Autopsy of HEAT-2X and TM65LE

HEAT-2X was taken down from the VTC3 test stand on 20th August. On that occasion, the engine section was separated from the tank section, which required separation of the pipes in the engine compartment corresponding to the interface between the two sections.

On August 23, we conducted the “autopsy” of HEAT-2X in order to study the cause and effect of the fire in details. CS is obviously mainly interested in the cause of the fire; but there is of course a very rewarding experiment in the fire itself, which can guide us with regard to future design, materials and protection of engine parts and structures.

The tank section is macroscopically intact, why we this time concentrated on the engine section. The main findings will be mentioned below, stressing that this was a structural survey. Coupling to the data available are not made here and videos are only superficially mentioned.

Motor casing and fins

The first task was to get the engine out of its casing (the fin section). This had been badly messed up by the heat, so it pressed hard against parts of the engine. We ended up having to make a longitudinal cut with a saber saw and grinder in order to get a crowbar in there to increase the diameter of the casing. With a crane arm on the forklift we managed to pry the engine out. I recall more elegant autopsy overtures; but it worked.

Casing and fins has a melting point of 660 oC (aluminum), but was only melted locally: Part of the Northwest fin was burned away from the lower edge (respectively 67 and 21 cm on the two plates). At the lower edge of the western engine hatch a 5 cm notch was melted off, corresponding to where a cable was passed through the not quite closed hatch. On the inside of the motor casing – next to the fuel main-stage valve upper edge (tank side) – was a clear, 4-cm wide mark from a small flame that has hit the aluminum surface. There was no soot; but the surface appeared melted, and in an upwardly directed oval above the point a whitening of the aluminum oxide can be seen. On the outside of this point the surface was more glossy than the surroundings.

The engine

The engine’s exterior was largely unaffected. A plastic tubing that protected the electrical cords to thermocouples was burnt away and had left a soot stain on the side of the engine.

Debris from damaged insulation, etc. had settled on horizontal surfaces such as the engine upper surface, battery boxes, air tanks, and on the engine compartment floor at the lower engine manifold.

FIG. 1 Peter Meincke (left) and Thomas Madsen. The engine's exterior was largely unaffected. The black spot comes from a burned plastic hose. There are burnt remains on lower manifold. Photo: Carsten Olsen/CS.

FIG. 1 Peter Meincke (left) and Thomas Madsen. The engine’s exterior was largely unaffected. The black spot comes from a burned plastic hose. There are burnt remains on lower manifold. Photo: Carsten Olsen/CS.

The engine’s top was taken apart and the injector plate taken out. This was undamaged. There were aluminum shavings in two of fuel injector canals as a result of insufficient cleaning of the fuel tank. There were no shavings from the LOX system, which had been extra thoroughly cleaned and degreased.

FIG. 2: The injector was fine and undamaged. Photo: Carsten Olsen/CS.

FIG. 2: The injector was fine and undamaged. Photo: Carsten Olsen/CS.

Inspection of combustion chamber close to the injector showed no damage, but several cm long welds (made from the outside) showed incomplete penetration.

The nozzle sports a severe bulging of the nozzle inner sheath comprising most of the length from the throat to the nozzle side and about 150 degrees of the circumference. The bulging is split into two vaults. Along one vault the nozzle edge and the manifold was pulled inward. Along the second the inner casing had instead been pulled from the nozzle edge (manifold). This has thus produced a free outlet of fuel from the cooling jacket through a several cm wide opening extending approximately 50 degrees along the nozzle side.

FIG. 3: The inner wall of the cooling jacket bulges inwards in two arches. Red arrow: The opening in the cooling jacket. Blue Arrow: The pulled-in nozzle edge/manifold. Photo: Carsten Olsen/CS.

FIG. 3: The inner wall of the cooling jacket bulges inwards in two arches. Red arrow: The opening in the cooling jacket. Blue Arrow: The pulled-in nozzle edge/manifold. Photo: Carsten Olsen/CS.

FIG. 4: The nozzle and the opening in the cooling jacket. Photo: Ruben Hansen/CS.

FIG. 4: The nozzle and the opening in the cooling jacket. Photo: Ruben Hansen/CS.

The broken weld was a single-stranded fillet weld with an A-measure of 1.3-1.5 mm on a 3 mm iron plate, where it – especially at a pressurized, strongly heated system – should have been at least 2.4 mm. (The A-measure is an expression of the weld thickness: the height of the largest right angle isosceles triangle that can be drawn in the cross section of the seam.)

FIG. 5: The broken weld with poor A-measure. Photo: Carsten Olsen/CS.

FIG. 5: The broken weld with poor A-measure. Photo: Carsten Olsen/CS.

As was the case in the combustion chamber the inner nozzle wall also sported several welds with incomplete penetration.

FIG. 6: Welding on the nozzle wall with incomplete penetration. Photo Carsten Olsen/CS.

FIG. 6: Welding on the nozzle wall with incomplete penetration. Photo Carsten Olsen/CS.

Because we want to preserve the remains in order to be able to display them later on, we decided not to cut open the engine, so the interior of the cooling jacket was inspected using a fiberscope. In here we saw tearing of the number of the spacers between the inner and outer wall, which should have helped prevent the collapse of the inner wall. The failure of these would also be necessary for the described bulging of the internal wall to take place. Several of the spacers showed signs of having been cold welded.

FIG. 7: The inside of the cooling wall was examined using a fiberscope. Photo: Ruben Hansen/CS.

FIG. 7: The inside of the cooling wall was examined using a fiberscope. Photo: Ruben Hansen/CS.

It may be assumed that a number of spacers were torn initially and that the bulging of the inner wall, pulling it apart and opening it happened subsequently.

There is no doubt that here we find the real cause of the fire: At main stage the cooling jacket blew open and large amount of fuel burst out. The fuel-injection into the combustion chamber has been proportionally reduced causing LOX/GOX to run unreacted through the nozzle into the ambient, ignited the fuel. The concrete walls of the test stand have helped to create an O2-rich environment for the fire. Various cables and electronics have been ignited; for example a fire lingered for a long time up and around the broken fin.

The engine compartment

All the plastic insulation of the cables was missing. The cable to the newly installed pressure sensor on the top of the LOX tank was the only cable insulated with Teflon and also sleeved in fiberglass. It wouldn’t have survived in working order; but its condition might indicate that it had survived longer than the standard plastic insulated cables. Teflon insulation and fiberglass sleeving was – somewhat paradoxically – adopted to protect it against the very cold LOX tank. Data from the engine controller (EC) may confirm this cable’s longevity.

The copper conductors of the cables, however, was preserved (melting point 1000 oC). So too the 6 mm copper tubes for pressurized maneuver air was intact.

The separation disassembly from VTC3 revealed an approximately 17 mm/45 degrees fault on the lower gasket (engine side) of the fuel pre-stage valve house. Because of the tight tolerances, the limited space, and the rigid tube, it has been a recurring problem installing the valves. We have been forced to loosen the engine casing and tank section and add 4 mm spacers in between them to make room to jostle the valves in between the flanges. It is most likely this operation, which has damaged the thin and vulnerable Teflon gasket. It has happened before, and there were no problems during disassembly.

FIG. 8: The defective gasket on the engine side of the fuel pre-stage valve. Photo: Flemming Rasmussen/CS.

FIG. 8: The defective gasket on the engine side of the fuel pre-stage valve. Photo: Flemming Rasmussen/CS.

At the separation it was noted that the stags, which pull together the pipes and valves seemed looser than expected. This may be due to severe temperature variations.

The above issues and the whole appearance of the engine compartment suggests the presence of an separate fire in the engine compartment. This is supported by the video, which shows a small puff from eastern engine hatch at T-1.1 s and perhaps later fire from this point. Cables were passed through the upper part of this hatch. Furthermore, an occurrence of an actual, downward directed aerosol or smoke jet and subsequent ignition from the left hatch from about T = 0 to T + 45 s was observed, corresponding to the cable entry at the bottom of this hatch, where, as mentioned above a notch in the casing had been melted.

Juxtaposing this with the defective seal on the fuel pre-stage valves on the engine side, it seems to indicate a fuel leak in the engine compartment already at pre-stage. A downward spray of fuel out of the slightly open left hatch is ignited shortly after the external fire, and the fire in the engine compartment may very well be due to blow back from there. The same can have happened at the eastern hatch, but were not convincingly visible beyond fire in the cables outside the hatch.

The described melt mark off the fuel tank main stage valve side may be from increased spacing later in the fire. Neither seals nor inspection on the test day with tank pressures up to 10 bar (where personnel left the rocket) indicated leakage here.

As for the fire in the engine compartment, it is doubtful whether the approximately ½ m3 of atmospheric air containing about 150 g of O2 could fuel this. However, there was an open, 15 mm cable hole in the bottom of the engine compartment and as mentioned two hatches slightly ajar so some chimney effect may have played a role.

There were no broken seals or other signs of leakage in the LOX system besides those slightly looser valve stags. Perhaps the fiercely cold LOX and subsequent heat from the fire have caused mechanical stress and increased spacing. Therefore, we can not reliably exclude leakage, but like I said; we saw no sign of it.

Engine Controller (EC)

All internal damage to the engine controller is apparently due to the use of plastic connectors, which was completely gone. (Type Buccaneer. Chosen because they are waterproof and cheap.) The cable fittings of metal at opposite end of the box has kept the fire out.

FIG. 9: Christian Ravn (left) and Peter Meincke. The Engine Controller can be seen with the missing connectors. Photo: Carsten Olsen/CS.

FIG. 9: Christian Ravn (left) and Peter Meincke. The Engine Controller can be seen with the missing connectors. Photo: Carsten Olsen/CS.

The Interface Control board was locally delaminated to a severe degree, while the CSduino card (manufactured by Printline) had fared much better. In here too, fiberglass stockings on cabling had reduced the damage to the wiring insulation.

And very important: Data stored in a flash memory IC on CSduino will most likely be saved. We have just managed to pry the IC undamaged from the board. It will be mounted on a new CSduino card in order to read out the data as soon as we obtain a new card.

FIG. 10: The EC-box at top. Note the preserved connectors at one end. On the left the CSduino card. On the right, the more damaged interface card. Photo: Carsten Olsen/CS.

FIG. 10: The EC-box at top. Note the preserved connectors at one end. On the left the CSduino card. On the right, the more damaged interface card. Photo: Carsten Olsen/CS.

FIG. 11: The other side of the CSduino was well preserved. The arrow shows the flash memory, which has been recovered and we later will attempt to read. Photo: Carsten Olsen/CS.

FIG. 11: The other side of the CSduino was well preserved. The arrow shows the flash memory, which has been recovered and we later will attempt to read. Photo: Carsten Olsen/CS.

Summary

The results of the autopsy are consistent with a scenario in which the engine compartment had a leak of fuel already in the pre-stage due to a damaged packaging under the fuel pre-stage valve. As the pressure increased nominally at the start of the main-stage, the undersized spacers between the inner and outer walls of the cooling jacket as well as the poor and undersized welds caused the cooling jacket to burst open. Fuel ran out into the open air without passing through the combustion chamber, while the LOX/GOX unreacted ran out through the nozzle. Both contributed to the outer fire, which in turn ignited the fuel which escaped through the hatches from the engine compartment. Blow back through hatches created a secondary fire in the engine compartment. We could not detect a leak in the LOX system.

More pictures from the autopsy

The (now famous) slow motion GoPro HERO3 video “Through fire and water for CS”.

4 thoughts on “Autopsy of HEAT-2X and TM65LE

  1. I would like to analyze the video stream of the fabulous camera HERO3 by correlating the slow-motion videostream with the data figures. But there is a misfit. Is the original video file made with 24 fps? The YouTube video is an 25 fps video. The slow motion is made with 240 fps.
    Could someone help?
    An another question: Where is the position of the camera in relationship to the vessel?
    Many thanks
    Andreas

  2. First analysis of the video file:
    (the time marks depends on the whole YouTube file)
    On 29 s the ignition of the pre-stage can be recognize. You can see the follwong 4 “real time” seconds a very turbolent and bumby combustion, follwed by the main stage at 1:18 minute. At 1:21 the burn-through occures. You can see it more clearly at 1:22 on the video stream.That is in the data figures the first glitch at over 16 bars of the fuel inlet pressure, migt be the real glitch is much higher but the measurement unit could not record it because of the slow sampling rate. It could be a pop that could create a very high pressure peak and very heigh temperature that was disturben the nozzle (as I explained eralier in my coment to the test data), more than ever the pressure drops down and sises up again in a short time, ahwtever the reasons are. It has something to do with resonace, refelction and high-frequency instability. I found a source: Page 352 of Sutton and Biblarz “Rocket Propulsion Elements”.
    Just all for now and a short fough analysis.
    Best regards
    Andreas

  3. Have some nights spent brooding and think it must have been a hard start, possibly an extreme pop.

    I think that the irregular burning of the pre-stage indicated that the cooling jacket of the combustion chamber was not sufficiently stable under pressure and therefore the cooling flow of the film cooling and in the cooling jacket was not uniform enough, partially even came to a halt. The extremely high gradient pressure peak (more than measured, as I explained earlier) at the beginning of the main stage lifted the injector head from the cooling jacket of the combustion chamber a bit (strange contours around the fuel holes on the injection head could indicate that) and then there was an intrusion of the excess oxygen mixture from the combustion chamber into the cooling jacket, being ignited in the cooling jacket and discharged to the weak wall of the nozzle and bulged it. This deformation broke up the weak welds of the nozzle edge.

    So, root cause could be the irregular combustion of the pre-stage due to low fuel pressure and the associated low fuel flow rate, leading to oscillation of the fuel flow in the film-cooling and the cooling jacket.

    So far, best regards
    Andreas

  4. Thinking about is ongoing… Hope, your investigation, too. My next theory:

    During the start of the engine (after ignition), the nozzle probably was working in an over-expanded modus during the increase of chamber pressure, in addition to other possible failure reasons.

    The pressure from exhaust dropped down along the nozzle axis from the throat to the middle and rises up again to the end of the nozzle. The pressure minimum was far below of the fuel pressure in the jacket and moves during the process of combustion stabilization and thrust buildup from the middle to the end of the nozzle, to the weld seam, like a shock wave inside the diverging nozzle section.

    So, this depression wave tumbled the nozzle wall and the fuel overpressure in the jacket blows the nozzle up because of the too thin wall plate and the poor welds (see Figure 3.9 in Rocket Propulsion Elements from Sutton/Biblarz).

    I think that this is not the only reason. It might be that this reason came together with some other reasons that I explained earlier.

    Andreas

Leave a Reply


*