code checking software

huangyhg

code checking software
i was just reading the discussion in
a lot of my colleagues use mathcad sheets with commercial software. we use staad in house and people have written macros that connect mathcad with staad results for code check. they use something called open staad. i am not a programmer, but i would check out their website on how to do it. the good part of having your own template is you are in control of the design sequence. staad has nice output so you can follow the sequence, but not every step.
what a great thread.
question 1 (is there a concern with using commercial programs?):
there's no problem with using a commercial program as long as you know precisely what the program is doing. most reputable commercial programs perform the majority of calcs correctly but will either have some strange or completely wrong calculations buried in there somewhere. should most people use commercial programs? absolutely. do most people dig in deep enough to be responsible in their use? who knows.
examples are:
risa's version from last fall did cb calcs completely wrong in some cases.
etabs' composite beam design had lots of problems the last time i checked, which was a couple of years ago. it spat out lots of cool-looking output, but some of it was either wrong or just silly.
risa's idea of column effective length calcs for rigid frames is not even in the ballpark of technically correct.
etabs' automatically generated wind load combination factors were wrong the last time i checked and had to be manually fixed every time.
the ram system used obscolete floor vibrations criteria until just recently, 10 years after aisc dg11 was published. i've seen this one actually cause a large problem on one structure.
the last time i checked (about 8 years ago) cfs did not determine the correct flat widths for the case of intermediate stiffeners along a large flat panel width.
with the use of a rigid diaphragm, braced frame beams get absolutely zero axial load in every single commercial program i've used. with both beam nodes attached to the diaphragm, there's no elongation, therefore no strain, therefore no stress, and therefore no force.
safe completely excludes the one-way shear check for slabs and footings.
i have about 18 more of these but think that's enough.
the bottom line is that you have to know in nauseating detail what your program is doing.
as far as in-house software:
i've seen excel used extensively over the years and think that it is great for simple calculations or calculations that are simple and fit well into tabular format. for more complex calcs, forget it. excel programming turns into unintelligible spaghetti code after it gets past some level of complexity.
i like mathematica for academic-type stuff but find it to be cumbersome for design-type stuff. it also has the huge downside that practically no structural people outside academia use it, making it impossible to trade files with clients. i really like its stability and speed. it is an extremely robust program.
mathcad is a fantastic program on which to base a structural engineering company's in-house software. it is not super-stable or fast like mathematica, but the output is truly excellent. it is easy to read, flows from top-to-bottom (unlike spaghetti-coded excel junk), and can be formatted to be very compact and attractive. this last one is why it's better than mathematica for design-office usage. mathematica flows line-to-line so very inefficiently uses the paper. at this point, i've tried out everything that i could find and think that mathcad is the best by a long shot for typical structural engineers.
vb? c++? mathcad has a monstrously-large advantage over actual programming languages in speed of development and ease of modification. i can write an entire design program by the time i set up the variables in a c++ program. these are suitable for rare occasions that are outside the scope of mathcad.
i kinda wandered off your subject -- sorry.
dbd
14159,
thanks for you extended response; i really appreciate your comments.  to me this is a very interesting subject.  years ago i wrote a code checking program in vb for use with a general purpose fea program.  it worked ok but was a lot of work to do.  i am currently working with multiframe with its code checking option, and while it seems to be pretty neat, it’s hard for me to get used to the black box approach.  i suppose with enough hand checking i will get there.  
one thing i really don’t like to do is calculating stresses for single angle struts.  the last time i looked into risa it only checked axial loads for kl/rz, i think.  multiframe checks bending about x and y axes, but in my opinion bending should be checked about z and w axes unless there is some kind of restraint.  still working on this.
for calculation software i use calculationcenter (think of it as mathematica lite) and my son uses mathcad.  your comments about these are pretty much on target.  mathcad produces nice looking output and mathematica doesn’t.  mathcad is widely used by design engineers and mathematica for some reason isn’t.  well the reason i guess was the price; it was way too expensive compared to mathcad about 5 or more years ago.  i still like vb for many things but find that i am using calculationcenter more and vb less.  
if anyone else wants to weigh in (or vent) on this subject please post your comments.  it should be interesting to hear what works well and what doesn’t, places to be careful, and so on.
regards,
-mike
i wouldn't bet anything on any of the commercial programs' abilities to correctly determine the strength of a single angle strut! in my experience, those guys think kl/r and that's it. flexural-torsional buckling? torsional buckling? what's that? some of the higher-end packages might do a little better -- i'll check the sap documentation later for my morbid curiosity.
about calccenter, i have to admit that i really like mathematica because of its speed and stability, but shy away from it because of the stuff that i typed earlier. i downloaded the calccenter free trial and found it to be similar to mathematica for relevant features. i will consider it in the future because it's $575 price tag is a bit more favorable than the $1200 mathcad and $1900 (!?)mathematica.
i've been in contact (some slightly heated debates!) with some of the mathematica people about their crummy output format. perhaps if more people express disappointment they will improve for future versions. there's no doubt that, at the very core, it's far better than mathcad. there's really no good excuse to be slaughtered in the marketplace for cosmetics, which is exactly what they're doing, at least in our discipline. i don't know what me, ee, etc use, but i can't help to think their needs are not that much different.
14159
i currently am using calccenter to do some basic code checking when necessary.  as i mentioned in my previous post i am trying to get used to a commercial package but sometimes i just need to do it myself.  the nice thing about calculation software is that you can calculate p/a if p is equal to a single value, or an array of hundreds.  this makes writing of the program very much easier in that the coding to do matrix manipulation is unnecessary.
the next thing that i am going to add to my program is the ability to check single angle struts, or to at least do it the way i think it should be done.  with this approach i expect to be able to include eccentricities wherever needed and calculate bending about any axis i want, with the number of loading conditions not really important due to the power of calccenter, or i suppose any other cas software.
i guess writing your own software is a hard thing to give up.
-mike
14159
re: your comments about mathematica output
i just spent two days working on some calculations in calccenter that i need to send to a customer, and it is driving me nuts.  i personally don't like full width justification and i don't like how calccenter does page breaks.  you are right, the cosmetics of this is terrible.  sometimes it is just plain ugly.
regards,
-mike
i have almost a pen-pal relationship going with one very nice fellow with wolfram right now. i'm trying really hard to convince them that they need to make the output in the form (using newton's second law as an example):
f = m a = 10 kips
where the user types "=" after the "a" and the 10 kips pops in there to the right. not below. not some other unnatural place. to the right! he's trying, but i don't think it's gonna happen. that would solve a large percentage of my issues with the program. i've been after mathsoft for this also, with even less luck.
i wish many people using these programs would bombard these folks with this suggestion. i'm just one complaining guy, i guess.
14159
14159
that looks like a good idea to me.  in fact i would even settle for this:
ma = 10 kips
which is better than
ma
10 kips
actually calccenter doesn't even have the units kips.  it's not that i mind a plain or spartan look, but sometimes it really is unattactive.  anyways, i didn't start this thread to complain about mathematica, but you know, i am feeling better.   
-mike
mike:
the following is the latest from my discussion with a fellow from wolfram. it might be helpful to you. i have not tried his suggestion yet, but it seems promising.
"i talked to the front end group and here is their reply to me.
the closest i can think of to something like this would be to use authortools' bilateral cells, where the $firstbilateralstyle is set to {"input"}, although that would remove the ability to re-evaluate the inputs.
also they told me that they might be looking at this type of format in a future release."
their product is so much more stable and robust than mathcad that i would personally really like to see them work on the cosmetics and compete successfully with mathcad in the structural engineering world. the situation reminds me of how windows beat out os/2 and mac back in the mid-90s even though it was technologically inferior. in that case, another non-technical issue, marketing, made the difference. the computing world would've been a better place if this did not happen, but i digress.
14159
14159,
about 18 months back before i bought calccenter i looked at several others in this category.  i recently posted links to these in