B737 fuselage large-scale barrel specimen fatigue examination

From the very beginning, an express FE analysis using COSMOS/M 1.52a code was made. The barrel structure was simulated linearly by 1300 3-D elements, including SHELL4 for skin and PIPE for stringers. The stress distribution obtained was used to set up lap joint boundary conditions for linear plane analysis. The rivet stiffness kr =1/Cr (Cr rivet flexibility) was determined using the Swift formula.

The express analysis was also used for crack initiation prediction, when it is planned to apply synchronized combination of pressure and bending load to the second barrel specimen. The general results of the express analysis for both cases of load, are summarized in Table 1.

Table 1. Express analysis general results

Zone
Rivet row
Force on a fastener, N
Stress component, MPa
Hoop
Axial
Hoop
Axial
Shear
Side, Top, Bottom (pressure)
outer
508
0
97.0
44.8
0.0
central
358
0
Side

(pressure+bending)
outer
495
116
97.2
51.9
19.8
central
346
51
Top

(pressure+bending)
outer
495
0
97.2
87.5
0
central
346
0
Bottom

(pressure+bending)
outer
492
0
96.9
3.3
0
central
343
0

Then contact problem for a pin loaded hole of an ordinary fastener from the outer row was solved. Tangential stress curves along hole edge are given in Figure.

This stress component is supposed to be fatigue-critical in the case that the interference fit can�t be guaranteed. Then crack initiation life was calculated (see results in Table 2) using the algorithm described earlier. The material parameters used were those for D16-AT aluminum alloy material that is considered as an analogue one to 2024-T3 alloy.

Table 2. Prediction lives from express FEA
Zone (Fig.2)
Maximum tangential stress, MPa
Life prediction, reversals
absolute
relative
S, T, C (pressure)
382
93,900
100%
S (pressure+bending)
404
72,200
77%
T (pressure+bending)
349
155,000
165%
C (pressure+bending)
395
79,800
85%

 

More accurate geometrically nonlinear FE analysis, as well as sub-modeling procedure, were implemented using an ANSYS54 code. The nonlinear analysis provided better correlation between numerical and experimental results than the linear analysis achieved with. Sub-modeling procedure, also known as cut-boundary displacement method or the specified boundary displacement method, ensures the maximum accuracy for boundary condition transfer available for a given FEA code. The Swift formula was exploited again to get rivet flexibility/stiffness for the lap joint model.

Table 3. Prediction lives from detailed FEA
Zone (Fig.2)
Maximum tangential stress, MPa
Life prediction, reversals
Initial fatigue damage
Crack initiation period
Side, Top, Bottom (pressure)
336
82,270
199,500

Back to "An approach for prediction..."