Designing by Hand vs. FEA 10

[jester86403]

I have just been curious lately about what the upper extent people feel is that can be designed by hand or at what point would designing a structure by hand be no longer close at all to using FEA as far as efficiency goes.

I would like to get responses for buildings, bridges, and other miscellaneous structures if possible.

Obviously almost anything can be designed by hand...I mean high rises were built long before computers, but when is it no longer worth the time to do it by hand?

 

[connect2]

not to mention, silos, chimneys, bunks, tanks, piers, wharves ..., and other misc. structures. About covers it.

 

[civeng80]

Its amazing how things have changed over the years in structural work. 25 years ago the steel codes and wind codes were very simple and things like portal frames just took a few pages of calculations to obtain member sizes. Now it probably cant be done by hand because the wind loads vary over the span of the portal. It takes about 60 to 70 pages of computer printout and eventually the sizes come out the same as 25 years ago. I actually did the exercise not long ago. Connections are probably a little better designed now and probably reflect more limit states rather than working stress.

Was it all worth the effort of changing from working stress design to Limit state design? More sophisticated wind codes? I dont know. It seems to me that it only came about because of more powerful computers becoming available. Computers made it happen, not good structural practical design.

Sorry if Ive gone off in a tangent here but these are my thoughts at least.

 

[allisch]

I use programs to model stuff on the complicated things.
Obviously simple beam calcs are by hand, but I always do a crude hand calc on the main features of the comples items to double check my modeling. Many times I have found errors in entering the model by double checking the basics by hand (wrong fixity, axis of member, braced/unbraced, etc) and I use both methods to double check the other.
I also have my own excel sheets that I use to do iterative tasks "by hand" more quickly.

 

[whyun]

Computers can't layout a good layout of the structural system. Only engineers can.

Quick hand calcs should always precede computer models to lay out a starting point.

 

[smb4050]

What a great post for engineers to discuss!

I'm a small firm owner and fall into the category that hand calc's must confirm the computer output. I contend an engineer can't understand a structural system they are designing without doing hand calcs. Just today I put a $(US)30k structural proposal together for an architectural firm for a commercial design which would be a really nice project, but I won't it because I'm way over priced for this reason. The computer is a tool not an engineer. The computer doesn't design structures!

 

[RobertEIT]

Computer can only help engineer, but can not replace engineer.

I just designed a back to back cold rolled form C-channel beam. Using software, it only tells me the yielding failure load is around 19.8 kips uniform load. By hand calcs, I found the failure mode is torsional buckling, the failure load is only 7.7 kips uniform load. I doubt any structural software can replace this kind of engineering judgement.

 

[csd72]

A simple question with a very complicated answer.

As many people have no doubt mentioned above, you should always have a good idea of what to expect before relying on an analysis program.

I would probably divide jobs into simple categories as far as the overall members are concerned.

1. Simple spans and pinned connections - This I would always do by hand because it definitely tends to be quicker. I would even include double spans and more complex loading patterns in this analysis. The only exception for me would be when significant second order moments may be encountered.

2. Framed structures such as portal frames with significant secondary moments - I would tend to do an analysis on the computer and member checks by hand.

3. More complicated fixed structures with significant restraint, indeterminacy and possibly torsion. An example of this would be a complex canopy made in one welded 'lump'. I would either do a quick hand calc to determine sizes and then back it up with a computer analysis or I would do the analysis and refine it to the optimum solution then check the output against hand calcs.

Basically, you should always understand the simplifying assumptions of your hand calculations so that you can make the judgement whether you can live with it. Generally indeterminacy, secondary load paths, restraint, and second order effects are the main reasons why a hand calc would not be accurate.

As far as connections are concerned then it is usually the best approach to us the simplest method, it is easier for someone to check and is also easier to get it right.

Personally I have never used FEM and have only had one or two situations where I thought it would be truly essential.

 

[whyun]

Structural engineering is all about determining the loads first, then selecting the material that can resist the loads. No computer can determine the first on its own.

Computer is useful only when all the load cases have been estimated correctly and all possible load combinations assigned.

I mainly rely on the computer model for "analysis" part. I look deeper into the capabilities and limitations of each structural software for the "design" part to ensure that computer is performing what I want it to perform (via hand calc etc).

 

[GregLocock]

RobertEIT - your FEA model wasn't sophisticated enough. Real (non linear) FEA is perfectly good at handling instability.






Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[csd72]

A simple question with a very complicated answer.

As many people have no doubt mentioned above, you should always have a good idea of what to expect before relying on an analysis program.

I would probably divide jobs into simple categories as far as the overall members are concerned.

1. Simple spans and pinned connections - This I would always do by hand because it definitely tends to be quicker. I would even include even double spans and more complex loading patterns in this category. The only exception for me would be when significant second order moments may be encountered.

2. Framed structures such as portal frames with significant secondary moments - I would tend to do an analysis on the computer and member checks by hand.

3. More complicated fixed structures with significant restraint, indeterminacy and possibly torsion. An example of this would be a complex canopy made in one welded 'lump'. I would either do a quick hand calc to determine sizes and then back it up with a computer analysis or I would do the analysis and refine it to the optimum solution then check the output against hand calcs.

Basically, you should always understand the simplifying assumptions of your hand calculations so that you can make the judgement whether you can live with it. Generally indeterminacy, secondary load paths, restraint, and second order effects are the main reasons why a hand calc would not be accurate.

As far as connections are concerned then it is usually the best approach to us the simplest method, it is easier for someone to check and is also easier to get it right.

Personally I have never used FEM and have only had one or two situations where I thought it would be truly essential.