Neutron Alt History

In this post, I'm going to try to describe what I would have wanted to see RocketLab pursue instead of Neutron.

It's important to preface this by saying that I don't have a strong objection to the Neutron architecture itself. I also think that RocketLab has a brilliant team of great executors. My concerns lie primarily with strategy:

1. I wish they had pursued the MLV market sooner; many outside observers, as well as other companies in the industry, have seen small launch as a stepping stone where there isn't very much opportunity to grow. RocketLab, on the other hand, resisted going into big rockets, and lost time (see time between Falcon 1 and 9 vs. between Electron and Neutron for a sense of the might-have-been).

2. I wish their stated goals for Neutron cost and pricing reflected the reality of SpaceX's very comfortable margins with Falcon 9, as well as the threat of a larger LV with aspirationally single digit million dollar marginal launch costs. RL is aiming to price the tad smaller Neutron at Falcon 9's price/kg in the hopes that the lower price/launch lets them scoop payloads that don't need F9's full payload capacity. But their stated target COGS of 20-25M per Neutron flight is *higher* than the range of rumored marginal costs of Falcon 9 ASDS launches, let alone RTLS launches that match Neutron's target ASDS performance, meaning SpaceX would already be in a position to outbid Neutron with Falcon 9 once Neutron is on the market. And if SpaceX gets anywhere near their cost goals with Starship, their margins and (by extension freedom to cut costs) will be even larger.

I'm talking about dedicated launches here, not rideshare. A lot of people assume that because Starship is being designed to be able to carry 100+t to space, it won't be competing in dedicated medium launch. But there's no reason to expect this. Once SpaceX is demand constrained (have enough launch capacity that they can launch all the Starlinks they want all the high value payloads they can get, and still have capacity to take on new payloads), there is no reason for them not to compete in dedicated medium and even small launch (if they can get their costs low enough) with discounted prices just for those weight classes that can be launched by other, smaller rockets.

With or without Neutron, with or without full reuse, if RocketLab had a long term plan to either match SpaceX's cost targets *or* was aiming more directly for the explicitly non-SpaceX market (customers who don't want to fly SpaceX or don't want to fly exclusively SpaceX), I'd find their strategy more defensible.

3. In line with the above, I wish RocketLab was more aggressively pursuing the kind of capabilities reflected in e.g. NSSL Cat C requirements. Higher value defense, science, and commercial GEO launches are less frequent but more lucrative, and if you can't compete with SpaceX on price, you might need to displace the likes of ULA to have a long term foothold in the medium+ launch business. And that is a safer bet if you can do everything ULA does. That seems like a much harder hill to climb at first glance, but essentially what you need is a heavy lifter with a long duration third stage. And if you're already aiming to build a new medium launch vehicle with vertical payload integration, a heavy lifter is not that big of a stretch, and a long duration third stage is not that different from what they've already achieved with Photon at a small scale.

With that context, let's get to alt history:

With how much I stock I put in the prospects of Starship launch dominance, you might think that I believe RocketLab should have dived right into developing a large fully reusable launch vehicle instead of Neutron. But while I do think that would have been a great play, I'm not a big risk taker myself when it comes to business, and I respect RocketLab's desire to avoid excess dependence on external capital.

My actual proposal looks to further reduce development time and risk, and more thoroughly leverage what RocketLab already had to pave a way forward, with a multi-phase roadmap all the way to fully reusable heavy launch.

For this bit of alternate history, let's roll the clock back to 2019.

In August of that year, after Electron's 7th launch since its debut in 2017, RocketLab announced they'd be pursuing first stage reuse after all, after emphatically saying they wouldn't. The reason cited was launch frequency.

RocketLab had been ramping up Electron's cadence -- from 1 in 2017 to 3 in 2018 and 6 in 2019, with a goal of eventually reaching 50-120 launches per year.

And in the midst of that ramp up, they seem to have quickly realized that recovery and reuse would be easier than increasing first stage production to those levels. And when we think about what part of the first stage production would be hardest to scale, it's not much of a stretch to suppose that it's the engines -- 3D printing of parts helps a lot, but the assembly, acceptance testing, and integration of 9 engines is still a substantial pain, with Peter Beck peferring fewer engines. Engines are generally expected to be roughly half the cost of a first stage even on large rockets with fewer engines, and they're the only part of the Electron first stage that's an order of magnitude more work to put together than its equivalent on the second stage.

Reuse makes the cost of putting together a just stage less relevant, but given that it took RocketLab 4 years from that point to refly a single Rutherford, and we're still waiting on full booster reuse, it stands to reason that there would have been justification even then to switch to a single engine first stage if it could have been done quickly with low development cost. It may seem dubious that they could introduce a new engine in just a couple of years, but what would this new engine be?

Imagine a Rutherford but scaled up by a factor of 9: 9x the chamber volume, 3x the throat & exit diameters, 3x the pipe and valve diameters and pump diameters (for the same flow speed), 9x the pump power. To save engineering effort you could even use sets of 9 Rutherford pump motors on each pump, with a planetary gear setup with a 3:1 ratio (if you want to keep the same pump blade velocity with 3x the diameter you'll need to reduce pump rpm by a factor of 3). You'd of course need some engineering work scaling up injectors and accounting for channel size changes and the like, but it should be nowhere near what you would need for a brand new engine with a completely different cycle and fuel.

Now the 9 Rutherfords become 1, and each one isn't much harder to build than a single original Rutherford. There aren't going to be a whole lot of manufacturing changes beyond needing to buy a bigger 3D printer and build a bigger test stand.

But what if we took this opportunity to increase thrust a bit more, without increasing pump power? Rutherfords have gas generator -like chamber pressure*, so there's actually a lot of room to reduce chamber pressure and exit area to increase mass flow with the same pump power. The battery and the motors and the rest of the electrical system remain the same, but suppose we increase characteristic diameters by ~4 instead of 3, increasing flow rate by a further ~80% and reducing pressure by ~45%. Widening the throat to compensate, we'd lose ~5% isp on the first stage, but gain ~70% thrust. We'd have an engine with specs not far from early Merlins.

On Electron, lower gravity losses would pretty much compensate for the lower isp even without a tank stretch, and I wouldn't expect much structural work to be needed to accommodate the higher thrust given the scale of the vehicle and tank pressures. (Obviously the thrust structure would change to switch to a single engine.) With a tank stretch, Electron could accommodate 1st stage recovery hardware *and* increase payload performance.

Would all this be worth the trouble with 1st stage recovery around the corner? Maybe or maybe not if just for Electron (though it would reduce inspection and cleaning time and allow reuse without reducing performance). But what if you were also planning to use this engine on your next, larger launch vehicle?ppp

A 9+1 configuration rocket (9 engines on the first stage, one vacuum optimized engine on the upper -- isp barely impacted by chamber pressure) built around this engine would be in the neighborhood of 200t liftoff mass with a body diameter around 3m. It could have a LEO payload in the 5-6t range with downrange booster recovery (whether propulsive or parachute). It could support a 4m or even 5m class fairing. The first stage would have one Electron first stage battery pack per engine, while the second stage could have two with hot swapping (expending two preflown 1st stage batteries per flight).

Fairing reuse could follow Falcon 9's lead. Total expended hardware per flight would compare very favorably to a fully expended Electron, and in contrast with current plans for Neutron, marginal cost per flight could more easily be kept well below Falcon 9's. This would enable this vehicle to serve much of that sub-8t market RocketLab originally had in mind for Neutron when they were baselining RTLS, and outbid Falcon 9 every time in a race to the bottom, with healthy margins. Including and especially with a Transporter-like rideshare program.

But I wouldn't stop there. I'd take that and cluster it.

First a cluster of 3. Not the way SpaceX did it with Falcon Heavy, clustering the first stage only and needing to do major structural work on the core and thermal work to account for hotter reentries. But clustering both stages, so that the load paths and staging velocities remain virtually the same.

A tri-core first stage that can be jettisoned as a single block also enables fixed struts for better stiffness, and opens up single barge propulsive recovery (if that is deemed preferable to fishing the cores out of the water). As a two stage vehicle, we'd expect 15-18t of LEO payload from this setup. But if we then parallel stage the upper stage, we can get even better performance, especially to higher energy orbits. In this setup, the lower three would fire and cut out together; then the upper three would light up together but the center would be quickly throttled and the outer two would be jettisoned first. We'd be adding a bit of complexity to the PAF to accommodate sending thrust through all three upper stage cores while being able to jettison the outer two later in flight.

With this 2.5 stage rocket partially reusable rocket, we can expect Falcon 9 like performance to LEO (upwards of 18t) and better than F9R performance for higher energy launches (6-7.5t to GTO, potentially up to ~3t direct to GEO). And while this tri-core version may or may not be quite as cheap to fly as Neutron or Falcon 9, it would be comparable, and in this payload class the not-SpaceX factor is the main advantage I'd expect Neutron to have anyway. Moreover, with this tri-core alternative, RocketLab would be able to compete for those more lucrative GTO & higher energy missions. And even fly some that would require SpaceX to fly a Falcon Heavy. It might even have lower cost/kg for LEO constellation launches than Neutron, given the ~40% higher payload capability, and it could accommodate larger fairings (with a tri-core upper stage through atmospheric flight, and a stubbier aspect ratio, it would have better stiffness than Falcon, and wouldn't have the constraints of Neutron style recovery on size and shape).

Shortly after the tri-core would come a cluster of 7. This would be similar to the tri-core, with a fixed 7-core first stage, followed by an outer ring of 6 on the second stage that's jettisoned, leaving the upper center as a third stage. In this configuration we would wait to light the upper center until the upper outer cores are jettisoned. This configuration yields north of 40t to LEO, upwards of 15t to GTO, and crucially, 7t or more to GEO -- the performance level required for the likes of NSSL Lane 2, and potentially even the likes of Europa Clipper, all with full first stage reuse.

The 7-core would be about as heavy as a Falcon Heavy (and with similar liftoff thrust), but with a nice squat profile and a 9-10m span. It could easily accommodate a 7m or larger fairing. Even 12m for those really bulky payloads. And it could stand on its own without a strongback just as easily as Neutron (the 3-core could also do this with 2-3 legs on each of the outer cores).

To further make the case for the sizing, while the 3-core would be good for launching crewed and cargo spacecraft to LEO, the 7-core could launch them to TLI.

So what's the upshot? We end up with three rockets, instead of one, with comparable cost/kg to Neutron. They are suited to a wider range of payloads, and as such have more of the launch market accessible to them. They are much closer in design to Electron, and the components are far more modest, meaning much lower development costs, at least in getting to the single core variant. And the three variants have a lot of parts commonality and can be designed from the get-go to also share ground infrastructure and leverage many of the optimizations intended for Neuron. So the iteration from single to three to seven core variants should also be cheap. We have lower risk, because each step is a smaller gamble with immediate payoff (cheaper and more easily scalable propulsion on Electron; a modest MLV filling a niche, a Falcon 9 competitor, and a a proper heavy launcher). But most importantly, the timeline would move forward a lot.

Which brings us to another reason it would have made sense to start working in this direction in fall 2019. A day before RocketLab announced they would pursue Electron reuse, SpaceX announced dedicated rideshares, and three weeks later, they cut prices.

It was obvious at that point that SpaceX was going to suck the air out of the room for small launch, and that they were positioned to capture most of the growth in smallsat launch demand. Dedicated small launch would not meet the growth projections of optimistic small launch companies. (See above with RL's 50-120/yr projection for Electron.) They would need other revenue sources (or at least the promise of them) ASAP to pay their fixed costs and to have capital with which to grow (and a market to grow into). A quickly developed medium launcher based on Electron, fitting in the Delta II niche that had been left open in the West, and extendable into a larger medium and even heavy launcher, seems like a pretty sensible choice in that context...

Now let's talk timing. Peter Beck himself has said that propulsion is always the thing that drives development schedule -- the last thing to be ready. His defense of Neutron timelines back at the end of 2021 was that they were going for a simple gas generator cycle. And he said that if you can start with a well-trodden path with propulsion you've solved half the problems. He went on to say that a lot of the other stuff like valves scaled really well from Electron to Neutron.

Fast forward to fall '22, and RocketLab was announcing that Archimedes was now going to be a staged combustion engine for the sake of reusability (ox-rich staged combustion would allow them to keep preburner temperatures low).

That means that RocketLab switched to working an ORSC design 2.5-3 years after they would have had good reason to work on a much simpler big-Rutherford. I contend that had they started working on such an engine in 2018, it would have likely been ready 3-5 years before Archimedes will be. Structures too would have been less ambitious, as would the recovery scheme. They could have pursued either Falcon 9-like propulsive landing or Electron-like parachute recovery, or just left both design directions open. But Big-Electron could have started out expendable with reasonable cost, and iterated on reuse as it ramped up from a few launches a year. The two larger variants could have followed quickly on the heels of the single core, and RocketLab would have been in a good position early on to serve the full range of NSSL Phase 3 launches. They may have gotten a number of Lane 2 launches (as a third or potentially fourth provider -- the rules may have been changed to onboard them to Lane 2, as they were for Blue (granted, with lobbying)), as well as many Lane 1 launches.

And even before such opportunities were realized, working towards them at an earlier date would likely have enabled capital raises on more favorable terms.

Then there's the matter of propellant choice. Reusability had also been given as the primary reason for switching to methane. Reusability, not propellant cost, which was considered to be approximately a wash given the higher fixed costs; not isp, which was icing on the cake; not having an overlapping temperature range in your propellants. It was about not having to clean your engine between flights. In Peter Beck's words, "Ultimately that's what drove the decision into methane, and nothing else. It was purely how do we actually get reusability in these engines reliably and just not have that coking issue."

Coking, it turns out, is not nearly as big of a problem with modern RP-1. For a big enough engine (with high enough surface to volume ratio), it's not hard to keep regen coolant channel temperatures low enough to prevent coking there. And injectors can likely also be kept clean by virtue of film cooling. As far as the main combustion chamber walls go, coking is actually desirable as a regenerative insulating & ablating layer that helps keep the walls cool. So if you can prevent coking in your powerhead, you should be good to go without cleaning between flights. Guess which cycle doesn't need hot fuel to power the pumps? Electric...

One more thing. I said I'd present a roadmap to full reuse, but so far I've only talked about reusing the first stage and fairings. What about the upper stage(s)?

This is where things get a bit more complicated. There are a number of options, but what I would focus on is a strategy optimized for the 7-core variant. This is because the gains are greater, and because I expect the market to gravitate towards heavier launch as prices come down.

There are a number of ways to recover and reuse the second and third stage, but in this context I favor a large HIAD (hypersonic inflatable aerodynamic decelerator) for the second stage. For the third stage, I'd use rigid TPS & leverage the existing thrust structure as part of an aeroshell.

Our second stage is composed of six cores arranged in a circle with an outer diameter around 10m. For launch and recovery ops, engine-out capability, and dry mass, it'd be better to keep the six cores joined together with fixed mechanical and plumbing connections, and recover it as a block. The third stage would slide out (release clamps, then slide on interior bushings). A ~30m diameter HIAD ought to be enough for aerodynamic stability for the second stage, and at that diameter and ballistic coefficient, even a fully ballistic reentry from near orbital speeds would be within the limits of flexible reusable TPS materials. An asymmetric HIAD would provide lift and enable even lower temperatures at the cost of higher total heating, longer recovery distance, higher complexity, and more failure modes.

The third stage has the central core, but we also want to get back the PAF and the upper thrust structure. The thrust structure would primarily comprise an inverted conical section flaring out to ~7m in diameter to transfer loads from the second stage to the PAF and the third stage (which would in effect be a hung stage, at least for second stage flight, if not also for first stageg flight), and from the third stage to the PAF. So we've got a natural shuttlecock shape with a ~3m diameter shaft leading up to a ~7m diameter inverted conical section (which is then capped with the PAF -- another conical section and a ~flat top). In between the two conical sections we've got a fair amount of volume which we can use for additional propellant to a) shorten barrel section, and b) accommodate higher propellant loads for higher energy launches while keeping the same staging velocity to aid in practical second stage recovery. Since the third stage isn't lit until 2nd stage burnout, we can fit a multi-petal rigid heat shield at the base of the 3rd stage. This shield would be closed when the rocket is stacked (to fit inside the second stage inner diameter) and would open up after SECO and second stage separation, and then close back up for reentry. TPS on the tank barrel section and inverted thrust cone, hinged covers for the second stage mount points, off-axis center of mass (battery placement near the 7m diameter) for lift, RCS thrusters for roll, and a net/wire catch at/near the launch site complete the picture.

Battery advances and thrust increases (can re-raise chamber pressures with higher capacity batteries and/or widen the throat) with stretched tankage would enable going to full reuse without compromising performance.

Then, from there, there is a path towards further optimization. Methalox would save on fluid costs -- fuel, ignition fluid, and pressurant (assuming a switch to autogenous pressurization. The lower density would require a tank stretch to keep the same performance, and that would be easiest for the first stage would be the easiest, in terms of changes needed for stage recovery. It would yield the biggest operational cost improvements, and would pair well with a switch to a new large engine with a different cycle. I'd go to 7 ~300tf FFSC methalox engines (one per core). This switch would be in the 30s, giving RocketLab time to hire a lot of people who worked on Raptor. After this, the next and final step would be a two-stage-to-orbit rocket with a larger diameter, using more of the same engine, as well as 1-3 vacuum-optimized engines on the second stage, and return to launch site on both stages. A smaller methalox engine could be developed for landing the second stage, and used on an optional third stage for high energy launches. Said third stage could also be the basis for an in-space tug. Derivates of either or both the second stage and third stage could pursue further in-space applications. A mini-Starship stack with an optional third stage, if you will.

In summary, this would be a 5-phase plan:

Phase 1 (2020-2021 first flight): Big-Rutherford (32-35tf sea level kerolox electric, same battery requirements as Electron octaweb) flying on Electron.

Phase 2A (2022-2023): Anti-Neutron-1 single-core. 5-6t LEO lifter using 9 Big-Rutherfords on the first stage (recovered downrange) and 1 vacuum-optimized one on the second, with 2 hot-swapped pre-flown stage 1 batteries. Transporter and light-medium dedicated launches. Decidedly lower cost per launch than Falcon 9.

Phase 2B (2024): Anti-Neutron-3 -- tri-core, 2.5 stages, monolithic tri-core first stage recovered downrange (ikely propulsively), upper stage with throttled center and jettisoned sides. Falcon 9 peer to LEO with better higher energy performance. Likely higher marginal cost per launch than Falcon 9, but still quite profitable at current F9 prices, targeting mainly customers not wanting to fly exclusively with SpaceX, capable of getting a range of GTO & higher energy contracts. In time for early ramp to NSSL Phase 3 Lane 1.

Phase 2C (2025-2026): Anti-Neutron-7 - 7-core, 3 stages, competitor to Falcon Heavy, Vulcan and New Glenn. In time to bid for NSSL Phase 3 Lane 2 (or to get Lane 1 expanded to include some higher value flights).

Phase 3 (2028-2032): Anti-Neutron-7-FR. Full reuse for Anti-Neutron-7 with thrust upgrades and tank stretch to retain partial reuse performance capabilities with the switch to full reuse. Downrange water recovery for the 2nd stage with a HIAD and launch site wire or net recovery for the 3rd stage, thrust structure, and PAF with shuttlecock aerodynamics and rigid TPS. Replaces Anti-Neutron-1 and Anti-Neutron-3, as well. 1st stage RTLS reduces recovery cost of 1st and 2nd stages for less demanding launches.

Phase 4 (2030-2035): AN7FR+. 300tf FFSC methalox engine (well-established tech at this point with lots of ex-SpaceX people available to work on it) and autogenous pressurization and moderate tank stretch on first stage cores.

Phase 5 (2035-2042): NextGen launcher; ~7m diameter single stick. All methalox, 9-13 RL-Raptor engines on the first stage with RTLS baselined. Second stage with 1-3 RL-RaptorVac and a number of smaller landing engines. Optional HIAD recoverable third stage with 1-3 vac-optimized mini methalox engines. Fully competitive in the Starship world.

I'm expecting that it will take SpaceX at least until the late 20s to realize their low single digit million marginal launch cost goals. And I expect it'll take way longer for the market to catch up enough to enable average (not minimum) launch prices anywhere near that level. So this plan could have given RocketLab a smooth, low risk ramp to a post-Starship world, with a progression of capability and cost reduction to keep up with evolving markets, while getting them into medium and heavy launch about as soon as possible.

----

*I don't know the exact chamber pressure of a Rutherford, but this article from 2015 (back in the prototype days) has 1400psi (~100bar) as pump output pressure. Chamber pressure would be lower, but probably not lower than 50bar, maybe higher than 70 bar. And pressures may have increased with the substantial performance increases relative to that pre-production engine.
There are also a lot of references around to 37kW motors on each of two pumps (one fuel, one ox), as well as a battery power rating over 1MW. Astute observers might note that 9 pairs of 37kW motors would have a total power requirement of 666kW, and a >1MW power supply might be overkill unless you have >30% losses between your battery and motors. You might also note that since there's a substantially larger volume of oxygen to be moved, you'd need a lot more power on the ox side to hit the same output pressure. Then again, maybe the ox output pressure is a lot lower than the fuel side, because the ox side is not subject to the pressure losses of regeneratively cooling the engine.
Rutherford has also experienced thrust and isp increases since those pump power figures came out.

As a side note, RocketLab provides a single isp figure for sea level Rutherfords, which seems too high to be sea level isp and too low to be vacuum isp, but does seem very much in line with expected isp at optimal expansion (with an exit diameter of ~25cm, exit pressure is likely in the neighborhood of 0.3bar).