Imminent Failure Criteria For Steel Beams

[jamie2000]

I am in the process of evaluating a roof that was designed in the 60's. The roof is constructed of steel bar joists supported on steel beams and columns. As far as I can tell, it was not designed for drifted snow, and the flat roof design snow load increased by 20 psf since the original design. My calcs indicate that the steel beams around a high roof area (drift area) are stressed beyond 0.60xFy. There is no visible sign that the beams are overstressed (i.e. no sagging was noticed, etc). I spoke with the local code enforcement officer and he said that typically they do not enforce roof structural upgrades unless the structural engineer determines that there is a possibility of "imminent failure". My first assumption was to use the 1.0xFy as the point of "imminent failure", but then I decided to go to 0.85xFy to leave some additional factor of safety. Does anyone have any other suggestions? Some criteria for bar joists would also be helpful.
Thanks

 

[Ron]

Jamie2000....great question!

Obviously when you exceed the established criteria you are "plowing new ground!" and opening up your liability. With that comes the premise that when you are "out there" you had better be right! I applaud your willingness to think outside the box and not just give the answer that it must be stengthened, without first considering its present capability.

For the beams, I would probably look at the 0.75fy to 0.85fy range for a little added safety against excessive deflection and progressive collapse. The predictability of the properties of rolled steel shapes is, in my opinion, sufficient to allow this. I would also carefully look at all the connections and interactions with other members/systems. One little screw-up here and you get "stress chasing its path of greatest resistance", thus leading to overload in unanticipated areas.

For the open web joists, I would not be so gratuitous. Open web joists are "empirically designed" and fabricated, so exceeding their load tables is risky. Keep in mind that welds on open web joists do not meet AWS standards and with that comes less true predictability in the joist's response to load.

Finally, is the code requirement in your area reasonable based on historical data? Have you prudently checked weather records and considered that the snow density in your area coupled with the intensity and frequency might not be appropriate for your specific locale? (whether high or low!) I think this level of research is necessary for self-protection in the event of problems later on.

One more aspect....document, document, document everything you are doing in this endeavor.

If you decide to leave in place or unmodified, I would suggest posting the structure to warn against modifications or additions in the future that might affect your premise. Further, you might want to issue your report as a matter of public record with the building department.

Good luck.
Ron

 

[JAE]

Jamie2000:

Concur with Ron and also a few other comments:

The Fy of your beams may be actually higher than 36 ksi, or whatever you are basing your calcs on. We have, in the past, taken coupon samples from beam flanges (near the ends of course) and tested to verify Fy. Many A36 steels reach Fy = 48 ksi or higher. This would possibly explain why you saw no permanent sag in the beams...they never reached their yield limit.

Also, on bar joists, we usually develop the shear/moment curves of the actual loading and compare those with a shear/moment "capacity" curve based on the bar joist size and span and the appropriate load tables. Keep in mind that, for older joists, the center portion of the web was typically fabricated for 50% of the end reaction instead of coming to zero at mid-span. More recent SJI specs call for only 25% of the end reaction. Check out the SJI 50 year catalog for the best estimate of what actual joist you're working with. We usually only feel safe exceeding the capacity by 5% or so (see Ron's reasons above).

Or....alternatively, you could get lucky and find a joist tag wired to the end of the joists. Some fabricators identify themselves and the joist size or project number. Many times we contact the particular joist supplier with their tag number and find them willing to run an actual design check of the joist based on my supplied load diagrams.

Finally, just to re-iterate Ron::::: connections connections connections....check them against the new loads.

 

[Gourile]

Hi!
Concured with whatever Ron and Jamie2000 said, I would like to suggest a load test on the structure with close eyes on sags and connections. Before that, as everyone else suggested, watch connections for any sign of corrosion or wearing.
Note that Fy reported as the characteristic figure of a material, say A36, is a statistical phenomena, from which 99% of specimens perform a higher yield strength. It means that in 99% of specimens that might be taken from a specific heat in the rolling mill, the actual Fy is higher than what is typically reported as yield strength. For the actual flocyuation from the Fy, in a statistical form ofcourse, contact the producer. You may afterwards conclude that ,say 95% of material, possibly show yield stresses above 1.2Fy. Then you may perform a risk analysis to judge scientifically about the integrity of your structure and say whether or not it should be strengthen.

 

[Taro]

As others have mentioned, there are many ways to sharpen your pencil to determine the likelihood of collapse (more accurate material properties, historical snow load intensity research, more sophisticated plastic collapse mechanism analysis, etc.). However, I would guess that the building official does not REALLY understand what "imminent collapse" means when he/she is talking about retrofit design criteria. I suggest that you make a reasonable effort to analyze the system accurately without undue conservatism, but do not reduce the factor of safety excessively. It's there for a reason.

 

[TowerEngineer]

Just a couple comments to add to this discussion. We typically use an allowable bending stress for beams with full lateral support: Fb = 0.66Fy for Strong-Axis - Compact Section designs.
In general, beams will not fail until a full plastic hinge has formed creating structural instability. The plastic strength of most steels (e.g., ASTM A36) is about 12% higher than the yield limit.
Therefore, the typical factor of safety designed into beams subjected to Strong-Axis bending and whose sections are compact is: F.S. = 1.12/0.66 = 1.7.
You can combine this F.S. with an actual yield stress that is somewhat higher than the specified minimum if you can perform a coupon test or have Mill Certifications indicating an actual yield stress that exceeds the minimum.
So, as you can see, you may indeed have a very large factor of safety with respect to any particular beam failure. How much you can encroach on this factor of safety is up to you. One particular guideline is a new FEMA publication for the evaluation of existing building structures. It is available for free from FEMA.
You should be careful with respect to roof snow loads in that these can turn into small water dams that tend to concentrate loads in the weakest portion of the roof. That is, water will tend to pond in the location where the roof deflections are the greatest. A very well known stadium roof failure occurred in Denver and in other buildings (post offices, super-markets, and schools) due to water ponding in the last 30 years.

Remember: No amount of planning and forethought will ever replace dumb luck.