Heatsink porthole die: an extrusion die rebuilt from a paper's drawings (3-axis)

Original source: T.-T. Truong (Hung Yen University of Technology and Education; National Kaohsiung University of Science and Technology), Q.-C. Hsu (NKUST, Department of Mechanical Engineering), V.-C. Tong (Korea Institute of Machinery and Materials), J.-J. Sheu (NKUST, Department of Mold and Die Engineering), A Design Approach of Porthole Die for Flow Balance in Extrusion of Complex Solid Aluminum Heatsink Profile with Large Variable Wall Thickness, Metals 2020, 10, 553 (doi:10.3390/met10050553).
Search keywords: porthole die heatsink large variable wall thickness Truong Hsu Sheu, met10050553
Licence: CC BY 4.0. The paper's figures are not reproduced; the pictures are HiNC renderings of the agent's model, and drawings and charts made by the agent's own scripts.
Attribution: Die and profile geometry after T.-T. Truong, Q.-C. Hsu, V.-C. Tong, J.-J. Sheu, Metals 2020, 10, 553, doi:10.3390/met10050553, CC BY 4.0. Die model and machining set-up by Tech Coordinate's agent.
About the case: A two-plate porthole die, Ø180 mm and 90 mm tall, that extrudes a twelve-fin aluminium heatsink with walls from 1.51 to 23.13 mm; the paper's authors designed it by simulation, had it made and extruded the profile on a 930-ton press.

The story

The paper publishes drawings and a handful of dimensions, not a model. An AI agent read the proposed die off the figures — its porthole areas came out within 2.3 % of the paper's — and decided what a die shop mills and what it leaves to wire EDM: the portholes, the welding chamber, the bridge's chamfers, the pocket and the relief behind the bearing are milled; the die orifice and its bearing are wire-cut and left out. It wrote four set-ups and eight Fanuc programs with its own CAM, wrote its pass criteria down before the first play, and drove HiNC through its web API.

The upper plate in HiNC from its joint face: a 16 mm end mill 55 mm out of a shrink-fit chuck roughing the first porthole, the second porthole and the welding chamber's outline beside it — The upper plate while its roughing played: the 16 mm end mill in its shrink-fit chuck is halfway down porthole 1, and the stock is shown as cut at that moment.

The lower plate in HiNC from its exit face: a 3 mm long-neck cutter reaching from its shrink-fit chuck into the run-out step and cutting a fin slot, the wide relief already cut — The fin relief: the 3 mm cutter's neck reaches 40 mm below the chuck, which stays 7 mm or more above the exit face.

Section through both plates of the agent's die: the upper plate with its portholes' outer walls opening 7 degrees towards the lower plate and the bridge between them, chamfered at the top; the lower plate with the pocket, the bearing and relief between the fins, and the run-out step on the exit side — The die the agent read off the paper's drawings, cut through the bridge and every fin: the portholes' outer walls and the bridge's top chamfers open opposite ways, so the upper plate is machined from both faces; below the joint lie the 5 mm pocket, the bearing and the two run-out steps.

Bearing lengths made from the back: the relief behind the wire-cut bearing is milled in 16 depth zones, each deep enough to leave its segment's bearing length from the paper's Fig. 10, from 1.2 mm at the fin tips to 24 mm under the block.

Tip deflection of the 3 mm long-neck cutter against its step-down in the two kinds of fin slot: the 99th percentile falls from 106 and 130 micrometres at 0.1 mm to 48 and 63 at 0.035 mm, every point above the 30 micrometre line set beforehand — The 3 mm long-neck cutter in the fin slots, read on a block cut down to two slots at a 0.03125 mm grid: even at a 0.035 mm step-down its 99th-percentile bending stays above the 30 µm the agent set beforehand.

Time of each of the eight programs, the agent's estimate beside HiNC's simulated time: the two agree for every program, and the 3 mm cutter in the fin slots takes 419.7 minutes, more than twice all the others together — Every program's time, the agent's estimate beside HiNC's: 10.1 h in all, 7.0 h of it the 3 mm cutter in the fin slots, the hours that send the fin relief to EDM.

Pictures rendered by HiNC from the die model Tech Coordinate's agent built after the paper's figures; not the authors' CAD. The section and the charts are drawn by the agent's own scripts from that model and from the plays.

Four of its ten dilemmas

Rapid moves through stock

HiNC's first play reported four rapid moves that cut steel, up to 27 mm deep. The crossings between cutting levels ran the tool's edge along the roughing wall with no clearance. The agent made every rapid keep 1 mm off the walls and added its own height-field replay of each program before every later play; the rerun reported none.

One side cannot reach everything

The portholes open 7° towards the lower plate, the bridge's top chamfer the other way, so one of them is an undercut from either face. The upper plate is machined from its joint face and turned over only for the two chamfers.

Bearing lengths made from the back

The bearing is wire-cut through the whole plate, yet its length runs from 1.2 to 24 mm along the profile. The relief behind it is milled from the exit side in 16 depth zones, each leaving its segment's length; at the fin tips the zone is narrower than a 3 mm cutter, which the criteria named beforehand as work for EDM.

A grid coarser than the step-down

At 0.5 mm HiNC read the 3 mm cutter's 0.1 mm step-down as whole layers and its bending as 0.44 mm. The agent cut the case down to two fin slots and played them at 0.03125 mm with four step-downs in one program.

The other six, among them two dies in one paper, a source note that took the extrusion ratio for the ram speed, a build setting that slowed the play to two or three steps a second, and a 14 mm plunge the agent's own check caught before HiNC played it, are in the full record.

The result

The four set-ups played every line at 0.25 mm with no alarm and no collision, and the milled faces of the upper plate and the pocket came within ±0.15 mm of the model on 99.9–100 % of the points. The porthole roughing used 0.194 of the spindle's 11 kW continuous power and 0.14 of the tool's yield stress, and took 38.73 min against the agent's estimate of 38.68. Both plates take 10.1 h of machining, 7.0 h of it the 3 mm cutter in the eleven fin slots; there the cutter bends 106–130 µm at a 0.1 mm step-down (48–63 µm at 0.035 mm) against the 30 µm the agent set beforehand, and about 2 % of the relief lies in bearing zones narrower than the cutter. By the rule written before the first play (over 4 h, recommend EDM), the fin relief goes to EDM, and the milling program stays as the evaluated alternative. Three criteria were not met — the cutter's bending, the relief's share within ±0.15 mm, and the deepest point past the model (−0.32 mm on edges no tool touched, −0.78 mm on the relief) — and the full record says so. The four acceptance plays took 73 minutes on a shared 32-thread server. Everything is simulated; no die was cut.

Key number	What it is
10.1 h	simulated machining of both plates: upper 1.9 h, lower 8.3 h, of which the 3 mm cutter in the fin slots 7.0 h; tool changes, turn-overs and wire EDM not included
1927 / 1663 mm²	the model's porthole entrance areas against the paper's 1960 / 1702 mm² (−1.7 %, −2.3 %): the check on the reading of the drawings
4 → 0	rapid moves through stock HiNC reported: four on the first coarse play, up to 27 mm deep, none in the acceptance; no collision
0.194	the porthole roughing's highest spindle power against the 11 kW continuous rating; the limit written beforehand was 0.8
100 % / 99.9 % / 100 %	share of the milled faces within ±0.15 mm of the model: the upper plate from its joint face, from its billet face, and the pocket
98.1 %	share of the relief outside the profile within ±0.15 mm; the rest lies in bearing zones narrower than the 3 mm cutter and deeper than their neighbours
106–130 µm	the 3 mm long-neck cutter's tip deflection, 99th percentile, at a 0.1 mm step-down on the two-slot block; the limit written beforehand was 30 µm
4.9 h + 2.1 h	the 3 mm cutter's 7.0 h in the fin slots: cutting, and feed through air where the tool leaves a slot and comes down again
73 min	the four acceptance plays at 0.25 mm together, on a shared 32-thread server

NC optimization of this case →

What it brought

For a machining engineer: before any steel is cut, HiNC shows that the porthole roughing sits far inside the tools' strength and the power of the generic BT40 spindle it was set up with (11 kW continuous), gives the time of every program, and shows where a long-neck 3 mm cutter stops paying: seven hours, a tenth of a millimetre of bending, and bearing zones it cannot reach. That is the evidence behind the usual choice to sink the fin relief by EDM and mill the rest.
For a teacher or a student: how a porthole die is split between milling and wire EDM, why a die is machined from both faces, how bearing lengths are made from the back, and why a simulation grid must be finer than the step-down it judges.
For someone deciding whether to use an agent with HiNC: from a paper's drawings to an accepted simulation of both plates, with its own model, CAM and checks; the pass criteria were fixed before the first run, and no second agent rebuilt the case.

Read the full case record: heatsink porthole die

The source and its backup

Case	Original files	Backup
Heatsink porthole die	the article (Metals 2020, 10, 553)	Showcase-Heatsink-PortholeDie.zip

The zip holds the case's SOURCE.md and, under Setup/, the agent's scripts that build the die from the figures and write the programs, the numbers files with every value marked read, measured, derived or chosen, the pass criteria and the accepted results. The publisher's 12.8 MB article stays at the source, under CC BY 4.0; the scripts rebuild the meshes.

All Showcase cases