3D Printing Liquid Rocket Engines: How SLM Parts Actually Hold Together

• 1
  Why a rocket engine is a natural fit for laser powder bed fusion
• 2
  Inconel 718 in an SLM build: what changes versus wrought
• 3
  Where the design rules bite: overhangs, thin walls and drain paths
• 4
  How you know the part is good before you light it
• 5
  When to print, when to machine, and when to do both
• 6
  From CAD to a fired engine: the sequence we follow
• 7
  Additive versus subtractive for propulsion hardware

A liquid rocket engine does two jobs at once: it has to survive a 3,000 K combustion zone on one side and cryogenic propellant on the other. In a regeneratively cooled design, the propellant runs through channels cut into the chamber wall before it reaches the injector. Those channels are the reason additive manufacturing shows up here first.

Machining those channels into a curved chamber wall means either splitting the chamber into segments and brazing or welding them back together, or cutting them from outside and closing them with a jacket. Every joint is a leak path and a thermal discontinuity. Laser powder bed fusion skips the joints: you grow the wall and the channels in the same build.

The trade is that you inherit a different set of problems. Surface roughness inside those channels runs Ra 8–15 μm as-built, which raises pressure drop. Powder can be trapped in internal cavities. Thin walls near the throat may warp. None of these are reasons to avoid the process, but they decide whether your first build is a test article or a paperweight.

• 1
  Joints are the enemyEvery weld or braze line on a cooled chamber is a potential leak at 200 bar.
• 2
  Roughness costs pressureAs-built channels need flow testing or an internal finishing pass before hot fire.
• 3
  Trapped powderAny cavity without a drain path is a risk; design it out before you build.

The SDSU team chose Inconel 718, which is the standard nickel superalloy for this class of hardware. It keeps strength to roughly 650 °C and resists oxidation in the combustion environment. In wrought form it is well understood. In a laser powder bed fusion build, the same alloy arrives with a columnar grain structure aligned with the build direction, plus residual stress from the rapid melt and cool cycle.

That anisotropy matters. A tensile bar cut vertically from the build often reads differently from one cut horizontally. For a chamber, the hoop stress runs around the barrel, so orientation of the build relative to the thrust axis is a design decision, not a printer setting. Rotating the part 45° on the plate can change both the stress state and the support structure needed.

Post-build heat treatment is not optional. A solution treatment plus double aging brings the microstructure closer to wrought AMS 5662 properties and relieves a meaningful share of the residual stress. Skip it and the first thermal cycle may do the stress relief for you, in the form of a crack.

Machining allowances should be left on any sealing face, injector interface or bolt circle. As-built surfaces do not hold a gasket. Plan 0.5–1.0 mm of stock on those features and cut them afterward.

Laser powder bed fusion builds on a powder bed, so anything hanging in space needs support or a shallow angle. A common shop rule is to keep unsupported surfaces above 45° from horizontal. Injector faceplates with deep orifices and chamber domes both push against that limit.

Internal channels need a drain path for loose powder. If a channel is closed on all ends, you either leave it open for the build and close it later, or accept that you cannot fully evacuate it. Neither is fatal, but both change the drawing.

Thin walls near the throat are the other pinch point. A 0.5 mm wall on a 40 mm diameter barrel is routine. The same wall on a large, hot chamber can distort enough to move the throat diameter out of tolerance. Thicker walls with more cooling margin are often cheaper than chasing the last gram.

Build orientation also drives support removal. Supports on internal surfaces are hard to reach and leave marks that disturb flow. Where possible, orient the part so critical internal surfaces face up and need no support at all.

Destructive cut-ups tell you what the process did, but they cost you the part. For flight-intent hardware, computed tomography is the practical check: it finds trapped powder, internal porosity and channel blockage without cutting. Resolution limits how small a defect you can see, so agree on the acceptance criteria before the scan, not after.

Dimensional inspection follows the drawing. Critical diameters such as the throat and injector bores are measured on a CMM, and sealing faces are checked for flatness. Wall thickness at the throat is often verified by ultrasonic or CT because a few tenths of a millimeter changes the thermal margin.

Flow testing is the step people skip. Push water or inert gas through the cooling channels and record pressure drop against the predicted value. A channel that is 30% blocked by powder will not show up on a CMM, but it will show up in the flow bench and, later, in a burned wall.

Finally, pressure test the cooling jacket at 1.5× the maximum expected operating pressure and hold it. A leak that weeps at 1.5× will become a spray at operating pressure.

If your engine has internal cooling channels that cannot be reached by a cutter, print the chamber and machine the interfaces. That hybrid route is what most small propulsion teams end up using: the geometry that only additive can make stays additive, and every surface that seals, threads or locates gets cut.

If the part is a solid manifold, a flange, a valve body or an injector plate with straight drilled orifices, machining is usually faster and cheaper. A 5-axis center can hold ±0.005 mm on those features and deliver in days without a heat treat cycle.

At GreatLight we run both. Metal additive for the geometry that needs it, and 127 CNC machines including 16 simultaneous 5-axis centers for the interfaces, plus the finishing and inspection steps that make the part usable. No minimum order quantity, from one prototype to 10,000+ parts.

Send the model and we return a quotation with a DFM analysis within 12 hours. Production can start within 24 hours, and parts ship in 3–5 days.

• 1
  Print the chamberInternal channels with no tool access belong in the additive build.
• 2
  Machine the sealsFlanges, bolt circles and injector bores need a cut surface, not an as-built one.
• 3
  Check bothCT for internal quality, CMM for interfaces, flow bench for the channels.