Twenty Parts, One Nozzle: What GE's LEAP Fuel Nozzle Teaches About Consolidation

A jet-engine fuel nozzle tip is about the size of a walnut. The version GE Aviation flew before additive manufacturing was an assembly of 20 separate pieces — machined, brazed and welded together, every joint sitting a few centimeters from combustor gas hot enough to melt the metal around it. Twenty part numbers, a rack of braze fixtures, and an X-ray check on every unit leaving the line. GE's answer was to stop building the assembly and print the tip as a single piece of cobalt-chrome.

Joints, not parts, set the cost of an assembly

Break down the cost of any small welded or brazed assembly and the parts themselves are rarely the expensive line. The money goes into interfaces: fixtures to hold pieces in alignment, weld and braze operations, then inspection to prove each joint is sound. A joint in a fuel system is also a leak path, and a leak path next to a combustor is a failure mode you keep inspecting for over the entire service life of the engine, not just at delivery.

Tolerance stack-up compounds the problem. Twenty parts means at least nineteen interfaces, each consuming a share of the total tolerance budget — which forces tighter, more expensive tolerances onto every individual piece. And the whole structure ossifies: once braze fixtures and inspection routines exist for one geometry, changing any single piece means touching all of them, so the design stops improving.

Twenty parts became one

The LEAP engine, built by CFM International — GE Aviation and Safran's joint venture — meters fuel into its combustor through nozzles whose tips contain intricate internal passages. Those passages route fuel so that it cools the tip on the way in, keeping it from coking in the heat. In the brazed design, the passage network was whatever twenty machinable pieces could form once joined. Printed by direct metal laser melting (DMLM) in a cobalt-chrome alloy, the network became whatever the combustion engineers actually wanted: the consolidated tip came out about 25% lighter and roughly five times more durable than the assembly it replaced.

The design move that matters: GE did not print a copy of the old assembly. The team redesigned the internal galleries as one continuous flow path, putting walls only where heat and pressure demanded them — geometry no machining setup could ever reach, because there is no way to get a cutter inside a closed passage.

This was never a prototype exercise. GE built a dedicated plant in Auburn, Alabama around the part, and by 2021 that plant had shipped its 100,000th printed nozzle tip. Each LEAP engine carries 19 of them, and LEAP powers the Boeing 737 MAX and much of the Airbus A320neo family — six-figure serial production of a flight-critical printed part, running for a decade.

What consolidation actually buys — and what it costs

The table below is the honest ledger. The famous wins are weight and durability, but for most assemblies the quiet wins sit upstream: procurement drops from twenty part numbers to one, braze fixtures disappear, and every joint you never make is a joint you never inspect.

Consolidation also sends a bill. Internal features cannot be measured with calipers, so you inherit CT scanning or borescope inspection. A defect scraps the whole part instead of one sub-piece. And a consolidated part is design-locked: change one region and you re-qualify the entire component. GE accepted all of that because joint count was the dominant failure driver in this part — that is the trade you are actually making. Note also what the trade does not require: volume. The fixtures, assembly steps and joint inspections you delete are deleted at quantity one, so consolidation pays on the first article — exactly where prototype-stage teams live.

Spotting consolidation candidates in your own BOM

You do not need an engine program to use this. Walk your assembly drawings and score candidates against the signals below — the strongest ones combine a sealing function with parts that only exist because machining forced the split.

For a Taiwan hardware team the shortlist is usually not a nozzle — it is the six-piece brazed coolant manifold on a test rig, the bolted bracket stack inside a machine frame, the fitting assembly that keeps failing pressure checks at the same joint. Metal printing often loses a naive per-piece price comparison and starts winning the moment you price in the fixtures, labor and inspections the assembly drags along.

Count joints before you count grams: the LEAP nozzle earned its place by deleting failure modes, and the weight saving rode along for free. When you have a candidate, price both routes side by side — quote the consolidated design and the original stack of parts through Orinovate's 3D printing service and let the totals argue.