Analysis Software

[RARWOOD]

My company is thinking about upgrading our analysis software. We are trying to decide about using OSI software which has open code that we can modify, or using proprietary software.

Since the advent of personal computers, all the structural engineering firms I am familair with use proprietary software.

What are the pros and cons of OSI and proprietary software?

Also what do you see being used?

 

[dik]

I use an early demo version of RISA on my laptop and use the full version at the office; they're proprietary.

I'm not familiar with OSI, but with open source programs, they can be a mix of very good... or very ugly, it depends on the program. Good open source like Firefox and Thunderbird (Mozilla) is excellent and is kept up to date and bug fixes are also quick. Often open source, if it has an active programming base is full of features and any errors quickly fixed... depends on the program.

Dik

 

[rb1957]

issues to consider ...

accuracy ... prop. s/ware can be reasonably considered to be accurate (within each codes' bounds). with upgrades, a few checks are enough to appreciate the changes and to understnad the bounds of accuracy. with open source, can't anyone fiddle with the coding of the elements ?

repeatability ... since prop. s/ware changes with increments, you can rerun an analysis with the same code (and get the same answers). this would be much harder with open source.

consider a third alternative, where you buy an open source code, or write one for ourself, then maintain it yourself, adding the features you want.

 

[miecz]

On the issue of accuracy. Many (most?) proprietary programs include a disclaimer in the license agreement that limits the liability of the vendor and places the responsibility for the accuracy of the program on the user! It's surprising to me, how many serious errors occur in proprietary software. I think fewer errors would occur if the vendors were held liable.

 

[rb1957]

jmiec, i'd have thought the the codes aren't accepting liability for your (or anyone else's) models. my concenr was some fiddling (possibly with the best of intentions) with the formulation of an element, which might not suit your purposes.

 

[RARWOOD]

Thank you to all the contributors to this thread. rb1957 one of my concerns is, the problem that anyone can fiddle with the code. Even though they have a good intent when they make their changes, as a desginer I may not agree with their intent.

Our current software modified in house instead of reducing the allowable stress where required, the program increases the moment so that the output shows a element stress as it relates to the unreduced allowable stress. However there are times when I need the true moment value, which I have to back calculate using the reduction factors.

The above practice was followed when the program was modified based on a long standing design practice in the office. Personally I want to see the actual stress compared to the reduced allowable.

One problem with using open software is the staff required to keep it functioning and up to date. Our engineering staff prefers to rely on an outside company who specializes in engineering software. Our IT people prefer open source software that they control over.