Machining large thin walled honeycomb welded assemblies

• 1
  Why large thin walled honeycomb behaves unlike solid stock
• 2
  Datums, clamping and how the weld seam is oriented
• 3
  Cutting sequence that keeps large thin walled honeycomb in tolerance
• 4
  What tolerance and finish are realistic on thin honeycomb
• 5
  Verifying a welded honeycomb assembly before shipment
• 6
  Which route fits which large thin walled honeycomb part

A honeycomb core is mostly air. Foil walls 0.05–0.10 mm thick carry shear between two skins, so the panel is stiff in bending but has almost no mass to absorb cutting energy. When that core is welded or brazed into a ring, the assembly becomes a closed loop of thin sheet with a stiff core inside it. Every cut you make changes where the stiffness sits.

Solid aluminium forgives a heavy pass. Honeycomb does not. A 12 mm end mill at 2 mm radial engagement on a 1.5 mm skin will push the skin away from the cutter before it shears. The deflection shows up as chatter marks, tapered walls and a surface that fails a Ra check. The usual fix is not a slower feed but a shorter, stiffer tool and a lighter stepdown.

The welded joints are the second problem. Welding puts residual tension into the sheet near the seam, and machining removes the material that was balancing it. The part moves after the clamps come off, sometimes by 0.3–0.8 mm over a 1,200 mm span. That is far more than the ±0.005 mm tolerance on a sealing face, so the sequence has to plan for relaxation rather than fight it.

Temperature matters more than most shops expect. A thin skin on a 4,000 mm assembly will grow roughly 0.05 mm per °C in aluminium. Cut the part warm in the morning and measure it cold at night and the numbers will not agree. Rough and finish in one thermal window where the geometry allows it.

Start from the weld. The seam is the least predictable feature on the part, so it should never be a datum. Pick three pads on the machined skin or on a cast boss, set them as A-B-C, and let the seam float. If the drawing dimensions everything from the weld centerline, ask for a datum change before the first cut, not after the first scrap part.

Clamping force is the quiet killer. A standard toggle clamp can flatten a 1.5 mm skin by 0.2 mm and spring it back when released, which leaves a part that measures correctly on the machine and wrong on the CMM. Use soft jaws, vacuum fixturing or low-pressure hydraulic clamps. Target clamp pressure that holds the part against a 0.02 mm shim without crushing it.

For the ring-shaped assemblies we see most often, a faceplate with six to eight support points and a Ø400 mm rotary table works well. The table lets you index the part between features instead of repositioning it, so the datum does not shift between operations. On a 4,000 mm part, plan two setups maximum and accept the re-datum error rather than chasing one perfect setup.

Support the core, not just the rim. Honeycomb panels sag between clamp points when the tool pushes down. Fill the unsupported span with a machined nest or a poured low-melt fixture so the skin has something behind it. This is the single change that removes most chatter on thin honeycomb walls.

Rough first, and rough generously. Leave 0.5–0.8 mm on skins and 0.3 mm on pocket floors, then unclamp and let the part rest for several hours. Welded assemblies move most in the first two hours after roughing. If the drawing allows, stress-relieve or vibratory-relieve before finishing. On 6061-T6 and 4130 parts we normally see the movement stop after that rest.

Finish the sealing faces before the pockets. Sealing faces carry the flatness and finish callout, usually 0.05 mm flatness and Ra 0.8–1.6 μm. Once the pockets are open, the part is far more flexible and the face will not come in clean. Cut the face, verify it, then open the core.

Open the honeycomb core with a small tool. A Ø6 mm three-flute carbide end mill at 8,000–12,000 rpm, 0.3–0.5 mm axial stepdown and 0.2 mm radial engagement cuts foil cleanly without tearing it. Higher radial engagement folds the foil over instead of cutting it. Air blast beats flood coolant here, because chips trapped in cells rub the walls on the next pass.

Cut the weld seam last, and only if the drawing requires it. Machining a seam relieves the tension that holds the assembly round. If the seam is not a functional surface, leave it as welded and save the risk. Where it must be cut, take it in two light passes and measure the roundness between them.

On a rigid boss or a solid flange, ±0.005 mm is routine on our 5-axis centers and we hold it every day. On free-standing honeycomb walls 1.0–1.5 mm thick, the same callout is not a machining problem, it is a metrology and handling problem. The wall deflects under the CMM probe tip. A realistic callout for a free wall is ±0.05 mm, and even that needs a light touch probe and controlled temperature.

Surface finish follows the same logic. Ra 0.2–0.8 μm is achievable on a supported face with a sharp tool and a finishing pass of 0.1 mm. On an unsupported thin wall, chatter sets the floor well before the tool does. Expect Ra 1.6–3.2 μm as-machined on free walls, and specify a supported face if you need better.

Position tolerance is where thin assemblies surprise people. A hole pattern drilled in the rough state can shift 0.2 mm after finishing because the part relaxed. Drill and ream after the finishing passes, or drill undersize and ream at the end. That single sequencing change fixes most hole-position complaints on welded honeycomb rings.

Do not over-specify. A drawing that calls ±0.005 mm on every surface forces the shop into slow light passes across the whole part and multiplies cost. Put the tight callouts on the sealing faces, bearing bores and mating pads, and let the rest sit at ±0.1 mm.

Inspect in the clamped and unclamped states. A part that measures well on the fixture and badly on the table has a clamping problem, not a machining problem. Record both numbers. That pair tells the shop how much spring is in the setup and whether the next part needs a different nest.

Use a light-touch CMM or a scanning head on thin walls. A 2 mm ruby stylus at low force is enough for most honeycomb skins. For sealing faces, flatness matters more than point thickness, so scan a grid and look at the form error rather than a few spot readings.

Check the weld after machining, not only before. Dye penetrant or a visual with a borescope through the open cells will show whether a light pass opened a seam. A 0.1 mm deep tear in a brazed joint will fail in service even though the part passed dimensional inspection.

We run raw material check, in-process monitoring and final inspection on these assemblies, with reports on request. Every part is inspected before shipment. Where the customer needs it, the inspection plan is agreed at the DFM stage so the first article and the production parts are measured the same way.