English

Turbo Manifold Prototype: Three Routes, One Part

Table of Contents
What did this manifold order actually need?
How was the first route chosen?
What did the wax print route prove?
When did tooling become worth it?
What does the route record contain?

This turbo manifold prototype was not built one way. The order used all three routes on one part: printed metal for fit in 4 days, a printed wax pattern for casting data in 14 days, and a 4-cavity tool for the run that passed 300 parts. Each route answered a question the previous one could not.

What did this manifold order actually need?

The order needed three different answers, and no single route gives all three. The first need was fit: whether the twin-port flange cleared the turbo, whether the six runners met the head at the right angle, and whether the bracket tabs landed inside their window. That is a geometry question, and geometry answers fastest in printed metal.

The second need was casting data. A printed part proves nothing about fill, and this manifold has a 2.5 mm wall running the length of six runners. Whether the metal reaches the far end before it loses heat is a gating question, and only a casting answers it.

The third need was repetition. Once the design froze, the programme needed parts at a rate that a printer cannot hold, and the design freeze record made a cavity safe to cut.

Written that way, the three routes are not alternatives; they are a sequence. The value of the sequence is that each step closes a question before the next step spends money on it.

How was the first route chosen?

The first route was chosen by the calendar. The test date sat inside the casting lead time, so the only route that could put a part on a bench in time was the printer. A 14-day casting lead time would have missed the window by a week, and an 8-week tool by far more.

That choice had a cost attached, and the cost was written down before the part was made. Printed metal at 1.0 mm of allowance and 800 °C for 1.5 h is a different starting state from a casting at 1.5 mm and 880 °C for 3 h, so the printed part could confirm fit and could not confirm machining or fill.

The team accepted that limitation explicitly, which is the part that matters. A printed prototype used to answer fit is evidence; a printed prototype used to answer fill is a guess. Marking which question a part answers is the discipline that keeps the sequence honest.

What did the wax print route prove?

The wax print route proved the two things the printed metal could not. It proved that the gating system fills a 2.5 mm wall at the end of a long runner, and it proved that the flange face cleans up inside the 1.5 mm allowance. Both are casting facts, and both changed decisions.

The fill result was measured, not judged. The part came out of the 7-layer shell with a 1600 °C pour and no cold shut at the last cavity, which is the check that a printed part cannot perform. The 880 °C for 3 h stress relief then left the flange flat enough to machine to 0.05 mm. That part also served as the scan baseline: a 3D scan against the model recorded where the casting sat before any machining, so the tool route had a reference point instead of a hope.

The pour itself ran through the standard vacuum investment casting line, which is the point of using a wax pattern 3D printing route rather than another prototype method. The part inherits the shell, the pour and the heat treat that the series will run.

The second result was cost of iteration. A printed wax pattern takes about 2 days to re-make, so a gating change costs days, not weeks, and the compromise between a hotter pour and a wider runner could be tested rather than argued.

What the wax route also proved is a limit. It produced one part per shell, so it could validate the process but could not carry the batch. That is the point where the tool stops being an expense and starts being the cheaper option.

When did tooling become worth it?

Tooling became worth it at 300 parts, and the number was known before the tool was ordered. Below that, the 4-cavity tool spreads over too few parts and the unit price stays above the printed routes; above it, the same tool pays for itself inside the first batch.

The tool decision also needed the design frozen, and this is where the sequence earns its keep. By the time the cavity was cut, the twin-port flange had been on a bench, the runner angles had been measured, and the gating had been proven in a real casting. The automatic wax press was booked while the cavity was still being cut, so the press was free the day the tool arrived.

Capacity was the other half of the decision. A 4-cavity tool returns four parts per cycle on an automatic wax press, which is what makes a 300-part run a schedule rather than a queue, and the cavity was cut to the same precision wax pattern rules as the printed pattern it replaced.

The first article from the tool was compared against the wax-print part, and the comparison is the deliverable. Same drawing, two routes, one set of measurements: that is what a first article comparison is for.

What does the route record contain?

The record contains one line per route, each with the question it answered and the number it produced. Route one, printed metal: fit confirmed in 4 days, wall 2.5 mm, machining allowance 1.0 mm, no statement on fill. Route two, printed wax: fill confirmed in 14 days, shell 7 layers, pour 1600 °C, stress relief 880 °C for 3 h, flange flatness 0.05 mm after machining.

Route three, hard tooling: cavity cut, first article at week eight, 4 cavities, same 0.05 mm flatness, M10×1.5 gauged with a go/no-go gauge, and every part inspected to CT6 under ISO 8062-3 with a 3D scan against the model at ±0.9 mm. Machining runs through the same superalloy CNC machining sequence for all three routes, so the only variable left is the blank.

The record is also explicit about what it does not contain. There is no customer name, no drawing number, no price and no bench result in it. Those belong to the parties involved, and the stainless casting case keeps the same boundary. What the record carries is the sequence and the measurements that made each step defensible.

That is the format worth copying for the next part: which route answered which question, with which number, and what the next route was allowed to assume because of it. It is also the format this order used to justify each spend before the money left, rather than after.

If a programme needs the same three answers, send the drawing, the test date and the annual quantity, and the reply will lay out the sequence with the number each step is expected to produce.