Gs. Szekely et al., PRACTICAL PROCEDURES FOR RELIABILITY ESTIMATION OF SPACECRAFT STRUCTURES AND THEIR COMPONENTS, AIAA journal, 36(8), 1998, pp. 1509-1515
The analysis of structural systems is mostly based on the assumption t
hat the parameters are fixed, i.e., deterministic values. This contrad
icts the experience that the structural parameters often may vary (ran
domly) considerably. A way to model these uncertainties is to apply pr
obabilistic methods. Two concepts are presented: The first is an engin
eering tool, which basically performs (weight-controlled) Monte Carlo
simulation strategies using the capabilities of highly developed finit
e element programs, e.g., ASKA. A second concept is the stochastic fin
ite element method using random fields that, combined with the respons
e surface method, is a versatile tool to calculate the probabilities o
f failure. This approach is coded within the software package COSSAN,
a tool for computational stochastic structural analysis. The feasibili
ty of the concepts is presented using examples from spacecraft design,
e.g., the influence of randomly distributed parameters on serviceabil
ity of devices mounted on satellites or the influence of randomly vary
ing thickness on the reliability of an axially mounted fuel tank. Both
concepts are shown to be practicable, especially for the design engin
eer.