We needed launches. Lots of launches. That much was clear, even if how to get there was not. It seemed it was always the same meeting, about a launcher real or imagined, a Shuttle upgrade or some vehicle post-Shuttle. Perhaps the rocket was expendable, the big dumb booster, or perhaps it was reusable. Perhaps it was not entirely a rocket at all, but rather a spaceplane where the engines and the airframe were all better understood as one. Predictably, the reliability analysis was at the end of the day, naturally following from what came earlier. My field of operations was usually nearby. This left the impression it was decisions first, consequences later as time permitted. Like reliability analysis, operations could be mistaken for ill-defined and tentative results coming from technology seen as real and final. Yet inside operations were launch processes and depending on how often a launch was likely there were launch rates, the purpose to it all. Similarly, inside those reliability numbers was technology maturity, a one-off now that might one day be common. And all this was connected.
US launches over time and by launcher, from the first Space Shuttle flight in 1981 through the LauncherOne launch June 30, 2021. Credit: Edgar Zapata, zapatatalksnasa.com [Update February 26, 2022: For the continuously updated data and graph of the above see US launches by launcher over time.]
This story begins with how complex projects like to break themselves up into a million small pieces. The notion is the small pieces are so much more manageable for being limited. Project management must have birthed the military strategy of “divide and conquer”. Still, well before it all meets up again in the real world, on the plant floor or at the launch site, an analyst with a penchant for putting the pieces back together does so, at least on paper. We can’t help but take a peek right now to see how it might add up in the future, or how it might not. Having checked trees all day, someone dives in to talk about the health of the forest. The picture may not be what’s expected from the earlier walk in the woods.
Plotting US launches by date and by launcher is not new. The data below going back to the 1960s, including the Apollo program and the Moon landings, shows some impressive launch rates (the Atlas D in 1960). Those were different times, especially as far as NASA budgets, urgency and the sense of national purpose. This was my inspiration for creating the graph above with US launches since the Shuttle’s first flight in 1981. There are much fewer failures in the recent decades of spaceflight compared to that first decade. Learning can do that.
It’s simple enough to say “practice makes perfect”. If only it were always a simple matter how to practice. In launch-speak this translates to “a high flight rate cures all ills”, leaving the problem of how to get the high flight rate. Spaceflight debates also simplify this to a “chicken and egg” problem. A launch system, meaning people, must launch often to improve by learning, but can’t because it is so expensive here and now. Reliability for rockets can never be aircraft-like until you’ve flown about as often. A variation on this problem is specific to reusable launch. Here a reusable system flying often can make up for a higher up-front cost, but the customers justifying that high flight rate do not exist because no launcher like that is available. (“No, you go first”.) The expendable rocket version of the trap shifts this vicious cycle over into manufacturing rather than turnaround. Low-rate production of complex technology is expensive, dampening demand, making the low-rate production a self-fulfilling prophecy. Although there is also the other matter – that throwing away your airliner each flight will never be cheap no matter how much you learn about manufacturing. If you learn anything, it’s to be reusable.
There is another hitch too – carrying crew. Reliability and safety are two sides of the same coin as parts that constantly act up on the ground can hardly be expected to behave in flight. Unfortunately, one way around these traps is unwarranted optimism. Take lots of flights, a billion dollars here and a billion dollars there and before you know it you assume the rest – that you must have made a massive dent in learning and making it all safer. The opposite is unthinkable. We want to believe massive expenditures reflect massive improvements, but sometimes they merely reflect the difficulty of modifying a system carrying crew every flight. This is what leads to the Space Shuttle’s safety numbers published just months before the loss of Columbia. The safety numbers below, from a Shuttle booklet handed out to employees in May 2002, proved to be woefully incorrect when reassessed.
“If you can’t explain it simply, you don’t understand it well enough.”Albert Einstein
Simply put, a hundred launches, or a hundred of anything for that matter, is just a drop in a statistical bucket. It’s never enough to chart a path to ever higher rates of launch that are increasingly cheaper and reliable. Or is it?
There are facts on the ground all around us that say these traps are not real. We are surrounded by complex technology people depend on with their lives, all quite reliable and inexpensive enough to be common. Planes, trains and automobiles figured out how to cross the valley from being scarce, expensive, and even dangerous, to being common, affordable and safe. A common retort here is the physics of getting to orbit makes these comparisons invalid (repeat mantra here, “space is hard”). And yet we have the beginning of a steep rate of climb in launch rates with the SpaceX Falcon 9. The dots are even atop each other, with no space in-between. This is a good thing. Are SpaceX and NASA going around the traps we thought were always there, dooming getting to orbit to only glacially slow improvement?
First, the NASA investment that resulted in the Falcon 9 and the Dragon cargo spacecraft was able to assume more risk, at first carrying cargo, not crew. Even better, backup options existed for delivering that cargo to the International Space Station. More risk opens the door to innovation. With innovation come new ways of doing business to be more affordable, attracting more customers, so more opportunities to learn. When the inevitable failure does occur (100 flights ago for Falcon), the improved tempo assures you quickly have “lessons implemented”, not just “lessons learned”.
Lastly, not enough demand? Provide it yourself with another innovation like a Starlink low Earth orbit satellite constellation of thousands of satellites. Demand gets the driver’s seat finally, as the egg decides a chicken is a great way to make more eggs.
Times have changed. None of the growth and launch rate we have seen in Falcon would be possible based only on NASA demand, every launch and payload from a limited budget. Ironically though, none of the growth and launch rate we are seeing on a system like Falcon would be possible without NASA, from up-front investments to buying services and providing technical assistance.
All this points the way to increasing space exploration that is ever more reliable and safer beyond launch, in low Earth orbit systems and systems on the surface of the Moon. Here’s to seeing lightning strike twice and then again, adding more lines on the launch rate graphs, climbing even faster. Some craters on the Moon will help too.