QUOTE (Paolo @ Dec 17 2010, 11:03 AM)
JAXA has published what look like some quite detailed report on the investigation so far, unfortunately (for me at least), it's in Japanese only and Google translate doesn't know how to handle pdfs
http://www.jaxa.jp/press/2010/12/20101217_...akatsuki_j.htmlThis report is indeed quite detailed - JAXA must be commended for its openness here.
Pandanenko's translations are very welcome.
As with other foreign literature, I can at least look at the pictures - these are rather telling.
First note the architecture of the propulsion system
Click to view attachmentThe history of these parameters (P1...P4) should be something as follows. P1 (the helium tank pressure) should
fall during the burn, as the pressurant fills the space above the fuel and oxidizer as these are used up. P2 is the
pressure upstream of the fuel and oxidizer tanks and is controlled by the regulator (which I see is redundant RG1
and RG2). This pressure basically determines the flow rate of fuel and oxidizer and thus the thrust.
P3 and P4 are the pressures downstream of the fuel and oxidizer tanks - these should not be too different from
P2 (and thus each other).
But now look what happens during the burn
Click to view attachmentP1 drops (not sure whether the amount by which it drops is reasonable - this will depend on the respective
volumes of the pressurant and fuel tanks). Odd that it seems to drop well beyond (to 3000s) the duration of the
(truncated) burn itself - 152s, we'll come back to this
P2 behaves more or less as it should - stays constant as the regulator meters the helium flow to fill up the space
as fuel/ox is used. Similarly P4 (oxidant downstream) follows P2
P3 - the downstream fuel pressure - is the anomaly. It drops substantially during the burn (and implies the fuel
flow also reduces). This could be a downstream leak, or an upstream constriction (which, since P4 tracks P2 pretty well,
must be between P3 and P2 - the CV-F valve or the pipework around there seems a likely culprit. Either way, the flow
from the regulated P2 is not keeping up.
Once the burn ends, P3 creeps back up - suggesting that gas is managing slowly to get to the fuel tank - it is not a
complete blockage. This is consistent with P1 dropping as the regulator allows some gas through to make up the
ullage volume opened up by the fuel usage. With knowledge of the tank and pipework volumes, one should be able
to calculate the leak flowrate.
So, the engineering forensics seems to make sense - regulator and everything upstream of there was fine, oxidizer
side was fine, but somewhere between regulator and fuel tank exit there was a constriction.
As the fuel pressure dropped, so did the fuel flow (by ~30%, to judge from the pressure record).
Thus the thrust declines. Normally, for a constant regulated pressure,
the thrust should be constant, and thus the acceleration of the vehicle will increase with time as the mass of the spacecraft
falls as the fuel/ox is used up. Yet here the acceleration drops, so the thrust must be declining along with the fuel pressure
Click to view attachmentThen at 152s, the acceleration drops suddenly, implying a massive loss in thrust. Not only that, but the attitude history
shows that something suddenly starts to torque the spacecraft around. This history
Click to view attachmentshows the rate about the X-axis (blue curve, orthogonal to the motor thrust axis) increases from 0 to 12 deg/s in about a third of a
second (presumably at this point the ACS threw a fit and cut off the burn and initiated safing). 12 deg/s is ~0.2 radians/sec
so to get there in 0.3 sec implies an angular acceleration of ~0.6 rad/s2. Taking the moment of inertia of the ~500kg
spacecraft as a cube of dimension 1.4m gives ~160 kgm2, and thus the torque must have been 0.6*160 = 100 Nm. If
the torque acted 0.7m from the center of mass (much more than that would be off the body of the spacecraft), then
a thrust component orthogonal to the line to center of mass must have been 130N, if my midnight math is right.
For a 500N thruster, this is a pretty severe misalignment. Note in the acceleration history above, at this point the axial
acceleration drops from 0.8 to 0.5 m/s2, suggesting the thrust component in that direction fell by 30%. If we take the
thrust as perhaps closer to 350N total than 500, since the fuelpressure was down, then if the thrust suddenly were
directed about 30-45 degrees to the side (if a large part of the nozzle broke off) this would give a component normal to
the axis of about 100N or so, and would drop the axial component by 30%, as required to match the torque and
acceleration record. None of this proves anything, of course - perhaps a meteorite hit the nozzle and the
fuel pressure just happened to be declining beforehand - but a coherent story is that the low fuel pressure caused
poor fuel flow; usually fuel is injected into the combustion chamber to provide film cooling; the poor fuel flow
interfered with this cooling action and a hot spot developed, cracking the nozzle and causing the thrust diversion,
which led to the attitude excursion, in response to which the spacecraft correctly safed; had it not, the spacecraft would
be spinning at some absurd rate - even without the autonomous termination of the burn, the mispointing would have
exceeded 90 degrees in a couple of seconds and so VOI wouldnt have succeeded)
Rather an interesting puzzle to solve (not that I'd not rather be working on Venus data from Akatsuki instead).
The exercise
shows (1) the value of engineering telemetry in diagnosing what happened, (2) that the fault protection on Akatsuki worked
well and prevented the situation getting worse. (3) If all the above is correct, any further
use of the main engine will cause large torques, as well as significant thrust losses. Furthermore, (4) unless the
constriction is cleared, thrust will fall appreciably for burns longer than a few tens of seconds.
It might nonetheless be
possible to use the engine in shorter bursts (and the pressure drop effect will become less troublesome as the fuel is
used up and so the ullage volume is larger), and by spinning the spacecraft during the burn to even out the torque.
I don't know whether VOI in 6 years will be possible given these constraints - presumably that is exactly the analysis that is being
done by JAXA now : as Hayabusa showed, if it is possible, JAXA will find a way.
(sorry for the long post)