# What happens next if the root cause of one rocket failure won't be identified?

Supposing that in a rocket failure engineers don't find what the root cause was, what will happen next?

In general if the root cause will never be fully understood is there a law, rule, or standard procedure in spaceflight that tells what to do next for moving forward and starting to operate again? If not and these are situations where the procedure depends by the case, then how will it be managed?

• Many times the exact root cause of an incident is never known. A fault tree is generated and all the branches are run down to closure if possible, but often it is not possible. Any open branches (possible causes) will be addressed, so more than one fix could be applied. Sometimes this process is executed and all the answers are wrong (like the initial try of fixing the foam shedding problem after the Columbia accident, and then there was a giant chunk of foam shed on the next flight). – Organic Marble Oct 31 '16 at 16:39
• How is this collection of six questions not a textbook example of a question which is "too broad" and requiring answers which are "opinion-based"? Somebody (anybody) tell me! (e.g. $N_{COULD}+N_{WOULD}=12$). I propose that an answer considered good by stackexchange metrics can't really exist, looking forward to seeing an attempt! – uhoh Dec 20 '16 at 3:47
• Organic Marble comment was a good start for the answer. But he hasn't explained explained completely. – J. Young Dec 21 '16 at 0:13
• Maybe the question here is simply (roughly) "Does anything like a 'standard procedure in spaceflight exist for moving forward after a disaster who's root cause was never fully understood? If so, has it ever been used?" An answer to that has a more yes/no + example format, the kind of verifiable answer that stackexchange encourages. Obviously with so much upvoting people find this to be a good and interesting question, can you adjust it a bit so that it asks for primarily verifiable fact-based answers? Questions that are 'closed for adjustment' are often re-opened after they are adjusted. – uhoh Dec 21 '16 at 2:24
• @J.Young I didn't expand my comment into an answer because I agree that your question is too broad or opinion based. For example, no one can know what SpaceX will do in the future. If you edit your question down to something like 'how have rocket accidents been investigated in the past and what happens if there is no clear root cause` then it could be answerable. – Organic Marble Dec 21 '16 at 17:45

## 2 Answers

Its hard to answer this in the "typical case" let alone the general case because it will depend upon the manufacturer and customer relationships. Failure review board terms of reference can quite broad and "simply" say that the board must determine the cause and prognosis without reference cost and schedule concerns.

One concept that often arises in commercial, and perhaps wider, circumstances is the "most probable root cause". At this point Organic Marble's description is a good start:

Many times the exact root cause of an incident is never known. A fault tree is generated and all the branches are run down to closure if possible, but often it is not possible. Any open branches (possible causes) will be addressed, so more than one fix could be applied. Sometimes this process is executed and all the answers are wrong (like the initial try of fixing the foam shedding problem after the Columbia accident, and then there was a giant chunk of foam shed on the next flight).

It is quite common in both satellite and launch failure analysis to see many branches of a fault tree given a "not credible" assessment and then to find several attributed to as either "credible" or "credible but unlikely". It may well seem uncomfortable to the outsider but it is not impossible to see lot of money spent on producing evidence against every branch of a fault tree through ground tests and generation of a range of corrective actions and yet the failure review board has only narrowed the range of possibilities for the root cause. Then the "most probable" root cause simply becomes a coin flip between a number of scenarios, all of which happen to be plausible but unlikely. It can be quite unsettling.

If anyone is wondering what I'm adding to this answer myself, it would be that I have taken part in plenty of failure review meetings at equipment/subsystem/system level (mainly for satellites) and I have come to the view that this is a far softer-edged and human process than many people are comfortable with. Satellites and launch vehicles aren't naturally occurring phenomena, every failure comes down eventually a human failing in design or workmanship. Correspondingly every decision to go ahead with a design change also has to be a human made compromise.

If no root cause for the failure is found it should be discussed what to do: another ground test, repeat the launch without any change, add more telemetry and repeat the launch or do more tests during manufacture and assembly of the rocket. The problem is what extra sensors for telemetry might help to identify the cause. Additional high speed cameras inside the rocket may help to understand what is going wrong. But even if the next starts are a full success the team does not know if the problem still exists.