Software

[todd2ny]

I am looking for some software that will design steel members as well as reiforced concrete walls and slabs. A lot of member design is repetitive, so I am looking for a software package that can design the individual members once I run the analysis in STAAD Pro.

Thanks

 

[Ussuri]

What's wrong with the steel design component within Staad?

 

[todd2ny]

I assume you are talking about the Code Check command? Also, I have heard that some engineers do not trust STAAD to do the actual design.

 

[JAE]

"I have heard that some engineers do not trust STAAD to do the actual design."..... i.e. me.

 

[271828]

If I had STAAD, then I would try to use it to help me design these members. Read the documentation and compare simple test cases to manual calcs to make absolutely sure you know what it's doing. I have to think it does an ok job or it wouldn't have survived this long. Just make sure you know *exactly* what it's doing. I do know for a fact that it used to have totally illogical and unconservative defaults.

I think we need to change the language of the thread. Software programs do not *design* members. They're sophisticated calculators with great potential for showing good-looking results that are wrong, and they have no judgment, so should be used as such.

 

[Ussuri]

You should not "trust" any software. You should be familiar with the way it operates, the process it takes to do the code check, and have an idea of what the answer will be before you start. If you follow what it is doing you are then in a position to determine if it is spitting out rubbish.

Just accepting what it tells you is dangerous.

 

[civeng80]

I second that approach Ussuri. Some idea of section sizes and/or dimensions of reinforced concrete rebar should be thought of before analysis. Analysis should then be used as a means to an end.

 

[abusementpark]

JAE, why don't you trust STAAD??

 

[miecz]

There is no way to completely check a program. It may be possible to compare results to manual calcs for a few simple cases, but the only way to be sure a program calculates everything correctly is to completely review the program code. Impossible of course, since the vendor doesn't provide the program code. So, whenever we use someone else's software, we're hoping (and praying) that the program is correct.

I trusted a program once after checking some cases against hand calcs. Turned out, the program didn't check some section of the code that rarely governed. I didn't pick up the error in my comparisons, and neither did anyone else using the software, apparently. The structure needed to be reinforced. We paid. The software vendor, who admitted the error, was liable for the cost of the software, a tiny fraction of the cost of repair. Many software vendors include verbiage in the contract that limits their liability in this way.

 

[hokie66]

miecz,

Without being too specific about the error which occurred, can you share with us how it was discovered, and if possible what the error was? We all learn by errors, and they should be reported insofar as possible.

 

[miecz]

The error was discovered when the state DOT rated the bridge with another software package. That software gave more conservative results, which prompted us to do a hand calc to determine the difference in program output. The hand calc agreed with the DOT's software.

The problem had to do with web buckling of a continuous composite stringer.

Since then, the DOT has challenged the strength of a second bridge we designed using another software package. In that instance, a hand calc showed that the DOT's software was incorrect. This was the same software that the DOT used to rate the first bridge. After much discussion between the three parties, the writer of that software agreed in writing that the software couldn't be used for our bridge.

My point is that there are a lot of people selling software that is incorrect. They cover themselves by including verbiage in the license contract that limits their liability to the price of the software, a pittance compared to the cost of repair. We're fooling ourselves if we think we can verify these packages by comparing them to a few hand calcs.

 

[abusementpark]

We use RAM Advanse (a great program, by the way) for analysis and design to do our code checks and in the help section they provide the flowchart that the computer uses for the code checks with reference to equation numbers and such. So, seems like if you went through that diligently, you could get a good idea what the program is doing. That being said, I am pretty sure no one in our office has taken the time to go through it.

 

[hokie66]

miecz,

Thanks for that. So instead of checking software by hand calculation, perhaps these packages should always be subjected to checking by comparison with the results from other software. If both have errors, hopefully they will be different errors, as in your case.

We made errors in the pre-computer age as well, but I think we had a better idea of what the answers should be before we started. In today's world, we attempt so many complex structures which defy hand calculation that we have to find ways of making these programs reliable.

 

[civeng80]

miecz

Very interesting and disturbing.

Hand calculations checks can only be done in simple cases and then it would not be viable to use computer analysis in any case. But for my last design a portal frame of span 25m I actually did an analysis and design by using the old allowable stress design just as a check and actually got the same section sizes as LSD. Wind loads are not uniform across the entire rafter section so it would be extremely hard to check by hand.

As hokie66 says maybe having another analysis program on hand as a check particularly if dealing with a completely new design where sizes are not easily assessed at the start.

 

[271828]

miecz, I think that software mis-use by black box users (not your firm, obviously, because there was effort to reproduce the calcs) is a more pervasive problem than errors within the programs themselves. Sounds like you guys just got really unlucky. The software company got some free troubleshooting. I bet you didn't even get a "thank you" LOL.

I think the bottom line is still to do the manual calcs (gotta be thorough and make sure that all the limit states are verified) and know what a reasonable end result should look like. It's definitely still possible to get slammed like you guys did, but this seems to work pretty well.

Does anybody have a better idea than this? Design tables, scientific calculators, manual calcs, spec. equations, etc. have been known to have errors, so I don't see why programs would be any different.

 

[civeng80]

But how would anyone do hand calculations for an analysis ? especially when the code requires secondary analysis? Its just not possible to do right??

 

[271828]

civeng80, it is very possible to verify whether a program is doing a second order analysis manually. There are two easy options.

The AISC 13th Ed. Commentary gives some benchmark problems to test the program against. The upcoming Stability Design Guide will have more, and a slew of examples.

More fundamentally, the B1, B2 approach given in the AISC Spec. Ch. C computes amplifications that are amazingly close to what one gets from a correctly done second order analysis using a program.

 

[miecz]

amusementpark-

Thank you for sharing that. I suspect that few engineers take the time to properly verify the software we use.

hokie66-

I agree that, for the reasons you mention, we need to make the programs reliable. The only way to do that is to make the software vendor liable for program error. If it can be proven that a design error with the software causes a structural issue, then the vendor should be held liable. The only way I see this happening is if engineers refuse to purchase software with a limited liability clause.

281828-

I agree that "the bottom line is still to do the manual calcs." Using one software package to check another doesn't do it for me. I just don't take on anymore work that I can't do or check by hand.

 

[271828]

" Using one software package to check another doesn't do it for me."

I agree with this. Comparing to another program can be a great part of the checking process, though, because of the sheer number of problems that can be compared very quickly.

 

[Ussuri]

One point I may add, is that in my experiance, not many companies have two different software packages for doing the same job. It can be expensive, especially if you only intend to use one to check the other.