by iamlucky13 » Mon Jul 16, 2007 7:21 pm
I guess that's a fair way of saying it. It was a simple mistake, but it happened deep within the minute details of a very complicated task.
The mission this happened on was the Mars Climate Orbiter. To save costs, much of the ground control software was re-used from the Mars Global Surveyor, but of course, modified where necessary to fit the particulars of the new mission. It worked fine for the MGS, but under time pressure to get the mission ready for the launch window, the conversion factor was missed when adapting the software. Testing was limited both because of the time and budget pressure, and because the MGS had already "proven" the code was safe. The accident investigation later concluded more testing would probably have showed the error this led to, which was very subtle.
Among the duties of the software was calculating velocity based on the forces of accelleration everytime they fired the reacton control thrusters. The reaction control thrusters control the orientation of the orbiter to keep it pointed the right way. To save mass and cost, the number of thrusters was kept minimal, which required a geometry such that some of their force was rotational (desired) and some was translational (in the direction of travel). This has to be accounted for to obtain accurate navigation. Although the effect is very, very small, it adds up over millions of kilometers.
This thrust was given by given by Lockheed in pounds, but NASA works in Newtons (they now require all vendors to provide everything in metric, I believe). The modified MGS ground-control software should have had the conversion factor added. The result was the calculated effect on velocity from using the reaction control thrusters was off by a factor of 4.45. It sounds big, but it was still small enough that it did not result in an noticeable error in the limited testing they did.
In flight, this showed up as monitoring the velocity from earth showed some discrepancy between the calculations. This is expected and all missions have mid-course corrections, so it didn't raise any flags. You calculate a thruster burn based on position and velocity to fix the error. However, the error in the spacecraft calculations meant the position was in error. This error went into the course-correction calculations, so while those fixed the velocity each time, the craft was already out of position.
The end result is that instead of entering orbit at about 220 km, the MCO probably dipped as low as 57 kilometers, disintegrating due to the force and heating of drag as it skimmed the Martian atmosphere.
I guess that's a fair way of saying it. It was a simple mistake, but it happened deep within the minute details of a very complicated task.
The mission this happened on was the Mars Climate Orbiter. To save costs, much of the ground control software was re-used from the Mars Global Surveyor, but of course, modified where necessary to fit the particulars of the new mission. It worked fine for the MGS, but under time pressure to get the mission ready for the launch window, the conversion factor was missed when adapting the software. Testing was limited both because of the time and budget pressure, and because the MGS had already "proven" the code was safe. The accident investigation later concluded more testing would probably have showed the error this led to, which was very subtle.
Among the duties of the software was calculating velocity based on the forces of accelleration everytime they fired the reacton control thrusters. The reaction control thrusters control the orientation of the orbiter to keep it pointed the right way. To save mass and cost, the number of thrusters was kept minimal, which required a geometry such that some of their force was rotational (desired) and some was translational (in the direction of travel). This has to be accounted for to obtain accurate navigation. Although the effect is very, very small, it adds up over millions of kilometers.
This thrust was given by given by Lockheed in pounds, but NASA works in Newtons (they now require all vendors to provide everything in metric, I believe). The modified MGS ground-control software should have had the conversion factor added. The result was the calculated effect on velocity from using the reaction control thrusters was off by a factor of 4.45. It sounds big, but it was still small enough that it did not result in an noticeable error in the limited testing they did.
In flight, this showed up as monitoring the velocity from earth showed some discrepancy between the calculations. This is expected and all missions have mid-course corrections, so it didn't raise any flags. You calculate a thruster burn based on position and velocity to fix the error. However, the error in the spacecraft calculations meant the position was in error. This error went into the course-correction calculations, so while those fixed the velocity each time, the craft was already out of position.
The end result is that instead of entering orbit at about 220 km, the MCO probably dipped as low as 57 kilometers, disintegrating due to the force and heating of drag as it skimmed the Martian atmosphere.