Fatigue Analysis - FAA Report 1

[NGone]

I'm new in the forum and I congratulate for the helpfull of your interesting thread.
I'm readign the FAA report "Fatigue evaluation of wing for small airplanes" (see

http://www.abbottaerospace.com/down...aluation_of_wing_etc_for_small_airplanes.pdf)

and I've not great experience with fatigue analysis.
The report is pretty clear but I can't understant the use of the first part of Table V (Appendix page 7) for the next calculations and I can't understand how the values in the table have been calculated.
In other words I've understood that 9650 and -900 are rispectively max and min stress of the mission but I can't understand how they have been obtained and how the values

.73, .105, .165, -, for +9650
.037, .007, .336, .59, for -900

have been calculated.
Moreover in the report "Load and Stress Spectrum Generation" (see

https://smartutsa.files.wordpress.com/2014/07/load-and-stress-spectrum-generation_ld.pdf)

at paragraph "Ground-Air-Ground Loads" is showed one more procedure for obtaining GAG stress but again I can't understand how the two curves (in green and in red) have been obtained.
Thank in advance for your help

 

[Dougt115]

They state that the 9650 and -900 are measured values, page 4 section 3. The other numbers are a percentage of the damage caused by each phase, landing, taxing, and so on and are somehow derived from the figures. But they don't explain how.

 

[NGone]

Hi Dougt115 and thank you for your reply! I wrote the tables of the example into an excel so that can be read more easily.
My doubts remain because the report in the Appendix, Page 4, Par 3, says "The GAG cycle is obtained by developing cumulative frequency curves for positive and negative stress exceedance from all sources and then determining the average exceedace per flight. The overall cumulative frequency curves are obtained by adding the cumulative frequency curves for gust, maneuvers, landing and taxy." but the problem is exactly the development of these curves...
The second report I attached last time "load-and-stress-spectrum-generation" shows some curves (Pages 10/12 and 11/12) but even here it is not clear how to obtain them.

 

[realtime2]

Hi NGone,
This goes back a bit but one of the first papers on recording flight data using accelerometer
data was : J.A.B.Lambert, "The Use of Counting Accelerometer Data in Fatiuge Life Predictions
for Aircraft Flying in Complex Roles," Fat. Life Predictions for Aircr. Struc. and Matls., AGARD-LS-62, 1973.
If I remember correctly in those days the accelerometer voltage signal was "tracked" by counters
where each counter added a 1 whenever the voltage exceeded its monitoring level. Thus
counter 1 at 1 volt, counter 2 at 2volts, 3 at 3volts etc, and a similar set of counters
on the negative side of the voltage (or acceleration) scale.
Thus a signal that rose from zero to 5.1 volts would trigger a count in all 5 level or
"exceedence" counters. When the history input is done the counts are plotted on semi-log paper and joined
together by lines. Your reference only shows the lines and not the individual counting levels.

In those figures the counts have been normalized on a per flight mile (by dividing by the total miles, probably)
The way to reconstitute the history from these summary curves is to use a "blocked" approximation of both the
+ve and -ve curves. The counting method is also described in the "SAE Fatigue Design Handbook," 3rd edition
AE-22 1997 pg. 141. The method is often called "Level Crossing Cycle Counting".
The F.D.E. Comm. of SAE lists a level counting program (run on a digitized signal) here:

http://fde.uwaterloo.ca/Fde/Software/lcross.f.txt

The present state of the art does not use Level Crossing Counts generally. The counting method,
although simple in concept, can make mistakes. Noisy signals are a problem, for example. To get around
that the method needs to set a "reset" level, before the counters are activated again.
Presently the best method to count cycles for fatigue evaluation purposes is the "Rainflow Cycle Count",
although it too is not perfect. It attempts to resolve closed stress-strain hysteresis loops as the
"fatigue event". Several versions are available on-line.