SMED

Running a Changeover Reduction Program That Pays

Every changeover hour is capacity you already paid for. How to value it, film it, cut it in half, and bank the savings in schedules and batch sizes.

Changeover time is capacity you already paid for. A press that changes over 6 times a week at 3 hours each gives up 18 hours, about 900 hours a year; at $110 per hour of machine burden that is $99,000, and on a sold-out asset the real number is the contribution margin those hours could have produced, often $250,000 plus. Cutting changeover in half either buys that capacity back or lets you run smaller batches and cut inventory 20 to 30 percent. Either way it is money, which is why changeover reduction is the rare improvement program a CFO will fund from a single page of arithmetic.

Value the program before you launch it. Annual changeover hours equal events per year times average duration: 300 events at 2.5 hours is 750 hours. Value each recovered hour three ways: labor released, machine burden, and, on constrained equipment, contribution margin of production gained. A plant at $40 labor, $95 burden, and $180 margin per bottleneck hour values a 40 percent reduction, 300 hours, at $40,000 to $54,000 on the first two layers and over $90,000 if the machine is the constraint. The Changeover Reduction calculator runs those layers so you can rank machines by dollar recovery instead of by whoever complains loudest.

The method is SMED, and it is 60 years old because it works. Film a full changeover, then classify every step as internal work requiring a stopped machine or external work doable while running. The first pass typically finds 30 to 50 percent of steps are external work being done internal: tools fetched, paperwork found, the next die located, all with the machine dark. Convert those and you cut 30 to 50 percent in weeks for under $1,000. The second pass streamlines what remains with quick-release clamps, standardized die heights, and preset tooling, usually another 20 to 30 percent for $5,000 to $30,000 per machine.

Benchmarks by process: stamping die changes go from 2 to 4 hours untreated to under 20 minutes with full SMED treatment, which is where the single-minute name came from. Injection mold swaps run 60 to 180 minutes typically, 15 to 30 with magnetic clamping and preheated molds. CNC job shops commonly cut 90 minute setups to 25 to 40 minutes with offline presetting and standardized workholding. Packaging lines get 45 minute changeovers under 15. First-year reductions of 50 percent are normal, and plants that keep going reach 75 to 90 percent over 2 to 3 years. If your first filmed changeover does not embarrass anyone, you filmed the wrong machine.

Failure modes: the classic is banking the savings as coffee time; if the schedule still plans 3 hours for a changeover proven at 90 minutes, you bought nothing, so update planning parameters within a week. Second, cutting changeover and leaving batch sizes alone, which strands the inventory benefit; revisit batch policy within a month. Third, hero dependence: one setup wizard hits 45 minutes and the other three techs take 2 hours; the fix is a written setup standard with a checklist, not applause. Fourth, skipping the video because people feel watched; without footage the team debates memories instead of facts and the analysis takes 4 times as long.

Cadence: daily, post actual changeover times against standard on the machine board, every event, no exceptions; misses over 25 percent get a two-line cause note. Weekly, the SMED team reviews the worst three events and converts items off the external-work list; one workshop per month per focus machine keeps momentum. Monthly, report recovered hours and their dollar value to plant leadership, and confirm planning parameters and batch sizes reflect current proven times. Quarterly, pick the next machine by the same dollar ranking, and re-film a past success, because changeover times creep back 10 to 20 percent a year when nobody is looking.

World-class changeover culture: every changeover has a published standard time, techs work from staged carts with shadow boards, external prep starts 30 minutes before the machine stops, and actual times sit within 10 percent of standard week after week. Numbers to aim at: setup under 5 percent of machine time, top movers changing over in under 10 minutes, and batch sizes cut 50 percent plus from where the program started. The strategic payoff is flexibility: a plant that changes over in 15 minutes can run to actual demand instead of forecast, which shows up as 20 to 40 percent less finished inventory and lead times measured in days instead of weeks.

Published 2026-07-02.