Putting the latest Starship crash into perspective—it’s “stupidly difficult”

Once again, on Tuesday afternoon, a Starship prototype soared into the clear skies above South Texas like something out of the pages of a science fiction novel. Once again, after reaching a high altitude, the spaceship leaned into a “belly flop” maneuver, making a controlled descent back toward the planet.

And then, once again, a problem within the last few seconds caused the Starship prototype to spectacularly crash near its launch platform.

Seven weeks have passed since the first full-scale Starship prototype, SN8, performed its high-altitude flight. Now, SN9 has met a similar fate. It appeared that one of the two Raptor rocket engines intended to power the final controlled descent failed to relight (see a great slow-motion view). As a result, when the vehicle began reorienting itself into a vertical position, it never stopped swinging. Then, BOOM!

So what are we to make of a second high-profile failure of the Starship program? Is this a program on the cusp of failure?

“Stupidly hard”

We should probably take a step back to better understand what SpaceX is trying to do with Starship. This vehicle is the second stage of a launch system that includes a large booster, named Super Heavy. With its Falcon 9 program, SpaceX has demonstrated the ability to launch and land a rocket vertically. But the real trick comes with Starship—specifically bringing it back safely from orbital velocity and through the atmosphere so that it can be launched with minimal refurbishment a short time later.

In late 2019, as I was working on a book about the origins of SpaceX, founder Elon Musk invited me to sit in on technical meetings so that I might get a sense of his leadership style. During one of these meetings with the Raptor engine team, Musk was pushing back on a decision his team wanted to make that would fractionally reduce the specific impulse (ISP) of the engine. He was not happy. As Musk well knows, when you build a rocket, if you’re adding mass or losing performance, you’re losing the battle against gravity.

“That is why we fight for mass, and we fight for every fraction of a second of ISP,” he told his team at one point. “Especially with a reusable upper stage, which nobody has ever succeeded in. Just FYI. It’s not like they were huge idiots who wanted to throw their rocket away all the time. One of the hardest engineering problems known to man is making a reusable orbital rocket. It’s stupidly difficult to have a fully reusable orbital system. It would be one of the biggest breakthroughs in the history of humanity.”

As he delivered this little speech to his engineers, Musk’s mood steadily mellowed. Soon, he was joking with the team. His point had been made. Yes, he understood what he was asking of them. It was damned hard. It would hurt their brains. It hurt his. But they had no choice but to push through the engineering challenges.

SpaceX has come a long way in the 16 months since then, moving through about 10 various Starship test vehicles and tank designs to reach SN8 and SN9. Much of the Starship program’s time, since late 2019, has been focused on building a factory in South Texas to churn out Starship prototypes. The company can now crank out these stainless steel spaceships at a rate of a couple a month. Already, SN10 is nearing readiness to take its turn at launching and landing. Its flight may come as soon as the third week of February.

This is a benefit of running a hardware-rich program, like what SpaceX has built with its assembly line for Starship. It also helps that Musk has adopted a failure-is-an-option mentality. He can still get angry about failures, of course. But they are an acceptable part of an iterative design program that allows him to move quickly.

Iterative versus linear design

It is fair to question whether we celebrate these SpaceX failures too often. After all, when NASA fired up its Space Launch System rocket for the first time in January, its failure to meet its test objectives was met with criticism rather than approbation. Is this hypocrisy? Another sign that the cult-of-Musk has run amok? Not really.

NASA and its prime contractor, Boeing, have literally spent a decade designing and building the SLS rocket, and they produced a single core stage during that time. US taxpayers have footed the bill for all of this, nearly $20 billion. NASA did not even need to build the engines—the SLS rocket uses space shuttle main engines that have flown multiple times already. That is not to say it isn’t hard—building any rocket is.

However, by following a linear design methodology and needing to please Congress, NASA cannot afford to fail. With linear design, years are spent designing and testing small pieces of a project, and only after very much analysis are the components put together and tested. This is the safest way to build a vehicle that has the greatest chance of succeeding the first time out. But it is also costly and drawn out.

Accordingly, the test program NASA is carrying out for its SLS rocket is not so much a development campaign but a validation campaign. If there is a problem with this core stage, the second one will not leave its factory in southern Louisiana until mid-2022 at the earliest.

Meanwhile, in South Texas, SpaceX has a half-dozen Starship prototypes. None is a refined or finished product like the SLS rocket. They’re rough prototypes. But each probably cost a few million dollars to build, plus the cost of the engines. This is part and parcel of an iterative design campaign—each vehicle improves on its previous model, incorporates learnings, and allows for failures. This process allows a company to move fast and make mistakes.

And SpaceX still does have so much more to learn about Starship. This upper-stage rocket need not simply fire its engines for eight minutes and then fall into the ocean, like the SLS core stage. It has to be capable of making multiple relights of its engines, surviving for weeks or even months in space, and re-entering through Earth’s atmosphere with minimal effects to ensure rapid reusability. And then it has to stick a landing. This is not an easy to-do list for something several times larger than a school bus that travels at 25 times the speed of sound.

So SpaceX has a long way to go. In the words of SpaceX engineer John Insprucker, who commentated on Tuesday’s launch attempt, “We’ve just got to work on that landing a little bit.” Yeah. That and a million other things before satellites, let alone people, fly into space on Starship.

It’s stupidly difficult work. But does anyone doubt SpaceX will get there?

Listing image by Trevor Mahlmann