Cataloguing reasons for AGNI & GSLV launch failures

arun
BRF Oldie
Posts: 10248
Joined: 28 Nov 2002 12:31

Post by arun »

Article by the one and only T.S. Subramanian.

Another GSLV strap on motor recovered, but unfortunately again, not the one that malfunctioned :

Another strap-on engine recovered
Arun_S
BRF Oldie
Posts: 2800
Joined: 14 Jun 2000 11:31
Location: KhyberDurra

Post by Arun_S »

After much research and simulation here is my Agni-3 maiden flight Failure Preliminary Analysis

The assessment is based on available news, publications and extensive use of BR Rocket Simulator.


Some basic data:

Agni-3 with an intrinsic maximum range of 5,5000Km with 1000Kg payload was launched on trajectory to splash near Indian Navy ships stages near Car-Nicobar islands, 1,700Km from Wheeler Island. A lofted flight trajectory (@74 degree) was chosen (as against using a depressed trajectory).

The lofted trajectory involves:
  • Sonic boom: @ 19 second, Altitude:3.1 Km
  • Max-Q: @ 27 second, Altitude: 6.5 Km, Speed: Mach 1.6, Drag: 8.4 Tonne.
  • Point-A: @ 36 second, Altitude:12 Km, Speed: Mach 2.4, Drag:6.1 Tonne.
  • End of Booster Stage: @65 Second, Altitude: 50Km, Mach 5.6
  • End of Second Stage@ 174 second, Altitude: 363 Km, Range: 98 Km
  • Apogee@ 1033 second, Altitude: 2353 Km, Range: 864 Km
  • Rentry & Spashdown@ 1963 second, Range: 1722 Km
Weak Links:
Aerospace systems are designed to take into account almost all single point of failures and large collection of –double/multiple point of failures too. Almost all aerospace crash has a Root-Cause, but crash occur when two or more (multi-point) systems fail together or progressively.

Of all new parts the two solid fuel motors are most easily and adequately tested. All indications are the weak links are one or more of the following:
  • 1. The new Booster Flex nozzle
  • 2. the new classical Inertial Measurement Platform (a component of Inertial Navigation System)
  • 3. Flight control software/actuator chain
  • 4. Sensors and software responsible for stage separation (This one IMO is a almost a certainty. All news reports problem with stage separation)
Apart from root cause that is the first failure the weak links mentioned above are the likely second point of failures that eventully doomed the flight.

Based on various news report, extensive BR Rocket Simulation runs to validate failure postulates with reported data points in the news stories.

Most likely sequence of failure events:
  • 1. Various DDM reports report a 12 Km event.
  • 2. Agni was launched in loft trajectory. It crossed Mach 1 at 19 second from launch at an altitude of 3.1 Km. Drag increases to 4,200Kg-force.
  • 3. Then came the most important Max-Q point at 27 seconds when the rocket encounters maximum air drag resistance. For the loft trajectory Agni-3 hit Max-Q 27 second after launch, at an altitude of 6.5 Km encountering a drag of 8,400 Kg-force largely at its tip. This is also the point the flex nozzle is working to aerodynamically orient the missile at an angle attack to change the rocket's direction. The missile enters a very stressful phase of the flight (that will last 30 seconds over 27 km additional altitude) involving tremendous drag and fine aerodynamic flight balance as it reorients the trajectory. This phase involves extreme vibration.
  • 4. First Event: 10 seconds after Max-Q, most news report the missiles got into flight control trouble. The flight stress of previous 10 seconds also degraded the inertial reference frame and/or the flex nozzle or its actuators. Clearly the software design did not anticipate/handle the compounding effect of secondary/tertiary failures and lose control of the vehicle.

    From the description, it appears the severe vibrations during Max-Q stress phase induced failure in either the new Inertial platform (in particular the rate gyro) and/or the flex nozzle actuator/sensor; developing excessive backlash.
  • 5. Second Event: Around 60 second into the flight the booster stage almost expended, in a normal flight the missile would have been at ~49 Km altitude, and speed of Mach 5.6.
  • 6. Separation failure: At about 70 seconds the second stage kicks into life but the separation system did not work as planned. Implying the second stage of the missile takes to sky with appendage hanging behind after frying what is left of the interstage & booster. A bad separation almost certainly affected the second stage motor nozzle integrity (it also happens to be another flex nozzle). Clearly the wounded missile does not make its mark hits Bay of Bengal 1000 Km from launch (added later: near Andaman Island) instead of planned 1,700Km location near Nicobar Island. The remnants of the Booster stage reach the 150Km apogee (as reported by some news report) before falling back in the bay approximately 300 Km from launch site.
Last edited by Arun_S on 04 Aug 2006 20:57, edited 2 times in total.
ramana
Forum Moderator
Posts: 59773
Joined: 01 Jan 1970 05:30

Post by ramana »

So the range shortcoming was not too off as it was in the general range of the Andamans if not the Nicobars.

Do Indian test missles have destruct features? This is a perfect example of for test missile self destruct function.
Arun_S
BRF Oldie
Posts: 2800
Joined: 14 Jun 2000 11:31
Location: KhyberDurra

Post by Arun_S »

Ramana: I agree, self destruct on test flight is not only critical for range safety but more importantly to prevent any one from knowing the construction of the nuclear warhead and RV. That is an extremely sensitive information for national security, as well as know-how proliferation danger.

BTW deep water mapping, surveillance and salvaging by remotely piloted or autonomous marine crafts has become mainstream in US Navy and intelligence services. India must ensure all small or big missiles fired into sea destroy at the end of the mission, irrespective if the mission was successful or not.
This is a critical national security issue.
putnanja
BRF Oldie
Posts: 4665
Joined: 26 Mar 2002 12:31
Location: searching for the next al-qaida #3

Post by putnanja »

"Design flaw behind Agni-III failure"
NEW DELHI: India's longest-range ballistic missile, Agni-III, which failed its maiden test on July 9, has more serious problems than have been reported. Unless rectified and successfully proven, the failed Agni-III test will continue to have serious implications for India's credible minimum deterrence, according to the strategic affairs magazine Force.

"The problem in Agni-III occurred in its first rocket itself after the propellant burnt for about 50 seconds. A top Defence Research and Development Organisation (DRDO) official confirmed this to Force. This implies a design flaw that could happen on many counts. The gimballed nozzles in the first stage of the rocket may not have functioned, there could have been an irregular flow of propellant through the nozzles, and the propellant itself may require an investigation," writes Force Editor Pravin Sawhney in the latest edition of the magazine.

The magazine goes on to point out that the more pertinent thing is that nothing of Agni-III as a whole system has yet been validated including stage separation, re-entry at high temperature, the composite (carbon-carbon) bonded material on the nose cone of the re-entry vehicle as the temperature at re-entering the atmosphere would be much higher than 3,000 degrees centigrade, the stability of the warhead within the payload and the guidance system which is imperfect for long ranges.

Probably, the ground-testing of various systems was not done successfully as otherwise the basic flaw in the design of the first stage of the propulsion system would have been detected. The DRDO had the time to do this task while it waited for the government's go ahead for the test firing.

According to the DRDO source, it would take upto eight months to a year for detecting flaws and producing a fresh prototype for test firing. It will require a minimum of three to four tests in rapid succession to validate a weapon system before its design is sealed for production.
SriKumar
BRF Oldie
Posts: 2243
Joined: 27 Feb 2006 07:22
Location: sarvatra

Post by SriKumar »

RaviBg wrote:"Design flaw behind Agni-III failure"
NEW DELHI: India's longest-range ballistic missile, Agni-III, which failed its maiden test on July 9, has more serious problems than have been reported............
.....................

"The problem in Agni-III occurred in its first rocket itself after the propellant burnt for about 50 seconds. A top Defence Research and Development Organisation (DRDO) official confirmed this to Force. This implies a design flaw that could happen on many counts. The gimballed nozzles in the first stage of the rocket may not have functioned, there could have been an irregular flow of propellant through the nozzles, and the propellant itself may require an investigation," writes Force Editor Pravin Sawhney in the latest edition of the magazine.

The 'Hindu' article makes for strange reading. The headline asserts that a design flaw is/was the cause when in reality it is only a speculative assessment from a magazine editor. The headline is mentioned in quotes as if some one made that statement, but no one made that assertion, not even PS. Per comment 'This implies that......', note the smooth segue from a single comment (by an anonymous DRDO man) to the unbridled speculation/guessing by PS, liberally peppered with the coulda's/mebbe's etc. I suppose PS has the freedom to speculate, but tone of the report in Hindu (Sandeep Dikshit) morphs uncertainity into fact. I suppose there is not much else one can contribute if you are writing a report on a report....:roll:
Probably, the ground-testing of various systems was not done successfully as otherwise the basic flaw in the design of the first stage of the propulsion system would have been detected.
Been wondering about that myself....how much ground-testing could have been done, and how much was done.
dxchen
BRFite -Trainee
Posts: 2
Joined: 19 Apr 2006 06:03

Post by dxchen »

The magazine is misleading. I doubt it is BASIC design failure. It would not have passed initial simulation if it is. Each component is tested extensively beyond normal operating range. The integration is tested by countless simulation run. I suspect it was one component failed bought the whole thing down. Remember all components work together, if one failed, chain reaction start. Those incidents happen on mature rocket too. Learn from failure is must, blow it out of proportion is psy op.
Arun_S
BRF Oldie
Posts: 2800
Joined: 14 Jun 2000 11:31
Location: KhyberDurra

Post by Arun_S »

SriKumar and other bhartiya bhailog: I know more than I could say on forum. For once you can read between lines of my post of 04 Aug 2006, and you will not be off mark.

And yes let me plain speak: The maiden Agni-III has many serious design flaws. And yes fixing the serious design flaws is not major rework, but can be rectified in ~3-5 months time.
SriKumar
BRF Oldie
Posts: 2243
Joined: 27 Feb 2006 07:22
Location: sarvatra

Post by SriKumar »

Arun_S wrote:
The maiden Agni-III has many serious design flaws
If true, that is a bit strange. One would think that thorough testing would have been done (as dxchen said), which would help work out many of the fundamental design issues, to not do so is almost unbelievable.

dxchen: I agree with your post, but I differ on one point, that a single malfunction could bring down the rocket. One can expect redundancies in critical systems.
ramana
Forum Moderator
Posts: 59773
Joined: 01 Jan 1970 05:30

Post by ramana »

SriKumar, One flight test equals a hundred ground tests. I am not being facetious. Flight test environment is so much different than ground tests. Ground tests can test only the individual components. Flight test tests the vehicle integration. Please read Arun_S analysis of the flight sequence and ponder. I think Arun has the first independent analysis of the AIII in the public domain.

From his analysis there are three issues- thrust vector control for first stage, stage separation process and third overall flight control logic as shown by S/S ignition. All these are trivial issues and can be resolved.

There is a bigger problem. Is the process being followed at DRDO? Have all the reviews and open items closed out? Are all the steps being taken prior to the flight to ensure success?

The pols have stuck their neck out and got an egg in their face. Difficult to soar like and eagle when flocked by turkeys.
bhavesh

Post by bhavesh »

ramana wrote:
Do Indian test missles have destruct features? This is a perfect example of for test missile self destruct function.
what????
I thought indian test missile have back up trajectory to self-destruct over islamabad in case of malfunction... something changed?

please inform ASAP
ramana
Forum Moderator
Posts: 59773
Joined: 01 Jan 1970 05:30

Post by ramana »

The interesting thing is the propellent business that keeps coming up. the charges are- its not compatible with the increased diameter of new F/S and airflow around the nozzle during combustion.
....there could have been an irregular flow of propellant through the nozzles, and the propellant itself may require an investigation
I think the propellent composition was same as before and what they are saying is it needs more specific impulse to account for the increased weight of F/S. That is more energetic propellent might be required. The second charge of airflow could be to ensure no re-entrant flow at high speed.
narmad
BRFite
Posts: 226
Joined: 10 May 2005 09:47
Location: Mumbai
Contact:

Post by narmad »

India blames controller
http://www.flightglobal.com/Articles/20 ... oller.html
The Indian Space Research Organisation (ISRO) says a malfunctioning engine thrust controller caused the failure of the Geosynchronous Satellite Launch Vehicle (GSLV) on 10 July, but the official report will not be published for another month, writes Radhakrishna Rao.

The three-stage GSLV was carrying the Insat-4C communications satellite when it veered off its predetermined trajectory. Investigators have yet to determine what caused the controller to malfunction.

Meanwhile, preparations continue for the flight of the ISRO four-stage Polar Satellite Launch Vehicle in October, which will loft three satellites - the ISRO Indian remote sensing satellite, the Cartosat space capsule recovery experiment and the Indonesian Lapan Tubsat probe.
arun
BRF Oldie
Posts: 10248
Joined: 28 Nov 2002 12:31

Post by arun »

Anyone, Is Agni III Liquid fuelled :?:

I ask as I think "gimballed nozzles" and "flow", terms used in the quoted Force article, are associated only with Liquid propulsion systems.

Would that be barking up the wrong tree ?
rakall
BRFite
Posts: 798
Joined: 10 May 2005 10:26

Post by rakall »

arun wrote:
Anyone, Is Agni III Liquid fuelled :?:

I ask as I think "gimballed nozzles" and "flow", terms used in the quoted Force article, are associated only with Liquid propulsion systems.

Would that be barking up the wrong tree ?
I think A3 uses flex nozzles and not engine gimbal... For second stage control Flex nozzles have also been used in A2 from 4th test (1994) onwards.. the first 3tests used engine gimbal control for second stage bcoz 2nd stage was liquid-fuelled (Prithvi)

By propellant flow, he may hae meant propellant burning rate characteristics & associated pressure changes etc..

Dont expect DDM (especially farce) to be so accurate.
Vick
BRFite
Posts: 753
Joined: 14 Oct 1999 11:31

Post by Vick »

India blames controller
The Indian Space Research Organisation (ISRO) says a malfunctioning engine thrust controller caused the failure of the Geosynchronous Satellite Launch Vehicle (GSLV) on 10 July, but the official report will not be published for another month, writes Radhakrishna Rao.
Arun_S
BRF Oldie
Posts: 2800
Joined: 14 Jun 2000 11:31
Location: KhyberDurra

Post by Arun_S »

As I asserted before it turned out to be a software problem, and point to inadequate (nonrigorous) software code review process and system integration review process.
Vidyarthi
BRFite -Trainee
Posts: 26
Joined: 16 Aug 2006 15:46
Location: Thiruvananthapuram
Contact:

Post by Vidyarthi »

Quoting Ramana

"From his analysis there are three issues- thrust vector control for first stage, stage separation process and third overall flight control logic as shown by S/S ignition. All these are trivial issues and can be resolved.

There is a bigger problem. Is the process being followed at DRDO? Have all the reviews and open items closed out? Are all the steps being taken prior to the flight to ensure success? "

To the list above, add fin-less aerodynamically unstable configuration and non-availability of control force due to earlier than expected burn out of the solid propellant engine. Remember, failure analysis of ASLV-D2?
ShibaPJ
BRFite
Posts: 146
Joined: 20 Oct 2005 21:21

Post by ShibaPJ »

Breaking News.. GSLV failure tracked to the fuel-control device..

Gurus, can you shed some more light :?: If I recollect correctly (from Arun_S, Ramana & other's posts), probably a software glitch or component design was to blame.

Faulty gadget brought down GSLV

A defective fuel-control device has been identified as the rogue component that propelled the GSLV (Geosynchronous Satellite Launch Vehicle), with Insat-4C communications satellite on board, off course and into the Bay of Bengal seconds after liftoff on July 10.

This has been confirmed by a team of 15 aerospace experts headed by K Narayana, former director of the Satish Dhawan Space Centre, Sriharikota Range.

They have said that the flawed gadget, on board one of the four strap-on boosters of GSLV F02, let in more propellant than required into the engine, setting off a chain reaction that resulted in erratic propulsion. The thrust generated by this booster did not match that of the other three strap-ons, forcing the rocket off its trajectory.

The experts analysed a heap of data passed on by the rocket before its crash and simulated the conditions on a computer, sources in ISRO told HT.

But the team was unable to erecover the rogue device or other parts of the strap-on booster from the seabed or the vicinity of the launch pad. One of the three strap-ons that generated the set thrust was recovered from the Bay of Bengal.

The sources said the booster was manufactured at ISRO's Liquid Propulsion Systems Centre, Mahendragiri, but the manufacturing of most of the components had been outsourced to Indian companies.

They said the malfunctioning device on board the Vikas engine (strap-on booster) caught ISRO scientists by surprise because the engine had been proven in several PSLV (Polar Satellite Launch Vehicle) and GSLV flights in the past.

According to the sources, the experts declared the GSLV's design "robust" and felt there was no need for a complete review of the rocket. The scientists would address the defect ahead of the next GSLV flight, scheduled in 2007, the sources said.

The GSLV's failure ended ISRO's run of 12 successful launches in as many years from the Satish Dhawan Space Centre.
Arun_S
BRF Oldie
Posts: 2800
Joined: 14 Jun 2000 11:31
Location: KhyberDurra

Post by Arun_S »

ShibaPJ wrote:Breaking News.. GSLV failure tracked to the fuel-control device..

Gurus, can you shed some more light :?: If I recollect correctly (from Arun_S, Ramana & other's posts), probably a software glitch or component design was to blame.

Faulty gadget brought down GSLV
Don't have to second guess, my views are still available on previous page.

So as I said before (on this thread, IIRC regarding Agni-III failure) that invariably all flight failures are due to multi-point failures, since all single point failures and most two point and muti point failures are already designed into aeronautical vehicles. In this case the F02 flight root cause was partial failure of the flow valve, however from what I understand the root failure was not in itself a catastrophic failure, and at worst it was a partial failure, a well designed control system should have captured this non-catastrophic partial failure scenario and operated the engine in graceful degradation mode, that yields reduced performance rather than take the extreme step to shutdown the engine let the vehicle fly out of control become a dangerous missile. The range safety officer saved the day by executing destruct command.

So yes the primary failure was component but the mission failed because the secondary system that was supposed to properly handle the primary failure did not work due to inadequate design, review and/or test regime in the process followed by ISRO.
geeth
BRFite
Posts: 1196
Joined: 22 Aug 1999 11:31
Location: India

Post by geeth »

very unlikely failure if proper QC was in place, or still worse, may be some spy work. The failure was of GSLV was preceded by a symbolic visit by NASA Chief.

Something similar happened diring the first launch as well. Then it was a pipe carrying the fuel.
ramana
Forum Moderator
Posts: 59773
Joined: 01 Jan 1970 05:30

Post by ramana »

ShibaPJ
BRFite
Posts: 146
Joined: 20 Oct 2005 21:21

Post by ShibaPJ »

Thanks, Arun_S. Yeah I checked the thread later & saw ur posts, but forgot to edit mine!!!
a well designed control system should have captured this non-catastrophic partial failure scenario and operated the engine in graceful degradation mode, that yields reduced performance rather than take the extreme step to shutdown the engine
The launch would still be a failure, wouldn't it? I don't think it is possible to rectify such situations mid-flight. Is there a specific reason to call it a design failure? I would rather think, this is a serious QC/QA failure that a faulty component managed to sneak in, as it was not a new design/ usage. or did I miss something else?

Geeth, I think it would be far-fetched to point to NASA chief's visit or any spy work or sabotage. I would rather go with slipshod QC (as previous usages in PSLV/ GSLV launches have been successful) till it is proved otherwise.
ramana
Forum Moderator
Posts: 59773
Joined: 01 Jan 1970 05:30

Post by ramana »

The faulty compoenet is the primary cause. However the system should have secondary defences/redundancies to handle such component failure. Again it probably was an unanticipated component failure and will be fixed.

The more interesting thing is the AIII failure. Nothing on that yet.
Arun_S
BRF Oldie
Posts: 2800
Joined: 14 Jun 2000 11:31
Location: KhyberDurra

Post by Arun_S »

ISRO chairmens comment in the IBN video and this fragment indicated it the low level (fast control loop) controller software was unable to converge. It is thus possible that modal coefficients would not match for the wrongly machined parts and even a good software will not work (since such systems do not require adpative control loops and are not designed to be over engineered).

http://www.hindustantimes.com/news/181_ ... 030009.htm
They have said that the flawed gadget, on board one of the four strap-on boosters of GSLV F02, let in more propellant than required into the engine, setting off a chain reaction that resulted in erratic propulsion. The thrust generated by this booster did not match that of the other three strap-ons, forcing the rocket off
ShibaPJ
BRFite
Posts: 146
Joined: 20 Oct 2005 21:21

Post by ShibaPJ »

Some more tidbits.. Talk of a Rs. 1 lakh component bringing down Rs. 250-Cr project down :evil: Also, they say the component was non-testable on ground :?: Arun_S, Ramana: what are your take on this?

Why INSAT-4C never made it to space

A manufacturing error in an engine component caused the sudden loss of thrust in one of the four liquid propellent strap-on stages leading to the failure of the Geo-synchronous Satellite Launch Vehicle-F02 mission on July 10, the Failure Analysis Committee of the Indian Space Research Organisation has concluded.

The propellant regulator in question was supplied by an Indian company, one of the industry-partners of ISRO, and it was an 'inadvertent human error,' ISRO Chairman G Madhavan Nair told a news conference, where the findings and recommendations of the FAC were released.

The 15-member FAC, constituted to investigate the failure of the mission from Satish Dhawan Space Centre Sriharikota, concluded that all vehicle systems, except one strap-on stage, were normal until 56.4 seconds.

"The primary cause for the failure was the sudden loss of thrust in one out of the four liquid propellant strap-on stages (S4) immediately after lift-off at 0.2 seconds," concluded the FAC, chaired by K Narayana, former Director of SDSC SHAR, which reviewed the performance of GSLV-02 from lift-off to the end of flight.

The loss because of GSLV failure was put at Rs 250 crore -- ISRO spent Rs 150 crore to build the rocket, and another Rs 100 crore to fabricate INSAT-4C, which was to have been put in orbit by the GSLV.

But Nair said the firm, which supplied the faulty propellant regulator, which cost Rs 1 lakh, would not be black-listed :eek: and that the space agency will continue to source the component from the same supplier.

The Bangalore-headquartered ISRO would not reveal the identity of the supplier citing confidentiality clauses. Stressing that it was an inadvertent human error, Nair said the mechanical component in question was one of the modules, 'which is non-testable on ground,' but added that penalty clauses will be invoked against the company.

"It is not ISRO's culture to go on witch-hunting. It's not the time to say 'x' is faulty and 'y' is faulty. At ISRO, we own up the responsibility (for GSLV failure)," he said.

"We have to tell them (the company) they have to be stricter in future and not commit the error of this type."

He said ISRO will build an identical GSLV and a satellite similar to INSAT-4C for a launch in June next year.

According to ISRO, with only three liquid propellant strap-on stages working -- after the sudden loss of thrust in S-4 -- there was significant reduction in the control capability. However, the vehicle could be controlled till about 50 seconds. At the same time, the vehicle reached the transonic regime of flight and the vehicle attitude errors built up large values, resulting in aerodynamic loads exceeding the design limits thus leading to break-up of the vehicle.

The FAC has concluded that the design of GSLV is robust and recommended implementation of strict control on fabrication, inspection and acceptance procedures. Among others, FAC recommended fabrication processes to be critically reviewed and updated.

It has also recommended independent inspection of all critical dimensions of components and sub-assemblies by in-house agencies.
ramana
Forum Moderator
Posts: 59773
Joined: 01 Jan 1970 05:30

Post by ramana »

Not going after the component maker is the right thing. No point in loosing the expertise built to date. So from the FAC recommendations, as corrective action there will be more stricter complinace to design disclosure and an independent audit process by inhouse people. I think they should have more operator training at the sub contractors to ensure that processes are being followed. That is how you reduce human error.

So the design was reviewed again which is good.
MN Kumar
BRFite
Posts: 393
Joined: 27 Jan 2002 12:31

Post by MN Kumar »

"It is not ISRO's culture to go on witch-hunting. It's not the time to say 'x' is faulty and 'y' is faulty. At ISRO, we own up the responsibility (for GSLV failure)," he said.
John Snow
BRFite
Posts: 1941
Joined: 03 Feb 2006 00:44

Post by John Snow »

But Nair said the firm, which supplied the faulty propellant regulator, which cost Rs 1 lakh, would not be black-listed and that the space agency will continue to source the component from the same supplier.
The chain is as strong as the weakest link. Napolean

Martin Thiakol was not black listed after the challenger disaster
abhishek
BRFite -Trainee
Posts: 70
Joined: 07 Jun 2004 11:31

Post by abhishek »

But Nair said the firm, which supplied the faulty propellant regulator, which cost Rs 1 lakh, would not be black-listed :eek: and that the space agency will continue to source the component from the same supplier.
If a vendor was blacklisted everytime he made a mistake, there would be no Lockheeds, Boeings, General Dynamics, Raytheons, blah, blah, blah, today....
bala
BRFite
Posts: 1975
Joined: 02 Sep 1999 11:31
Location: Office Lounge

Post by bala »

In this report some more details:

ISRO, reviewed a huge volume of data and concluded that the failure was due to the malfunctioning of a regulator in the fourth liquid propellant strap-on motor. "With only three strap-on stages working, there was significant reduction in the control capability," Dr. Nair said.

The thrust of the liquid engines used in the strap-on stages is controlled by a set of regulators. Analysis indicated that in S4 engine, the thrust control was not effective and the engine failed at 0.2 seconds after lift-off, which is five seconds after ignition. "Instead of stabilising at 5.85 mega pascal (MPs) chamber pressure, it reached 7.11 MPs at 2.8 seconds. This was much beyond the design limit and the engine failed."

The performance of all sub-systems of the vehicle, except the fourth strap-on, was normal until 56.4 seconds. "It was a hidden problem that led to the failure of the vehicle 62 seconds into the flight," Dr. Nair said.


Dr. Nair said the FAC concluded that the GSLV design was robust and recommended the implementation of strict control on fabrication, inspection and acceptance procedures. It recommended fabrication processes to be critically reviewed and updated. ISRO had also been asked to go in for independent inspection of all critical dimensions of components and sub-assemblies by in-house agencies. The report also recommended long duration hot test on one out of every 20 engines fabricated to ensure that the production process was under control. It was also for the strengthening of the process of clearance of launch during the Automatic Launch Sequence phase.
ShibaPJ
BRFite
Posts: 146
Joined: 20 Oct 2005 21:21

Post by ShibaPJ »

Guys, coool. Blame the wrong smiley at the wrong place :)

What I was pointing to was the need for much better inspection/ QC etc for the entire sub-assembly/ assembly. Good to seem that it is being taken care of; also that ISRO did not indulge in buck-passing. Welcome steps from ISRO in tying up the loose ends and improving process mgmt as a whole.
SaiK
BRF Oldie
Posts: 36424
Joined: 29 Oct 2003 12:31
Location: NowHere

Post by SaiK »

abhishek wrote:
But Nair said the firm, which supplied the faulty propellant regulator, which cost Rs 1 lakh, would not be black-listed :eek: and that the space agency will continue to source the component from the same supplier.
If a vendor was blacklisted everytime he made a mistake, there would be no Lockheeds, Boeings, General Dynamics, Raytheons, blah, blah, blah, today....
quite right.. at the same time, we should have better regulation and inspection regime (just like in the USA where vendors need to be certified something and that thing security type pass etc) that reports on the vendors audit.

how can one say if its not a sabbotage? may be some yankee doodle pro scribbled something and paid some guy.

hey its only 1lakh stuff.. for all you know a paki or an indian traitor could have done some mischeif.

how do we setup a system that verifies into all these. of course black listing is not the answer. what is the correct procedure that is followed in advanced countries that we are following too.

may be i don't know. lets get that on the table. may be i don't know how the agni-3 failed.. same supplier supplying that faulty regulator? we need more info. its not the question of money wasted into agni and gslv, but catching the errorneous process, part or that person.
SriKumar
BRF Oldie
Posts: 2243
Joined: 27 Feb 2006 07:22
Location: sarvatra

Post by SriKumar »

I wonder if they were able to determine what specifically was wrong with this part, and why. I think that should be a part of the analysis.

So this part was 'not testable on ground'? This raises the question of what other rocket parts are not testable on ground; and are additional QA checks needed for them. Surely they should be able to devise something to make sure it works before they put it on. 'Testing' them by flying them is expensive, to say the least. I mean, you lose not just the rocket but the payload as well.
jayam
BRFite -Trainee
Posts: 15
Joined: 01 Jan 1970 05:30
Location: USA

ISRO's GSLV failure analysis report

Post by jayam »

ISRO's GSLV failure analysis report http://isro.org/pressrelease/Sep06_2006.htm
symontk
BRFite
Posts: 920
Joined: 01 Nov 2001 12:31
Location: Bangalore

Post by symontk »

SriKumar wrote:I wonder if they were able to determine what specifically was wrong with this part, and why. I think that should be a part of the analysis.

So this part was 'not testable on ground'? This raises the question of what other rocket parts are not testable on ground; and are additional QA checks needed for them. Surely they should be able to devise something to make sure it works before they put it on. 'Testing' them by flying them is expensive, to say the least. I mean, you lose not just the rocket but the payload as well.
Yes. Many critical rocket parts are non testable since any intensive testing would destroy the part itself. Especially parts associated with propellants. motors and engines. So only Non - destructive testing is applied against these parts and results we see for ourself.

The production has to ensure that the parts are like the "same" ones used for previous tests / development and would have to assume that it would "behave" similarly in flight.

It is tough but thats the only way until someone figure out how to find the faults by some other way.
shiv
BRF Oldie
Posts: 34982
Joined: 01 Jan 1970 05:30
Location: Pindliyon ka Gooda

Post by shiv »

symontk wrote: Yes. Many critical rocket parts are non testable since any intensive testing would destroy the part itself.
Chidambaram as part of an IISc talk some years ago had stated that with unmanned rockets - it is cheaper to test the rocket itself than conduct exhaustive tests of every single part. Some parts cannot be tested until you use them. In this particular case the part was a "known and tested" part and it may have to be retested or QC checked.
geeth
BRFite
Posts: 1196
Joined: 22 Aug 1999 11:31
Location: India

Post by geeth »

From the report, it appears that the part failed 0.2 seconds after lift-off. If it had 'failed' within 4 seconds after ignition, the software was supposed to shut off the engine and abort the mission - it would have saved the rocket & payload. But it 'failed' after the mission could not be aborted by the onboard computer.

Some reports say it pumped in more fuel than required. Some says it developed less chamber pressure....Unless we see more detailed report on failure analysis, it will be difficult to conclude anything.

Afterall, what is published for the consumption of the general public may not be the exact cause. In that case we will only be able to do some guesswork.
rakall
BRFite
Posts: 798
Joined: 10 May 2005 10:26

Post by rakall »

geeth wrote:From the report, it appears that the part failed 0.2 seconds after lift-off. If it had 'failed' within 4 seconds after ignition, the software was supposed to shut off the engine and abort the mission - it would have saved the rocket & payload..

Read carefully... the failure occured 5sec after ignition.. so if indeed there is a 4sec window for "abort", it had already passed..

the vehicle had already lifted-off when the failure occured.. so there was no scope for abort.. abort at this stage would only have destroyed the launch pad etc..

One question is does the ALS process only vital parameters before lift-off such as thrust etc or does it monitor & process everything down to details like chamber pressure etc.. if it is designed to monitor Pc also, then probably it wud have detecded abnormality and aborted like it did in GSLV D1..
geeth
BRFite
Posts: 1196
Joined: 22 Aug 1999 11:31
Location: India

Post by geeth »

>>>Read carefully... the failure occured 5sec after ignition.. so if indeed there is a 4sec window for "abort", it had already passed..


Yeeah, I read it carefully before I made the comment. You are essentially saying the same as that I said (pls re-read).

My point was, why fail immediately on lift-off? Is it just a co-incidence? any corruption on the software? Or did someone try to outsmart the software? It is the software that gives the command to shut -off (within 4 Sec), if it notices any abnormality within those 4 secs. In the first launch also, when the fuel pipe leaked, the sensors reported a lower pressure in one of the chambers and the software shut off everything else.

Now, if it is a manufacturing defect, the increase in flow would have been there from begining, leading to a higher chamber pressure (ISRO site gives more details), isn't it?. Logically, the software should have been programmed to cater for a higher chamber pressure also...that is my contention. May be I am wrong, because it is difficult to come to any conclusion based on partial data, or what is made available to public.

>>>the vehicle had already lifted-off when the failure occured.. so there was no scope for abort.. abort at this stage would only have destroyed the launch pad etc..

Ofcourse, you are right. For a moment think a bit tangentially - the failure doesn't occure within 4 secs. So, for sure the vehicle lifts-off. Immediately after that the failure occurs. Now the software cannot be programmed to retrieve the situation, because the vehicle is already in motion. At the same time, if there was not sufficient extra control, can you imagine a situation where the vehicle loses control immediately (not after 56 seconds) and falls inside the space centre (or still worse falls in land)? Even if you give the command to explode, the fall of debris would have been catastrphic. That way, hats off to the designers for putting sufficient redundancy in control forces, so that the vehicle could fly without problems for 56 secs.

That is why this failure after 0.2 secs after lift-off sounds alarming to me...leading to my "as-usual" conspiracy theory.

>>>One question is does the ALS process only vital parameters before lift-off such as thrust etc or does it monitor & process everything down to details like chamber pressure etc.. if it is designed to monitor Pc also, then probably it wud have detecded abnormality and aborted like it did in GSLV D1..

IMO it does check the vital parameters and chamber pressure is one of the parameters. Why it did not find abnormal chamber pressure within 4 secs, or why the abnormality occured only after 4 secs (by which time the software could do nothing) is a point to ponder.
Locked