A tale of two product ideas
Ideas don't become products at scale just because we'd like them to; and sometimes they can be worth investing in for the technology spillovers alone.
Two products show us how to think about futuristic ideas.
The first product, or product class, is that of solar-powered electric vehicles. The idea is to take advantage of free sunshine to recharge the batteries on [battery] electric vehicles. Here's a CNET video on them:
The second product is payload yeeting, or the SpinLaunch centrifugal space launcher. The idea is to dispense with the rocket booster and launch the second stage of a rocket by imparting speed in a centrifuge and letting go. Real Engineering has a good video on it:
We'll use these two product proposals, for that's what they are now, to think about the issues of developing technologies, futuristic products, and where to find value.
To get us all on the same page regarding the process of going from ideas to deployed products (roughly), here's a picture that's worth 40,0001 words:
Solar-powered cars
Starting with the solar-powered cars, how much range can we get from solar panels? A few numbers can get tell us a lot.
A large sedan, like a Mercedes S-class, has approximately 10 square meters of horizontal-ish surface. At around 150 W/m^2 maximum power rating of the best current non-experimental solar cells and a consumption of around 300 Wh/mile, used here as a reference for BEV sedans, every hour in the sunshine adds 5 miles of range, at 100% efficiency.
For comparison, in that same hour, the 5 kW free chargers available in some malls and office buildings add 17 miles; the 20 to 60 kW paid chargers add 67 to 200 miles; and the 125 kW Tesla Superchargers could theoretically add 417 miles.
Still, the Sun is free and not everyone has access to the 5 kW free chargers...
Yes, but the solar panels aren't free, and replacing or complementing the flat surfaces of a car with solar panels is neither a free engineering proposition nor usage-neutral.
Starting with the obvious, solar panels are heavier, more expensive, and require more engineering and production work than aluminum or composite panels. They also need additional structural support and electrical wiring and other components.
Even at current grid prices (around 30 cents/kWh in California), the break-even point for a 100 W (nominal) current-day mass-deployed solar panel, priced at around $200/unit and operating at 100% efficiency, is at 6,667 hours of operation. That would be around 5 years of 250 days per year of charging 5 hours per day or 3 years and 8 months of 365 days charging 5 hours/day. (That's break-even for cost of the panels alone. Doesn't include the cost of the reengineered structure and additional electrical wiring and components.)2
Second, in many urban and suburban environments, cars tend to be parked in garages or parking structures; furthermore, in some areas with high insolation, what used to be open-air parking lots are being covered in solar panels for power generation and to provide shade to keep the vehicles cool. And in much street parking, buildings, trees, and other obstructions reduce the amount of light hitting the panels. There are also some areas where outside parking is increasingly less safe from property crime.
Okay, but technologies get better and cheaper, right?
Yes. Solar panels are currently very low-efficiency, relative to insolation; we expect that to improve. But there's a limit: energy from the Sun is low-density, peaking at around 1000 W/m^2; and inside parking structures, that density is much lower.
Some of the prototypes shown in the CNET video have clearly been optimized for aerodynamics; there are also ongoing developments in batteries (making them lighter and smaller for the same capacity) and in electric motors; all of these impact the numbers above, making the solar solution more attractive. But all of these are independent of the source of power: they apply similarly to cars that are solar-charged, charged from an onboard combustion engine (hybrids), or from the grid (current BEVs and PHEVs).
So, unless we're talking about order-of-magnitude changes in these building-block technologies — which are unlikely, though not impossible, in the near future — these technological changes, by themselves, aren't enough of a counter to the engineering and in-use experience shortcomings of solar-powered vehicles. At least for now.
Now, the yeeting of payloads, replacing the first-stage booster.
Yeeeeet!
While solar-powered EVs are putting together technologies that already exist, SpinLaunch is dependent on technologies that have yet to be created (to the level needed). This contrast is important, because the latter offers a separate potential source of value: the development of the technologies themselves (in addition to the value of the launch services SpinLauch would offer).
First two simple calculations: using a linear payload velocity on release of 2 km/s, and a 45 m radius for the centrifuge, we’ll determine the rotation speed and the centrifugal acceleration. These determine the necessary precision and strength of the payload clamp-and-release mechanism, so they’re key numbers for SpinLaunch.
Because there are many wrong numbers for SpinLaunch circulating on social media, let's do those calculations in detail: with a radius of 45 m, the circumference of the centrifuge is 283 m, so for a linear velocity of 2000 m/s, it must complete 7.07 rotations per second, or 424 rotations per minute. The centrifugal acceleration is the square of the linear velocity divided by the radius, 4,000,000/45 or 88,889 m/s^2, a little over 9,000 gee.3
The precision timing required for those airlock doors to operate is quite tight. Let's say the whole airlock is 50 m long (eyestimated from the renderings) and SpinLaunch is yeeting a 5 m rocket moving at 2 km/s. Then the outer door has to open within 45/2000 = 0.0225 s or 22.5 milliseconds of the rocket clearing the inner door, presumably at the same time the inner door is closing or, better yet, after it closes.
Given that these are airtight doors holding one atmosphere (10 tonne per square meter) and with an apparent area (also eyestimated from the renderings) in excess of 4 square meter, they need significant support structure to withstand 40 tonnes, which increases the inertia and the difficulty of fast and precise movement.
The precision timing of the release mechanism is even more critical, and the forces involved are significant (because of the 9,000 gee).
Let's say the target area for the payload at the outer airlock door is 1 m in diameter. With a 50 m airlock and assuming the inner door is very close to the payload at release, the needed angular precision for release is 1.15 degrees (arc tan of 1/50). At 424 RPM (that is 360*424/60 = 2,544 degrees per second), that gives an interval for the clamping system to release the payload of 1.15/2544 = 450 microseconds.
NOTE: A previous version of this post had an error that made this timing off by a factor of 3600. That made the feasibility of the mechanism appear much lower than the correct number. Thanks to David Wrenn of SpinLaunch for catching that error. The numbers in the post are now corrected.
The following table shows the actuator speeds needed for different payload velocities.
Because the clamp-and-release mechanism is such a key success factor for the system, it's understandable that the company doesn't want to show it, or discuss its details, until it's ready for demonstration. And here lies a problem for those interested in the company, call it the Apple-to-Theranos continuum.
Apple famously guards its secrets until product launch, which makes sense from a competition standpoint, but also lets Apple create expectations (“buzz”) that there will be something surprising at product announcements. Apple has been known to do this and has a proven track record of delivering product, so the secrecy doesn't work against it.
Theranos, a name that shall live in infamy as the poster child of unjustifiable tech investments, was also secretive about their allegedly revolutionary technology; but their secret was that they were faking the blood tests that their technology had putatively improved.4
Given no specific information about a company, here SpinLaunch, each of us should assess where in the Apple-to-Theranos continuum we'd be comfortable placing our interpretation of the secrecy.
But we should take into account that SpinLaunch has already demonstrated a smaller system, which required a less precise and less strong, but still impressive, clamp-and-release mechanism. (The demonstration didn’t have an airlock: it relied on a single-use membrane that was ruptured by the payload; not having an airlock also makes the timing of the release less tight.)
Given how mechanically challenging the systems for the airlock doors and the clamp-and-release mechanism are, how key to the success of the product they are, and how applicable outside of SpinLaunch's specific service they may be, it might be worth investing in SpinLaunch for the development of the technology alone: the spillovers of these very strong very fast mechanical actuators to other mechanical products might be worth more than the actual service of launching satellites.
Sometimes the potential spillovers are worth the investment even if the product never comes into existence. (CYA statement: and sometimes they aren’t.)
Ironic extra
Interestingly, one of the common online criticisms of the SpinLaunch system is one for which there's plenty of counter-evidence: that the payload (presumably electronics) wouldn't be able to survive the acceleration. But there are plenty of artillery munitions that include electronic proximity fuses (since WWII, for what it's worth) and the acceleration in an artillery barrel is of the same order of magnitude:
For example, for a M777 howitzer, with a barrel length of 5 m and a muzzle velocity of 827 m/s, the acceleration of the payload is just below 7,000 gee (68,392.9 m/s^2; assuming constant acceleration throughout the barrel, which is the assumption that yields the lowest peak acceleration).
That's the current word count for the chapters describing and detailing this process in the current draft of my next book.
The 250 days year represents leaving the car in the parking lot during work days; the 5 hours/day is a shorthand way to account for the different angles of the Sun depending on the season, the latitude, and the time of the day.
Since the speed is “in excess of 2 km/s,” the 10,000 gee number in the Real Engineering video is reasonable. And this is a distinction without a difference for practical purposes.
As some people, including investors specializing in medical devices and WSJ journalist John Carreyrou, pointed out, the company was claiming orders of magnitude improvement on the quantity of blood required relative to competitors that had had entire research campuses on this precise problem for more than a decade, and their (Theranos's) only argument was essentially "Silicon Valley startup magic."