VI
Mark học AILaunched 11 December 1998 · signal lost 23 September 1999 · at Mars orbit insertion
Source: MCO Mishap Investigation Board, Phase I report (10 Nov 1999), pp. 6–7, 13–17.
The interface spec said metric. Navigators saw numbers that did not add up for months. So how did a unit error slip through the handoff between two teams, all the way to the Martian atmosphere?
Source: MCO Mishap Investigation Board, Phase I report, pp. 13–18.
The milestones below come from the Phase I report of the investigation board (chaired by Arthur Stephenson, NASA).
Source: MCO Mishap Investigation Board, Phase I report, pp. 6, 13–14, 17–20; mass of 629 kg: NASA/JPL MCO arrival press kit (Sep 1999), p. 6.
Source: MCO Mishap Investigation Board, Phase I report, pp. 7, 22; Report on Project Management in NASA (Mar 2000), p. 18.
12 things often said about Mars Climate Orbiter, checked against the primary record. Click each row to see why.
The board's finding (Phase I, pp. 6, 16): the "Small Forces" file did not use metric units and was low by a factor of 4.45.
Aimed at 226 km, final estimate about 57 km, below the 80 km survival limit (pp. 13–14).
The navigation team ran 3 missions at once and MCO had a single navigator (board, p. 24). MPIAT judged Mars '98 "underfunded by at least 30 percent".
Broadly right: NASA's 30 Sep 1999 release says a team in Colorado (industrial partner Lockheed Martin) and a team in California (JPL) used different units. But it was not "NASA metric, the company imperial": the shared spec required metric, and the board does not pin the blame on one company.
Nobody knows for sure. The board wrote that it "either was destroyed in the atmosphere or re-entered heliocentric space" (p. 7).
This figure comes from 1999 press reports (Reuters, 1 Oct 1999); it is not in the investigation reports. NASA's official numbers cover the orbiter and lander together: $193.1M development and $42.8M operations (NASA/JPL press kit, Sep 1999).
MPIAT said the slogan works "if properly applied". The culprit was applying it badly: fewer people, less money, fewer checks.
Not doing it is one of the 8 contributing causes, and it was designed to raise the orbit to a safe altitude. But it had never been prepared, and the board does not claim it would surely have worked (pp. 19–20).
The board listed 8 contributing causes. NASA's Edward Weiler: "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 AMD software installed on the spacecraft used metric units for the computation and was correct." "Mismodeling only occurred in the ground software" (p. 17).
From spring 1999 there were working-level concerns about orbit solutions that did not match. They were "only informally reported" and "were not resolved" (p. 13).
The lander was lost on 3 Dec 1999 for a different reason: the "probable cause" was premature shutdown of the descent engines after a spurious touchdown signal at leg deployment (JPL Special Review Board, Mar 2000).
Sources: MCO Mishap Investigation Board, Phase I (Nov 1999); NASA release 99-113 (30 Sep 1999); MPIAT (Mar 2000); NASA/JPL press kit (Sep 1999); Reuters (1 Oct 1999); JPL D-18709 (22 Mar 2000).
Scores are Mark's assessment based on the sources in Part 8, not figures from a study.
A very ordinary error that had to pass through four layers of checks to become a disaster. That makes it far more worth studying than a joke about units.
Mars Polar Lander was lost on 3 December 1999. JPL's review board concluded the software probably mistook a spurious signal at leg deployment for touchdown, allowing engine shutdown at about 40 m altitude. A completely different bug, but the independent MPIAT team put both losses in the same picture: Mars '98 was short of money, short of people, and light on checks.
The MCO board's second report (March 2000) set out the vision "Mission Success First" and called for an environment where "problems may be raised without fear of reprisal".
Sources: NASA release 99-113 (30 Sep 1999); JPL D-18709 (22 Mar 2000); MPIAT (14 Mar 2000); Report on Project Management in NASA (13 Mar 2000), pp. 6, 37.
Applies wherever one team passes numbers, files or an API to another: finance and sales, backend and frontend, supplier and customer.
Source: Report on Project Management in NASA (Mar 2000), p. 40; MCO board, Phase I, pp. 22–23, 30.
Primary documents first. Page numbers follow the PDF and may be off by one. Where only a press source exists, it is flagged in the Fact check section.