Lessons learned from a sample of 20 launch vehicles failures
Motet, Andre
Arianespace

A sample of about 20 European and French launch vehicles (or equivalent) failures dating back to the mid-sixties was analysed. For each failure case of this sample, the technical inquiry allowed to locate, beyond any doubt, the piece of HW or SW which failed and led to the loss of the launch vehicle. Then, changes were successfully applied either to the design or to the manufacturing and acceptance process. However, no useful lesson can be learned from the technical inquiries alone. That is why, in a second step, each failure was submitted to a ”quality” analysis in order to find out its root cause. The objective was to derive lessons which could be useful to improve launch vehicles reliability. Root causes can be classified in a very few categories. The three most important, representing about 75 % of the sample, are:

Qualification tests on ground did not represent flight conditions

The design had no margins and the qualification process (by analyses or by tests) did not allow to demonstrate margins (or lack of margins)

The qualification process did not cover the whole domain of operation
Even though the size of the sample is rather small, three important lessons can be learned from the “quality” analyses. For more than 25 years, experience has shown that, whenever the guidelines derived from the aforementioned lessons were implemented in the development process, the probability of success of the qualification flights as well as the first operational flights improved dramatically and the asymptotic reliability was reached much faster.