Element Analysis subgroup.
Regarding the 35% number given--based on my experiences in benchmarking and comparing against elasticity results, I have found consistently that codes are within single-digit accuracy, presuming reasonable user assumptions.
Many of the better codes are within 1-2% of "exact" numerical answers, subject to proper element selection and element assumptions. This is not me parroting the marketing of the FEA companies--this is me stating my own findings.
The only time I have seen errors on the order which anthae states is when there are poor meshes, or when the "closed-form" problem is not appropriate for elasticity problems (either St. Venant's does not apply, or the calculated problem does not fit the elasticity assumptions for it--i.e. euler buckling assumptions used on squat columns).
Before anybody wrings their hands over anthae's statements, I would encourage you to look at the ancillary discussion of this on the Finite Element Analysis area.
Brad
There is currently a relevant discussion which came out of anthae's comments going on in the Finite Element Analysis subgroup.
Regarding the 35% number given--based on my experiences in benchmarking and comparing against elasticity results, I have found consistently that codes are within single-digit accuracy, presuming reasonable user assumptions.
Many of the better codes are within 1-2% of "exact" numerical answers, subject to proper element selection and element assumptions. This is not me parroting the marketing of the FEA companies--this is me stating my own findings.
The only time I have seen errors on the order which anthae states is when there are poor meshes, or when the "closed-form" problem is not appropriate for elasticity problems (either St. Venant's does not apply, or the calculated problem does not fit the elasticity assumptions for it--i.e. euler buckling assumptions used on squat columns).
Before anybody wrings their hands over anthae's statements, I would encourage you to look at the ancillary discussion of this on the Finite Element Analysis area.
Brad
Sorry for the delayed response...
The only place in FE that I know with significant inaccuracy is I beam torsion. Warping is simply not modeled in most FE packages. Maybe in this area you might have 35% inaccuracy for an unlucky beam under a lot of torsion.
Of course a high level of torsion is uncommon in I beams because they are designed this way, so this is normally not a problem. You need to be careful if you are designing a curved girder though.
Warping is also significant in lateral buckling. Lateral buckling analysis software normally considers warping in individual elements, but there is no continuity of warping at the joints.
Michael
The only place in FE that I know with significant inaccuracy is I beam torsion. Warping is simply not modeled in most FE packages. Maybe in this area you might have 35% inaccuracy for an unlucky beam under a lot of torsion.
Of course a high level of torsion is uncommon in I beams because they are designed this way, so this is normally not a problem. You need to be careful if you are designing a curved girder though.
Warping is also significant in lateral buckling. Lateral buckling analysis software normally considers warping in individual elements, but there is no continuity of warping at the joints.
Michael
ERNESTOBARRERA
Structural
the problem with the Research engineers people is the lack of respect for thier costumers. For example, the MASTER/SLAVE specification has been finally been corrected (at least they claim that)in the STAAD PRO2001 version. the people that bought STAAD PRO 2000 and earlier versions obtained WRONG results. This is not only unjust, but also risky for an engineer that uses the software to then realize that his results are wrong. Also when a firm buy a software expecting to solve its structural needs, realizes that the software dont do what thier developers claim, therefore having to buy another software, losing a lot of money. the only question that you can ask with STAAD PRO is " WHAT IS GOING TO BE THE NEXT BUG"?
-
1
- #51
There are many people who earn a living by developing, writing and supporting good software. I do not see a reason to deprive them of a living by trying to obtain their software for nothing. If one is a student, then it is often possible to obtain a demo version for free or cheap. If one is going to use the software to do paid work, then one should be prepared to pay to purchase it. That is my view. I hope I am not being too harsh.
Richard Beneke.
Richard Beneke.
SteamJetPE
Mechanical
Robot was my choice, after a month of comparing features and costs. I just couldn't find a better package for what we needed anywhere else. We design and fab small industrial non-building structures (<500 elements), and do load cases and FEA on fabricated metal structures (VRU ducts, hoods, stacks, etc). I love everything about it. But my opinion might have been different if we were using it for larger, more complex structures. GTStrudl and STAAD looked impre$$ive. SteamJetPE
Ejectors, LRVPs, Hybrids
Troubleshooting, startup, design
Ejectors, LRVPs, Hybrids
Troubleshooting, startup, design
If you only design at concrete, I wish recommend you RCBE their e mail is RCBEINFO@ACI.COM vertion 5.1.1 is a great software combine vertical load, horizontal load like wind, earthquake.Where you can work with vertical,horizaontal and oblique elements all together. I am sure that is the best of the best.
Guest
How about RAM Advanse . This product is very good and costs a lot less than STAAD , SAP and a few others mentioned . It is a general analysis and design FEA .This use to be Avansse/AVwin98 and was bought by RAM Intl. The software has been improved and in use for over 10 years. It is especially good for concrete detailing of beams and columns.
ReBeneke
Anthae
Austim
I just want to add another horror story about STAAD ...by the way ...I agree that the people who makes a living writing software.. should not be deprived of selling a commodity..but I think that in addition to pretend "we are the best"..I think they claim "our program does it all"...gibberish.. Because many grand-father clause engineers partners of firms believe them... I was working as a Sr Structural Engineer in bridge design, and then company was running low of work..even thought we did not use STAAD for bridges..I have used STAAD off and on since 1989. Contrary to Anthae's findings in the past ( in previous versions) I have been able to double check statics and accuracy withn 5%...by hand.. in my practice I just don't accept any output unless I can check it.. ( by the way in this firm I was working I was shocked seeing binders of calculations for bridges ( done before me) .. containing only printouts .. without absolutely no checking or elaboration on results.) Well. they asked me to check a 150' tower for new antennaes with ICE formation using STAAD suite , first time for me..I asked a junior engr to draw it in Microstation and we transferred the dxf file into STAAD..thinking that using this advantage would save time..the results were disastrous...because STAAD did not create hundreds of joints tying at mid point of other members.. and connectivity issues..it took me two days to renumber joints and coordinates to make it run ..finally it did .. but I was two days late.... I also was doing a QA/QC on a bridge design by another younger engineer at the H/Q office I found similar issues like Austim (above)...I was puzzled that STAAD gave member flexural stresses on H-Piles, and nowhere in the input specifies the orientation of the Major and minor axis...STAAD assumed one...and similar issues...( STAAD "thinking" for the Engineer)... After a MEMBER displacement command an FZ was input , and STAAD did not produce an error message, (FZ is a force not a displacement)...and variety of issues.. I included those comments in my review...Well, the young engineer said he was right and that I was wrong , and that STAAD was perfect...cannot make mistake.. I was laid off.... the firm concluded that I must have lied saying I had experience using STAAD.. becasue the software vendor told them the program can make an engineer out of a high schooler... I guess STAAD works fine for the partners of the firm, who only want to see the results, and don't care if the results are correct....
Anthae
Austim
I just want to add another horror story about STAAD ...by the way ...I agree that the people who makes a living writing software.. should not be deprived of selling a commodity..but I think that in addition to pretend "we are the best"..I think they claim "our program does it all"...gibberish.. Because many grand-father clause engineers partners of firms believe them... I was working as a Sr Structural Engineer in bridge design, and then company was running low of work..even thought we did not use STAAD for bridges..I have used STAAD off and on since 1989. Contrary to Anthae's findings in the past ( in previous versions) I have been able to double check statics and accuracy withn 5%...by hand.. in my practice I just don't accept any output unless I can check it.. ( by the way in this firm I was working I was shocked seeing binders of calculations for bridges ( done before me) .. containing only printouts .. without absolutely no checking or elaboration on results.) Well. they asked me to check a 150' tower for new antennaes with ICE formation using STAAD suite , first time for me..I asked a junior engr to draw it in Microstation and we transferred the dxf file into STAAD..thinking that using this advantage would save time..the results were disastrous...because STAAD did not create hundreds of joints tying at mid point of other members.. and connectivity issues..it took me two days to renumber joints and coordinates to make it run ..finally it did .. but I was two days late.... I also was doing a QA/QC on a bridge design by another younger engineer at the H/Q office I found similar issues like Austim (above)...I was puzzled that STAAD gave member flexural stresses on H-Piles, and nowhere in the input specifies the orientation of the Major and minor axis...STAAD assumed one...and similar issues...( STAAD "thinking" for the Engineer)... After a MEMBER displacement command an FZ was input , and STAAD did not produce an error message, (FZ is a force not a displacement)...and variety of issues.. I included those comments in my review...Well, the young engineer said he was right and that I was wrong , and that STAAD was perfect...cannot make mistake.. I was laid off.... the firm concluded that I must have lied saying I had experience using STAAD.. becasue the software vendor told them the program can make an engineer out of a high schooler... I guess STAAD works fine for the partners of the firm, who only want to see the results, and don't care if the results are correct....
Quite right,
IMO You should not be let near a computer until you at least have a feel of what the results should be. Computer analysis is a tool for speeding up the design process and justifying the engineers 'assumptions and conclusions'.
I once worked with a junior engineer who made a very slight error of input into an analysis package. He was perplexed when a joint deflection was stated as 4km and he asked me if I thought this was correct for the type of structure!
IMO You should not be let near a computer until you at least have a feel of what the results should be. Computer analysis is a tool for speeding up the design process and justifying the engineers 'assumptions and conclusions'.
I once worked with a junior engineer who made a very slight error of input into an analysis package. He was perplexed when a joint deflection was stated as 4km and he asked me if I thought this was correct for the type of structure!
- Status
Similar threads
- Question