First, Aerospikes only have better efficiency over staged rockets over part of the launch to orbit. At sea level it has vurtually no advantage, it’s not until Tye rocket ge s to thin air that the Aerospike efficiency is higher. But guess what happens shortly thereafter? The first stage is dropped and a vacuum optimized second stage takes over, again reducing the aerospike efficiency advantage to near zero. And that slightly better efficiency over part of the flight isn’t free, it costs extra weight over bell shaped nozzles,
So the big win is for Aerospikes is SSTO, that’s where it’s far more efficient than fixed nozzles. But because of the rocket equation, SSTO doesn’t make sense. Dropping off a big chunk of mass half way up (the first stage) provides substantial efficiency gains.
Musk has shown that the key to making space flight affordable isn’t breakthroughs in engine performance, and its not reusability, at least not yet. Before SpaceX ever landed a booster they were 30x cheaper than the Shuttle, and 20x cheaper than the SLS. How did they do it when the Falcon 9 engine has very average performance, and when they didn’t reuse any of them?
The reason is same as Henry Fords, assembly line manufacturing. They mastered making Merlins cheaply by building them in high volumes and constantly improving at it. Nine engines per rocket made that possible, just as it’s a key to their making reuse work.
> The reason is same as Henry Fords, assembly line manufacturing. They mastered making Merlins cheaply by building them in high volumes and constantly improving at it. Nine engines per rocket made that possible, just as it’s a key to their making reuse work.
This is one ingredient, and may be the largest, but there are several others as well. To start with, engine R&D costs were significantly reduced because Elon was able to build atop a well-established foundation by Tom Mueller whom he hired from TRW. That's not a dig - it has proven to be a very smart move. Vertical integration, and not having to pay Aerojet Rocketdyne their margin on each engine also helps.
Another is that a lot of his initial R&D was done via the Falcon 1, sponsored in significant part by DARPA.
Yet another is that the legacy rocket companies do business with the traditional "Mission Assurance" mindset that accompanies national security space missions as well as high-value commercial missions (e.g., GEO sats). Starting in the late 2000s free of that legacy enabled a completely different culture that was far less plodding, risk-averse, and legacy-driven. Additional mission assurance was then driven in after failures (like the early Falcon 1 flights) and more when SpaceX put themselves through the USAF launch certification process.
Don't underestimate the impact of the legacy mission assurance mindset. The US has a major Federally-Funded Research and Development Center, The Aerospace Corporation where likely the largest component of its income is being involved in DoD mission assurance.
> At sea level it has vurtually no advantage, it’s not until Tye rocket ge s to thin air that the Aerospike efficiency is higher.
There's more to it than this. Almost all practical rocket engines will have nozzle area ratios sized for altitude rather than sea level. That is to say, actual design nozzle area ratios will be significantly higher than the optimal ratio for ground level. This implies overexpansion losses at low altitude. The higher the area ratio, the worse this effect is. Rockets are designed this way because when you consider the mission-average Isp, it's better to reduce the underexpansion losses at altitude through higher area ratio and eat the overexpansion losses at low altitude than to optimize for liftoff conditions.
An ideal aerospike lets you run a very high geometric area ratio with reduced overexpansion penalties. So it's conditional as to whether the aerospike "has an advantage" at sea level. If you are running a large enough area ratio, then compared to an equivalent area ratio bell, the aerospike will minimize the overexpansion losses at lower altitude, and additionally it will have reduced unsteady pressure loads due to avoidance of internal shock-induced separation, the way you would with a bell. The limiting area ratio for a bell nozzle is usually this internal shock separation criterion, whereas that constraint is removed for the aerospike. So at high enough area ratio, the aerospike could indeed have a low altitude benefit over an equivalent area ratio bell.
As you point out, in practice, this effect doesn't tend to buy you a large net difference in mission-averaged Isp, particularly with staged rockets. Also as you correctly indicate, the largest mission-average Isp benefit for an aerospike is when the engine fires over a large enough range of altitude (e.g., SSTO) and when you can maximize the geometric area ratio. Such a case would also include the Shuttle, where an aerospike could have made a meaningful mission-average Isp difference, though for the same cycle parameters as the RS-25/SSME it likely would have increased engine weight.
Also note that assuming you truncate the spike (as all practical aerospike engines would be designed), a high area ratio aerospike configuration would be shorter than the equivalent area ratio bell.
It's not that easy an answer. This point about maximizing efficiency to maximize payload only really matters if you are trying to increase the payload delivered for an existing system of fixed size. If you have the latitude to design a new /derivative vehicle, then you have the latitude to size the tanks larger to account for increased propellant. Propellant is cheap compared to the other costs of manufacturing and operating a rocket-propelled vehicle, though as re-use becomes more and more common, the sensitivity to propellant costs might be higher.
Another thing to keep in mind is that efficiency in the form of increased Isp can be traded for engine thrust-to-weight. The SpaceX Merlin engine for example, uses an "inefficient" (i.e., low Isp) gas generator kerosene cycle, but has incredible thrust-to-weight (so is very mass-efficient).
Aerospike engines will tend to be heavier due to distributing the thrust chamber exhaust circumferentially as opposed to through a small central throat.
In practice, with a multi-stage system it is very hard to realize a system-level performance benefit from an aerospike configuration, which is the principal reason why they have not been used. Worth noting, the original Space Shuttle Main Engine configuration Rocketdyne developed was an aerospike, which makes sense given the engine fires from ground level to orbit. NASA Marshall apparently felt this was excessively high risk, so mandated a bell nozzle. The high area-ratio SSME nozzle is operating at the hairy edge of internal separation, and visibly "twangs" as the engine starts up and the internal shock blows out of the nozzle. Very low-throttle operation at sea level involves running with the shock and separated flow inside the nozzle and results in debits against the nozzle life.
Interest in the aerospike configuration waxes and wanes within the U.S. DoD & NASA, but currently the Air Force Research Lab is considering pursuing a modular aerospike engine program. The benefits are not in the realm of performance, but rather the aerospike Isp benefit is used to offset the lower thrust-to-weight. The real motivation is the desire to gang smaller thrust chambers and turbomachinery sets to provide more of a "Lego-like" low-cost way to design, test, and scale up these engines.
I wonder what the physics looks like for landing a linear spike. One of the problems that SpaceX mentions in their voiceover work for landings is that the thrust for the rockets is so high that even at the lowest thrust it's almost too much to land.
With a linear spike I wonder if they could turn off most of the pumps. Might give them more pitch control if only the middle ones were running?
For fixed gross mass. The point is, for a given payload mass fraction, you can increase the gross mass of the system to increase payload. And indeed for larger systems, you get a subsystem mass amortization effect that tends to decrease the dry mass fraction.
> I wonder what the physics looks like for landing a linear spike. One of the problems that SpaceX mentions in their voiceover work for landings is that the thrust for the rockets is so high that even at the lowest thrust it's almost too much to land.
The difference in nozzle performance shouldn't change this much at all. To make this problem more tractable, you need more net throttling capability. The problem is the gross mass at liftoff vs. at landing. When you size the engine(s) for liftoff mass, it's difficult to throttle them down enough to keep Thrust / Weight (T/W) low at landing when the system is not much more than first stage dry mass. This is especially true when a propellant is liquid as it leaves the main injector (such as with the Merlin, injecting liquid RP). Gas/gas injection (such as with an expander cycle) can help to some extent. See the CECE testbed engine for an example.
As per my comment about the prospective AFRL project, one of the ideas that tends to go along with contemporary aerospike engines is the idea of modular thrust chambers of smaller size. Depending on how granular the turbomachinery is and the chosen cycle, this approach could potentially permit deeper throttling. The XRS-2200 engine for the X-33 was designed with this modular approach.
One system-level benefit for the aerospike is the ability to use an "easy" cycle, like gas generator, and use the gas generator exhaust to improve the aerospike performance, such as by plug base pressurization. This is more constructive than what is done in conventional bell nozzle systems with gas generator exhaust.
Linear spikes exacerbate the mass inefficiency problem, BTW, and are not likely the way any operational aerospike engine will be designed in the future.
Also, when you think of throttling modular thrust chambers for steering, realize that you are either reducing the net thrust coming from the engine, which is not what you want to do, or you are jacking up chamber pressure on some modules while reducing it on others. Also not something you want to do, as if you have the capacity for higher chamber pressure, you want to use it for the whole flight, not just leave that mass margin there for steering. A number of studies have been done on steering using fluidic or hinging/flapped aerospike configurations, and again, it's hard to beat old-fashioned gimbaling. Yet another system-level reason why bells continue to be the status quo.
Surely it's a massively important cost in terms of weight? Engine efficiency isn't just about cutting costs, it's about maximising payload - and the majority of a rocket's mass is fuel. Even a 5 percent efficiency gain is a huge win for payload.