Safety Factor

[BearcatEngineer]

I have a general question regarding safety factors. If you have sufficient test data for products/structural elements to failure, what value of safety factor would you recommend? I know some are dependant on the mfg., but if no recommendations are given, it is acceptable to us a factor of safety of 2, similar to the safety factor used with ulitimate loads in steel design?

 

[UcfSE]

You should check in the building code or national design standard for the needed factor of safety. What is the type of product to which you are referring?

 

[BearcatEngineer]

It is a walk in cooler roof panel. MSR lumber with 26 ga. steel skin on each side. I have analyzed the system as a composite panel system, neglecting insulation material, and have determined allowable stress has been satisfied. I have been provided with test data for these panels as well and I am around a 2.88 safety factor against failure for the loading I am reviewing. I feel more than comfortable with this since test data is available, but I just wanted to see if there are any rules of thumbs or industry standards to go by.

Thanks.

 

[Lutfi]

It depends on several factors such as:

1. Fatigue considerations
2. Use/application
3. Product and material
4. Current code requirements
5. Who did the testing (self serving testing is not very credible in my opinion unless it is done by independent third party)

You can always use engineering judgment. However, I would not depart from normally accepted practices. Few points to ponder.


Regards,
Lutfi

 

[JStephen]

Also depends on consequences of failure (ASME pressure vessels currently use 3.5, formerly used 4.0). Depends on the uncertainty in the applied loads and in the tested/calculated strength. Depends on economics.

 

[DaveAtkins]

In reality, I believe the safety factor is a function of the coefficient of variation of the test data (any statistics whiz out there?). Hence, products with a high degree of variability, such as expansion bolts, have a high safety factor. But in your situation, if you are comfortable that this system would REPEATEDLY give you the same result if tested, then 2.88 should be OK.

DaveAtkins

 

[pba]

It comes down to what is reasonable. This is a personal judgement as none of us here has the full information.

There are some questions which may help to inform on the correct judgement:

What is the consequence of failure?
What is the certainty of operation?
Is there a danger to life and limb?
Is there likely to be on-going deterioration?
In establishing a unique factor of safety, are any industry guides being overlooked?

Now as DaveAtkins correctly points out, factors of safety are based on probabilities of exceedence. Limit state concepts (LRFD)utilise partial safety factors on materials, workmanship, service conditions and loading (probably there are others). These all combine to give you a single overall factor of safety. Other design approaches make the assessment of factor of safety more difficult as the individual factors are hidden. It is generally understood that the actual safety factor is higher than the one which is calculated as part of a design. As an industry we are certainly not perfect, and accidents happen far too frequently as a result of incorrect design / detail / material / installation etc.

I would propose that the original question may be somewhat misplaced. It might better to investigate if, through design of the panel, some of the other variables affecting safety could be reduced or eliminated.