Third Stage Design Problem Cause of Most Recent Proton Failure

schwit1 writes: The Russian investigation into the latest Proton rocket failure has concluded that the failure was caused by a design failure in the rocket's third stage. The steering third stage engine failed due to excessive vibration as a result of an imbalance in a rotor of a pump unit. While it is always possible for new design issues to be discovered, I wonder why this problem hadn't been noticed in the decades prior to 2010, when the Proton began to have repeated failures.

Comments

two envelopes

[roman_mir]

So I read that this problem dates back to 1988 (so they say). Reminds me of a two envelope joke. A president steps down due to scandals, gives his replacement 2 envelopes. Tells him to open the first one when there is the first serious problem he cannot handle and the second one in case of another problem.

The replacement starts on the job, eventually there is a serious political problem he cannot solve. He opens the first envelope and it says: blame everything on the previous guy. So he does and the problem goes away. Later there is another problem that cannot be solved, the guy opens the second envelope and in says: prepare 2 envelopes.

I think somebody opened the first envelope.

Re: Redesigned at some point, obviously

[Anonymous Coward]

The real reason behind the switch from Proton-K to Proton-M was that the M one had a digital guidance computer, that could've been programmed by a rookie engineer, while Proton K relied on analog circuits that had to be rebuilt for every trajectory/payload combination.

--Russian vodka engineer

Re:It's always Stage III

[Anonymous Coward]

I think too, not sure, but with the third stage usually you're fully ballistic when it's running. (don't need to fight gravity)

Absolutely right. Play Kerbal Space Program for a few hours, and you really feel this. All your upper-stage engine needs to do is give you enough horizontal velocity to stay in orbit. So you want it to be as efficient as possible, and as weak as possible (since a weak engine is lighter), while still allowing it to finish its burn before you fall back into the atmosphere.

The first stage is quite different. As you said, the first 1g of acceleration it gives you is wasted fighting gravity, so you want the thrust-to-weight to be as high as possible to minimise this fractional loss. On the other hand, go too fast too soon, and you're losing energy to drag. You don't want to go really fast until you're above most of the atmosphere. The mathematical formulation of this is called "Goddard's problem", and the optimum solution is something like: accelerate flat-out until you reach the speed where atmospheric drag becomes significant, then cut your thrust back, and gradually ramp your acceleration back up to max again as the air thins out.

Slapped by The Tail of the Failure Distribution

[Irate Engineer]

The reasons why this flaw was not identified previously was because it was a low probability occurrence. The shaft was just barely adequate to survive most of the launches, but sometimes it failed before engine cutoff. Since the debris is hard to access, gathering evidence that this was indeed the culprit was very difficult, especially when they didn't know what to look for. The engineers got some hints from previous failures that caused them to put vibration sensors in an area of the rocket that allowed them to identify the current failure mechanism.

This is a problem in rocket design where you have two opposing constraints - they need a pump that works reliably all the way to orbit, but since the rocket is disposable and extra mass reduces payload, overbuilding the pump is not ideal either. This pushes one toward a design that is just barely good enough and no better. It turns out that they wanted a pump that would work for 99.9% of the flights, but they got one that worked 86% of the flights instead.

This was actually a pretty challenging problem in engineering forensics. I hope this fixes their issue. The Proton is a pretty solid rocket otherwise.