Space Shuttle fatal event risk

If the remaining fleet of three Space Shuttles continue to fly until the end of their operational lives, there is from a 40% to 99% chance that there will be at least one more fatal event involving a Space Shuttle mission.

From the inception of the Space Shuttle program in 1981 to the loss of Columbia in 2003, there have been a total of 113 missions and two fatal events. Two ways to estimate future risks would be to use a formal engineering or scientific assessment of risk or to use an estimate based on observed data. While there have been many formal assessments of Shuttle accident risks, there has also been wide variations in estimated accident risks. Two representative examples are one from Dr. Richard Feynman, a member of the Rogers Commission that investigated the Challenger accident, who estimated the likelihood of failure as about 1% per flight [1]. A second example is from a 1996 report from the U.S. General Accounting Office that identified a NASA-sponsored study that had a median estimate of a catastrophic shuttle failure of one in 145 launches [2].

One could also take the experience of the first 113 flights to come up with an estimate. Because the last resulted in a fatal event, it may be overly biased to include that outcome. If one were to err on the optimistic side and exclude the 113th mission, one could estimate the likelihood of a fatal event as one in 112. If one were to err on the conservative side and include both events, the estimate would be one fatal event per 56 missions.

The formal assessments and the basic estimate using observed data are varied, with about a factor of three difference from the lowest to the highest. One could take the extreme estimates, one fatal event per 145 flights and one in 56 flights, and use a basic probability model to estimate the risk of a future fatal event.

Risk Model
Risk is defined as the combination of a specific hazard and the likelihood of that hazard. The specific hazard in this context is a Space Shuttle flight that results in at least one crew fatality. The likelihood of this event was estimated to be between one fatal event per 56 flights and one fatal event per 145 flights. A simple probability model is the binomial model, where the likelihood of the hazard is constant.

Assumptions and Estimates
The likelihood of at least one more fatal event would simply be (100% - the likelihood of no fatal more events). The following table presents the likelihood of a least one fatal event for a combination of two estimates of the probability of a fatal event on each flight (1/56 and 1/145) and two estimates of the total operational life of the shuttles. There are three remaining Shuttles and those shuttles have flown 76 times. If it is assumed that each orbiter will fly no more than 50 times, then there would be 74 flights in the remaining lifetime of the fleet. If the assumption is a 100 flights per shuttle operational lifetime, then 224 flights would remain.

Table 1: Likelihood of More Fatal Events as a Function of Remaining Fleet Life and Fatal Event Probability
74 Flights 224 Flights
1 per 56 Flights 73.7% 98.8%
1 per 145 Flights 40% 79.8%

[1] Feynman, R.P., 1988, What Do You Care What Other People Think?, Appendix F.
[2] Science Applications International Corporation, 1995, Shuttle Probabilistic Risk Assessment, Washington, DC, Center for Aerospace Information

Space Shuttle fatal event risk -- Revised: 15 June 2015