STS-114: The Question That May Have Saved Discovery
In 2005, NASA was preparing to do something it had not done in more than two years.
Launch a Space Shuttle.
The loss of Columbia and its seven astronauts on February 1, 2003, had grounded the fleet while NASA examined virtually every aspect of Shuttle safety. The first Return-to-Flight mission, STS-114, would be flown by Discovery. NASA had spent two and a half years modifying hardware, changing procedures, improving inspection techniques and scrutinizing the enormous orange External Tank that had played a central role in the Columbia accident.
Everything about STS-114 was supposed to demonstrate that NASA had learned from the past.
Then an outside supplier asked a simple question.
And NASA discovered something it apparently had not known for years.
Some of its mission-critical Shuttle components had been manufactured incorrectly.
Hidden Inside the Orange Tank
The Space Shuttle External Tank was much more than a large fuel container.
It held the liquid hydrogen and liquid oxygen consumed by the Shuttle’s three main engines and also served as the structural backbone connecting the Orbiter and two Solid Rocket Boosters during launch.
Inside the tank’s pressurization system were components called diffusers.
There were two—one associated with the liquid hydrogen tank and another with the liquid oxygen tank.
Their job was to disperse pressurization gas uniformly inside the tanks as propellant was consumed. Maintaining the proper pressure was essential to making the entire propulsion system behave as designed.
And inside each diffuser was something that hardly looked like a spacecraft component at all.
Wire mesh.
Small.
Ordinary-looking.
And mission critical.
NASA’s own Lessons Learned system later described the diffuser as controlling the uniform dispersal of gases used to maintain positive pressure as the levels of liquid hydrogen and liquid oxygen dropped.
For years, a supplier had manufactured the mesh used in those diffusers.
Then, during the Shuttle’s post-Columbia grounding, that supplier went out of business. NASA needed another source.
So NASA began looking for a replacement vendor.
And that is when this story became extraordinary.
“Why Are There Two Different Kinds?”
The prospective supplier asked NASA for samples of the material it was supposed to reproduce.
NASA sent them.
The vendor examined the samples.
Then came the question:
“Why are there two different kinds, and which one is the one you want?”
NASA’s response, according to the case documented in the Shuttle lessons material, was essentially:
What are you talking about? They’re all the same.
Except they weren’t.
The new supplier had noticed something that apparently had escaped the system that produced and accepted the components.
The mesh samples were different.
NASA’s later technical description is more precise than the shorthand “single weave” and “double weave” often used in retelling the story.
The engineering requirement called for a plain Dutch weave wire screen.
But some diffusers had instead been manufactured using a duplex Dutch weave.
That seemingly minor manufacturing substitution changed the resistance of gas flowing through the diffuser. NASA’s later propulsion-system review concluded that the incorrect mesh altered the pressure drop through the component and changed the response of the tank’s pressurization system.
A tiny difference in wire.
Deep inside an enormous rocket tank.
Then NASA Started Looking
Once engineers understood what the new vendor had found, the obvious question followed:
How many diffusers had already been made with the wrong material?
NASA began tracing the hardware.
The numbers recorded in the Shuttle case study were astonishing.
There were 31 diffusers in the population examined.
Sixteen were wrong.
And one of the incorrectly manufactured diffusers was associated with the External Tank already mated to Discovery as NASA prepared for the first Shuttle mission after Columbia.
NASA’s formal Lessons Learned record independently confirms the essential point: the unacceptable material substitution affected multiple diffusers already installed in External Tanks, including one on the tank mated to Discovery for Return to Flight.
Think about that for a moment.
This was not a defective component sitting on a shelf waiting to be installed someday.
The spacecraft was being prepared to fly.
The Tank Was Already Talking
There was another clue.
During tanking tests of Discovery’s original External Tank, ET-120, engineers had seen unusual behavior in the liquid-hydrogen pressurization system.
A vent valve was cycling more frequently than expected.
NASA reported that ET-120 contained a diffuser with a tighter woven mesh than specified and that analysis showed the out-of-specification diffuser may have contributed to the liquid-hydrogen pressurization problem.
A later NASA propulsion review was even more specific: the incorrect screen material changed the diffuser’s pressure drop and system response. NASA ultimately replaced the out-of-configuration diffusers and identified stronger supplier controls and material acceptance inspections as the lesson from the event.
Discovery’s Return-to-Flight processing eventually included something highly unusual: rollback to the Vehicle Assembly Building, de-mating the Orbiter, and swapping its External Tank and boosters for another stack as NASA worked through several External Tank and propulsion concerns.
The diffuser was no longer a theoretical quality problem.
NASA had found hardware that did not conform to its engineering requirements in a system whose performance had to be precise.
The Most Frightening Part
There was apparently no inspector standing at the factory who simply looked at the mesh and said:
That’s wrong.
The difference was not obvious that way.
The process depended upon the supplier manufacturing the material to specification—and upon NASA’s supply and acceptance system ensuring that what had been ordered was actually what had been delivered. The case study concluded that there had been no adequate process for detecting this particular difference through ordinary visual inspection.
That makes the discovery profoundly uncomfortable.
The problem was not uncovered because NASA deliberately decided to reexamine the diffuser mesh.
It was uncovered because the original supplier disappeared.
NASA needed someone else to manufacture the component.
The replacement vendor wanted to know exactly what NASA wanted.
NASA sent samples.
The samples disagreed with one another.
And somebody asked why.
The Supplier Who Went Out of Business
There is a remarkable irony here.
Normally, having a critical supplier go out of business is considered a problem.
In this case, it may have created the circumstances that exposed a much more dangerous problem.
Had the original supplier remained in business, production might simply have continued.
No new manufacturer would necessarily have requested reference samples.
Nobody would necessarily have laid those samples beside one another.
And nobody might have asked the question:
Why aren’t these the same?
That is the haunting counterfactual at the heart of the story.
We cannot say with certainty that an incorrectly configured diffuser would have destroyed a Shuttle. NASA’s technical record does not support making that claim as an inevitability.
But we can say something significant.
The component was mission-critical.
It did not conform to its engineering requirement.
It affected tank pressurization behavior.
Multiple units containing the incorrect material had entered the Shuttle hardware population.
And one had reached the Return-to-Flight vehicle.
NASA itself characterized the event as a lesson in controlling complex supplier networks and ensuring mission-critical products are manufactured consistently.
That is serious enough.
Discovery Finally Flies
On July 26, 2005, Discovery finally left Launch Pad 39B.
STS-114 became the first Shuttle mission since the loss of Columbia.
The Shuttle carried a different External Tank than the one originally assigned to the mission, and NASA had removed the out-of-configuration diffuser hardware from the flight population.
Discovery reached orbit and returned safely 14 days later.
The mission is remembered primarily as NASA’s first step back into space after Columbia.
But hidden inside the long story of STS-114 is another story.
One involving no astronaut heroics.
No dramatic emergency in orbit.
No countdown stopped at the final second.
Just some pieces of wire mesh sitting on a table.
And a supplier who noticed they didn’t match.
One Question
The great disasters of spaceflight tend to be remembered because something failed.
This incident deserves to be remembered for the opposite reason.
Someone asked a question before the failure occurred.
NASA believed the samples were identical.
The supplier did not accept that assumption.
It examined what was actually in front of it.
The difference eventually led investigators backward through the supply chain and into Shuttle hardware that had already been manufactured.
That is what makes this such an important spaceflight story.
The original supplier going out of business looked like bad luck.
In retrospect, the timing may have been extraordinarily fortunate.
Because sometimes catastrophe is prevented by a sophisticated computer.
Sometimes by an astronaut.
Sometimes by thousands of engineers working a problem.
And sometimes—
it is prevented by the person willing to hold up two supposedly identical parts and ask:
“Why are these different?”