Design Software QA/QC

[jonk97]

I've recently been looking to purchase new connection design software and was very surprised to find a lack of QA/QC documentation available to prospective customers. I'm not necessarily looking for the level of GT Strudl where everything in their program is documented and approved by the DOE, more or less wanting some verification that a PE has at least reviewed the software code and output.

I specifically was looking into RAM Connection and was surprised that they had no idea what I was talking about when I asked for some documentation.

How do others deal with this? Do you find it useful to have the software if you have to check each run anyway?

 

[LobstaEata]

Documentation of software can be elusive thing. Yes, specific test cases are used by DOD to inspire some level of confidence or justify use of a certain software program. Programmers are quite capable of writing code for those specific test cases to complies with DOD requirements. However, any diversion from those test cases may uncover other flaws that were not anticipated. For example, STAAD provided that level of documentation early on for several test cases, but later found that bug reports still kept on flooding in for specific models. Seldom do our projects perfectly fit test documentation, so there is always the possibility that other bugs will be discovered along the way that the programmers didn't expect.

Regardless of the skill of programmers writing any software code, there is the possibility of unexpected results. As such, all of us accept that as professional engineers, we are ultimately responsible for the accurate results of any software package. Seems that there will never be a sufficient replacement for an engineer's due diligence in manually checking calculations.

 

[JoshPlumSE]

The kind of testingt that is done during software development is often differnet from the type that you would do for DOD verification. There should definitely be some overlap. But, it might not be as much as you think. Plus, there is a big difference between the type of verification that they do "in-house" and the type that they are willing to publish and make available to their users.

My guess (and it's a complete guess because I have zero knowledge of the internal workings of that company) is that this is a situation of left hand not knowing what the right hand is doing. The support engineers over at Bentley may not have ever met the folks who developed their connection program. The support engineers may be in the US and the development staff may be in south america. Therefore, the support engineers may not know what was produced during internal testing or what is available now.... not unless it is something that they have published in their standard documentation.

I work for another software company (RISA) and we have all kinds of un-published verification materials. If you asked us that same question, we'd probably tell you that we don't have any published verification problems similar to what the RAM Connection folks told you. However, I would hope that we would follow this up by e-mailing some specific example models based on some AISC published calculations.... maybe even with an e-mail summary of the hand written notes we have about any discrepancies between the AISC calcs and our calcs.