Is this ok for design?? 1

[phatso86]

hi, i wanna make this short
i work for a company and it seemed to me that the way they work things are quite inefficient

i attached a picture of what they have..
looking at it, it seems VERY basic; in reality its even a 2D frame

anyways, i wanted to know if this is sound engineering..
i was thinking of using a finite element program using an estimate on what the member sizes should be. I plan on running the case twice, once with aasumed pinned connections and once with assumed fixed connections. I planned on taking the worst case for internal stresses and use that to size the members accoringly. Then maybe repeat analysis until i can get a fairly decent size on members.
Then finally in the last run, i want to take the worst reaction moments and using this for the sizing/spacing of connections (bolts, welds, etc.)

as for boundary conditions, i wanted to take the worst reactions out of all the cases and use that for design of footer/piles/ etc.

can anyone tell me if this is flawed in any way?

thank you

 

[GBor]

It is conservative, but doesn't sound flawed. You will probably need to do more than 2 runs for the trial and error approach to sizing the members.

For this system, I would think you could hand calculate it at least as quickly. Or set up an Excel spreadsheet that would calculate this for you with rapid succession.

 

[EdR]

If you are concerned about efficiency then you should do the design by hand. Even if you haven't done any statics recently this simple truss should not take more than an hour or so to completely do the design (and I'm being liberal with the time)...and if you can't do the design by hand you certainly have no business doing it with software.

By the time you check out whatever software you are using in sufficient detail to have confidence in the results, make and check out the model(s) and then run how ever many runs you have in mind you will have much more time spent than it will take to do it by hand......

Ed.R.

 

[GregLocock]

As drawn there is one redundant member, even with pinned joints.

Analysing the remaining structure by hand with fixed joints would take me longer than an hour, analysing it with pinned joints would take maybe 15 minutes using a Bowes diagram or guessing, as I haven't used Bowes for 25 years.

I suggest you need to think about boundary conditions.





Cheers

Greg Locock

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

 

[civilperson]

The bottom horizontal member seems redundant. Are you trying for minimum cost?

 

[phatso86]

hello again

i would like to thank you for all your imput

Gbor - i meant two runs as in pinned/fixed. I did plan on running it multiple times for the member sizes tho

EdR/Greglock/civilperson - these cannot be done by hand because i need to analyze for dynamic effects as well, earthquake/vibrations/etc.

and i was also thinking of maybe, once i have the member size, creating a 3D mesh of the actual members for my analysis, that way i can get somewhat accurate von mises stresses on the members.

i am not sure how my company has been analyzing this, but it has been taking too long and i wasnt sure if i could do what i have been suggesting, that is to assume fixed conditions and use the moment reactions to design the bolts/welds

i would really appreciate some more input

 

[VibeAC]

I'm not sure what the intended purpose of your structure is, but I would be careful about just dabbling in vibration analysis of complex civil structures; unless you or your company have some experiences/test data that you can use to at the very least benchmark your results.

If you simply want to look at the modal frequencies/shapes of this system that might be fine, but more in-depth analysis of civil engineering structures such as serviceability and assessing floor vibrations due to human activities such as dancing, running, etc., should be done carefully as a poor dynamic design (that functions well statically) has the potential to result in excessive fatigue.

For earthquake analysis, doesn't IBC have a thorough treatment of design procedures?

http://www.iccsafe.org/e/prodshow.html?prodid=9650S00

Sorry to be a downer, but just wanted to make sure this is all put in perspective.

Andrew Gorton, MSc
Noise & Vibration Consultant

http://www.PapadimosGroup.com

 

[EdR]

Just to stir the pot!!!

Label the truss joints top to bottom left side as A,B,C,D and sloping side as E,F,G (top to bottom) then using method of joints....

1. Joint A solves for members AB and AE
2. Joint B solves for members BC and BE
3. Joint E solves for members EF and EC
4. Joint C solves for members CF and CD
5. Joint F solves for members FG and FD

then if both joints D and G are pinned then there is no horizontal deflection between D and G and thus the force in DG is zero (delta = F L/(A E)) and the horizontal forces at joints D & G can be solved using a free body diagram of each joint. If either D or G is rollered then equilibrium of the entire structure gives the horizontal forces....Of course the vertical reactions at D & G are obtained from equilibrium of the structure.....and even though no external forces are shown the given procedure still solves the structure.....

As to being concerned about dynamic effects, etc. you should be able to apply dynamic amplification factors (from the applicable codes, etc. )to get design sizes...

With regard to considering the joints as fixed rather than pinned I think you will find that a maximum difference of about 5% will result from these assumptions...

By the way Greg it took me 5 minutes to write out the solution.......but then I've been teaching and doing this stuff for quite some time.....

Ed.R.

 

[ThomasH]

You say that the company you work for do things very ineffecient. I can't help wondering how they do it.

Regards

Thomas

 

[IRstuff]

Hunting and pecking is NOT efficient. Unless you already have a clear idea about how each element behaves and contributes to the strength of the system, running FEA is a waste of time and money.

TTFN

FAQ731-376

 

[phatso86]

hopefully i won't get in trouble for giving details

but i work for a nuclear powerplant. These are already in place, but they proposed to make some additions (see pic, lines in red)
There are two sets of these, of either 3 or 4.
I know for a fact that nuclear powerplants need to be analyzed for seismic activity (though they are in florida)

There is also a beam (well truss, two total) going across the top point in each of these in the z plane
One of these beams goes across one set of 3 (of the figure i shown) and the other beam goes across the other set of 4.
I'm pretty sure these are all static, there really shouldnt be any movement.
The reson they have these is so they can transport spent fuel casks, so i take it that they somehow go along the truss beam.

anyways, my company contracts other engineers to do the analysis and we look over their work and either give the thumbs up or ask for revision.
I've looked at these calcs and they seem to just pick somewhat random numbers and use that as loads. Perhaps they are not even using the seismic loading. It just seemed to me that my company is spending way too much time dwelling on what the contractor may have done in their analysis. (we have seen this "richard factor" and spent a long time figuring out what that meant)
I just thought it would be easier to, rather than do hand calcs, make a FEA model and analyze it like that.
Then i guess hire someone else to independantly verify our results...

i just wasnt sure if it is a valid assumption to assume fixed connections and design bolts/welds based off the moment/shear reactions at each connection point.
Oh and i wanna point out that the contractor company assumed no moment on the bolt group (seemed like a non-conservative assumption). I understand why they did it, but it still feels unconservative. We are spending way too much money on engineering alone and i feel that we are better off overdesigning components.

 

[phatso86]

there is supposed to be a third pin connection at the bottom right

 

[EdR]

So your real problem is not technical analysis but rather the fact that you don't trust your consulting firm. In that case you need to hire a firm you do trust to do this type analysis or do it yourself. In any case if you are trying to check the analysis the consultants should be able to provide you with justification for the loads they are using and indeed should go over the entire analysis and justify anything that is in doubt. It is much easier to ask questions than it is to spend lots of time guessing what was done.....

As for doing it yourself I'm sure the NRC will accept your feeling that engineering this type of structure is much to expensive and that whatever you come up with is "overdesigned" so much that it is OK.....

These darn consultants are always trying to cheat their clients and never provide useful information for their extremely high fees.....(Just make sure that when you do it yourself that you keep good cost records so you can compare your in-house costs with the cost of the consultant at the end of the project...)

Ed.R.

 

[DRC1]

I am not trying to throw rocks at the OP and I hope phatso86 does not take offense, because my comments are directed more at his firm and the consultant's and the profession in general than at a particular individual.
First this is not a complicated structure and if pinned joints are assumed (and detailed) then it can be analyzed as apinned structure. By hand. Structural analysis existed long before FEA. A Sr. undergaduate engineering student should be anble to complete t6ha analsis. Second if we are modifing the original strucure, we should only analze the modified structure, not the orginal strucure. Finally the main questions appears to be not what are the member stress but what are the applied loads. Until the loads are agreed upon, the analysis is pointless.
The fact that this simple strucure is being reviewed by two firms (I assume at least one is a large firm)designing a nuclear power plant and between them they can not seem to get the horse in front of the cart is very frightning. Even more scary is that this is becomming more common.
Done Ranting.

 

[transmissiontowers]

From what I understand on Nuke Plants ( I have a friend that works at one and another that works for the NRC), the structural software used has to pass rigid QA/QC NRC requirements.

_____________________________________
I have been called "A storehouse of worthless information" many times.

 

[phatso86]

DRC1 - i do not agree with you. A undergraduate CANNOT solve this problem when it has vibration and earthquake motion. Dynamics of structures is a 6000 level course, as is Earthquake engineering along with Finite/boundary elements.
Also, since there has been much research (AISC references this) on just the connection of multiple members on a single node, it should be obvious that this cannot be a simple sum of forces problem.


my point wasnt so much about the member analysis, but rather the bolt/weld group design

but anyways, thanks for the help guys, i'll take transmissiontower's advice and just ask the NRC resident when i get a chance

 

[transmissiontowers]

I design Transmission Towers and we have multiple bolt connections of multiple members framing onto a single joint which we analyze as a pin ended truss. We do modify the L/r of the members for eccentricities and fixed ends based on many full scale tower tests. We also do not use earthquakes and vibrations.

I do use GTStrudl for frame analysis and it is NRC QA/QC qualified. They have a free student version with limited nodes and members but your problem could be solved. Give Georgia Tech a call and ask for the student version for a technical evaluation.

_____________________________________
I have been called "A storehouse of worthless information" many times.

 

[GBor]

Feel free to call GA Tech and get the student version, but if they find out you are using it for a business problem, you are opening yourself up to legal action. Student versions are supposed to be for learning how to use software, and the licensing agreement with which you must agree before installing anything is generally very clear on its purpose...just an observation.

 

[rb1957]

if this is a search for efficiency, why run the model twice ? the gussets are no doubt multi-fastener groups, presumably quite capable of reacting moments. in traditional structures (ie non-nuke) i think these end-moments are typically ignored, conservatively. in your case they probably have a significant affect on the dynamic response of the structure.

if you want to make a "research project" out of this, you could, instead of modelling the joints as fully fixed (whihc is a theretical limit) you could model the fasteners as (shear) springs.

i take it that the long chords are continuous ?

presumably you've seen in your analysis that the lower chord is very lightly loaded; i think all it does is help redistribute the end-moments of the lower elements.

does the structure have fail-safe requirements ?

 

[DRC1]

Phatso86

For the structure you describe, I do not see tremendous complication to solve for static or for dynamic loads such as earthquake, psudostatic loads distributed through this truss. I think the problem is that your team may be searching to answers before they know what the questions are.
If you are looking for the strucures response to a dynamic load, such as the period and natural frequency of the strucure, that is a different question than how loads distribute through the member. Although for this structure, I don't see what you expect to find. Earthquakes can be analyzed in a variety of ways and complexity. However once you detrermine the loads, they are simply distrubeted through the truss. You talk about about a 2-D strucure, with a 3-D mesh and then talk about tying members together with some cross pieces that may or may not effect the analysis. Finally, you are reviewing someonelse's design and want to check it by using your own assumptions and own analysis method. Doubtful you will get any meaningful answers from that unless you are ditcing the consultant's design.
I think your team needs to foucus on what you are trying to analyze and what the consultant has selected for loads and if you agree with that. Then you can proceede to review the analysis. Remeber structural analysis is not the accurate determination of the stress in the structure. It is determining what the maximum apllied stress might be and ensuring that it is less the the minimum allowable stress.
To answer your question, whatever it costs you to detail economic structure, will be far less than the cost of construction. Designing members based on pin connection, and connections based on moment conections would probably be wastfully conservatie. Pick reasonable assumptions and design based on standard details, but first define what loads you are applying to the strucure.