Adam Hibberd
During the 36th Flight of Ariane, Vol 36 (V36), and long before I joined the Ariane 4 Project, one of the strap-on boosters mounted onto the side of the 1st stage of the Ariane 4 rocket developed an extremely low thrust perturbation.
Consequently, shortly after lift-off the launcher strayed off course and, due to extremely high aerodynamic load on the body of the launcher, began to break up in flight.
Unfortunately, for any launch vehicle, when this sort of thing happens there is nothing for it than to send the destruct command from ground to destroy the vehicle and so prevent the vehicle disintegrating or coming down on populated areas for example.
Now the French, Aerospatiale, did some investigations and found the low thrust in this strap-on booster was due to a dirty rag left by a workman in the piping, and recommended some procedural changes to prevent this failure ever happening again.
However (and relevant to my work), they also concluded that an alternative guidance algorithm design for the Ariane 4 flight software in 1st stage would have avoided the high aerodynamic loads, in turn leading to survival of the launch vehicle under this severe thrust deficit, and what’s more eventually allowing orbital injection of the satellite payload!
The problem was that the old guidance algorithm performed a straight-forward interpolation of tables of demanded yaw and pitch against flight time. Thus, since you know the current time (the On-Board Computer had an internal oscillator), then you know the current yaw and pitch. However, this is no good at all if you have a severe thrust problem as struck V36.
What they concluded was that instead of time, the software should employ a thrust or trajectory dependent parameter to look-up the current tabulated yaw and pitch values. A full investigation was conducted and they eventually discovered that ‘Relative Velocity’ (which is the current speed of the launcher with respect to the atmosphere) is an ideal quantity to use to look-up yaw and pitch in this way, as it is quite sensitive to thrust performance.
As Ariane 4 guidance lead, I was responsible for the design/implementation/validation of this new algorithm into the Ariane 4 flight software. It was first flown in V52 and was called Back-Up Mode (BUM).
BUM gave mission planners the reassurance that if a similar rocket engine failure to V36 happened during 1st stage of a future real flight, then rather than losing the entire mission, the vehicle would be steered by BUM along a trajectory with low aerodynamic load and would even eventually result in attainment of some kind of orbit.
Of course, in V52 no such thrust failure transpired, but I did not hesitate to open a bottle of Veuve-Clicquot to celebrate the successful maiden flight of the BUM Guidance Algorithm.