The number that cost NASA the Mars Climate Orbiter was 4.45. That is how many newtons there are in one pound of force.
The orbiter launched on December 11, 1998. Lockheed Martin Astronautics in Denver built and operated it. NASA’s Jet Propulsion Laboratory in Pasadena navigated it. On the way to Mars, small thrusters fired routinely to keep its flywheels from spinning up too far, and each firing nudged it slightly off course. A ground program wrote the size of every nudge into a file for the navigators. The interface specification said newton-seconds. The program wrote pound-force seconds. Every nudge went into the navigation model 4.45 times too small.
One small error, repeated 10 to 14 times more often than the navigators expected, because a daily maneuver meant to reduce the firings had been dropped, and the decision was never passed on to them.
The error rode along for nine months, and people noticed.
In April 1999 the navigators finally got usable versions of the file, and within about a week the data suggested the nudges were being underestimated. Through the spring and summer, people at the working level worried about the discrepancies. The investigation board later found they were “only informally reported.” When the data conflicted, the team traded emails instead of using the project’s formal anomaly process, whose purpose some team members did not recognize.
The board found that the navigators had not communicated their trajectory concerns effectively to the spacecraft team or to project management, and that the spacecraft team did not understand them. The navigation team had come on board shortly before launch and had not taken part in testing the ground software or in the design reviews. It was somewhat isolated from the spacecraft team and even from its own line organization. Before arrival, MCO’s navigation was handled by two people, the team lead and a single navigator, in an operations project flying three Mars missions at once. It did not help that the nudges were mostly at right angles to Earth’s line of sight, which Doppler tracking can barely measure.
And there was a cross-check sitting in the data. The spacecraft calculated each nudge itself, in the right units, and radioed the result to Earth. The navigators had assumed this craft was much like an earlier one, Mars Global Surveyor, and did not learn about that data until long after launch. An early comparison, the board wrote, might have uncovered the units problem.
In the final week, the predicted closest pass kept dropping, from a planned 226 kilometers to as low as 110 an hour before arrival. An emergency correction burn was discussed verbally and never carried out. The team had not prepared for it.
On September 23, 1999, the main engine fired, and four minutes later the orbiter passed behind Mars, 49 seconds earlier than predicted. Its signal never came back. Rerun with the right units, the navigation put its closest approach at 57 kilometers. The orbiter needed at least 80 to survive.
“People sometimes make errors,” said Edward Weiler, NASA’s associate administrator for space science. “The problem here was not the error, it was the failure of NASA’s systems engineering, and the checks and balances in our processes to detect the error.” The Los Angeles Times put the orbiter’s cost at $125 million.
The board’s first report came out on November 10, aimed at the Mars Polar Lander, due to land on December 3. Among its recommendations for the lander’s team: a 15-minute stand-up meeting every morning, so that everyone could hear what was happening.