Potentialities of theoretical and experimental prediction of life support systems reliability

To develop and design Life Support Systems it is necessary to evaluate their reliability. However direct experiments take much time, are very expensive, and therefore are practically impossible. Promising way is to use approximate estimates of reliability, which need essentially fewer amounts of experimental data. Two types of estimates of Life Support System reliability - additive and multiplicative ones are considered in the paper. Additive estimate is based on the assumption that total system failure probability is low and therefore it can be considered as the sum of failure probability of separate units. Additive approach allows obtaining near lower-bounded estimate of failure probability. Multiplicative estimate allows evaluating the possibility of system catastrophe due to simultaneous effect of several factors when each of them separately is not dangerous. Evaluation shows that the possible error of reliability forecast increases with the increasing of number of external factors faster than exponential function. An illustration of the ecological similarity approach as promising tool for providing estimation of full-scale system reliability by means the set of small similar experimental models. (C) 1999 COSPAR. Published by Elsevier Science Ltd.