The most critical consumables were water, used to cool the CSM and LM systems during use; CSM and LM battery power, the CSM batteries being for use during reentry and the LM batteries needed for the rest of the mission; LM oxygen for breathing; and lithium hydroxide (LiOH) filter cannisters used to remove carbon dioxide from the spacecraft cabin atmosphere. These consumables, and in particular the water and LiOH cannisters, appear to be extremely marginal in quality shortly after the accident, but once the LM was powered down to conserve electric power and to generate less heat and thus use less water. the situation greatly improved. Engineers at MCS developed a method which allowed the crew to use materials onboard to fashion a device allowing the use of the CM LiOH cannisters in the LM cabin atmosphere cleaning system. At splashdown time, many hours of each consumable remained available.

With respect to the steps taken after the accident, Mission Control and the crew worked, under trying circumstances, as well as was humanly possible, which was very well indeed.

The Board's conclusion that the Apollo 13 accident resulted from an unusual combination of mistakes, coupled with a somewhat deficient and unforgiving design, is based on the Board's in-depth analysis of the oxygen tank, its design, manufacturing, test, handling, checkout, use, failure mode, and eventual effects on the rest of the spacecraft.

OXYGEN TANK #2 HISTORY

On February 26, 1966, the North American Aviation Corporation, now North American Rockwell (NR), prime contractor for the Apollo command and service modules (CSM), awarded a subcontract to the Beech Aircraft Corporation (Beech) to design, develop, fabricate, assemble, test, and deliver the Block II Apollo cryogenic gas storage subsystem. This was a follow-on to an earlier subcontract under which the somewhat different Block I subsystem was procured.

Manufacture

The manufacture of oxygen tank #2 began in 1966. In its review, the Board noted that the design inherently requires during assembly a substantial amount of wire movement inside the tank, where movement cannot be readily observed, and where possible damage to wire insulation by scraping or flexing cannot be easily detected before the tank is capped off and welded closed. It does not appear, however, that these design deficiencies played any part in the accident. Several minor manufacturing flaws were discovered in the oxygen tank #2 in the course of testing. A porosity in a weld on the lower half of the outer shell necessitated grinding and rewelding. Rewelding was also required when it was determined that incorrect welding wire had been inadvertently used for a small weld on a vacuum pump mounted on the outside tank dome. The upper fan motor originally installed was noisy and drew excessive current. The tank was disassembled and the heater assembly fans, and heaters were replaced.

Following acceptance testing at Beech; during which the tank was filled and detanked without apparent difficulty, oxygen tank #2 was shipped to NR on May 3, 1967, for installation, which was completed on March 11, 1968, on a shelf to be installed in service module 106 for flight in the Apollo 10 mission.

From April 27 to May 29, 1968, the assembled oxygen shelf underwent standard proof pressure, leak, and functional checks. One valve on the shelf leaked and was repaired, but no anomalies were noted with regard to oxygen tank #2, and therefore no rework of oxygen tank #2 was required.

On June 4, 1968, the shelf was installed in SM 106.

Between August 3 and August 8, 1968, testing of the shelf in the SM was conducted, including operation of the heater controls and fan motors. No anomalies were noted.

Due to electromagnetic interference problems with the vacuum pumps on cryogenic tank domes in earlier Apollo spacecraft, a modification was introduced and a decision was made to replace the complete oxygen shelf in SM 106. An oxygen shelf was approved modifications was prepared for installation in SM 106. On October 21, 1968, the oxygen shelf was removed from SM 106 for the required modification and installation in a later spacecraft.

During the initial attempt to remove the shelf, one shelf bolt was mistakenly left in place; and as a consequence, after the shelf was raised about two inches, the lifting support broke, allowing the shelf to drop back into place. At the time, it was believed that the oxygen shelf had simply dropped back into place. and an analysis was performed to calculate the forces resulting from a drop of two inches. It now seems likely that the shelf was first accelerated upward and then dropped.

The remaining bolt was then removed, the incident recorded, and the oxygen shelf was removed without further difficulty. Following removal, the oxygen shelf was retested to check shelf integrity, including proof pressure tests, leak tests, and fan and heater operation. Visual inspection revealed no problem. These tests would have disclosed external leakage or serious internal malfunctions of most types, but would not disclose fill line leakage within oxygen tank #2. Further calculations and tests conducted during this investigation have indicated that the forces experienced by the shelf were probably close to those originally calculated, assuming a 2-inch drop only. The probability of tank damage from this incident, therefore, is now considered to be rather low, although it is possible that a loosely fitting fill tube assembly could have been displaced by the event.

The shelf passed these tests and was installed in SM 109, the Apollo 13 service module, on November 22, 1968. The shelf tests accomplished earlier in SM 106 were repeated in SM 109 in late December and early January, with no significant problems, and SM 109 was shipped to KSC in June of 1969 for further testing, assembly on the launch vehicle, and launch.

47-476 0-70 4

Testing at KSC

At the Kennedy Space Center the CM and the SM were mated, checked, assembled on the Saturn V launch vehicle, and the total vehicle was moved to the launch pad.

The Countdown Demonstration Test (CDDT) began on March 16, 1970. Up to this point, nothing unusual about oxygen tank #2 had been noted during the extensive testing at KSC. Cryogenic oxygen loading and tank pressurization to 33 psi was completed without abnormalities. At the time during CDDT when the oxygen tanks are normally vented down to about 50 percent of capacity, oxygen tank #1 behaved normally, but oxygen tank #2 only went down to 92 percent of its capacity. The normal procedure during CDDT to reduce the quantity in the tank is to apply gaseous oxygen at 80 psi through the vent line and to open the fill line. When this procedure failed, it was decided to proceed with the CDDT until completion and then look at the oxygen detanking problem in detail.

On Friday, March 27, 1970, detanking operations were resumed, after discussions of the problem had been held with KSC, MSC, NR, and Beech personnel participating, either personally or by telephone. As a first step, oxygen tank #2, which had self-pressurized to 178 psi and was about 83 percent full, was vented through its fill line. The quantity decreased to 65 percent. Further discussions between KSC, MSC, NR, and Beech personnel considered that the problem might be due to a leak in the path between the fill line and the quantity probe due to loose fit in the sleeves and tube. Such a leak would allow the gaseous oxygen being supplied to the vent line to leak directly to the fill line without forcing any significant amount of LOX out of the tank. At this point, a Discrepancy Report against the spacecraft system was written.

A "normal" detanking procedure was then conducted on both oxygen tanks, pressurizing through the vent line and opening the fill lines. Tank #1 emptied in a few minutes; tank #2 did not. Additional attempts were made with higher pressures without effect, and a decision was made to try to "boil off" the remaining oxygen in tank #2 by use of the tank heaters. The heaters were energized with the 65 volt DC GSE power supply and, about 11⁄2 hours later, the fans were turned on to add more heat and mixing. After 6 hours of heater operation, the quantity had only decreased to 35 percent, and it was decided to attempt a pressure cycling technique. With the heaters and fans still energized, the tank was pressurized to about 300 psi, held for a few minutes, and then vented through the fill line. The first cycle produced a 7 percent quantity decrease, and the process was continued, with the tank emptied after five pressure/vent cycles. The fans and heaters were turned off after 8 hours of heater operation.

Suspecting the loosely fitting fill line connection to the quantity probe inner cylinder, KSC personnel consulted with cognizant personnel at MSC and at NR. It was decided that if the tank could be filled, the leak in the fill line would not be a problem in flight, since it was felt that even a loose tube resulting in an electrical short between the capacitance plates of the quantity gage would result in an energy level too low to cause any other damage. Replacement of the oxygen shelf in the CM would have been difficult and would have taken at least 45 hours. In addition, shelf replacement would have had the potential of damaging or degrading other elements of the service module in the course of replacement activity. Therefore, the decision was made to test the ability to fill oxygen tank #2 on March 30, 1970, 12 days prior to the scheduled Saturday, April 11, launch, so as to be in a position to decide on shelf replacement well before the launch date.

Flow tests were first made with gaseous oxygen on oxygen tank #2 and on oxygen tank #1 for comparison. No problems were encountered, and the flow rates in the two tanks were similar. In addition, Beech was asked to test the electrical energy level reached in the event of a short circuit between plates of the quantity probe capacitance gage. This test showed that very low energy levels would result. Then, oxygen tanks #1 and #2 were filled with LOX to about 20 percent of capacity on March 30 with no difficulty. Tank #1 emptied in the normal manner, but emptying oxygen tank #2 again required pressure cycling with the heaters turned on.

As the launch date approached, the oxygen tank #2 detanking problem was considered by the Apollo organization. At this point, the "shelf drop" incident on October 21, 1968, at NR was not considered and it was felt that the apparently normal detanking which had occurred in 1967 at Beech was not pertinent because it was believed that a different procedure was used by Beech. In fact,

however, the last portion of the procedure was quite similar, although at a slightly lower pressure.

Throughout these considerations, which involved technical and management personnel of KSC, MSC, NR, Beech, and NASA Headquarters, emphasis was directed toward the possibility and consequence of a loose fill tube; very little attention was paid to the extended heater and fan operation, except to note that they operated during and after the detanking sequences.

Many of the principals in the discussion were not aware of the extended heater operations. Those that did know the details of the procedure did not consider the possibility of damage due to excessive heat within the tank, and therefore did not advise management officials of any possible consequences of the unusually long heater operations.

As I noted earlier, each heater is protected with a thermostatic switch, mounted on the heater tube, which is intended to open the heater circuit when it senses a temperature of about 80°F. In tests conducted since the accident, however, it was found that the switches failed to open when the heaters were powered from a 65 volt DC supply similar to the power used at KSC during the detanking sequence. Subsequent investigations have shown that the thermostatic switches used, while rated as satisfactory for the 28 volt DC spacecraft power supply, could not open properly at 65 volts DC with 6-7 amps of current. A review of the voltage recordings made during the detanking at KSC indicates that, in fact, the switches did not open when the temperature of the switches rose past 80° F. Figure 18 shows a thermostatic switch welded closed after application of 12 amperes of 65 volts DC. Further tests have shown that the temperatures on the heater tube subsequent to the switch failures may have reached as much as 1000°F during the detanking. This temperature can cause serious damage to adjacent Teflon insulation, and such damage almost certainly occurred. Figures 19 and 20 show the condition of wires, such as those used in the fan motor circuit, after they have been subjected to temperatures of about 1000° F.