Musk Lists Raptor Engine Parts Removal Checklist
Elon Musk responded to how complex piping in the first generation of Raptor engines was removed or merged in the second and third generations by listing five categories of parts: secondary propellant and purge lines, power and sensor harnesses, sensors, avionics boxes, and main combustion chamber spark igniters. Many bolt flanges have been replaced with welded joints, reducing weight and leak paths, but some parts are now harder to replace individually.
He emphasized that while removing parts is theoretically simple, practical implementation requires multiple iterations to determine alternative paths, and details are not disclosed. The thrust of the Raptor 3 sea-level variant has increased from about 230 tons to about 250 tons, and the vacuum variant from about 258 tons to about 275 tons; the sea-level engine weight has decreased from about 1630 kg to about 1525 kg. Sensors and controllers are integrated into the airframe, the fairing has been removed, and a new ignition scheme has been implemented across all models, saving about one ton of weight per engine for the entire rocket.
The ignition architecture has shifted from main chamber spark plugs to preburner hot gas ignition. This means that ignition is now more sensitive to temperature and propellant timing: if the preburner does not reach the threshold, the entire system may fail to ignite. Flight 13 was automatically aborted due to some booster engines failing to ignite, and two Raptor 3 engines were subsequently replaced. Early flights also experienced rear chamber leaks and self-destruction after communication was interrupted, leading the company to modify fuel lines, propellant temperatures, ventilation, and purging systems.
The Starship 3 version is also undergoing simplification: the engine fairing has been completely removed, CO2 fire suppression has been simplified, the rear chamber cavity has been reduced, and propulsion and avionics are more tightly coupled. The flaps have been changed from dual actuators to single actuators with three motors. The avionics side has about 60 custom units, with a peak electrical power of about 9 megawatts, and includes microgravity liquid level RF sensors. These removals and welding follow the same logic: fewer interfaces, less leakage, and less disassembly.
SpaceX's long-standing design mantra has been "remove parts or processes first; if the proportion that cannot be added back is less than about 10%, it indicates that the removal was not sufficient." The Raptor 1 to 3 has applied this principle to the methane full-flow staged combustion piping. The welding replacing flanges aligns with the stainless steel rocket body approach, and the company has stopped using 301 stainless steel, opting for self-developed alloys instead.
In market mechanisms, the buyer is the engine hours needed for the deployment and reusability of Starlink 3, while the seller is the McGregor production line and Starbase test stand. Funding has shifted from replaceable maintenance parts inventory to a production rhythm of one-time welding and complete replacements. Beneficiaries are those who can reduce the weight of 33 sea-level and vacuum engines together; the pressure bearers are the traditional engine supply chains that still rely on flange replacements on-site. Event-driven factors stem from continuous flight test feedback on the ignition reliability of the third-generation engines and rear chamber leaks.
Additionally, he did not provide the weight in grams and leakage rates for each category of removed parts, nor did he explain how to recover from preburner failures in flight after the spark igniter was removed.
Source: Public Information
ABAB AI Insight
Musk's list reflects the same algorithm he has repeatedly discussed at Tesla and SpaceX: the first version aimed to work around sensors and flanges, the second version integrated functions into structures, and the third version treats detachable interfaces as adversaries through welding and integration. The Raptor 1 resembles laboratory piping, while Raptor 3 looks like a single cast piece; this transformation is not a single inspiration but the result of multiple test runs that turned leaks and ignition failures into a removal catalog. SpaceX early on set the goal for Raptor 2 as "removing a lot of piping," and Raptor 3 welded sensors and controllers into the shell, effectively turning avionics from an external box into the engine skin.
The capital path follows the maintenance philosophy. Flanges allow for parts replacement at the factory, quick repairs at the launch site, and inventory turnover; after welding, the economic equation shifts to "decommissioning the entire unit and introducing a new one." McGregor is responsible for mass production, Hawthorne focuses on research and vacuum variants, and Starbase uses static firing for selection. The motivation is to eliminate interface failures of 33 engines from the launch window, trading weight for payload and reducing leakage for reusability. The approximately 1-ton savings per engine, multiplied by the number of booster engines, translates into additional kilograms that Starlink satellites can carry.
Analogous developments can be seen in jet engines moving from detachable accessory casings to more additive whole parts, as well as Tesla reducing wiring from kilometers to structural battery packs. Blue Origin and ULA still rely more on traditional accessory layouts, while SpaceX is in a phase of forcing parts to disappear through flight data. Replacing spark plugs with preburner ignition is a typical case of swapping one failure mode for another: losing one ignition hardware introduces stringent requirements for timing and thermal state.
Structural judgment belongs to technical substitution. The mechanism is that the cost of a rocket does not lie in the metal itself, but in every leakable joint and every breakable wire. When reusability becomes the pricing unit, the convenience of detachable repairs turns into a weight tax and a leakage tax. Whoever can migrate functions from parts into structures simultaneously reduces dry weight and enhances launch reliability. Removing parts may seem like engineering fastidiousness, but it fundamentally reclaims maintenance rights from the launch pad back to the design room.
ABAB News · Cognitive Laws
What is removed is not parts, but points of failure.
Welding saves weight but also sells off reparability.
Theoretical subtraction must be validated through testing.