← Index

Novelis Internship

Novelis makes aluminum. Somewhere in that process a small thing called a grease pack gets assembled by hand, hundreds at a time, and it had never been automated. Eight of us spent the summer building a machine that does it.

A pack is five layers. Butyl rubber in the middle, about three grams of lanolin on each face, a PTFE disc pressed onto each side to seal it.

PTFE lanolin, about 3 g butyl rubber lanolin, about 3 g PTFE one pack, in section
By hand that is five steps, and the third one is a person scooping grease onto a disc.

The lanolin problem

The first thing that broke was weight. Operators scooped lanolin into the dispenser tube by hand, which trapped air in the tube, and trapped air meant every pack came out weighing something different. melting it was the whole trick

Lanolin turns out to be very thermosensitive. Melt it on a hotplate, pour it in as a liquid, and the bubbles are gone. That fixed the weight and immediately caused a leak, because what was now in the tube was a liquid. Most of the summer had that shape. One fix, one new problem.

My part, the controls

Last year's interns ran their build off a simple board. Ours had three servos, two linear actuators, a vacuum pump, solenoids and a pneumatic slicer, so it moved onto a CLICK PLC with a C-More touch panel on the front of the electrical box.

C-More touch panel pre-check, batch count, progress, manual menu Ethernet CLICK PLC sequencing, one slot per kind of hardware Servos flip, pivot, clamp Actuators lanolin, PTFE Solenoids vacuum, slicer
The panel sits on the outside of the box, which is the only part of this an operator ever touches.

The panel is the part I care about. An operator gets a five-screen pre-check before a batch, a field for how many packs they want, a progress bar that tells them when to load more rubber, and a manual menu that drives each actuator on its own for calibration. Someone can be shown the whole thing in a minute. That was the goal, not the screen being clever.

What didn't work

The rubber dispenser. We tried slicing the stack and we tried blowing it apart with compressed air, and the conclusion was that rubber does not separate mechanically. So it became a magazine holding fifty pre-separated pieces, moved along by a belt and screw system.

It worked on the bench and then would not fit. Too large for the space we had left for it, the belt too loose for the torque it was asked to carry, and we ended up adding finicky ramps to get a piece from where it came out to where it needed to be.

We designed the replacement, a gravity-fed slot dispenser sized to the space that actually exists, and ran out of summer before building it. It is the first thing next year's team should make.

What's live

The machine assembles packs. Lanolin dispenses to weight, the wrist flips and places, the vacuum head picks a PTFE disc up cleanly and sets it down. Rubber still gets loaded by hand each cycle, and the panel is what tells the operator when.