The End of an Error

The last landing of a Space Shuttle provided a poignant moment for those interested in space exploration. For the foreseeable future U.S. astronauts will be riding expendable Russian rockets into space at some $60,000,000 a pop. That this represents a considerable savings reveals quite a story of innovation gone awry. The Space Transportation System (STS) was intended to make frequent access to space on a reusable vehicle cheaper, safer and more reliable than previous launches on expendable vehicles.

A very significant aspect of this organizational disaster is that no individual could be held responsible.

The original proposal was a way of investing in development of new technology with a large one-time development cost to cut recurring costs of space exploration. The first two configuration designs were rejected for political reasons. It turned out that the ideal political vehicle should have low initial investment and high recurring costs. It should also parcel out work to as many important political constituencies as possible.

In this way, the nation ended up with a configuration which some experts in structural dynamics and configuration design found painful to watch at launch. This can hardly be blamed on configuration designer Max Faget, who had a solid record in configuration design before this, and made the two earlier proposals that were rejected. He retired immediately after the current design was adopted.

The initial design had problems, but they were being solved, slowly and painfully, with punctuation by explosions. Then the political decision to make this the sole launch vehicle for all large payloads made it necessary to stretch a design which had never flown.

Three aspects of the stretch worsened problems: increased payload weight, increased payload size and greatly increased cross-range capability. Engine development was already in unknown territory, with throttled turbopumps handling enormous quantities of liquid hydrogen and liquid oxygen. (Why throttled? Had the motors run flat out, like most rocket engines, the structure would have collapsed at maximum dynamic pressure.)

The thermal protection system became another problem child. Increased payload and range meant heavier and longer thermal loads during reentry. Increasing the thickness of the tiles could help that, but there was already a problem most people don't realize.

Orbital kinetic energy is greater than the energy required to completely melt the orbiter and all its payload. Unless most of this is radiated away during reentry, it will destroy the vehicle. Heat radiation is strongly dependent on temperature. Together these facts mean that the top of those thermal tiles must become red-hot during reentry, then be allowed to cool down without flowing enough to change the aerodynamics before the next cycle of heating. This happens on each S-turn during reentry.

At the bottom of the tiles we encounter another problem. The structure of the Space Shuttle is not titanium, (as implied, for example, by a comment in "Forrest Gump"). A decision to use ordinary aircraft aluminum was made even before the design was finished. This allowed project management to buy a large supply of aluminum while the market was down -- a "cost-saving" measure. Aircraft aluminum begins to weaken at 400 degrees, but the structure was sufficiently overdesigned to remain rigid up to some higher temperature.

The molded surface of those tiles resembles Pyrex, Corelle or Corningware cooking dishes. When you realize this was heated to red heat, becoming somewhat able to flow, like taffy, only two or three inches from the aluminum skin you get an inkling of how close to the edge normal operation of the Space Shuttle always was.

The problem was compounded by the need for space between tiles to accommodate thermal expansion. These gaps had to be filled to prevent hot gases from reaching the skin. The material of choice was Nomex, in the form of cords, pillows, felt and bars. (You may be familiar with Nomex in the flame retardant suits worn by racing drivers. This was never designed to survive prolonged exposure to thousands of degrees.) Keeping these in place was like caulking a boat with many thousands of seams. Missing gap filler and charred filler bar a quarter inch from the skin was commonly found after flights. (If you wonder about my use of inches here, measurements on the Space Shuttle were made in inches to the nearest 1/100 th of an inch, even on large structures like the external tank. European collaborators found this amusing.)

Actual attachment of these custom-molded, non-interchangeable tiles might also be a surprise. They were glued on with RTV, room-temperature vulcanizing rubber. This might be a higher grade than you find in auto parts stores, or it might not. Kits to service the tiles included plastic gloves, tongue depressors, cotton swabs, alcohol, acetone or methyl-ethyl-ketone solvent, razor blades, etc. A repair costing less than $1,000 was cheap. There were tens of thousands of tiles, though the exact number and configuration depended on the individual orbiter, and changed over time. Tile maps were constantly being updated.

Those reusable turbopumps needed to be pulled and inspected after each flight. I believe consideration of the optimum number of flights before rebuild ultimately came down to rebuilding for every flight. This is why there was never a Shuttle lost to engine explosion.

The reaction control system and orbital maneuvering system also had to be pulled and overhauled for each flight. Auxiliary power units (APU) also needed to be overhauled. The fuel for all these subsystems had to be removed from the orbiter before it was safe for personnel to work on other subsystems.

What was left of the reusable vehicle after you removed all these things was a kind of hulk. The result was a long turn-around process with high costs. The reusable part of the design criteria was partially met, cost savings never materialized.

At the point where the system was originally certified as out of engineering development some $35 billion had gone into the program. Redesign of the external tank (twice), redesign of the engines, redesign of the solid rocket boosters, and constant fiddling with the design of the thermal protection system all took place later.

Nothing I've said above was unknown 20 years ago. Why, prior to the collapse of the entire program, was there no way to escape from historical decisions made early in the history?

Does any of what I've said here about problems of innovation in government-funded research sound remotely applicable to the problem of backing out of decisions made decades ago about research on ME/CFS?
Likes: Sam Carter


Nothing I've said above should be taken as attacking the thousands of people who really wanted to advance space exploration and put their hearts into trying to make it work, or the astronauts who risked their lives, with 14 losing them. A major part of this tragedy was the misdirection of all this skilled effort.

Now that the program is winding down it is possible to evaluate the program in a way accounting tricks and political rhetoric cannot hide. The time to develop the Shuttle was roughly the same as the time to develop the Saturn V system used in the Apollo Moon missions. The cost was higher. You can avoid arguments over amortized costs by putting everything in a lump and dividing by 135 missions. This will give you the surprising estimate that Space Shuttle missions to low-Earth orbit (LEO) cost more apiece than full-scale Apollo Moon landings, even adjusted for inflation.

From the moment it was said to have left engineering development to the very end schedules regularly slipped and costs escalated. There never was a return on initial investment. Most groups who attempted to perform missions using the Shuttle came to regret it. What the program accomplished, more than anything else, was to soak up funds which might have gone toward developing alternative launch systems. This country is now decades behind in this field.

One more point about safety and reliability, several times in the course of the program people from outside the agency made estimates of the probability of catastrophic failure as low as one in 100 missions. Internal analyses were always much better. After the Challenger disaster, physicist Richard Feynman submitted a minority opinion as part of the Rogers commission report. He felt that, despite recommended improvements, there were so many paths to failure that the probability remained around one in 100. Today, we know the answer: two catastrophic failures in 135, or one in 67.5 missions. His 'pessimistic' assessment was actually optimistic.
At the moment when Challenger exploded, without any clear warning displayed on consoles in the firing room at KSC, three different engineering groups all thought it must be the result problems with subsystems for which they were responsible. One obvious candidate cause was an explosion of a turbopump driving the main engines. (Such explosions took place repeatedly during engineering development before 1980.) A second plausible failure was a structural failure in the external tank as it passed through maximum dynamic loading. A third was a failure of a solid rocket booster (SRB), which turned out to be the true cause.

Even that didn't exhaust the possibilities. Those huge SRBs are unusual as solid rockets with gimbaled nozzles. (These are driven by a monopropellant hydrazine motor.) When telemetry showed all gimbaled motors, both solid and liquid fueled, way off in their thrust vectors just prior to the explosion, there was concern that a failure in one of those nozzles had unbalanced thrust of the entire system. It turned out, the system was doing its best to compensate for reduced thrust in the SRB that had burned through its casing, and was losing pressure.

My own first guess, as an outsider, was that there had been a hydrogen leak. These plagued the program on most tankings, even though the vast majority of leaks were much too small to cause fires. Against a blue sky, a hydrogen flame is almost invisible. It took some time to establish that the flame from the SRB preceded any problem with the external tank.

Even with all these possible single-point failures in life-critical systems, we have not exhausted the possibilities. When I saw the 'red team' examining the ice on the launch platform on television, I turned off my TV. It was inconceivable to me that they would launch under those conditions. I left the radio on while I showered to hear when the next launch attempt would be made. The news that they had launched brought me out dripping.

I might even claim to have known that they were outside the temperature range under which the SRBs had been tested. (Qualifying solid rockets for use below freezing is a tricky business, which developers had not attempted.) My reasoning was actually different. With all that solid ice on the launch platform, there was a very real chance the thermal protection system would be damaged too badly by flying debris to allow reentry. Even without conspicuous problems a normal launch generally produced minor damage ('dings') to around 100 tiles. These were acceptable as long as they remained minor, the problem was the sheer number of independent impacts. There was always the possibility a cluster of impacts would produce damage the system could not survive. Maybe this happened, we simply don't know because Challenger didn't survive long enough to reenter.

All this was known many years before the demise of the program. Why didn't anything change?

If stupidity got us into this mess, why can't it get us out of it?
Will Rogers

Blog entry information

Last update

More entries in User Blogs

More entries from anciendaze