Launch failures: the boring stuffby Wayne Eleazer
|
| Component specifications and material standards are probably the most boring aspect of engineering. They may be vital but they certainly are not exciting. |
For decades, Air Force engineers and private companies struggled with a stiff requirement for anti-G suit valves, the valves that feed the correct amount of pressurized air to pilots’ G suits and thus enable then to stay alert in tight turns and sharp pull-ups. The specification for the valves required that they leak no more than 600 cubic centimeters of air per minute when operating. Now, air pressure regulators need to flow air in order to operate, and if that air is not going into the suit after it is inflated it has to go somewhere. A mere 600 cubic centimeters per minute is not much air, and it leaking into a cockpit that in jet fighters is already being cooled and pressurized by large quantities of air is of no real consequence.
So for years the engineers struggled with producing hardware that could meet that leakage requirement. They never questioned why it was so important, and it was not until the late 1970s that anyone figured it out. A team reviewed the original specification to see if changes were required to better support the new high-G fighters such as the F-15 and F-16 and found the origin of that tough leakage specification. The first aircraft using G-suits were prop-driven fighters such as the P-51 Mustang, and unlike jet fighters the earlier airplanes did not have copious quantities of pressurized air. Some models of the P-51 used a rubber bag with the aircraft battery placed atop it; you could not afford much air leakage and have the system continue to operate. In other words, that tough specification that had people tearing their hair out had been invalid since around 1946.
This was a nearly absurd real world example but not an isolated one. In the 1980s, someone pointed out that the nested and interlocking specifications for a piece of US Navy ground support equipment required that it be able to operate at 50,000 feet (15,000 meters) when in reality it would be used only on the ground or on a carrier deck.
When rocket engines were developed, a whole new set of specifications and standards were required and surprises would pop up regularly. For example, they found out the hard way that leather did not make a good seal for liquid oxygen. Making sure the right specifications were used and that the materials were of the required type became a big deal, often a big painful deal, and a major part of the space launch mission assurance process. Examination of the test results of both components and of materials became a boring but necessary effort. A test failure in a factory could shut down our entire space launch capabilities. Scrutiny of serial numbers and material lot date codes was an everyday task. If a company went out of business and a new supplier had to be qualified, that was a big deal as well.
Electronic components had to be tested rigorously and to standards far beyond that for other applications. At one pre-launch briefing in the early 1980s, the spacecraft system program office stated that some of their semiconductors had a 75 percent test failure rate; this was a good thing, they explained, as it showed the test was really tough. The Air Force general they were briefing replied that it sounded more like to him that their components were junk. The mere fact that the failure rates for some specific transistors were being discussed at the general officer level indicates just how seriously the top was taken.
| The “30 pound” specification on the package was the minimum strength for the line, and no one had ever complained before about their fishing line being too strong. |
Surprises still cropped up, especially with new innovations. In the early 1980s, Thiokol tried a new material, G-90, for the threat area in its solid rocket motor nozzles. It looked good at first but then some test failures occurred. It turned out that G-90, used for high voltage insulators for years and produced in the thousands of tons, was manufactured in too wide a range of properties to be sure it would work in rocket motor nozzles. And they did not know how to tighten the specification to ensure that it would work with rocket motors.
Even well established specifications could be found to be suspect. It was not until 1980 that the industry found out that the solder they had been using in liquid oxygen (LOX) tanks for the last 30 years could explode due to being in contact with LOX. In the same time period they also discovered that the standard repair method for cracks in Thor and Delta LOX tanks was not proper; the bolts used were not properly rated for use with the temperatures associated with LOX. There were no known actual failures associated with those discoveries, but the findings demonstrated how rigorous the testing needed to be; just because it had worked in the past did not mean it was suitable.
And there occasionally were problems with specifications that were, shall we say, less rigorous than needed. Following the launch of a new model of a spacecraft in the late 1970s, Air Force engineers puzzled over why the solar array panels did not deploy properly. They asked the spacecraft manufacturer what was used to restrain the solar arrays. The response was that they used ordinary fishing line. “Okay, what is that line rated at? What load causes it to break?” The answer was 30 pounds (133 newtons). “How do you know that?” The company engineers pointed out the 30 pound test specification was written right on the package. Actual testing showed the line broke at something like 100 pounds. The “30 pound” specification on the package was the minimum strength for the line, and no one had ever complained before about their fishing line being too strong.
After the loss of the Space Shuttle Challenger, investigators discovered something shocking about some of the material used in the solid rocket boosters. The shuttle SRBs used a zinc chromate putty to help seal the joints; the putty was filled with asbestos. However, the Environmental Protection Agency was trying to stamp out the use of asbestos and, as a result, one of the companies making the SRB putty quit using the original heat resistant asbestos material and instead substituted cotton—without telling NASA. As it turned out, the putty was not deemed to be a factor in the Challenger SRB failure, and its use was discontinued, but it served as a reminder of how closely materials needed to be monitored.
What was amazing is that by the time US launch systems matured in the mid-60s, virtually no actual launch failures occurred due to specification errors or substandard materials. Emphasizing the boring stuff was not exciting but it worked. Such was not the case everywhere. Catastrophic failures due to unexpected material incompatibilities or contaminated propellants occurred in both the USSR and later in Russia. In fact, in recent years launches of the Proton were shut down for an extended period because cheaper materials had been substituted in the engines and resulted in failures.
But the boring stuff in the US came under pressure. Following the loss the Shuttle Challenger in 1986 and the subsequent requirement for commercial payloads to launch on expendable boosters, the commercial space era truly began. For the first time the private firms building rockets really had to compete with one another, as well as with foreign companies for the first time. Desperate to recover from the serious damage done to the industry by the shuttle-era policies, they looked for ways to cut costs and simplify manufacturing. The specifications required for space launch systems were one target of their ire. Which ones? “Well, er, ah, all of them,” was what the consolidated industry reply amounted to. Established companies had chafed under the requirements for years and new firms trying to innovate did not want to even have to deal with learning the existing specifications.
When companies went to commercial-style operations, the layer of government oversight and inspection that had been there previously virtually disappeared. Did the private firms have to do that surveillance themselves? Most did not think so, or at least not as much oversight as the government did before.
However, the specifications and standards they counted on still applied, right? Well, yes, theoretically, but only if they actually applied the requirements and they and their suppliers were diligent in ensuring standards were maintained.
| In reality, it turned out that the companies were just as diligent as they knew how to be, but all too often that was not good enough. |
“Commercial-style” contracting, begun with the Titan IV program, decreased the degree of knowledge the government had over its procurements and therefore its control over them. Total System Performance Responsibility (TSPR) was inherent in the concept, the idea that the contractor was responsible for making all the parts work. One Air Force officer summed up the underlying belief, “Surely, the private firms will be just as diligent as they were under government control because they know their reputations depend on it.”
In reality, it turned out that the companies were just as diligent as they knew how to be, but all too often that was not good enough. After asserting that government oversight was a costly nuisance, General Dynamics suffered three Atlas failures and got out of the business. Lockheed Martin canceled its Commercial Titan III program after only four launches, which included a failure due to inadequate oversight. The McDonnell Douglas Delta III program died after its first two launches were failures.
The companies learned that someone had to be diligent, even if the government was not around. And the government learned that simply relying on private firms protecting their reputations was not enough. But, as always, the boring stuff still lurked in the background.
On February 24, 2009, an Orbital Sciences Taurus XL 3110 launched NASA’s Orbital Carbon Observatory. The mission was a failure; the fairing failed to separate. This was a puzzling failure, as seven previous Taurus launches had not experienced that problem. The company undertook a robust and thorough investigation and, lacking the failed hardware to identify the specific failure mechanism, even used the traditional “shotgun” approach of fixing everything that could have caused it, including redesign if necessary.
Two years later, on March 4, 2011, a Taurus XL 3110 launched the Glory environmental monitoring spacecraft mission, along with three small cubesat spacecraft. The mission was another failure; once again the payload fairing failed to separate. The launch marked the effective end of the Taurus program: no one would ever trust one again.
Another such failure was baffling. But years later something else came to light. The company that made some of the aluminum parts for both Taurus mission had been falsifying test results on the material since 1996. It was entirely possible, or even likely, that both fairing failures were due to the deliberate alteration of the lab reports.
| ou may end up struggling to meet a specification that has been invalid for 30 years. But you may also find out that some of those now-discounted old engineering requirements still really are needed. |
On June 28, 2015, a SpaceX Falcon 9 carrying supplies to the International Space Station exploded during ascent. The cause was found to be the failure of a strut that held a pressure vessel inside the second stage LOX tank. For corrective action SpaceX changed the design of the strut and switched to a different supplier. But further investigations by both NASA and Aerospace Corp revealed that the strut was made of a commercial grade steel alloy rather than an aerospace standard. The commercial grade steel might be able to handle the combination of loads and low temperatures, but like the G-90 rocket nozzle material of years before, there was no guarantee of that; the allowed range of properties was too broad.
The Boring Stuff is still out there, lurking unknown in file cabinets. You may end up struggling to meet a specification that has been invalid for 30 years. But you may also find out that some of those now-discounted old engineering requirements still really are needed. And you need to realize that just because there are lab reports or a label on a package that says “30 pounds,” or “passed tests,” that may not be accurate.
Recently a review team stated that the Air Force requirements for SpaceX launches are overly restrictive and do not fully take into account the commercial aspects of the procurement. This may be true; experience has shown that long since irrelevant requirements may still be in effect and there’s a tendency to pile on new requirements without restrictions. Regardless, someone has to be concerned with the Boring Stuff. Given that government agencies, non-government organizations that may develop and revise the specifications, private launch firms, and many individual suppliers all bear some responsibility, that “someone” appears to be just about everyone involved.
Note: we are temporarily moderating all comments subcommitted to deal with a surge in spam.