WIKI - Modeling of Real world structural engineering problems

[slickdeals]

Folks,
I was wondering if all the excellent members of this community with sharp minds and gray hairs of wisdom would be interested to contribute in a project that I am thinking of undertaking. It is just a vision of mine and I want to use this forum as a sounding board.

My graduate advisor used to tell me "Nothing can generate wrong results faster than a computer". I see this happening every day at work. Some of it due to engineer's similar to my age not wanting to question the validity of a computer answer, and some of it just to do not knowing how to accurately represent a building in a finite element program.

I propose to create a structural modeling WIKI that would have input from structural engineers on how to model simple things such as slabs, walls, moment frames, panel zones, braced frames and simiar. Things that would help an engineer to have a starting point.

I am sure there is proprietary information on modeling certain complex things that companies typically don't share, but I am sure as an engineering community we can only help watch each others backs.

I have a checklist for structures that I believe was created by Ohio Board of Professional Engineers that was given to me in school, which has a bunch of items to check off to ensure that run-of-the-mill stuff has been accounted for in design. I would post it online soon and you may want to help me to update it or add more pertinent information to it.

I would much appreciate it if anyone wants to offer me feedback or if anyone else is interested and willing to spend some free time to put together a community that helps engineers to understand and question the validity of computer generated results.

Computer programs are Garbage In Garbage Out, I wish all you folks will help to prevent Garbage from going in.

Let me know what you think.
Best Regards.

 

[kslee1000]

You have very good point, I think will be support by manys.
Due to the increased capability in computing and speed, the program has became more sophisticate and complicate, also, the project itsself has became larger and much more complicate. The brain in a typical structural office has been consistently shrinking while assignment increasing, all of above causing the engineers to lose grab on validity of modelling vs basic engineering fundamentals, and lose sight on how structures behave, or are expected to behave.
I wish you succeed in going forward.

Here is my thinking about computer, and old days practices:

- Computer = Calculators, just that, no more.
- Calculator
Before punching number into calculator, an equation is writen with variale identified, then the result is examined (does it make sense), and double checked.
- Computer
Before a computer is turned on, goals and rules around the assignment on hand are set, a model coresponding is drawn on paper with assumptions ident

 

[slickdeals]

@Kslee, you are absolutely right with your last point. I don't think one should even begin a computer model without having a order of magnitude of forces to expect after analysis.

The reason for this post is because I had to run numbers to verify a design in which the system self weight was not included in the original design. In RAM Structural System, I believe there is an option to specify the mass (dead) that will contribute in a seismic event. The engineer had no understanding of this concept and applied the system self weight (wt. of 12" joists on 4-3/4" slab = ~70 psf) as a mass load and not gravity load.

The only saving grace was that live load reduction was not used and the structure was oversized by more than 40%. Between LLR and overdesign, my results indicated that the structure was 95-100% of the design capacity and in some cases 2-3% overstressed. It was a BIG DEAL, but thank god for us reviewing it and taking corrective measures.

I hope all you folks will support this vision. Let us help each other to keep people safe.

 

[kslee1000]

sorry, hit wrong key.

...a model that corresponding to underlying assumptions is drawn on paper with all necessary geometries shown and checked before input. The results are examined against expectations, and selectively checked by simplified hand cals. The process is repeat to refine the model until the results are agreeable with the assumptions.

Due to all said previously, I wondering the above way of computer use is still the norm of practice, it takes a lot of time before and after the computer runs, and a lot of efforts to learn your TOOL.