Load testing

[onteng]

We have completed an analysis of an existing pre-eng building built in 1975. The frames were overstressed by 150% under unbalanced loads. The purlins were overstressed by over 100% under normal snow load and finally the girts were overstressed by 100% under wind. Once reported to the client, two separate engineering firms completed peer reviews, both agreeing with our conclusions.
The problem is the second firm recommended a load test. I have been asked to comment on their procedure. How is it possible to load test such a building without failing the structure or if local testing is done how you prevent adjacent frames from contributing to the resistance.
I have talked to other engineers who all agree that load testing does not seem reasonable. Your comments would be appreciated.
Onteng

 

[scottiesei]

I would be interested in what the cost difference between repairing and load testing the structure is.

 

[SlideRuleEra]

I agree that a meaningful load test is unreasonable. Being a pre-engineered building, I assume it is primarily constructed from steel. Since the allowable bending stress was 66% of the nominal yield strength (per AISC Manual of Steel Construction, Sixth Edition), then testing to the calculated stresses (2 x 66%, or more) should cause yielding (but not collapse). Of course yielding would most likely make the building unusable.

Your insight into the complications presented by have unloaded adjacent frames is true. That is one of the times when engineering judgment on the interpretation of test results in essential. The true precision of load test results is low, it makes slide rule accuracy (3 significant figures) look good.

http://www.SlideRuleEra.net

 

[JAE]

The only way a load test would work on a thin-engineered metal building would be if the engineer somehow knew or suspected that there was excess strength in framework that would allow loading up to a desired "ultimate" level. For these types of buildings its almost given that they were or are designed to the nat's ass and as SlideRuleEra states, it most likely would go past yielding.

Your separate analyses also suggests that a load test would be fruitless.

 

[Ron]

The only reasonable approach to such a load test would be to do a load deflection curve up to, say, 75% of design and see how well the actual curve follows the predicted curve. Even then, it is difficult to perform a load test on a pre-engineered (don't ya just love that term...pre-engineered...as if other buildings were not!)steel building.

 

[GregLocock]

At uni I compared predicted and actual load deflection curves for single welded trusses. Off the top of my head the results were Ok, within say 10%. But that is for a single,welded, truss with a point load and defined terminations.

If you add multiple members, bolted joints, foundations, missing bolts, etc etc I think you would be very lucky to get within 20%.



Cheers

Greg Locock

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

 

[onteng]

Thanks for all the replies.

If the problem with load sharing is eliminated, I still see problems. The webs are very thin with no web stiffeners and I am concerned that they buckle and cause collapse. I one frame collapses, it may pull the other frames down.

At this point I am not sure how they expect to load the frames and keep the techs safe during the test. I don't think I want to be in the building when they do this test.

We have had discussions with one engineering firm who indicated hundreds of thousands of dollars for testing. This does not include any costs for replacement frames once they yield. Our budget for reinforcing is about $1.8M.

 

[Ron]

onteng....the cost estimate for the load test is probably in the right range. I've done load tests that exceeded $100k.

I have gone to remote monitoring of load tests for safety.
We used to set dial indicators and read each one as the load was applied, often being under the load with catch shoring (set catch shoring to a level below the expected deflection x 2).

Since about 1992 or so, all load tests I've done have been done using remote sensing and strain gages/electronic dial indicators connected to laptop computers, where monitoring can be done outside the loaded area. Loading can be done hydraulically, either pushing or pulling on the members or bents.

In any load testing scheme on a building, it is important to be able to instantaneously drop the load if failure is imminent. That is often much more difficult to do than to apply the load.

 

[scottiesei]

Are you friendly with the firm that suggested the load test? Maybe it is time to discuss with them why they felt that would help. It could be that it is just something standard that they typically say to limit liability. Although it would probably be pretty interesting to load test the structure, it sounds like a definite waste of money with members being overstressed by %100-150.

 

[onteng]

Ron, thanks for the testing input. I will discuss this with the company overseeing the testing.

I had discussed this testing previously with the engineer from the firm who recommended it. At that point, he felt the load testing may not be practical. I think the final say may have been taken out of his hands.

I thank everyone for their contribution. My reservations to testing have been reinforced.

This project is a political issue and I just want to be out of the way when this thing blows up, so to speak.

 

[dik]

A couple of comments:
Load testing should mimic that various load conditions the building should be designed for.
Hydraulic load cells work well because the load falls of quickly on pending failure.
Digital 'dial gauges' as Ron noted with laptops; I've used both these and LVDT's for distance measuring.
Pre eng buildings are generally designed elastically but are contoured such that yielding (and plastic collapse) can occur at many points almost simultaneously.
You might want to look at "shakedown pheonmena' for plastic structures. It's possible for parts of the structure to yield and that under a different loading, the residual stresses causing a collapse load at a smaller load than anticipated.

Dik