The Space Reviewin association with SpaceNews

NSRC 2020

launch failure
The failure of the Atlas 76E/Navstar 7 launch in December 1981. (credit:USAF)

Launch failures: management

Bookmark and Share

Management is a factor in any kind of a failure. Failures may be due to sins of omission or definite actions, specific flawed decisions, or more vague philosophical approaches. In any case, management is usually a factor, obvious or not, but there are some noteworthy cases in which management decisions are the ultimate cause of launch failures.

Space Shuttle

One of the most outstanding examples of a management-driven failure occurred with the loss of the Space Shuttle Challenger on January 28, 1986. As previously described (See “Weather and launch failures”, The Space Review, July 12, 2010), NASA had not established temperature limits for shuttle launches, a serious failing in its own right. But in reality, a much broader series of management decisions drove the loss of the Challenger.

The Space Shuttle had to work, a fact that had been a given for well over a decade, and making sure that vision came true had become even more urgent.

The Space Shuttle program was designed first and foremost to make NASA’s human space program bulletproof by making it essential. The shuttle would be the only American launch vehicle, replacing all existing expendable launch vehicles (ELVs). It supposedly would be far more capable than the ELVs and much cheaper as well. Ordered “off the drawing board” with no proof-of-concept prototype or X vehicle, and no fly-off competition, the shuttle had to work. Long before the first shuttle ever flew, existing Atlas rockets sitting in warehouses were scrapped, run over with bulldozers. ELV launch pads and launch support equipment were allowed to decay in anticipation of their abandonment. Most new production of rocket engines was stopped; from the early ’70s on left over bits and pieces would have to suffice for the remaining launches. Development of any new launch vehicles except for the shuttle was cancelled.

In 1977, NASA confidently said that each shuttle flight would cost only $14 million each, or about $6 million less than a Delta 2914 launch at the time. The shuttle theoretically would be capable of carrying up to two medium-sized spacecraft on each flight, and those users would be charged $18 million each, for a hefty $22 million in “profit” on each launch. The system would be capable of flying 50 missions a year, or about one per week, with a two-week Christmas vacation built in. The maximum payload of 40,000 pounds (18,100 kilograms) to LEO would far exceed that of even the Titan III series of boosters.

The first shuttle launch was on April 12, 1981; there were a total of two that year. There were three shuttle launches in 1982, four in 1983, five in 1984, and nine in 1985. This represented definite progress, but it still was a long way from the advertised 50 missions a year. In addition, among the first 25 shuttle launches were four mission failures, where the shuttle made orbit and returned safely but all mission objectives were not successfully achieved.

Meanwhile, the shuttle’s mandatory monopoly was falling apart. In 1984, the US Air Force began the Complementary Expendable Launch Vehicle (CELV) program, originally designed to “complement” the shuttle for just ten military missions; the Titan IV won the competition. The Titan IV builder, Martin Marietta, announced it would use the new Air Force program as a springboard to create a Commercial Titan III program. General Dynamics’ entry into the CELV competition was not selected but the company had already made the decision to commercially market its Atlas Centaur booster. Internationally, the European Space Agency was still proceeding with its Ariane booster, which had launched 15 missions through the end of 1985, including three mission failures. Like the shuttle, the Ariane was a heavily subsidized program, and like the shuttle, some missions would fly on it, regardless.

And perhaps most significantly of all, the Commercial Space Launch Act of 1984 (CSLA) required and encouraged the US government to facilitate, enable, and support purely private space launch efforts at its launch ranges.

NASA planned to increase its number of shuttle launches considerably, and key to this was the activation of a new launch pad, LC-39B. The first mission planned for the new pad was STS-51L, with a launch date in late January 1986.

Meanwhile, disturbing data had been collected on the shuttle’s Solid Rocket Booster design deficiencies. While test requirements had mandated firing tests at as low as 40°F (4°C), those tests had never been accomplished. Examination of dismantled SRBs showed that problems occurred with the O-ring seals at temperatures well above even 40°F. Some engineers expressed concerns, especially about the very cold temperatures that existed in late January 1986. But still no launch temperature restrictions had been implemented or even defined.

There are procedures for addressing such technical concerns, but the fact was, NASA management could not afford to do so. The Space Shuttle had to work, a fact that had been a given for well over a decade, and making sure that vision came true had become even more urgent. At one of the teleconferences that discussed the SRB problem immediately before the STS-51L launch, the engineers expressing concerns were told, “Quit thinking like an engineer and start thinking like a manager.” NASA “managed” the problem out of existence and made the decision to launch.

US Air Force Atlas failures

The Air Force suffered three Atlas launch failures in 1980 and 1981. The first of these, the NOAA-B mission (see “Launch failures: engine out”, The Space Review, December 31, 2012) experienced a low-performing booster engine followed by an improper payload separation. While the engine problem precipitated the failure, ultimately it was because NASA and NOAA made the management decision to not employ a payload interface that featured a booster-generated separation signal. That decision was based on a desire to meet the NOAA mission call-up requirements without making the investment to keep a spacecraft in readiness at the launch site and thus reduce the time required to get the payload mated to the booster. That approach required a truncated processing schedule and that drove the booster/payload interface requirements. Even after the NOAA-B failure, NASA and NOAA decided to maintain the same booster interface approach for later TIROS NOAA missions.

It turned out to be a battle between management egos and real mission assurance needs—and the egos won.

The next Atlas failure was on December 9, 1980. A booster engine shut down a few hundred milliseconds early due to a loss of lubrication. The investigation concluded that the engine lubrication system featured a number of possible failure modes, including ICBM-vintage quick disconnect fittings that were no longer needed as well as various fittings that were susceptible to stress corrosion. These were known concerns and represented simple and inexpensive problems to fix, but Air Force management was focused on phasing out expendable boosters and there was little or no interest in fixing the older hardware. They did not manage the problems out of existence; they just ignored them.

The third Atlas failure, on December 19, 1981 (see “Launch failures: the ‘Oops!’ factor”, The Space Review, January 31, 2011), was due to an error made during reassembly of one of the booster engines following a corrective action. Normally the pre-launch review of the vehicle history would have addressed possible problems associated with that repair, but the Air Force System Program Office (SPO) was in a state of turmoil in that time period. The SPO chief had been fired; he had never supported the shuttle-only mandate and those that did deeply resented the fact. The new SPO chief was highly qualified, but middle management made briefing the new chief the top priority rather than focusing on preparing for the upcoming mission. At the same time some retired Aerospace Corporation senior officials raised concerns about the previous two Atlas failures and pushed for modifications to the booster that had yet to be even tested. Numerous meetings and discussions on the subject ensued and that took up a great deal of time and energy. A simple boroscope inspection of the repaired booster engine would have revealed the problem but the SPO was so occupied by the other issues that no one pushed for that prudent action. It turned out to be a battle between management egos and real mission assurance needs—and the egos won.

Commercial ELV programs

The 1984 Commercial Space Launch Act meant, among other things, that the US space launch industry was going to have to operate on its own for commercial missions, with far less government oversight than even the aviation industry. Prior to the CSLA, private firms seeking to launch payloads went through NASA to obtain Atlas Centaur or Delta boosters and launch services, using NASA’s existing contracts and launch facilities. Of course, the same was true for shuttle launches carrying commercial payloads, but commercial use of the shuttle ended after the loss of Challenger.

US companies would have to learn to stand on their own two feet, without relying on government quality assurance or even simply the “second set of eyes” that accompanied government contracts. The companies had chafed under what they often thought was unnecessary and extreme government reactions to minor problems, but when things went fully commercial it was time for them to put up or shut up.

The private firms’ reaction to this situation was in part a reflection of how well they were doing competing for government contracts. General Dynamics was stung by the loss of the Titan IV, Titan II, and Delta II Air Force contracts and widely touted what it claimed were the advantages of little or no government involvement. Martin Marietta and McDonnell Douglas voiced far fewer such claims than GD but no doubt were pleased at easier operations and lower costs for commercial launches.

All three of the companies would soon find out that the lack government oversight may have reduced their costs but also increased their responsibilities. General Dynamics suffered three commercial Atlas Centaur failures: one each in 1991, 1992, and 1993. Two of the failures were due to the same cause. The company then exited the business, selling its space launch division to Martin Marietta.

All three companies had made the conscious management decision not to replace government oversight with their own equivalent “second set of eyes,” and all paid for it dearly.

In 1990, one of Martin Marietta’s Titan III Commercial missions failed to deliver the payload due to an error in connecting the wiring harness to the spacecraft, a mistake that a government inspector almost certainly would have spotted. The company launched its last Commercial Titan III mission in 1992, after only a total of four, and then placed its commercial focus on its newly acquired Atlas program.

McDonnell Douglas did not find out how tough going it alone without government oversight could be until later, but the knowledge came with a vengeance. The first two Delta III missions were failures and that effectively ended the program.

All three companies had made the conscious management decision not to replace government oversight with their own equivalent “second set of eyes,” and all paid for it dearly.

The operationalization of Air Force space programs

Air Force Space Command was created in 1982, reenergizing a long-term internal Air Force argument concerning which command should have control of space assets. Air Force Systems Command, responsible for acquiring all of the Air Force’s new weapons systems, had been conducting all the service’s space launches and operating the Air Force Satellite Control Network since the beginning of the space age. It seemed to some that the advent of regular Space Shuttle launches as well as the promise of the Global Positioning System meant the Air Force had to be prepared to conduct space operations on a different basis than the research and development style approach that had been used since the first launches.

Air Force Space Command took over the launch centers at Cape Canaveral and Vandenberg AFB in 1990 and soon began making changes to those operations, touting making things “operational.” And “operational” meant in practical terms the type of people to accomplish the mission. An ‘operational” command had to use “operational” people and that did not mean the heavy emphasis on engineering and program management functions that marked Systems Command’s activities. The Commander of the 45th Space Wing even said to the Eastern Test Range Chief engineer, “We have to set this place up so that a history major rather than an engineer can run a space launch.” The reason was not that the Air Force had a severe shortage of engineers but rather that it had an excess of people like him, history majors. This led to a vision of a range that could be configured for a given launch simply by inserting a floppy disk into a console, a concept that eventually proved to be utterly unworkable.

This emphasis on changing the nature of operations so that the type of people employed could be changed gained new momentum in the early 1990s after Operation Desert Storm. The disintegration of the Warsaw Pact and collapse of the USSR as well as new arms control treaties led to an enormous downsizing of the US military. The entire Air Force would eventually be reduced by 40 percent and a number of aircraft types taken entirely out of service. The Air Force ICBM force would be reduced by more than 50 percent. The Air Force Chief of Staff, Gen. Merrill McPeak, thought that given the need to downsize the Air Force, the emphasis should be on retaining as many operational personnel as possible. He frequently cited the World War II German Luftwaffe as an example of efficiency that the Air Force should emulate.

Demanding to know why the first launch of a new type of spacecraft on a new model of booster from a newly modified and rebuilt launch pad was taking so long, an admiral said the problem was, “There are too many guys in white lab coats walking around on the launch pad.”

And schedules became of premier importance. Gen. Chuck Horner, commander of Air Force Space Command, asserted that the spacecraft should be able to be deployed within 60 days, the time required for military units to deploy overseas. Studies showed that were very few satellites under the control of Air Force Space Command where such an approach was possible. One was the Defense Meteorological Satellite Program (DMSP), and it had been operating on a call-up basis for years, although not actually as short as 60 days. The National Reconnaissance Office would make decisions to launch its spacecraft when required. The Global Positioning System had a very robust constellation that would not require such a 60-day deployment and could not be allowed to decay due to the very large number of civilian users. The Defense Support Program satellites required around a year to prepare, process, launch and accomplish early orbit checkouts, making a 60-day call up impossible.

Even so, there was belief within Air Force Space Command that the schedule could be maintained just by adopting a proper “military” attitude. A US Navy admiral on the Joint Staff in the Pentagon summed up this idea very succinctly. Demanding to know why the first launch of a new type of spacecraft on a new model of booster from a newly modified and rebuilt launch pad was taking so long, he said the problem was, “There are too many guys in white lab coats walking around on the launch pad.”

Finally, after numerous delays during the late 1980s and 1990s, Air Force Space Command could show what it could do. They would launch four Titan IV missions from Cape Canaveral AFS in less than 12 months. “Operationalization” would at last triumph.

The first launch, a Titan IVB with a classified payload, flew successfully on May 9, 1998.

The second launch, a Titan IVA Centaur with a classified payload, launched on August 12, 1998. At T+40 sec the vehicle pitched over, broke up and exploded.

The third launch, a Titan IVB IUS with a DSP payload, launched on April 9, 1999. At T+23,314 sec the second stage of the Inertial Upper Stage failed to separate properly, stranding the spacecraft in a useless elliptical orbit.

The fourth launch, a Titan IVB Centaur with a Milstar payload, launched on April 30, 1999. The Centaur stage wandered off course at T+540 sec and the payload attained a useless orbit.

The August 1998 failure was found to be due to a momentary short circuit in the line that fed electric power to the guidance system. The guidance system reset to T-0 and went into its pre-programmed pitch maneuver, which caused the vehicle to break up. The vehicle had a long history of numerous wiring problems and electrical anomalies, but these were overlooked in order to meet the schedule.

The April 9, 1999 failure was caused by the separation connector for the IUS being over-wrapped with tape designed to protect the wiring harness. The close-out photos of the IUS clearly showed the problem, and review of previous mission’s telemetry data and close out photos revealed the problem had occurred in a less severe manner on those flights as well.

The April 30, 1999 failure occurred because an incorrect roll constant had been entered into the Centaur guidance program. Problems were noted during flight simulations but pressure to meet the schedule led the anomaly being ignored by management

These were the kind of problems that should have been detected by “the guys in white lab coats” but their efforts had been scorned, discounted, and literally decimated for the previous ten years.

The Titan IV failures were viewed by the Air Force as even worse than the two Titan 34D failures of 1985 and 1986. A special very high-level review team was formed to consider just what had gone so terribly wrong with Air Force launch efforts. A retired Air Force Chief of Staff on the review team summed things bluntly: “We used to be able to do this mission. Now we can no longer. The problem is obvious. You made changes in the management structure a decade ago. Fire the people you put in charge and go rehire the ones you fired.”

Many failures clearly can be traced directly to management action and inaction. And most ultimately have been due to high-level management initiatives, such as the shuttle replacement of ELVs, commercial companies discarding proven mission assurance methods, or careerist concerns that overruled practical experience.

While doubtless an accurate assessment, that simplistic approach was impossible to implement. The people ultimately responsible for the debacle were mostly senior officers who had already retired. As for rehiring the people who had been fired, they were off leading new lives and in the late 1990s the aerospace industry as a whole was only able to attain about an 80 percent staffing percentage. The 1990s shakeout and downsizing in the industry had caused many experienced people to seek employment elsewhere; they were not coming back.

In reality the Titan IV failures simply showed that “operationalization” was an abject failure as a management approach, but rather than state that openly the conclusion of the review was that “better team work” was required. Changes were made and years later the Air Force once again separated the space and missile career fields.

It is tempting to say that management is the cause of all launch failures, and in one sense that is true. But as we have seen in the Launch Failures series, failures do still occur even with proper and prudent management. However, other failures clearly can be traced directly to management action and inaction. And most ultimately have been due to high-level management initiatives, such as the shuttle replacement of ELVs, commercial companies discarding proven mission assurance methods, or careerist concerns that overruled practical experience.

Note: we are temporarily moderating all comments submitted to deal with a surge in spam.