Aluminum Extrusion

Aluminum Extrusion Billet Yield: Cutting Scrap on the Press Floor

Billet yield is won or lost on butt length, setup scrap, and saw trim. A few pounds per billet matter fast when the press runs all shift.

Billet yield is the primary variable cost driver in aluminum extrusion, calculated as: yield % = (shipped weight / billet charge weight) x 100. Losses come from four sources: butt discard (the tail end of the billet left in the container, typically 8% to 15% of billet weight), crop ends and saw trim (run-out length removed from the front and back of each extrusion length, typically 2% to 5%), setup or trial lengths before the run is in-tolerance dimension (0.5% to 3%), and in-line or post-extrusion rejects (0.5% to 2%). A well-run press with an optimized butt length policy, good die condition, and tight saw programs achieves overall billet yield of 82% to 88%. A poorly managed press may run at 70% to 76%, representing a difference of 10 to 15 lbs of aluminum per 100 lb billet, at $1.20 to $1.50 per lb material cost.

Butt length is the single largest controllable yield loss in extrusion. The butt is the plug of aluminum left in the container at the end of each press cycle because the press cannot extrude the material against the die container without leaving a minimum tail thickness. Standard butt practice is to maintain a butt of roughly 10% to 15% of starting billet length, which on a 24-inch billet is 2.4 to 3.6 inches. However, some press crews run conservative butts of 20% to 25% of billet length to protect against surface inclusions reaching the die, particularly on critical surface profiles. Tightening butt length from 3.5 inches to 2.8 inches on a press running 500 billets per day at 5 lbs per linear inch of butt cross-section saves roughly 350 lbs per day of aluminum, or $420 per day at $1.20 per lb. Over a 250-day year, that is $105,000 from one process standard change.

Saw trim yield loss depends on finished length variation, cut plan optimization, and operator overcutting habits. A 20-foot profile run cut to 16-foot pieces should generate 1.25 lengths per profile with a 2.5-foot crop on each end if the process is running perfectly. In practice, dimension variation at run start requires discarding the first 12 to 24 inches of production until the die reaches thermal equilibrium, which adds to crop loss on every billet transition. Cut plan optimization software that matches required cut lengths to available run length, including taper in the back end of the extrusion, can recover 1% to 3% of total run yield, worth $15,000 to $45,000 annually in a high-volume shop.

Recovery value from scrap, butt, and trim does not offset material cost dollar for dollar. Primary aluminum billet may cost $1.20 to $1.50 per lb, but scrap returns at secondary aluminum prices of $0.50 to $0.80 per lb. The difference, $0.40 to $1.00 per lb, represents the permanent cost of material that went through the press but did not become a shipped product. Converting this loss to dollars requires knowing the actual yield, the volume of material lost, and the spread between primary and secondary aluminum price. In a vertically integrated shop with its own melter, the spread may be smaller, but the energy and labor cost of remelting still reduces the net recovery versus virgin metal.

Tracking billet yield at the die or profile family level reveals patterns that aggregate plant-level data hides. Some profiles with complex die geometry, thin walls, or tight tolerances require longer startup crops and generate more discard per billet than standard structural shapes. If a complex profile runs at 71% yield while standard structural runs at 85%, the complex profile's true conversion cost is significantly higher and must be reflected in the quote. Shops that price all profiles at the same yield assumption systematically underprice complex work. A billet yield calculator that takes profile family, billet length, butt target, and crop allowance as inputs gives the quoting and scheduling team the precision needed to price accurately and compare actual versus standard performance in real time.

Published 2026-05-28.