Factory Capability: Equipment and Lines
An equipment list records what a plant owns. It says nothing about what leaves the dock each month. Output is five numbers multiplied together: machines, cavities, cycle time, OEE, paid hours. A quote missing any one of them is a guess dressed as a figure. What follows is the arithmetic used to schedule real orders, with the losses left in instead of netted out: changeover, trial runs, tool repair, waiting on material. Every figure is a working range taken from ordinary production weeks, not a guaranteed ceiling. Read it as a way to question a quote, not to accept one.
01Five Inputs, One Equation
Start with the cycle: one full shot, clamp open to clamp open, counting cooling and take-out. Case tools usually land between 60 and 120 seconds. Cavities are counted per part — one for a shell, two to eight for fittings. Divide 3,600 by the cycle for shots per hour, then multiply by cavities for parts. Teams that skip the cavity step report shots as parts and inflate capacity several times over.
OEE is three factors, not one. Availability times performance times quality: 95% × 88% × 97% lands near 81%, while quoting availability alone gives a figure roughly a fifth too high. The 65–80% band exists because changeover, trials, tool repair and waiting for material never reach zero. A plan built on 95% OEE is usually a plan that moved its downtime into someone else's shift.
02Changeover Eats Two Hours of Every Two Shifts
Two shifts give 20 to 22 paid hours once handover, checks, meals and cleaning come out. A changeover runs 40 to 90 minutes: pulling the tool, lifting the next one, coupling water, dialling parameters. First-article approval on a new tool adds 30 to 60 minutes more. Two changeovers a day remove 2 to 3 hours, which is a tenth to a sixth of the available time.
Cost it in shots instead of minutes. At 60 minutes per changeover and 40 shots an hour, one change costs 40 shots; twenty a month costs 800, close to a tenth of a single machine's output. The fix is not a faster operator but work moved outside the press: pre-heated tools, quick couplings, a staged parameter card. Internal time can fall under 30 minutes, and that saving repeats every time.
03Finding the Bottleneck Without a Stopwatch
Load every station: demand divided by available capacity. The highest ratio is the constraint, and a healthy line runs it at 85% to 95%. Anything under 70% means the schedule is uneven and people are waiting. Line output equals the bottleneck's output. Whatever piles up behind it is not progress, it is stock queuing for a slot that does not exist.
A walk through the hall beats a spreadsheet. Find the stack that is both tallest and oldest: queue in front starves the downstream stations, a pile behind means blockage, and the two need opposite responses. Once identified, only two moves exist — add capacity there, or take part of the work off that station. Buying kit for a non-bottleneck changes nothing except work-in-process.
04Outsourcing 10–30%: What Leaves the Gate
Outsourced content typically runs 10% to 30%. Under 10% suggests slack; over 30% and schedule plus consistency both get hard to hold. Three categories travel well: surface work such as painting, printing and plating; standard hardware; odd parts too small to justify a machine slot. Moulds, shell forming and final assembly stay inside, because those three set fit, and a return trip doubles rework cost.
Price the trip into the promise. A round of outsourcing costs 3 to 7 days, and incoming inspection tightens from a routine 5% to 20%, which is normal rather than paranoid. Forty to 120 active tools is a common book of work, and the more tools there are, the more scheduling resembles an equation. One ratio shows whether it has gone too far: defects traced to outside parts as a share of all defects. Past 15%, bring work back.
05Where Expansion Stops Paying
Another press does not add output in a straight line. Three ceilings come first: power, tooling, downstream capacity. A mid-tonnage machine with its auxiliaries draws 60 to 120 kW, and a service upgrade is measured in months. Without enough tools the machine sits idle; without trimming, assembly and packing, finished parts simply block the aisle and count as nothing.
Monthly output swings about ±15% on overtime alone, and beyond that you need outsourcing or stock built ahead. The trigger for capital is specific: the bottleneck above 92% load for two straight weeks while outsourcing already sits at 30%. Below that, a purchase mostly converts cash into idle tonnage. At 80% bottleneck load, a third shift on the constraint is the cheaper move by far.
| Output formula | machines × cavities × (3,600 ÷ cycle) × OEE × hours |
|---|---|
| Cycle per shot | 60–120 s |
| OEE band | 65–80% |
| Paid hours, two shifts | 20–22 h |
| Bottleneck load | 85–95% |
| Active tool count | 40–120 tools |
| Outsourced share | 10–30% |
| Monthly output swing | ±15% |
FAQ
Why not estimate capacity from press tonnage?
Tonnage says what tool will fit, not how many parts come out. On one machine, a 30 s cycle difference and four extra cavities can multiply monthly output. You need machines, cavities, cycle, OEE and shift hours together; leave one out and the result is a random point inside a range.
How much does OEE 65% versus 80% matter?
That single factor is a 23% difference. On a machine running 10,000 shots a month, the two ends are more than 2,000 shots apart. Low OEE usually means frequent changeovers, trials and waiting; whether you sit at 65% or 80% follows order mix and tool condition, not effort alone.
Is a high outsourcing share a bad sign?
Not on its own. Ten to thirty percent is normal; past 30% lead time and consistency get hard to control. What matters is which steps leave: surface work is replaceable, shell forming is not. Send the second one out and a single rework usually costs more than the machine hours saved.
Does expansion always mean buying machines?
Check the constraint first. Capital makes sense when the bottleneck holds above 92% for two weeks and outsourcing is already at 30%. At 80%, a new machine only feeds inventory. An extra shift on the constraint, or splitting a step off it, usually lands faster than a service upgrade.