GSLV-D3 Failure Analysis Report

GSLV-D3 Failure Analysis Report by sanman on 09 Jul, 2010 14:00
http://www.isro.org/pressrelease/scripts/pressreleasein.aspx?Jul09_2010

July 09, 2010

GSLV-D3 Failure Analysis Report

The third developmental flight of Geosynchronous Satellite Launch Vehicle (GSLV-D3) conducted on April 15, 2010 from Satish Dhawan Space Centre SHAR, Sriharikota, primarily for the flight testing of indigenously developed Cryogenic Upper Stage (CUS), could not accomplish the mission objectives. Consequently, ISRO had instituted a two-tier process to carry out an in-depth analysis of the flight performance, identify the causes of failure and recommend corrective measures.

The Failure Analysis Committee comprising multi-disciplinary experts completed the analysis and its findings were further reviewed by a National Group of Eminent Experts. These reviews have brought out that:

1. Following a smooth countdown, the lift-off took place at 1627 hrs (IST) as planned. All four liquid strap-on stages (L40), solid core stage (S139), liquid second stage (GS2) functioned normally.

2. The vehicle performance was normal up to the burn-out of GS-2, that is, 293 seconds from lift-off. Altitude, velocity, flight path angle and acceleration profile closely followed the pre-flight predictions. All onboard real time decision-based events were as expected and as per pre-flight simulations.

3. The navigation, guidance and control systems using indigenous onboard computer Vikram 1601 as well as the advanced telemetry system functioned flawlessly. The composite payload fairing of 4 metre diameter inducted first time in this flight, also performed as expected. Performance of all other systems like engine gimbal control systems and stage auxiliary systems was normal.

4. The initial conditions required for the start of the indigenous Cryogenic Upper Stage (CUS) were attained as expected and the CUS start sequence got initiated as planned at 294.06 seconds from lift-off.

5. Ignition of the CUS Main Engine and two Steering Engines have been confirmed as normal, as observed from the vehicle acceleration and different parameters of CUS measured during the flight. Vehicle acceleration was comparable with that of earlier GSLV flights up to 2.2 seconds from start of CUS. However, the thrust build up did not progress as expected due to non-availability of liquid hydrogen (LH2) supply to the thrust chamber of the Main Engine.

6. The above failure is attributed to the anomalous stopping of Fuel Booster Turbo Pump (FBTP). The start-up of FBTP was normal. It reached a maximum speed of 34,800 rpm and continued to function as predicted after the start of CUS. However, the speed of FBTP started dipping after 0.9 seconds and it stopped within the next 0.6 seconds.

7. Two plausible scenarios have been identified for the failure of FBTP, namely, (a) gripping at one of the seal location and seizure of rotor and (b) rupture of turbine casing caused probably due to excessive pressure rise and thermal stresses. A series of confirmatory ground tests are planned.

After incorporating necessary corrective measures, the flight testing of Indigenous Cryogenic Upper Stage on GSLV is targeted within a year.

In the meantime, the next two GSLVs would fly with the available Russian Cryogenic Stages.
#1 by sanman on 09 Jul, 2010 17:00
India Traces GSLV Crash to Turbo Pump Failure

http://www.spacenews.com/launch/100709-gslv-crash-turbo-pump-failure.html
#2 by Art LeBrun on 09 Jul, 2010 17:33
Is that an extreme example of a seized pump failure - so abruptly?
Are there similar examples in the US programs either in flight or on test stands?
#3 by sanman on 09 Jul, 2010 18:15
Gee, I guess rocket turbopumps work at such extreme speed, temperature and pressure, that just about any failure is going to be abrupt and catastrophic for that particular component.

ISRO says they had done plenty of ground-testing, but just not testing under vacuum conditions. I assume that the difference between sea-level and vacuum most affects ignition, but perhaps there are repercussions on the turbopump as well.

They're certainly supposed to be one of the most critical and most difficult to develop components of a rocket. I hear they're also the most expensive component on a rocket, or otherwise perhaps there would be more redundant backups for them.

Maybe it would be worth it to put extra redundant turbopumps on a new rocket when it's first being tried out. That way you'd stand more chance of success. Then you can drop them on the later launches, when you have a better idea of how well everything is performing.
#4 by kevin-rf on 09 Jul, 2010 19:36
Quote from: sanman on 09 Jul, 2010 18:15

Maybe it would be worth it to put extra redundant turbopumps on a new rocket when it's first being tried out. That way you'd stand more chance of success. Then you can drop them on the later launches, when you have a better idea of how well everything is performing.

Huh? They require a particular startup environment, how do you achieve that after the first set failed. You would have to monitor the first set for failure, shut it down, and start the second set... And besides it will likely be of the same design, so it is just as likely to fail as the first set.
#5 by sanman on 09 Jul, 2010 19:56
Okay, then how about multiple concurrently operating turbo-pumps, to distribute the load that would otherwise be on just the one?

If one fails, then the others would have to take on the full load, with a rotary valve quickly sealing off the malfunctioning channel.

Consider that the use of more turbopumps per launch could improve economies of scale that could gradually lower the costs of turbopump manufacturing.
#6 by hop on 09 Jul, 2010 21:47
Quote from: sanman on 09 Jul, 2010 19:56
Okay, then how about multiple concurrently operating turbo-pumps, to distribute the load that would otherwise be on just the one?
Spontaneous turbo pump failure is not a major problem for existing, mature designs.

The answer is to find the exact cause of the problem and fix it, not to re-design the whole system and hope that it miraculously doesn't have any problems.
#7 by Danderman on 09 Jul, 2010 22:08
Could be ingestion of a foreign particle.
#8 by sdsds on 09 Jul, 2010 22:08
Quote from: hop on 09 Jul, 2010 21:47
Quote from: sanman on 09 Jul, 2010 19:56
Okay, then how about multiple concurrently operating turbo-pumps, to distribute the load that would otherwise be on just the one?

The answer is to find the exact cause of the problem and fix it

Right, and for GSLV that's clearly the approach ISRO will take.

But one would also like to glean from a failure analysis some kind of "lessons learned" about the engine design process. Trying to stay close to topic, why did SSME designers choose to distribute the fuel pressure build-up across two turbo-pumps (high and low pressure)? These aren't redundant: failure of either would lead to failure of the engine. But presumably this made sense because design of a single pump that would provide the necessary pressure was too challenging. In retrospect could ISRO have used a similar, dual-pump approach to ease the engineering challenge of their CUS engine?
#9 by sanman on 10 Jul, 2010 01:18
Here's something I just read about SpaceX's the Merlin engine -- it uses a turbopump with dual impellers:

http://en.wikipedia.org/wiki/Merlin_(rocket_engine)

Quote
Propellants are fed via a single shaft, dual impeller turbo-pump. The turbo-pump also provides high pressure kerosene for the hydraulic actuators, which then recycles into the low pressure inlet. This eliminates the need for a separate hydraulic power system and means that thrust vector control failure by running out of hydraulic fluid is not possible. A third use of the turbo-pump is to provide power to pivot the turbine exhaust nozzle for roll control purposes.

It seems to me that having dual impellers would reduce the stresses on the blades, since there are more of them to take the load. I'd presume that if turbopumps suffer mechanical failure, it's more likely at the blades than at the shaft, which should be the more solid part of the mechanism.
#10 by Zpoxy on 31 Jul, 2010 13:47
Quote from: sanman on 10 Jul, 2010 01:18
Here's something I just read about SpaceX's the Merlin engine -- it uses a turbopump with dual impellers:

It seems to me that having dual impellers would reduce the stresses on the blades, since there are more of them to take the load. I'd presume that if turbopumps suffer mechanical failure, it's more likely at the blades than at the shaft, which should be the more solid part of the mechanism.

I think you are misinterpreting the meaning of dual impellers. One pumps kerosene, the other LOx.
#11 by Salo on 13 Jul, 2011 12:44
http://www.livemint.com/2011/07/08213841/K-Radhakrishnan--Flight-stage.html?h=B
Quote
Why is developing a GSLV so difficult?

We’ve bought seven cryogenic engines from Russia, of which we’ve used six. The results coming out of GSLV have been mixed. Sometimes unforeseen obstacles don’t emerge until it’s actually launched. For instance, when we tried to launch last April using (an) indigenous cryogenic engine, all the preliminary stages were fine and our cryogenic engine ignited—and ignition in vacuum is a difficult thing. But after a few seconds, it stopped. For it to keep going, another device called a two-steering engine (or turbo pumps, which keep the launcher steady) ought to ignite, too. This will ignite only if hydrogen and oxygen are present in exact amounts. When we looked into it, there are several possible explanations as to why the turbo pumps stopped: There are three bearings for these turbo pumps; the bearings must rotate without being (distorted) out of shape by the liquid hydrogen fuel it is submerged into. It could also be that the turbo pumps were blown out of shape. There are several things that can go wrong, and each time we have to test from scratch and develop new solutions. While all these have been looked into, we have to undertake a full ground test, before we can be sure that this will work in flight. Hopefully, this flight stage should be ready for testing in March 2012.