Investigation and Response
Immediately upon the crew's return, NASA Administrator Paine and Deputy Administrator George Low appointed a review board – chaired by NASA Langley Research Center Director Edgar M. Cortright and including Neil Armstrong and six others – to investigate the accident. The board's final report, sent to Paine on June 15, found that the failure began in the service module's number 2 oxygen tank. Damaged Teflon insulation on the wires to the stirring fan inside Oxygen Tank 2 allowed the wires to short-circuit and ignite this insulation. The resulting fire quickly increased pressure inside the tank and the tank dome failed, filling the fuel cell bay (SM Sector 4) with rapidly expanding gaseous oxygen and combustion products. The escaping gas was probably enough by itself to blow out the aluminum exterior panel to Sector 4, but combustion products generated as nearby insulation ignited would have added to the pressure. The panel's departure exposed the sector to space, snuffing out the fire, and it probably hit the nearby high-gain antenna, disrupting communications to Earth for 1.8 seconds. The sectors of the SM were not airtight from each other, and had there been time for the entire SM to become as pressurized as Sector 4, the force on the CM's heat shield would have separated the two modules. The report questioned the use of Teflon and other materials shown to be flammable in supercritical oxygen, such as aluminum, within the tank. The board found no evidence pointing to any other theory of the accident.
SM Sector 4 cover panel just prior to ejection during a test
SM Sector 4 cover panel being ejected during a test
Mechanical shock forced the oxygen valves closed on the number 1 and number 3 fuel cells, putting them out of commission. The sudden failure of Oxygen Tank 2 compromised Oxygen Tank 1, causing its contents to leak out, possibly through a damaged line or valve, over the next 130 minutes, entirely depleting the SM's oxygen supply. With both SM oxygen tanks emptying, and with other damage to the SM, the mission had to be aborted. The board praised the response to the emergency, "The imperfection in Apollo 13 constituted a near disaster, averted only by outstanding performance on the part of the crew and the ground control team which supported them."
Oxygen Tank 2 was manufactured by the Beech Aircraft Company of Boulder, Colorado, as subcontractor to North American Rockwell (NAR) of Downey, California, prime contractor for the CSM. It contained two thermostatic switches, originally designed for the command module's 28-volt DC power, but which could fail if subjected to the 65 volts used during ground testing at KSC. Under the original 1962 specifications, the switches would be rated for 28 volts, but revised specifications issued in 1965 called for 65 volts to allow for quicker tank pressurization at KSC. Nonetheless, the switches Beech used were not rated for 65 volts.
At NAR's facility, Oxygen Tank 2 had been originally installed in an oxygen shelf placed in the Apollo 10 service module, SM-106, but which was removed to fix a potential electromagnetic interference problem and another shelf substituted. During removal, the shelf was accidentally dropped at least 5 centimeters (2 in) because a retaining bolt had not been removed. The probability of damage from this was low, but it is possible that the fill line assembly was loose and made worse by the fall. After some retesting (which did not include filling the tank with liquid oxygen), in November 1968 the shelf was re-installed in SM-109, intended for Apollo 13, which was shipped to KSC in June 1969.
The Countdown Demonstration Test took place with SM-109 in its place near the top of the Saturn V and began on March 16, 1970. During the test, the cryogenic tanks were filled, but Oxygen Tank 2 could not be emptied through the normal drain line, and a report was written documenting the problem. After discussion among NASA and the contractors, attempts to empty the tank resumed on March 27. When it would not empty normally, the heaters in the tank were turned on to boil off the oxygen. The thermostatic switches were designed to prevent the heaters from raising the temperature higher than 27 °C (80 °F), but they failed under the 65-volt power supply applied. Temperatures on the heater tube within the tank may have reached 540 °C (1,000 °F), most likely damaging the Teflon insulation. The temperature gauge was not designed to read higher than 29 °C (85 °F), so the technician monitoring the procedure detected nothing unusual. This heating had been approved by Lovell and Mattingly of the prime crew, as well as by NASA managers and engineers. Replacement of the tank would have delayed the mission by at least a month. The tank was filled with liquid oxygen again before launch; once electric power was connected, it was in a hazardous condition. The board found that Swigert's activation of the Oxygen Tank 2 fan at the request of Mission Control caused an electric arc that set the tank on fire.
The board conducted a test of an oxygen tank rigged with hot-wire ignitors that caused a rapid rise in temperature within the tank, after which it failed, producing telemetry similar to that seen with the Apollo 13 Oxygen Tank 2. Tests with panels similar to the one that was seen to be missing on SM Sector 4 caused separation of the panel in the test apparatus.