At the beginning of May 1990, the shuttle program was on its uppers. Four years since the catastrophic loss of Challenger and her crew of seven, the reusable fleet of orbiters had just completed their tenth post-51L mission by depositing NASA’s scientific showpiece—the $1.5 billion Hubble Space Telescope—into orbit. In less than three weeks’ time, another mission, STS-35, would roar aloft with a payload of ultraviolet telescopes to observe the Universe in unprecedented detail. ASTRO-1 had been long-delayed; it was supposed to fly in March 1986, as the next mission after Challenger, but had been delayed repeatedly. Those delays had even led to rumors that it would be canceled outright, and this informed Commander Jon McBride’s decision to resign from the mission. A year after McBride’s departure, with Vance Brand now in command, ASTRO-1 seemed ready to go. Columbia stood resplendent on Pad 39A at the Kennedy Space Center, Fla., with no problems being tracked … until the issue with the disconnects.
Deep inside the belly of the shuttle were a pair of 17-inch fuel lines, through which liquid oxygen and hydrogen flowed from the External Tank into the cluster of three main engines. Both “sides” possessed mechanical disconnect fixtures, and shortly before the separation of the External Tank, on the edge of space, a pair of flapper valves were commanded shut by pneumatic helium pressure to prevent further propellant discharge and contamination. The criticality of the disconnects and their flapper valves could not be underestimated; for any inadvertent closure whilst the main engines were firing would have prevented propellant flow from the External Tank and precipitated a catastrophic failure. In the days preceding the launch of STS-35, these disconnects took center stage in one of the most difficult engineering challenges ever faced by the shuttle program.
Before each mission, a Flight Readiness Review scrutinised all pertinent documentation, before formally reaching an agreement on a target launch date. At the beginning of May 1990, Columbia was provisionally scheduled to fly on the 16th, but a problem was noted with a proportioning valve on her Freon coolant loop, requiring its replacement. A second review was convened and settled on the 30th for launch. Then, during the lengthy process to load liquid hydrogen into the External Tank, a tiny leak was detected, close to the tail service mast on the Mobile Launch Platform. Further investigation revealed a much more extensive—and far more worrisome—leak, apparently coming from the disconnect hardware in Columbia’s belly. The launch attempt was called off and the External Tank was emptied and inerted. Since the crew would be working in two 12-hour teams in orbit, they had already begun “sleep-shifting,” and the “blue” team was awake when the news of the launch delay came through. Blue team member Jeff Hoffman recalled waking up his “red” team counterparts with the unwelcome news.
A few days later, on 6 June, a miniature “tanking test” was conducted to identify the exact location of the leak. It soon became clear that Pad 39A was not an appropriate place for exploratory work to be undertaken, and on the 12th the STS-35 stack was rolled back to the Vehicle Assembly Building (VAB). Columbia was removed from the stack and returned to the nearby Orbiter Processing Facility (OPF) for repairs on the 15th. Launch was rescheduled for mid-August. The shuttle “side” of the disconnect hardware was replaced with a set borrowed from the new orbiter, Endeavour, which was then undergoing final construction as Challenger’s replacement, and apparatus for the External Tank “side” arrived shortly thereafter from its assembly facility at Michoud in New Orleans. Jeff Hoffman remembered that it was a tough time on him and his crewmates. They were all fully trained, at the peak of mental and physical readiness, and yet were forced to sit through endless simulator sessions with no new launch date.
As this work was being undertaken, Atlantis progressed smoothly—or so it seemed—towards her own launch on STS-38 in July. As a precautionary measure, shortly after the STS-38 stack arrived at Pad 39A on 18 June, NASA decided to perform a tanking test on Atlantis to verify that she was not similarly affected. On the 29th, liquid hydrogen was pumped into her External Tank … and to engineers’ dismay the same problem appeared: after the fuelling process moved from “slow fill” to “fast fill,” concentrations of gas were found in the vicinity of her disconnect hardware. It was small and was described by NASA as “both temperature and flow-rate-dependent,” but the agency was convinced that the two leaks in Columbia and Atlantis represented nothing more than coincidence.
To identify the source, more instrumentation was fitted around the disconnect and a second tanking test was performed on 13 July. Sealants were added in a bid to stop the leak, but a third test on the 25th revealed that the problem persisted. Two weeks later, in the pre-dawn darkness of 9 August, the STS-38 stack rolled back to the VAB, providing a unique photo opportunity in the process, as the STS-35 stack passed it on its way back out to Pad 39A. Despite the similarities between the problems, NASA managers were convinced that the hydrogen leaks were unrelated. “Incorrect torqueing of bolts around the flange interface between the tank and the orbiter caused the Atlantis mishap,” Flight International told its readers on 8 August. “The Columbia leak was caused by a faulty seal in the drive mechanism used to close the flapper valve in the disconnect.”
STS-35 was rescheduled for launch on 1 September. Misfortune, though, was far from finished with the unlucky flight. Two days before launch, an avionics box on ASTRO-1 failed and required replacement, prompting a delay of several days. Then, on the evening of the 5th, technicians began to load propellants into Columbia’s External Tank … only to discover hydrogen gas again leaking into the aft compartment of the orbiter. The maximum allowable rate was 660 parts per million; but the actual rate of leakage was upwards of 6,500 ppm. …
It was now apparent that two separate leaks had evolved; the data indicated that the leakage from the disconnect was gone, but a new one had emerged, somewhere in Columbia’s aft engine compartment. A package of three hydrogen recirculation pumps in Columbia’s aft compartment were removed and replaced, together with a damaged Teflon seal on one of the main engines, but to no avail. In the hours preceding another launch attempt on 18 September, leaking gas was again detected, emerging at a rate of 6,700 ppm. STS-35 was indefinitely postponed until the leak was resolved. Space Shuttle Program Manager Bob Crippen assembled a “tiger team” to investigate the disconnects and totally “retorque” Columbia’s entire liquid hydrogen system.
Crippen assigned veteran engineer Bob Schwinghamer, from the Marshall Space Flight Center in Huntsville, Ala., to lead the investigation. In his NASA oral history, Schwinghamer remembered NASA Deputy Administrator J.R. Thompson telling him, without a hint of humour, that he had a one-way ticket to Florida; he was not to return to Alabama until the hydrogen leak was solved. Schwinghamer’s team spent three months at the Kennedy Space Center, setting up an intricate fault tree and co-ordinating across a huge number of personnel, spread across several NASA centers. By the time they completed the final tanking test on 30 October 1990, Schwinghamer could confidently declare that Columbia was now “the soundest leak-free orbiter at that time in the fleet.” The explosive nature of hydrogen meant that a leak of any sort could not be tolerated, and it certainly surprised Vance Brand, the commander of STS-35, that an orbiter was being grounded for such a long period. It marked a change in attitude from the way in which NASA management had worked before the Challenger accident.
The September 1990 delay was clearly remembered by Jeff Hoffman, many years later. Rather than risk having their children out of school for a lengthy spell, several of the crew decided to pay, personally, for air tickets from Houston to Florida and asked neighbours to take them to the airport. In the hours preceding the 18 September attempt, the Hoffman children—aged 11 and 15 at the time—were seated aboard the aircraft, with one engine in the process of starting up, when the Continental departures desk received a call from Hoffman himself. Another leak had emerged; there was no point in the children flying down to Florida. At the last possible moment, the engines fell quiet, the doors opened, and the disappointed children got off. Not until December would they see their father launch. By this time, the effort to fix the hydrogen leaks had cost NASA $3.8 million, and the agency suspected that Columbia’s problems originated from a complete disassembly of her main engines following the STS-32 mission in January 1990 to remove polishing grit from her fuel lines. When the engines were reassembled, the seals were imperfectly fitted and minute glass beads contaminated the disconnect hardware. From the beginning of June, the attention of engineers was drawn solely to the disconnect leak, which posed a more serious problem, and the seals were overlooked. “As a result,” explained Flight International in mid-November, “NASA has introduced a new processing programme in which key engine components will be checked for leaks before the engines are finally assembled.”
In late November 1990, NASA announced that a new hydrogen dispersion apparatus would be added to the Mobile Launch Platform for future missions, beginning with STS-39. The system would provide a nitrogen-rich air flow around the disconnect hardware, thus helping to disperse hydrogen concentrations. As for the disconnects themselves, NASA was already working on plans with Rockwell International to develop an upgraded system, somewhat narrower at 14 inches in diameter, and contracts worth upwards of $27.6 million were awarded in February 1991. It was expected that newer disconnect hardware would help to prevent inadvertent closure of the flapper doors during ascent, which threatened catastrophe. Ultimately, the new disconnect project was cancelled in 1993, but much of the technology behind it was employed to improve the safety and reliability of the existing 17-inch hardware. Furthermore, the space agency instituted more rigorous rules around the issue of hydrogen leakage, embedding them more firmly within its Launch Commit Criteria.
By 2 December 1990, the prophets of doom were noting that STS-35 would be the 13th post-Challenger mission—surely a portent of ill-fortune—but for the astronauts the impending launch after such a long wait came as a blessed relief. Jeff Hoffman knew that comets, like Halley, were supposed to bring bad luck, but by the end of 1990 there were no cometary targets for ASTRO-1 to observe. “We all know comets are harbingers of bad news,” Hoffman rationalised. “This time, we have no comet. So we’re going to go.”
And go they did. At 1:49 a.m. EST on the 2nd, Columbia rose gloriously into orbit on a mission which lit up the darkened Florida sky for hundreds of miles. The gremlins were not yet done with ASTRO-1, for the flight was plagued by technical difficulties and payload problems. However, the resurgence of this mission from the ashes of despair in the wake of Challenger, and its revival from the hydrogen leaks, makes it quite remarkable and an astonishing triumph over adversity.
This is part of a series of history articles, which will appear each weekend, barring any major news stories. Next weekend’s article will focus on Gemini 3, a five-hour voyage in March 1965 which vindicated the little spacecraft which would pave the way toward landing Americans on the Moon.