Investigating the SpaceX Rocket Explosion of September 1, 2016

[Jonathan Scott]

However, the ejected material was burning - even at the leading burst front...

I don't discount the possibility that the explosion started inside the stage, but I think it's more likely that material was ejected without being visible before it caught light (mostly oxygen with a much smaller amount of fuel) but that in the oxygen-rich environment the flame propagated extremely fast through it, immediately igniting much of the mixture. I also suspect that the apparent extent of the flames in the first couple of frames may be misleading, in that there is clearly such a bright light from the initial explosion that there is a lot of indirect light.

I'm aware of the junk yard; do we know if it was active in any way at the time? I've not heard of any other junk yard sounds being picked up.

Of course, if there were people anywhere near the camera, they would have seen the explosion before that sound was recorded, and could well have reacted in some way, causing the first sounds.

 

[jim hardy]

We don't know what a view from the other side would show.

Is this about where the camera was ? about 2½ miles SSE ?

 

[Jonathan Scott]

By using Google Earth 3D view and moving things around until the towers line up with the footage (matching the ratio of the spacing between the tops of the towers and also noting that the sphere to the right also matches in that case), then moving out for about 4km, I make it that the camera was further out to the west, around location 28.551189, -80.618837 which is about 4.2km from the site.

 

[jim hardy]

I think i like your location better than mine.
I hadn't noticed the power poles over there, had been looking for a spot with a high building . I only roughed in direction visually.

 

[.Scott]

I don't discount the possibility that the explosion started inside the stage, but I think it's more likely that material was ejected without being visible before it caught light (mostly oxygen with a much smaller amount of fuel) but that in the oxygen-rich environment the flame propagated extremely fast through it, immediately igniting much of the mixture. I also suspect that the apparent extent of the flames in the first couple of frames may be misleading, in that there is clearly such a bright light from the initial explosion that there is a lot of indirect light.

Reports are that the air was very moist - even accounting for the Florida locale. So any venting should have become immediately visible.

 

[mheslep]

We know that whatever caused the explosion is not a fundamental issue with every rocket. In the worst case, it is something that happens occasionally. Every rocket has things that go wrong occasionally. As discussed before, a 5% failure rate is typical for rocket launches. If they keep that rate (the worst case: if they don't manage to find the issue and if they do not improve anything), it means another rocket will blow up at some point, probably within 2-3 years given the high launch rate. Yes. But ultimately: so what? They won't get contracts for manned missions then, but for many unmanned missions a 1 in 20 risk is fine if the launch is cheap enough. Note that this is the worst case, I don't expect that to happen.

You make it sound like this specific lost rocket is the worst thing that ever happened. Hundreds of rockets failed in the history of spaceflight. It is something that happens, despite great care of every launch service to reduce the risk as much as possible.

Falcon 9 failed catastrophically in 2 of the last 11 launches, and no there are not enough launches yet to say this series of 11 is the worst possible case. This could be the coin toss series that happens to show heads only twice in 11 flips. Atlas II launches by contrast were successful for all 63 attempts. The other Atlas models have similarly superior records. I'm not sure what point is to be made by reporting on the history of rocketry since it's inception, any more than one would judge acceptable passenger aviation by going back to the Wright Flyer, or since the era of the de Haviland Comet passenger jet when aircraft fell out of the sky once per week.

https://en.m.wikipedia.org/wiki/Atlas_II

Cheap launch costs may make failures tolerable to SpaceX, but it is no help to the payload customer who just had his satellite or resupply destroyed and his research or business time lost.

It may be the case that "great care" is always taken by SpaceX to "reduce the risk as much as possible " but I don't know that to be the case, and I think neither do you.

Again though, mistakes of the past are correctable with a reasonable acceptance of self evaluation. What might block change is a leadership imposed state of denial, which I see hints of in Shotwell's comment.

 

[.Scott]

By using Google Earth 3D view and moving things around until the towers line up with the footage (matching the ratio of the spacing between the tops of the towers and also noting that the sphere to the right also matches in that case), then moving out for about 4km, I make it that the camera was further out to the west, around location 28.551189, -80.618837 which is about 4.2km from the site.

I have it further to the east.
28.525209, -80.575614.

Here's how I'm spotting it:
1) Per Google map photo: There is one tower that is closest to the spherical tank. The towers are of triangular cross section, and the closest tower is pointing one vertex to the tank. The the distance between the tank and the closest tower is about 3.5 times the height of the tower triangle at ground level.
2) From the video: If we number the towers from left to right 1, 2, 3, then only tower 2 matches the criteria for the closest tower. This is further demonstrated by the way the towers are lit up by the explosions. Tower 3 is in the background. Tower 2 is in the foreground.
3) This makes things pretty easy. Tower 3 lines up with the center point between the tank and tower 2 - almost due south. Following that to a building about 2.5 miles away.

 

[mfb]

Falcon 9 failed catastrophically in 2 of the last 11 launches, and no there are not enough launches yet to say this series of 11 is the worst possible case. This could be the coin toss series that happens to show heads only twice in 11 flips.

If you want to cherry-pick, why don't you take just the last rocket and claim 100% loss rate? Starting the sequence directly at a loss introduces a huge bias. In addition, just a single launch failed. Amos-6 didn't get launched. For the next few launches, I expect that they will do the static fire test without payload.

Atlas limited to II and V is one of the few examples to be significantly below the 5% loss rate (1 partial failure in 128 launches). It is a very mature system, and you are cherry-picking again by taking the rocket type with the lowest loss rate. Why don't you include the previous Atlas rockets? Atlas E: 2 losses out of 23 lauches. Atlas H: 0 out of 5. Atlas G: 2 losses out of 6 launches. Atlas I: 3 losses out of 11 launches. All those were after 1980, the earlier versions were much worse. With Atlas II they finally got it working properly, 63 lauches without a loss.
7 losses in 45 launches for the early Atlas rockets. That's the number of launches they needed to fix all the various issues with the rocket. SpaceX is not at 45 launches yet and has a lower failure rate already.

I'm not sure what point is to be made by reporting on the history of rocketry since it's inception

I don't do that, the 5% are the current rate.

Cheap launch costs may make failures tolerable to SpaceX, but it is no help to the payload customer who just had his satellite or resupply destroyed and his research or business time lost.

It is. They have insurances taking this into account, and rocket plus insurance is still cheaper than other rockets. SpaceX gets so many launch contracts for a good reason.

It may be the case that "great care" is always taken by SpaceX to "reduce the risk as much as possible " but I don't know that to be the case, and I think neither do you.

You'll never launch a rocket if you don't do that.Edit:

And because it exceeded the flame propagation velocity, it had to be ignited before being ejected.

Apparent flame propagation velocity can also go up if the material is moving.

 

[Jonathan Scott]

Reports are that the air was very moist - even accounting for the Florida locale. So any venting should have become immediately visible.

I must admit that if the venting would need to have started several frames before the explosion, that would have been visible. I had initially been assuming an abrupt rupture of the tank followed immediately On the other hand, there is already quite a bit of venting visible, and it might not be easy to see venting in the general direction of the camera, only to the sides. So I'm definitely considering the alternative that it did start with a small actual explosion inside the second stage.

 

[Jonathan Scott]

I have it further to the east.
28.525209, -80.575614.

Here's how I'm spotting it:
1) Per Google map photo: There is one tower that is closest to the spherical tank. The towers are of triangular cross section, and the closest tower is pointing one vertex to the tank. The the distance between the tank and the closest tower is about 3.5 times the height of the tower triangle at ground level.
2) From the video: If we number the towers from left to right 1, 2, 3, then only tower 2 matches the criteria for the closest tower. This is further demonstrated by the way the towers are lit up by the explosions. Tower 3 is in the background. Tower 2 is in the foreground.
3) This makes things pretty easy. Tower 3 lines up with the center point between the tank and tower 2 - almost due south. Following that to a building about 2.5 miles away.

Look at the direction of the steps or whatever they are on the spherical tank. They are clearly on the right of it, not on the front of it, from the viewpoint. They could of course have moved them, but there are other clues such as the buildings on the left which exactly match the view from the west, including for example the sloping roof whose ridge line is under the middle of the leftmost tower.

 

[Dr. Courtney]

Doing frame by frame, this is what I see:

50 to 71.73: Nothing of any significance.

71.77: In 40 msec or less, the "explosion" is already 17 feet high and wide. This is excluding illuminated sections of the booster that, on first glance, might appear to be part of the fire. Presuming that we started with a point source of ignition, this means that the illumination border has been moving at 8.5 feet in no more than 40msec or >210 feet per second or more.

71.81: In the next 40msec, the flames have continued to expand laterally, but the bottom of the flame has actually shortened! Moreover, a "cloud" created from cold temperatures along the bottom half of the booster is left undisturbed during this time - and for at least 10 frames that follow.

So at first, it would seem that what we are seeing in these first two frames is purely a flame propagation front. It would seem that an invisible and combustible mixture of gases was already there, hanging as a cloud in the air, when it became ignited.

But there's a problem. Flame propagation rates are measured in cm/sec, with our 210 feet per second being over 5300 cm/sec. Among the fastest propagation rates in air is a 38% mixture of hydrogen which tops out at roughly 480 cm/sec (http://www.comtherm.co.uk/CT-7a%20Fig.pdf). So, unsurprisingly, we are definitely working with something more potent than air.
Perhaps the chemists among us can come up with a mixture that will propagate at 6400 cm/sec. For comparison, the source above rates methane/air at about 70 cm/sec while this http://www.sciencedirect.com/science/article/pii/S0360544212005579 shows a oxy-methane mixture at about 295 cm/sec. Even assuming that hydrogen gets the same boost (haven't found a source on that), it would still fall short of our 6400 cm/sec. Besides, our gases should be on the cold side - slowing the burn rate down a bit.
So did something, perhaps turbulence or heat, enhanced the flame speed? Or perhaps the flame spread without incandescing for a few frames before becoming visible?

The flame propagation theory seems to have a problem.

Alternatively, both the the material that is burning and the oxidant could be ejecting laterally from the tank - carrying the flame with it. From what I can tell, this is exactly what is happening. The combustion started within the tank, created a lateral crack in the wall, and sprayed out.

I've worked a bit with high speed video of deflagration to detonation transitions in oxy-fuel mixtures. Summary: a wide variety of velocities are possible between the deflagration velocities one might look up in a source and the detonation velocities. Things can "sputter" like they are approaching detonation, but then slow back down. Lots of environmental factors change DDT transition issues and whether a reaction gets to detonation. A lot of this field is more art than science, and if all you have is a bound on the flame velocity, you can't say much. I do agree that a higher concentration of oxygen than found in air was most likely involved.

You know better than me, but c'mon, this is space age stuff and Elon Musk. I've seen spy satellite footage better than what I've seen. In this thread are we just seeing footage from visitors watching?

Agreed. There should have been better and more systematic monitoring of the system including high speed cameras with much better frame rates and resolution. Some nearby blast pressure sensors (sampled at 10 MHz) in the near field and some pzt based microphones in the far field sampled at 100 kHz would be valuable also. It is well known that a significant percentage of launches fail. There should have been due diligence to collect the event data for diagnosing the hows and whys when it happens.

 

[.Scott]

Apparent flame propagation velocity can also go up if the material is moving.

Exactly.

 

[jim hardy]

I have it further to the east.
28.525209, -80.575614.

i'm not in same league as you guys with imagery...
Using Google Maps' "measure distance" i am able to put a straight line from launchpad to proposed camera sites without losing resolution.
That's how i came up with my spot
which is halfway between .Scott's and Johnathan Scott's
so i feel successful !

i was looking for that distinctive three phase power pole

but there's a ring of them around the complex and it looks like it's nearer the camera than that.

 

[.Scott]

Look at the direction of the steps or whatever they are on the spherical tank. They are clearly on the right of it, not on the front of it, from the viewpoint. They could of course have moved them, but there are other clues such as the buildings on the left which exactly match the view from the west, including for example the sloping roof whose ridge line is under the middle of the leftmost tower.

I've tried to do some finer measurements. I can't nail it, but I have convinced myself that it is further west than I thought.

 

[liometopum]

The software used by Philip E. Mason to make his "Detailed analysis of Spacex Rocket Explosion" video was Sony Vegas Pro. I downloaded it.

To get a copy of the YouTube video to insert into Vegas Pro, this site was great: http://en.savefrom.net/1-how-to-download-youtube-video/
I placed the ss in front of the address, as recommended, and got a copy that worked perfectly.

You can play the video and listen clearly to the sounds.

The squeak and pop are further support, to me, that the whole thing started on the surface of the rocket. I don't think that the squeak and pop could have come from inside the rocket.

 

[Dotini]

Video of successful launch shows details of strongback, fueling process,

 

[jim hardy]

The squeak and pop are further support, to me, that the whole thing started on the surface of the rocket. I don't think that the squeak and pop could have come from inside the rocket.

I played it back with good headphones and volume turned all the way up.
If the sound is 12 seconds behind the video

there are several squeaks 1:06, 1:09, 1:12 at moment explosion on video , the big squeak at 1:17 followed by pop at 1:18, followed by what sounds like a bird singng at 1:20, .
1:12 through 1:23 there is a rapid clicking or tinkling sound that is easy to miss because the video is so pyrotechnic.

If I'm right about the 12 second delay, the explosion sound is at 1:24 just before payload hits the ground. Thereafter booms dominate.

I'd say from the sounds something started 18 seconds before visible explision. Could be birds between mike and rocket, or it could be thin aluminum crinkling as it gets cooled rapidly by a LOX leak. Ever listen to an air cooled motorcycle engine cool down? It clicks and ticks quite a lot...

I've forgot now which site had sound synchronized with video to remove the delay.
If this Rowvid one is synchronized then I'm dead wrong (again) , it's just the explosion woke up the neighborhood birds.
http://rowvid.com/?v=_BgJEXQkjNQ

old jim

 

[Dotini]

Video presents case for ignition promoted on the strongback preceding the explosion of the Falcon rocket carrying AMOS-6.

 

[Jonathan Scott]

As I've said before, it appears that ignition occurred outside the rocket, either on its surface or further out.
If the indicated pipe is indeed for LOX (others have suggested it is an air duct) then it does look vulnerable, but I don't understand why a LOX pipe would be going higher than the LOX umbilical.
I haven't viewed the entire video, but what is the suggested fuel source? Yes, many things will burn strongly in pure oxygen, but the initial flame spreads so rapidly it suggests a significant amount of pre-mixed fuel and oxygen.

 

[Dotini]

I haven't viewed the entire video, but what is the suggested fuel source?

Haven't viewed the video? It's only 9 minutes. The authors of the video suggest the fuel source as leaked kerosene or pipe insulation, seals, gaskets, or even the metal itself, potentially saturated or caked in recondensed LOX. In a previous video it is discussed how strongback kerosene pipes may at times flow in the reverse direction.

 

[Jonathan Scott]

Haven't viewed the video? It's only 9 minutes. The authors of the video suggest the fuel source as leaked kerosene or pipe insulation, seals, gaskets, or even the metal itself, potentially saturated or caked in recondensed LOX. In a previous video it is discussed how strongback kerosene pipes may at times flow in the reverse direction.

I'd like to hear something definitive about the locations of the LOX and RP-1 pipes in the strongback. Does anyone have any official information?

The umbilicals going across to the rocket are certainly lower down than the ignition point so there doesn't appear to be an obvious point in having pipes higher than that. I would not have expected leaking kerosene (which is of course still liquid at ambient temperature) to mix enough with any form of oxygen to explode initially with that much speed and energy. One can see how it burns later; the part which has been hit with LOX burns fiercely, but the part which is mixing with air looks more like a rain of fire. If it were leaking as a spray from a small aperture under pressure, it's possible that it could mix well with the oxygen, but as RP-1 loading was supposed to be complete and the wind was away from the strongback it would seem strange to me that fuel under pressure was still in the vicinity.

 

[Dotini]

The specific configuration of the Falcon and its strongback may have changed between JCSAT-16 and AMOS-6. But Mr Musk seems loathe to share such details.

 

[jim hardy]

How do they get the hydrazine into the payload ? Is it in the countdown sequence ?

I read hydrazine self ignites with LOX .
In that climate a squirt or drip of it down into a cloud of O2 boiloff seems way more plausible ignition source than static electricity.

 

[mfb]

Hydrazine should have been in the satellite before they put it into the rocket. It is liquid at room temperature, no need to have some fancy fueling steps later.

 

[mfb]

http://www.spacex.com/news/2016/09/01/anomaly-updates, bold from me:

At this stage of the investigation, preliminary review of the data and debris suggests that a large breach in the cryogenic helium system of the second stage liquid oxygen tank took place. All plausible causes are being tracked in an extensive fault tree and carefully investigated. Through the fault tree and data review process, we have exonerated any connection with last year’s CRS-7 mishap.

They want to have the launch pad repaired and ready by November.

Here is a video of those helium bottles bursting (on purpose in the video), 6 psi and 18 psi should be 6000 psi (~40 MPa, about the pressure SpaceX uses) and 18000 psi (~120 MPa):



COPV = Composite Overwrapped Pressure Vessel, titanium with carbon fiber. Fragments of carbon fiber in liquid oxygen is a very explosive mixture.

 

[.Scott]

SpaceX reports that the explosion was caused by Helium overpressure, but the cause of that is still being investigated.
http://www.spacex.com/news/2016/09/01/anomaly-updates

The timeline of the event is extremely short – from first signs of an anomaly to loss of data is about 93 milliseconds or less than 1/10th of a second. The majority of debris from the incident has been recovered, photographed, labeled and catalogued, and is now in a hangar for inspection and use during the investigation.

At this stage of the investigation, preliminary review of the data and debris suggests that a large breach in the cryogenic helium system of the second stage liquid oxygen tank took place. All plausible causes are being tracked in an extensive fault tree and carefully investigated. Through the fault tree and data review process, we have exonerated any connection with last year’s CRS-7 mishap.

 

[.Scott]

For those following the NASA Space Flight (NSF) forum, they have used this SpaceX announcement to start a new thread on the subject (and closed the previous one). It is now: http://forum.nasaspaceflight.com/index.php?topic=41252.0

By the way, the NSF forum had focused on external causes to the incident - so most of those posters were way off. A notable exception was "Jim" - not surprisingly, someone with quite some background in the field.

 

[Dotini]

Apparently they think the root cause could yet be traced back to the pad.

https://www.yahoo.com/news/spacex-releases-details-recent-rocket-181133330.html
The root cause of the accident may also change the amount of time SpaceX has to stand down from launching. If the issue is related to the rocket itself, it's possible the redesign and testing process would force a more extended grounding, while if the problem is traced back to the pad, it might be a quicker fix, industry experts have said.

 

[mfb]

Sure, ground equipment has no strong weight and size limits, that is way easier to change.

 

[jim hardy]

I did some reading about aluminum and titanium at cryogenic temperatures.
The exotic alloys of titanium you have to use are very sensitive to how hot you make the weld joint during fabrication. The heat affected zone adjacent weld must remain below 1800F or it's brittle and cannot be annealed by slow cooling. Something about alpha and beta phases in the metals crystal structure.

The article was dated 1974, long enough ago that some new little fabricator shop might not be aware ? All their gray-hairs retired, or it got outsourced overseas ?

I'm no metallurgist just had one course about 1966 and remember a few terms so won't embarrass myself by trying to explain further.

If i can find the article again will link it. Who made that COPV tank ?

 

[Jonathan Scott]

Who made that COPV tank ?

I believe SpaceX now make their own COPVs.

 

[jim hardy]

found the article

https://app.aws.org/wj/supplement/WJ_1974_03_s117.pdf

Conclusions
Welding procedures for Ti-6211 were established using the pulsed GMAW process and sound quality welds were produced. Hot ductility tests made with the Gleeble did not show any anomalous behavior and did not indicate a susceptibility to hot cracking. Detailed examination of the heat-affected zone using synthetic specimen techniques revealed that some problems may be anticipated. Poor impact strengths, lower than those of the weld fusion zone are developed in the areas of the heat-affected zone immediately adjacent to the fusion zone; specifically those regions which reach temperatures above 1800 F. The structure developed in these regions in equilibrium a-platelets with a' martensite within coarse former beta grains produced as a result of rapid cooling from a temperature at which it is fully beta.

i don't know if that's even the same alloy they used for their tank

what it does tell me is welding those specialty alloys is an acquired skill.

Maybe they'll find some pieces of the tank and tell us how it failed..

 

[Jonathan Scott]

i don't know if that's even the same alloy they used for their tank

The SpaceX COPVs are made of aluminium (or aluminum, depending on your local preference), which is commonly used for COPVs but not usually for ones which are immersed in LOX.
There have been some comments which suggest that aluminium could cause problems in this case, which I understand is due to aluminium's thermal expansion/contraction profile being less compatible with that of the overwrap layer.

 

[jim hardy]

i'm just curious about it. Part "old fire horse" reaction of a guy who spent too many years as a troubleshooter to not be curious, and i had that one metallurgy course 50 years ago...

LOX is -193C at 1 atmosphere, i don't know what temperature Spacex keeps theirs.

surely Spacex engineers know their cryogenic alloys
A quick search turned up
http://www.totalmateria.com/Article23.htm

Low-Temperature Properties. Aluminum alloys represent a very important class of structural metals for subzero-temperature applications and are used for structural parts for operation at temperatures as low as -270oC.

Below zero, most aluminum alloys show little change in properties; yield and tensile strengths may increase; elongation may decrease slightly; impact strength remains approximately constant. Consequently, aluminum is useful material for many low-temperature applications.

The chief deterrent is its relatively low elongation compared with certain austenitic ferrous alloys. This inhibiting factor affects principally industries that must work with public safety codes. A notable exception to this has been the approval, in the ASME unfired pressure vessel code, to use alloys 5083 and 5456 for pressure vessels within the range from -195 to 65oC. With these alloys tensile strength increases 30 to 40%, yield strength 5 to 10% and elongation 60 to 100% between room temperature and -195oC.

The wrought alloys most often considered for low-temperature service are alloys 1100, 2014, 2024, 2219, 3003, 5083, 5456, 6061, 7005, 7039 and 7075. Alloy 5083-O which is the most widely used aluminum alloy for cryogenic applications, exhibits the following cooled from room temperature to the boiling point of nitrogen (-195oC):

• About 40% in ultimate tensile strength
• About 10% in yield strength.

Retention of toughness also is of major importance for equipment operating at low temperature. Aluminum alloys have no ductile-to-brittle transition; consequently; neither ASTM nor ASME specifications require low-temperature Charpy or Izod tests of aluminum alloys. Other tests, including notch-tensile and tear tests, assess the notch-tensile and tear toughness of aluminum alloys at low temperature characteristics of welds in the weldable aluminum alloys.

Compared with other alloys, alloy 5083-O has substantially greater fracture toughness than the others. The fracture toughness of this alloy increases as exposure temperature decreases. Of the other alloys, evaluated in various heat-treated conditions, 2219-T87 has the best combination of strength and fracture toughness, both at room temperature and at -196oC, of all the alloys that can be readily welded.

Alloy 6061-T651 has good fracture toughness at room temperature and at -196oC, but its yield strength is lower than that of alloy 2219-T87. Alloy 7039 also is weldable and has a good combination of strength and fracture toughness at room temperature and at -196oC. Alloy 2124 is similar to 2024 but with a higher-purity base and special processing for improved fracture toughness. Tensile properties of 2124-T851 at subzero temperatures can be expected to be similar to those for 2024-T851.

Several other aluminum alloys, including 2214, 2419, 7050 and 7475, have been developed in order to obtain room-temperature fracture toughness superior to that of the other 2000 and 7000 series alloys. Information on subzero properties of these alloys is limited, but it is expected that these alloys also would have improved fracture toughness at subzero temperatures as well as at room temperature.

Fatigue Strength. Results of axial and flexural fatigue tests at 106 cycles on aluminum alloy specimens at room temperature and at subzero temperatures indicate that, for a fatigue life of 106 cycles, fatigue strength is higher at subzero temperatures than at room temperature for each alloy. This trend is not necessarily valid for the tests at higher stress levels and shorter fatigue lives, but at 106 cycles results are consistent with the effect of subzero temperatures on tensile strength.

It'll be interesting to see if they used one of those mentioned..Looks like cooling the tank ought to shrink it more than i'd imagined
http://www.noao.edu/ets/gnirs/SDN0013-02.htm

i'm no expert, just surfing net for background.old jim

 

[Dotini]

IMO the problem is solved by TechX. They have identified the explosive foam insulation on the strongback as combining with venting LOX, with ignition promoted from corona discharge. Then a couple of the COPV tanks were dislodged from their bosses from the concussive explosion and flying strongback shrapnel. A much safer insulation is discussed. And there you have it. SpaceX will be back in business soon!