Considering Fatigue in the Safety Factor of an Analysis

[BryanS]

I am trying to determine how the industry handles fatigue when designing to a pre-determined safety factor. If I want to design to a SF=3, it would seem purtinent to ensure this safety factor exists over the full life of the system. In order to determine this, you would need to know that the failure stress of the material after being loaded by a lower stress (<1/3 UTS), 1E6 times. For example you would end up knowing the following: After being loaded to 1E6 to 10ksi, the failure stress drops from 36ksi (static UTS) to 30ksi thus resulting in a SF=3 (30ksi/10ksi) over the life of the system. I've derived a way of obtaining conservative approximation of this from S-N curves, but I'm quite sure this isn't generally how it is typically handled in the industry. Does anyone have a basis for how fatigue is handled in the industry and possibly a resource to back it up? Some examples I've seen treat fatigue independently of the system safety factor - as long as the system doesn't fail in fatigue over the useful life of the system, the designer is happy. It seems that this is overlooking the safety factor of the system near the end of the useful life of the system. On the other hand the information is not readily avaible to obtain the numbers that I mentioned above.

 

[davefitz]

The safety factor is handled a number of different ways .

In ASME section VIII div 2 the safety factor is 2 on stress or 20 on cycles , plus the base values are based on 2 standard deviations worse than average properties, but based on unnotched specimens, parent material. While this might seem like a high safety factor, in real life the actual failure likely occurs at the heat affected zones of welds whcih typicaly have pre-existing microcracks , so it is not that conservative.

In the European Union PED, they recognize the likely location of failure is the weld HAZ and account for this in a manner similar to the BS. IN those codes which assume failure in the weld at a preexisting crack, then the failure mode is actually crack growth and not fatigue per se. Yu can freely choose the safety factor in those design codes.

 

[BryanS]

Thank you davefitz for the reply. It is interesting to see how this is handled in the pressure vessel standard. Do you have recommendation on how to correlate this to the ASME B30.20 standard (Below-the-Hook Lifting Devices)? The only design guideline that I can find presented here is: "A lifter shall be designed to withstand the forces imposed by its rated load, with a minimum design factor of 3, based on yield strength for load bearing structural components." This is not a design standard and thus does not try to address the nuances associated with this statement. Are you aware of a design standard that should be used in combination with the one listed above?

 

[davefitz]

I am not familiar with that code, but if the stress is limited to 1/3 yield then there is no significant fatigue damage. If you acces the s/n curves with an alternating stress range of 1/3 yield I am sure you will find no fatigue issues.

 

[boo1]

The way we handle this issue, is by using the design strength at the estimated cycles, then apply the appropiate SF. This decrease the premature failure will happen in high cycle loadings.

I will review the ASTM code Monday when I am back in the office.

Cheers

 

[boo1]

ASME B30.20-2003 20-1.2.2
"The load-bearing structure components of a lifter shall be designed to withstand the stresses imposed by its rated load plus the weight of the lifter, with a minimum design factor of three, based on yield strength of the material, and the stress ranges that do not exceed the values given in ANSI/AWS D14.1 for the applicacable condition. ...."

This paragraph determines the maximum allowable stress for materials used as load-bearing structural components. Analysis or actual max working stresses must be determined based upon working conditions including impact or other factors pertinent to the application. Actual max working stress must be equal to or less than the allowable stress.

The intent of Section 20-1.2.2 is that the load suspension parts of a lifter shall be designed so that the static stress calculation for the rated load shall not exceed 33% of the yield strength.

 

[BryanS]

Thanks boo1 for the reply - I concur with your evaluation. Unfortunately those who wrote ASME B30.20 and if I'm not mistaken ANSI/AWS D14.1 did not include a vast array of building materials that may be suitable for the applications at hand. Using a low to medium carbon steel with a SF=3 shows no real fatigue concerns, thus it is not otherwise addressed. What about when following the CE requirements of 2:1 based on failure and using a variety of aluminum alloys? I'm looking for an industry standard to handle these other types of inputs with regard to fatigue that we can use rather than my personally determined conservative techniques.

 

[boo1]

The carbon steel allows endurance timits depend on the steel microstructure. Typically Se~0.5*Sut Ref Mechanical Engineering Design, Shigley eq 7-2.

Aluminiums orinarly run from about 30 to 40 % of tensile strength.

I have beed involved with several failure investigations where the designers used the static strength/sf in fatigue applications. The prudent designer uses the design strength at the estimated cycles, then applies the appropiate SF.