The Space Reviewin association with SpaceNews

Thor launch vehicle
Two Thor launch vehicles suffered different failures but with the same root cause: a lack of end-to-end testing. (credit: US Air Force)

Launch failures: two Thors, one problem

The US Air Force launched hundreds of Thor space boosters over a period of more than 20 years, and the vast majority used the traditional space launch approach. Contractors, mainly the companies that built the flight hardware, would design, integrate, assemble, and test the vehicles, all under the control of an Air Force System Program Office (SPO) and Aerospace Test Group. However, a significant minority of Air Force Thor launches used a completely different approach, termed “Blue Suit Launch.”

Over most of the period that the Air Force used the Thor an Air Defense Command unit, the 10th AERODS, launched Thors—all of which were converted SM-75 missiles—on both ballistic and orbital space missions. The 10th AERODS was the unit that maintained a limited anti-satellite capability from the mid 1960s to the early 1970s, based on nuclear-armed Thor missiles located on Johnston Island in the Pacific. The unit even launched live nuclear weapons on Thors, detonating the bombs at very high altitudes for research programs. It seemed logical to use the 10th AERODS to launch orbital space missions using converted Thor missiles as well. The main payloads of these launches were Defense Meteorological Satellite Program (DMSP) missions, basically military versions of NOAA’s polar-orbiting weather satellites.

The failure investigation showed that F-34 had not carried enough fuel for a very simple reason. There literally was a typo in the data that accompanied the engine.

There never really was a true Air Force System Program Office for the Blue Suit Thor program. The Thors launched by the 10th AERODS were considered to be “operational” and followed the lead of the Air Force’s aircraft and missile systems. Air Force Logistics Command provided supplies via the Sacramento Air Logistics Center. An Air Force SPO procured the upper stages, if there were any. The Air Force DMSP SPO procured the spacecraft, and was responsible for payload integration. This unusual and rather non-systematic approach all worked pretty well for years, and then came the DMSP F-34 mission.

F-34 was to be the last of the DMSP Block 5C missions before the Block 5D-1 satellites ushered in major changes. The vehicle consisted of a converted IRBM Thor LV-2F booster, a Burner IIA upper stage, and the DMSP spacecraft. The Thor guidance system was quite rudimentary and depended on the Burner upper stage for the inertial data required for the mission.

On February 19, 1976 the vehicle lifted off from the SLC-10W pad at Vandenberg AFB and appeared to climb to orbit normally. However, the spacecraft never made it to the correct orbit; it made, at most, one very low orbit before it reentered.

A full formal mishap investigation revealed the reason for the failure; it was a simple problem with a rather complex cause. The Thor ran out of fuel before it could provide the required velocity to the upper stages.

Like most space boosters, the Thor had no propellant utilization system. Instead, the data recorded from the engine’s test runs was used to calculate the required amount of RJ-1 (ramjet fuel, like RP-1, based on kerosene) for the mission. A fuel counter at the launch pad, one no more sophisticated than that used when you fill up your car, showed the amount loaded in the booster, and that was that. The LOX tank was always filled all the way up for each launch, but depending on the characteristics of the specific serial numbered engine to be flown, the fuel load was varied. Some engines burned more fuel than others, which was not good for performance, since you had to carry that extra fuel along with you. And you could not simply fill the fuel tank up to brim full, because that extra fuel represented weight that had to be carried along with the upper stages and payload. Too much fuel left at the end of the burn and you would not make it to the required orbit. Too little fuel to reach the required velocity and you would not make the required orbit. Each mission walked a tightrope with only a very small safety net—the net being the margin between what absolutely was needed to fly the mission and how much extra fuel could be carried. The switch to the more dense RJ-1 fuel rather than the more common RP-1 had even been done earlier in the DMSP history to pick up some extra performance margin.

The failure investigation showed that F-34 had not carried enough fuel for a very simple reason. Analysis of the data on the engine revealed that the supplied mixture ratio information was wrong. There literally was a typo in the data that accompanied the engine. No one had questioned that data; not only was it wrong but there was no way it could be right. It was the equivalent of looking at new cars at an auto dealership and noting that for two identical models one was shown as getting 30 MPG and the other one 35 MPG—and then being told that the data was correct; one car was just put together a lot better than the other one, somehow.

The worst part was that the F-34 mission should never even have been attempted in the way it was. As with most spacecraft, the DMSP 5C series tended to gain weight as newer versions came off the assembly line, and the last one flown would be the heaviest of all. So, the top performing engine was chosen quite deliberately for that last mission. Given that no such exceptional engine was theoretically possible, the mission could never have flown successfully even if the engine data had not contained a typo.

To detect the problem that killed the F-34 mission required the kind of analysis and modeling that was beyond what the program had available. AFLC did not do that kind of thing, neither did the 10th AERODS, and nor did any of the supporting contractors. The spacecraft program office and the SPO that procured the upper stages both had limited areas of responsibility, and that did not include modeling engine performance.

The Air Force Mishap Investigation Board recommended that the Thor program henceforth be run like all the other Air Force booster programs. A contractor, McDonnell Douglas Astronautics Company (MDAC), should be given responsibility for the overall program. This meant that the vehicle also should be subject to the kind of mission assurance scrutiny as were all other booster programs. This did not happen.

The Air Force Mishap Investigation Board recommended that the Thor program henceforth be run like all the other Air Force booster programs. This did not happen.

The “radical new approach”—which was simply doing things the way all other Air Force booster programs did them—was not embraced by everyone. In particular, it would have meant the demise of the 10 AERODS, and there was some feeling in the Air Force that the service needed at least one launch organization that did not rely on contractors. Instead, the Blue Suit launch crew was retained and an Air Force Systems Command SPO created to support it—or sort of, anyway. The launch vehicle SPO’s responsibilities were primarily to replace AFLC in the logistics area. Payload integration, such as it was, remained in the hands of the DMSP SPO. What all this meant was that there still really was no one looking at the total stack: the entire flight vehicle and its supporting equipment.

Following DMSP Block 5C came Block 5D-1, and it represented a radical change. The spacecraft was much larger and far more complex. Among other things, the spacecraft would guide both the booster and its own upper stages during ascent.

Block 5D-1 proved to be a rather checkered program. The first mission, F-1, attained the proper orbit, but the spacecraft spun out of control shortly thereafter. Amazingly, it was recovered and made fully operational some six months later. The F-2 mission did not spin up on orbit like F-1, but a programming error in the guidance software led it to enter an orbit with the wrong inclination, resulting in the spacecraft drifting into less useful orbits. The F-3 mission got it all right and worked pretty well, the spacecraft even outliving its own primary sensor. Mission F-4 attained the proper orbit but died early due to a wiring error that prevented proper battery management.

Then came the DMSP F-5 mission, the last one that would fly on a Thor. Weight growth in the follow-on series, Block 5D-2, led to the decision to abandon the Thor for converted Atlas E ICBM’s starting with mission F-6.

A couple of weeks before the scheduled launch in July 1980, the assembled Thor LV-2F DMSP F-5 vehicle was being erected at the launch site, SLC-10. The Thor was unique among US boosters in the 1970s and ’80s in that it was assembled in a horizontal position, unfueled, and erected a few hours before launch. A hydraulic system would push the vehicle to the vertical position, including its launch mount. It was a normal everyday erection being done as part of the usual tests, the vehicle held in a shallow angle before going vertical—and then the booster suddenly dropped much of the way back down. Examination revealed that one of the motorized launch pins had broken; it was the pin at the top, the one in tension. The degree of sideways impact was impossible to judge; there was no active instrumentation to measure it. After considerable head scratching, visual examination, and electrical tests, the vehicle was cleared to fly.

Liftoff on July 14, 1980 looked good but was accompanied by a loss of telemetry from the booster. Fortunately, it was a clear day and the vehicle could be observed climbing skyward just fine and data from the spacecraft showed it was riding smoothly. The Thor completed its burn and the solid second stage took over. All looked good. And then the second stage burn was done and the third stage started. Seconds later, all telemetry from the spacecraft was lost.

Obviously the third stage motor had exploded, said some. But eventually investigation proved that the spacecraft had descended in a virtually uncontrolled death spiral with the rocket motor burning normally.

Finally, after some false starts and falser stops, the conclusion reached was that the jolt imparted by the broken launch pin had misaligned the electrical connectors between the second and third stages. The connectors were of a design known to be sensitive to such misalignment, and they failed to separate. The spacecraft, with its integral third stage motor firing, had tried to tow the second stage behind it. The resultant loads had ripped the wiring out of the spacecraft, causing massive short circuits that drove the guidance computer nuts.

A key principle discovered at great cost in the space program was that of the “End-to-End Test.” Hardware not only had to be tested as individual components and subassemblies but also when it was all hooked together.

Once again a DMSP mission had been lost because no one was looking at the total stack. No one was looking at it because no one office was responsible for it. And that deficiency included both examination of the design of the electrical staging connectors and the possibility that the launch mount pins just might need to be examined after being used on dozens of launches and test erections.

A key principle discovered at great cost in the space program was that of the “End-to-End Test.” Hardware not only had to be tested as individual components and subassemblies but also when it was all hooked together. This was not always possible in launch vehicles, and so the areas where a full end-to-end test could not be accomplished deserved special scrutiny. However, intrinsic to the end-to-end test concept was the existence of an organization that viewed the whole vehicle as truly a whole. No one had that job for the Thor program; it was unique in that respect. Ultimately that was the real cause of both the DMSP Thor launch failures.

Lessons learned from launch failures cover not only specific hardware problems but also the philosophical approaches and management techniques required to ensure success. And these more nebulous concepts are both the most important things to learn and also the hardest to discern.