Question 2 of 7: Additive manufacturing and the fuel injector nozzle

Question 2: Additive manufacturing and the fuel injector nozzle (15 marks)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Product selected: the fuel injector nozzle for a gas-turbine combustor. It is the clearest case in industry, because the part that additive manufacturing replaced was not a simpler version of itself but an assembly of about twenty machined and brazed pieces.

Part A — Impact on the functionality of the product (6 marks)

Internal passages that cannot be drilled. A nozzle's job is to deliver fuel at the right flow number, break it into a spray of the right cone angle and droplet size, and keep its own tip cool enough not to coke. All three depend on internal geometry: swirl passages, a shielding air circuit, and cooling channels that follow the surface they are protecting. Conventional manufacture builds those from drilled holes and brazed sleeves, so every passage must be straight, must start at an accessible face, and must be plugged afterwards. Powder-bed fusion removes that constraint entirely. Passages can curve, can vary in section along their length, and can be conformal to the hot surface, so the swirl can be tuned for atomisation instead of for drillability and the tip can be cooled where it actually runs hot.

Part consolidation and the leak paths it removes. The widely reported production example, the LEAP engine nozzle, consolidated roughly twenty pieces into a single printed component, and the reported outcome was about 25 per cent lower mass and around five times the service life. The life improvement is the functional point and it is not mainly a material effect: every brazed joint in the old assembly was a potential leak path, a stress raiser and a place where fuel could stagnate and coke, and consolidation deletes all of them at once. This is Boothroyd and Dewhurst's minimum-part-count criterion arriving from an unexpected direction — the parts were separate only because of how they were made.

Mass, and what mass is worth here. A quarter off the mass of a part that flies, multiplied by the number of nozzles in an annular combustor, is fuel burn for the life of the engine. Additive processes also permit local wall-thickness variation and internal lattice stiffening that a machinist cannot reach, so mass can be removed from the places a topology optimisation says are unloaded rather than uniformly.

The functional limits are equally real and belong in the answer. As-built surfaces inside small passages are rough, typically several micrometres Ra, which changes discharge coefficients and cannot be polished if the passage is inaccessible; properties are mildly anisotropic with build direction; residual stress from the thermal cycle must be relieved; and any powder left in a blind internal channel is a foreign-object hazard in a flight part. These are design constraints, not objections — they are why design for additive manufacturing exists as a discipline.

Part B — Enhancing the design process (5 marks)

The iteration loop shortens from months to days. A nozzle variant that previously required tooling, fixtures and a brazing schedule can be built overnight, so the team tests many concepts instead of arguing about two. Ulrich and Eppinger's point about prototypes applies directly: the value of a prototype is the uncertainty it retires, and cheap prototypes let you retire uncertainty early, when changes are still cheap.

The prototype can be in the production material and by the production process. This is what separates additive from other rapid prototyping. A cobalt-chrome nozzle printed for a spray-rig test is metallurgically comparable to the shipped part, so flow number, spray cone and thermal behaviour measured on it are meaningful, and the design does not have to be revalidated when it moves from prototype to production.

The design freeze can move later. Because there is no hard tooling to commit, the geometry can keep improving while the rest of the programme proceeds, which changes the shape of the whole development schedule rather than merely speeding one task.

Simulation and topology optimisation become directly usable. The output of a topology optimisation or a conjugate heat-transfer study is normally an organic shape that has to be re-drawn into manufacturable features, losing most of the benefit. With an additive route the optimised shape can be built nearly as computed, so computational fluid dynamics on the swirl passage and finite-element work on the thermal stress feed the geometry directly. Design for additive manufacturing rules take the place of the old manufacturability rules — self-supporting overhangs beyond roughly 45 degrees from horizontal, minimum passage diameter for powder evacuation, deliberate powder-removal ports, build orientation chosen so critical surfaces are up-skin, and support structures placed where they can be reached and removed. A build-process simulation predicts distortion and lets the geometry be pre-compensated before any powder is spread.

Part C — Two major impacts on the manufacturing process (4 marks)

Impact 1 — the process chain and its quality system are replaced, not shortened. Twenty machined details, their fixtures, their inventory, the braze furnace, the braze alloy qualification and the assembly operation all disappear. What replaces them is a different and unfamiliar chain: powder receipt, sieving, reuse tracking and lot control; the build itself with in-situ melt-pool monitoring; stress relief; wire-EDM removal from the build plate; hot isostatic pressing to close internal porosity; abrasive flow or chemical finishing of internal passages; computed-tomography inspection because there is no other way to see inside a consolidated part; and flow testing. Inspection moves from dimensional checks on twenty details to volumetric inspection of one, and qualification moves from qualifying a part to qualifying a machine, a powder lot and a parameter set. Powder handling also introduces a genuine occupational hazard — fine reactive metal powder requires inert handling, grounding and dust control under the applicable provincial occupational health and safety regulation.

Impact 2 — the cost structure inverts, so the economic batch size changes. The conventional route carries large fixed cost in tooling and fixtures and a modest variable cost per part; the additive route carries almost no tooling but a high variable cost driven by machine time and powder. Cost per part therefore stops falling with volume in the way a manufacturing engineer expects, and the two routes cross at a definite volume that has to be computed rather than assumed.

Given. The machined-and-brazed route needs 240,000 CAD of tooling and fixtures and costs 1,100 CAD per nozzle; the additive route needs 30,000 CAD of build-plate and parameter development and costs 1,850 CAD per nozzle.

Find. The annual volume at which the two routes cost the same.

1. Write each route as a fixed plus variable cost. With $T$ the fixed cost and $u$ the unit cost, the total for $n$ parts is $C(n)=T+un$, so the cost per part is $c(n)=T/n+u$.
2. Set the two totals equal and solve. $T_{\text{mach}}+u_{\text{mach}}n = T_{\text{AM}}+u_{\text{AM}}n$ gives $n^{*}=\dfrac{T_{\text{mach}}-T_{\text{AM}}}{u_{\text{AM}}-u_{\text{mach}}}=\dfrac{240{,}000-30{,}000}{1{,}850-1{,}100}=\dfrac{210{,}000}{750}$, so $$\boxed{n^{*}=280\ \text{nozzles per year, at a common total of }548{,}000\ \text{CAD}}$$
3. Read the result the way a manufacturing engineer would. At 100 units the additive route costs $215{,}000$ against $350{,}000$; at 600 units it costs $1{,}140{,}000$ against $900{,}000$. Below 280 units a year additive wins outright, above it the conventional route wins on cost alone — and the functional gains of part A are what must justify the difference at high volume.