Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations GregLocock on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Buggy Software

Status
Not open for further replies.

dozer

Structural
Apr 9, 2001
503
Is it just me or is the quality of engineering software going downhill? We received an update to a popular(?) FEA program last week. With hopes high I installed the update and promptly imported a solid model that came from a CAD package.

I started to sketch something on the part but something looked out of whack. I measured an edge on the part and it was way bigger than it should be. I double checked my work and became convinced it was the software, so I called technical support. Sure enough, they have a bug in the import routine. It seems that an inch in the solid model comes in as a meter in the FEA package. (Is anyone thinking Mars probe?)

Here's my question about overcoming obstacles in the workplace. Is there nothing we engineers can do to stop this crap or at least slow it down?? I don't know about y'all be we run into crappy bugs like this all of the time. All we ever get is how complicated these programs are and you've got to expect some mistakes. I would keep prattling on but I've got to go.
 
Replies continue below

Recommended for you

Most of the times these bugs appear because of poor management...

Either high management gave the customer impossible deadlines that the software designers couldn't keep and something missed checking.

Or maybe the software designer didn't have all the information to begin with...

Another common mistake is that in the middle of software development, high management changes one little thing that wrecks everything done to that point.

Or it could be poor software writing and debugging skills.
I would bet on the first three as most likely to occur ;)
 
Normally the units used in FEA are decided by the user and material properties chosen to suit. It sounds like the software developer has tried to develop a piece of software that requires no thought, but a simple click of the mouse. The problem lies in that people no longer expect to think. Mistakes are inevitable.

corus
 
Mistakes are inevitable, but if my car crashed as often as my lotus notes or my excel, I would have dumped it long ago! This is definitely an obstacle to get work done.
 
You definitly made a point there. Cars are as complex or more as any software, because they are a mix of software and hardware.

But the caring, knowledge, testing, and more testing that cars are submited to makes a difference.
Software on the other hand is made and tested only once or twice... And not all conditions are verified, or if verified, they are by people that don't know the purpose of the software.

Software writers can build programs for chemical or other engineering branch, but they don't have the knowledge to tell if all the possibilities came as expected.

Another thing might be the fact that, a car is intended to last, ideally, through a lifetime. Software isn't. :(
 
There is no mechanism in software that wears out.

There is no mechanism in software that behaves in a nondeterministic fashion.

There is no reason on God's Green Earth why computers should have, or need, a Reset button.

Well, except for Billy, who always sells stuff before it exists, then hires people to produce it, rushes it out the door, and induces every computer user in the world to work for him, and pay him for the privilege, for at least a few minutes a day, cleaning up the mess he made.

Nineteenth Century Robber Barons were putzes.




Mike Halloran
NOT speaking for
DeAngelo Marine Exhaust Inc.
Ft. Lauderdale, FL, USA
 
My boss gave a gear analysis problem to an FEA type. The root stresses were comparable to Lewis equation results and other gear specialty programs.

But the guy had difficulty modelling the load interface. When the deflections were magnified in the graphics, the interface appeared to be a point load instead of a line or narrow rectangle. What a waste.

FEA was prescribed by QC types in the QS9000 system.
 
FEA is more desireable to hand calculations as it tends to show secondary effects that hand calculations miss. A point load between contact surfaces isn't unusual when you're using elastic analysis. An inelastic analysis and a finer mesh may have shown a slightly larger area of contact.
Far from being a waste the analysis confirmed the hand calculation approach and also assured engineers that there were no other secondary effects to be worried about. FEA gives confidence in the results and design, engineers using F/A doesn't inspire confidence at all.

corus
 
I forgot to mention that there is no mechanism like intelligence in software, either. Apply the wrong load condition, you get the wrong answer, consistently.

Giving a powerful tool to someone who doesn't understand the tool or the problem the tool is intended to solve will produce unfortunate consequences. Lawyers at the defense table would call them unforeseen consequences, but engineers know better.

Introduce a toddler to a running chainsaw, and you should expect blood, not stump sculpture.




Mike Halloran
NOT speaking for
DeAngelo Marine Exhaust Inc.
Ft. Lauderdale, FL, USA
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor