All-up systems testing is risky but saves time.
The recent launch of SpaceX Starship, one of the biggest rocket ever built—bigger than even the legendary Saturn V from the days of Apollo—ended in failure. However, much was learned. The gigantic booster got the stack 24 miles up, to the point of staging, before the vehicle failed catastrophically. Although the mass media labelled the loss of the launch vehicle as a success, the simple fact is that it wasn’t. It was a failure—a spectacular failure, but not an unexpected one. Testing complex engineering systems, especially those where multiple systems interact with each other, is expensive and time-consuming business. The next test flight will be the critical one.
And, access all episodes of End of the Line on Engineering TV along with all of our other series.
* * *
Episode Transcript:
Last week, SpaceX launched the biggest rocket ever built—bigger than even the legendary Saturn V from the days of Apollo.
The gigantic booster got the stack 24 miles up, to the point of staging, before the vehicle failed catastrophically. While the mass media was working overtime to label the loss of the launch vehicle as a success, the simple fact is that it wasn’t.
It was a failure—a spectacular failure, but not an unexpected one.
Testing complex engineering systems, especially those where multiple systems interact with each other, is an expensive and time-consuming business. In rocketry, the original Saturn program from the early 60s was managed by a corps of engineers that cut their teeth on the German rocket programs of World War II, and was tested in the traditional stepwise fashion.
The first Saturn 1 was launched with the dummy second stage and no third stage. Each successive test flight added more complexity, qualifying the first stage, then the second, and finally the entire launch vehicle. The result was a spectacularly successful program, with the Saturn 1 series eventually launching five crewed Apollo capsules successfully.
What makes this incremental testing methodology work is unlimited time and money. The alternate strategy, the one used with the gigantic Saturn V moon rocket, was all-up testing. The idea was to assemble the entire vehicle, all three stages plus spacecraft, and launch the entire vehicle right from the beginning.
The risks for Saturn V were enormous. Everything about the system was new, with cutting-edge technology and none of the computer-aided engineering available to design teams today. The first all-up test was Apollo 4, in November 1967—and incredibly, it worked. A follow-up test, Apollo 6, had multiple problems, but nothing that prevented the first crewed flight of a Saturn V, Apollo 8, in 1968.
The need for speed in racing the Russians to the moon forced all-up testing, and frankly, NASA got lucky. SpaceX launched an even bigger vehicle, and they weren’t so lucky. The price of all up testing is a high risk of system failure, in exchange for more data on the entire launch, faster than will be possible with conventional test methods. If you want to make an omelette, you have to break some eggs, and SpaceX is obviously prepared to break many eggs to get the large and complex Starship program flying reliably.
Was its first flight successful? Clearly not. But we can’t determine overall program success until we see how many flights are needed by the SpaceX team to qualify the vehicle for crewed flight. NASA set the bar: two uncrewed tests. To match that standard, SpaceX will have to launch the next one successfully. Will they?
With modern engineering design, simulation and testing tools at their disposal, plus half a century of large spacecraft design heritage to draw from, I expect the answer will be “yes.”
