Detail Fatigue Rating (DFR) 3

[Kakalip]

I would like to understand the Detail Fatigue Rating (DFR) in Aerospace application and how its determined? Just thought some one should be able to throw some light on it.

 

[FastMouse]

Hi Kakalip,

Say that you select a couple of fatigue reference quantities. For the sake of example let's use

N (fatigue life) = 200 000 flight cycles, and
R (stress ratio) = Smin / Smax = 0.15.

Smin & Smax are, respectively, the minimum and maximum stresses in a stress cycle.

The Detail Fatigue Rating (DFR) is the just the value of Smax, with R = 0.15, that will result in a fatigue life of 200 000 FC for your particular structural detail. You could calculate DFR for a range of structural details, always using N = 200 000 FC & R = 0.15, and the details with the higher DFR values will tend to have better fatigue performance, ie a longer life.

Furthermore, if you have a quantitative relationship involving R, Smax, and N, then once you know the DFR, you can find N for any value of Smin and Smax. For a complex stress sequence, you could then do a rainflow analysis and a Miner's law calculation using d = 1 / N for each resulting stress cycle.

The method used to derive specific values of DFR for structural details is closely guarded intellectual property within organisations that use a DFR approach and it is unlikely that you will find significant, publicly-available data concerning this. However, the methods are typically the result of a great number of fatigue tests combined with theory and in-service experience. Be aware that design organisations might use reference values of N and R that differ from those mentioned here when calculating DFR values.

For further publicly-available information, for starters you could search for papers by Ulf Goranson who implemented a DFR methodology at Boeing. One of his presentations in which he discusses fatigue damage models and the use of DFRs is available at:

http://www.asipcon.com/2006/06_proceed/Tuesday/0800_Goranson.pdf

I hope that this information is useful.

FastMouse

 

[rb1957]

DFR is part of Boeing's DTA procedure (well i think at leat they started it, and others copied). It is not universally appliciable, like Kt. And even if you have a plot of your geometry vs DFR there Could be some many other factors that the OEM considers that the results could be misleading.

 

[FastMouse]

Apologies for the tardy response!

I personally prefer using a Kt, SN curves, and the pertinent design data to perform an analysis. However, if you include the Kt as one of the input parameters when calculating a DFR, then you should, hopefully, come up with about the same results. Alternatively, you can calculate the DFR based on the peak stress around a hole, and then include the Kt in the stresses when doing the damage calc.

I feel that the DFR approach is best suited to "periodic" structure like fuselage bays, where the stringers & frames can all have similar geometry. You can quite quickly get a lot of justified results. For highly varying structure such as a continuously varying wing spar+skin joint, with different skin and spar flange thicknesses, different bolt diameters, etc, for different spanwise locations, the DFR method stumbles a bit as an individual DFR has to be calculated for almost every separate location, and so the time saving that the DFR approach is supposed to give is lost. Perhaps the DFR approach will be less useful in the future as aircraft structures become more and more optimized, with very few components being identical to their neighbors.

Another weakness with the DFR approach is that it often masks how the structure is working. The temptation, especially amongst engineers just starting to accumulate experience, is sometimes think "Alright! I've got a DFR, let's calculate stuff!", neglecting to consider whether or not the DFR is correct for all the structural details for which the "stuff" is being calculated. Using a Kt with SN curves makes this attitude less likely as each detail needs to be considered separately.

Also, because DFRs are based on a specific R, occasionally everyone gets hung up on the value of R in the actual stress cycle and it needs to be pointed out that if the max and min stresses of the GAG stress cycle both increase, and the GAG damage ratio remains the same, then there MUST be a reduction in fatigue life. You do not need to give a damn about what is happening to the value of R (which is really only a mathematical artifact, not a physical property), it is enough to know that the slope of a S_alt vs N curve is negative. If an engineers are comfortable using fatigue curves rather than a "black box" program to navigate their way through the data, then this is immediately apparent. If they use a DFR with a "black box", then occasionally the fundamentals can be temporarily forgotten.