People don't grasp what an unbelievably complex engineering problem that is. It's at least an addition of difficulty at the same magnitude as building a steady-state 63000 hp turbine pump in the first place.

Mechanically implementing it in the inherently steady-state design rocket of most rocket cycles. Having variable controls able to work at those pressures. Testing structural dynamics for a range of harmonic conditions instead of one. And do all of that with materials that need to tolerate temperatures going from cryogenic to white hot, without allowing thermal expansion to affect the mechanical tolerances of parts running at thousands of RPMs. And now you have varying flow rates and negative pressures in the lines coming from the external tanks, so have to design such that cryogenic liquids (that normally would require immense positive pressure to keep liquid) don't spontaneously boil or cavitate or cause a shock-like wave (think water hammer turning off your bath faucet) under changing negative pressures.

It's really difficult even for seasoned engineers to grasp the scale of difficulty involved.

Its rocket science

Turbopumps are finicky, but at least heat and cryogenics are separated with the turbine shaft. In combustion chambers, you have both mechanical stress from high pressures (and thin walled constructions) and thermal stress - difference of a few thousands K, so chambers aren't that simple either.

Still - now we have enough knowledge to repeatedly design flyable rockets from scratch, different teams, periods of time, countries. A lot of work still remains - and the plumbing is a good manifestation for that.

The engineering of the seals that can operate at those speeds and pressures is a whole specialised field. If I remember correctly, the SSME design uses a labyrinth seal pressurised with Helium.

In other words, the various gases are kept separate only by more gas.

Similarly, the Saturn V main engines were unlubricated because no lubricant could be found that could tolerate the extreme conditions. They were just designed to wear out slow enough to keep operating over their operating lifetime, which was measured in hundreds of seconds (including static test firing).

What do today's reusable engines use?

Those pressures, rates of fluid flow, and shaft rpms is going to result in the mechanical surfaces being separated by hydrodynamic forces and/or boundary layer flow.

Technically, lubrication is the same physical principles, but viscosity, weight, chemical stability/ durability and other factors become more important for lubricants that are recirculating.

(Oil in eg, a 4-stroke engine operates on these principles, with the rotating shaft causing pressure differences in the oil that "lift" the shaft from contact with the surrounding metal, until pressure (and therefore spacing) is roughly equal on all sides.

A ping-pong ball floating on a column of air self-stabilizes in the same way.)

The primary concern for these turbopumps becomes heat. So heat transfer and temperature of the "lubricating" fluid become more important than its other nominative qualities as a lubricant. Fluids being pumped at cryogenic temperatures can obviously help here.

So instead of a normal "lubricant", most of these turbopump designs just run a portion of the fuel/oxidizer fluid through the critical areas to provide the surface separation and cooling required.

Ball bearings require "lubrication", and the lack of good lubricants is exactly why they use fluids that end up behaving as I described, and why "lubricant" is the wrong way to think of it in this case.

If anything, the existence of a good lubricant would prevent the surface wear. Temp control and fluid dynamics are what they have to work with, and yes it results in surface wear because it's suboptimal.

But so does running your 4-stroke too hot.

Edit: They used carbon seals. The turbopump uses ball bearings cooled by the fuel. Detailed description here: http://web.archive.org/web/20150509173846/http://agentdc.uah...

"Liquid oxygen is [the high pressure oxidizer turbopump's] only lubricant and a poor one at that."

So it's some lubricant and I imagine also significantly a coolant.

This https://ntrs.nasa.gov/citations/20100023061 goes into a lot more detail. But exotic materials and designs bathed in LOX which provides cooling (and minimal lubrication) seems to be how the space shuttle engines achieved their long bearing lifetimes.

Don't forget the foundations!

On the one hand, you have https://xkcd.com/793/ responses:

> You're trying to predict the behavior of <complicated system>? Just model it as a <simple object>, and then add some secondary terms to account for <complications I just thought of>. Easy, right? So, why does <your field> need a whole journal, anyway?

Just pressurize the tanks, and meter the flow with some valves. Easy, right?

They move to the other side of the distribution after a little more thought, when they they realize it's simply infeasible to put thousands of horsepower in a pump that size, and declare the whole endeavor completely nonsensical and impossible.

Ran through this on a recent project involving an automated sewing machine. At first, it seems ludicrous that you could tie knots thousands of times per second. Oh wait, it's a single motor and old cam-driven tech from the 1800s, available off the shelf for a couple hundred dollars?

To give a far less glamourous example, I am a mech eng who works in rail. We needed some more orecars to complement an existing fleet that only had about 10 years' operational life left in them. Thus, they wanted me to design for a shorter life than the 25-30 year standard that we target, to save cost. However, trying to thin out the structure so it only had 10-15 years' of fatigue life in it meant that it fell well short of the proof load requirements needed to stop it ripping in half in a worst-case shock load. Put differently, the constraint around peak loads effectively baked about 20 years of operational life into the structure, and in turn made it difficult to save money on a shorter-life design.

It would not surprise me if many of the non-ablative parts in a rocket are in fact fairly durable without the stop-start cycles. So whilst a launch may only take 5-8 minutes, I could totally believe that a 15-20 minute launch wouldn't demand heaps more from the parts.

Of course, this absolutely doesn't translate to ablative parts, or items that undergo stop-start cycles. The latter of course is where the devil is for reusable equipment. Depending on the failure modes in question for turbo pumps and throttling controls, those may or may not apply.

Whereas "all the power stations in the UK" keep on trucking, day in, day out, year after year.

That's not the power output of the rocket (which someone quoted below). That's the power required to move the fuel and oxidizer into the rocket engines. Pumping the same amount of liquid can move a 100,000 ton floating city across the ocean at 40mph.

As I understand it, a huge factor in the power requirement of the pumps is raising the pressure to the required level which is very high (combustion chamber pressure might be hundreds of atmospheres).

https://twitter.com/CopSub/status/1468280164404666373?s=20

And gas generators using main propellants are better, certainly, but less than you might think because they have to haul all their oxidizer with them (unlike aircraft) and are also usually run very far from stoichiometric (maybe just 0.3 O:F ratio compared to a stoichiometric 3.4) to keep the temperature down. So unless you have a pretty high temperature turbine, you might not beat peroxide by much!

So the easiest gas generators have worse energy density (keep in mind RocketLab does stage off batteries if necessary…), and the next easiest, while better, aren’t MASSIVELY better without careful efficiency improvements. The real efficiencies come when you use like an expander cycle or a staged combustion cycle or you feed the gas generator exhaust back into the nozzle like Merlin Vacuum or F-1. And those are all much more complicated. A level of complication that is not worth it for small rockets but is for larger.

So it’s really not about electric scaling poorly (electric scales just fine) but about the greater complexity of better engine cycles being worth it at larger scales.

If I remember correctly, with F-1 the gas generator exhaust was sent to the nozzle to cool the nozzle, not to add efficiency to the engine main cycle.

We're still trying to get more efficiency from isochoric combustion, but the expected wins aren't too big. It's good that full-flow combustion becomes more of a norm.

A .22 LR rifle bullet might acquire 200 J in 2 ms, which means the firing gun is producing 100 kW mechanical, plus probably another 300 kW thermal. So another way of thinking of this is that an engine dissipating 10 GW is equivalent to something like 25000 handguns firing at once, without ever stopping.