For an engineer, it's interesting to get the opportunity to find out what went wrong (and presumably less exciting to be on the receiving end of an investigation). Many times, failures are due to a combination of factors. In this case, the fin cracked at a point where stresses were highest, and good weld quality was difficult.


Sample of cracked surface plate removed   Fin model   Stresses in area where cracks occurred


It's also possible to attribute some failures to the combination of lack of experience and lack of control. Structural analysis might look easy, but it isn't! There are traps all the way for young (and even old) engineers. We have seen cases of major reports and expensive projects that had to be dumped because of simple but wrong assumptions that were not checked by a 3rd party.


Plots of even the simplest calculations often look relatively impressive. However it can be disastrous to plot centreplane stresses instead of surface stresses in plates subjected to bending. Or to use wrong stress limits, or averaged stresses when they shouldn't be. Or not consider panel or stiffener or frame buckling.


Simple model of a plate with stiffeners   Deflections in the plate   Bending stresses in the short edges   Bending stresses in the long edges