The plant runs around the clock five days a week, and most Saturdays as well. The only maintenance window is the one you take from production, and the weekend premium is already in the standard cost. There are no more hours to add. And there is a quote on the desk for a new line: seven figures, nine to fourteen months from purchase order to validated output, plus the people to run it.
The question nobody has answered with data is a simpler one. Are the lines you already own producing what they could?
Most plants’ honest answer is “we think we’re at about 70% OEE.” That number comes from shift reports, a target speed somebody set years ago and a downtime log that captures the stops people remember. It is not a measurement. It is a belief, and the capex decision is about to be built on it.
This article gives you a method to replace “we think” with a number: how much capacity is hidden in your existing lines, how much of it is recoverable, and what it is worth. It takes weeks, not quarters, and it should be finished before the capex request goes to the board.
Every production line has three capacities, and most capex cases confuse them.
Nameplate capacity is rated speed multiplied by available hours. It is what the line was sold as, and nobody has ever achieved it on a sustained basis.
Demonstrated capacity is what the line actually produced last quarter. It is the number finance plans on, the number the S&OP meeting uses, and the number the new-line business case compares against demand.
Effective capacity is what the line could produce at a realistic, sustained OEE with today’s equipment, today’s crews and today’s product mix. Not a stretch target. The output you would get if the recurring, avoidable losses were removed.
Hidden capacity is the gap between demonstrated and effective. When a capex case compares demand to demonstrated capacity, it silently assumes that gap is zero. On a line running below 65% OEE, it almost never is. In our experience across FMCG, food and component plants, the gap is typically 10 to 20 points of OEE, and the first half of it needs no capital at all.
Hidden capacity is not one thing. It is six categories of loss, and each one is invisible in your reports for a different reason.
1. Unplanned stops. These are logged, so they feel accounted for. But a recorded stop is the stop as the operator remembered it at the end of the shift. When the same stops are timed by a sensor, the logged total is often 20 to 50% short in the plants we have measured, and the undercount is worst on the categories nobody likes to write down.
2. Micro-stops. Stops under two to five minutes: a jam, a misfeed, a sensor fault, an operator clearing a label. Almost never logged, because logging takes longer than the stop. On packing lines they are often one of the largest loss categories, and they do not appear in the monthly OEE report at all.
3. Speed loss. The line runs “fine” at 85% of rated speed because that is the setting a shift found stable in 2021 and nobody has touched since. Unless the system compares actual cycle time to the rated cycle time per SKU, this loss is baked into the baseline and treated as normal.
4. Changeovers. The standard says 25 minutes. The data says 41, and the spread is driven by which crew is on. This is the best news in the whole Pareto, because the 25-minute version already exists in your plant: one crew does it. Measure the changeover by stage (run-down, clean, format change, start-up, ramp to speed) rather than as one block, and one or two stages will carry the whole spread. Compare crews on those stages, make the fastest crew’s sequence the written standard, move every step that can be done while the line runs (staging format parts, tooling, materials) off the critical path, and treat equipment readiness as a maintenance task with a named owner. Then schedule to the demonstrated time, not the standard, and sequence orders to minimize format changes. The target is the spread collapsing first; the average follows.
5. Starvation and blocking. The filler is up but waiting on bulk. The packer is up but backed up by the palletizer. In most systems the wait is logged as the filler’s or the packer’s own stop, or left uncoded. The loss is real, it is often large on lines with a bulk-to-pack handoff, and it is attributed to the wrong machine or to nothing.
6. Quality and rework. Good count versus total count, measured on the machine rather than on the QC sheet. Rework that goes back through the line consumes capacity twice and is usually counted once.
None of these are visible in a monthly OEE average. They are visible per shift, per SKU and per stop. That is the resolution the method below works at.
Step 1. Pick one representative line. Not the worst line and not the newest. The one most like the line you are about to buy, running the product mix that is driving the capex request. One line is enough to establish the method and the ratios; you extrapolate later, with the assumptions stated.
Step 2. Establish a machine-verified baseline. Take the stop and count signals from the PLC or from a sensor on the line, not from the operator log. Run it for three to four weeks minimum, so every SKU and every crew appears in the data at least twice. State the OEE basis explicitly: planned production time, not calendar time. Without that, the number cannot be compared to anything, including your own target.
Step 3. Build the loss Pareto. Rank loss reasons by minutes lost until you have explained at least 80% of lost time. On most lines the top five reasons carry it. If the top reason is “other” or “unknown”, you have a data-coding problem to fix before you have a capacity number; do that first.
Step 4. Sort every loss into one of three buckets. This is the step most plants skip, and it is the one that makes the number credible to a CFO.
Be honest in this step. Some of what the Pareto shows is not loss but data hygiene: stops coded to the wrong machine, two names for the same asset, a target speed that was never updated after a format change. Clean it before you count it, and say in the report how much you excluded.
Step 5. Convert hours to units to money. The arithmetic is simple and should be shown in full:
Recoverable units per week = recoverable OEE points × planned hours × rated speed per hour
Annual value = recoverable units × contribution margin per unit × weeks run
Use contribution margin, not sales price, and show production value and achievable margin as two separate lines. The board will ask whether the recovered output can actually be sold; the two-line format answers that before the question is raised.
Take a packing line planned for 120 hours a week, rated at 200 units a minute, measured over four weeks at 58% OEE. Nameplate is 1.44 million units a week; demonstrated is 835,000. The 42 lost points break down like this.
| Loss reason | OEE points lost | Bucket |
|---|---|---|
| Changeovers | 11 | 5 discipline · 6 accepted at standard |
| Micro-stops | 9 | 3 targeted fixes · 6 accepted |
| Unplanned stops | 8 | 2 targeted fixes · 4 structural · 2 accepted |
| Speed loss vs rated | 7 | 1 discipline · 2 structural · 4 accepted |
| Starvation and blocking | 5 | 1 discipline · 4 accepted (upstream) |
| Quality and rework | 2 | 2 accepted |
Sorted: 12 points recoverable within six months (7 through discipline, 5 through targeted fixes), 6 points structural, and 24 points that a realistic plant would still carry at a sustained 70% OEE. Effective capacity for this line is therefore 70% OEE, not 58%.
The money:
Across six similar lines, that is roughly 1.04 million units a week recovered, which is the output of a seventh line running at 70% OEE. No capital, no 12-month lead time, no new crew. The structural 6 points remain, and they tell you precisely what the eventual equipment spend should target: the bottleneck machine, not a whole line.
This is not a theoretical ratio. One FMCG group running Fabrico across its plants increased output by 14%. The full story is in the Ficosota customer case.
Trusting self-reported downtime. It undercounts, and it undercounts the embarrassing categories most. A capacity case built on the shift log will be challenged by the first engineer who has stood on the line.
Counting every loss as recoverable. The 42 points in the example are not 42 points of opportunity. Claiming them sets a target the plant will miss, and the next capacity case loses credibility before it is read.
Measuring for one week. You will miss the bad SKU, the weak crew and the Monday start-up. Three to four weeks is the minimum; a full product cycle is better.
Averaging across SKUs. A point of OEE on a high-margin format is worth several times a point on a low-margin one. Value recovered capacity per product, then sum. The average hides exactly the decision you need to make.
Treating the number as the finish line. The measured hidden capacity is a hypothesis. The next 90 days of interventions either prove it or revise it. Report it that way, with the proof plan attached.
Once you have the recoverable number and the structural number, the capex question becomes a comparison rather than a judgment call.
| Situation | Decision | What to tell the board |
|---|---|---|
| Recoverable capacity covers the demand gap | Recover first | Defer the line 12 months; fund the discipline and targeted-fix program; re-measure at month 6 |
| Structural bucket is the bottleneck and recoverable does not close the gap | Build, but build narrow | Spec the investment against the measured bottleneck (one machine, one station), not a whole line |
| Demand gap exceeds recoverable plus structural | Both, sequenced | Recover now to buy time and fund part of the line; place the order with the loss profile written into the acceptance criteria |
The measurement also de-risks the build. A new line installed next to an unmeasured one inherits the same crews, the same changeover habits and the same coding gaps. Running the method first means the line you buy starts at the effective OEE, not the demonstrated one.
What is hidden capacity in manufacturing? Hidden capacity is the difference between what a production line actually produces (demonstrated capacity) and what it could produce at a realistic, sustained OEE with its existing equipment and crews (effective capacity). It is made up of recurring, avoidable losses: micro-stops, changeovers over standard, speed loss, unplanned stops, starvation and blocking, and rework.
How do you calculate effective capacity? Effective capacity = planned production hours × rated speed × achievable OEE. Achievable OEE is the measured OEE plus the loss points that are recoverable through operational discipline and targeted fixes, excluding structural losses that only equipment can remove.
How much capacity can typically be recovered without capital investment? In our experience, on lines running below 65% OEE, 10 to 20 points of hidden capacity is common, and the first 8 to 12 points usually need no capital. The figure depends on the line, the product mix and the quality of the baseline data, which is why it has to be measured rather than estimated.
How long does it take to measure hidden capacity? Three to four weeks of machine-verified data on one representative line is enough to build a credible loss Pareto and a quantified capacity case. A full product cycle is better if changeovers are a major loss.
What is the difference between OEE improvement and capacity recovery? OEE improvement is a percentage change in a ratio. Capacity recovery converts that change into units, hours and money, per SKU, and separates what is recoverable from what is structural. The board funds capacity, not ratios.
Run the method on one of your own lines first. If the recoverable number covers the gap, you have just deferred seven figures of capital and a year of lead time. If it does not, you now know exactly what to buy and where the new line will lose capacity if the same habits follow it.
The fastest way to run it is a fixed-scope pilot: one representative line, six weeks, an industrial sensor and hub installed with your team, and an executive readout with the machine-verified loss profile, the recoverable capacity quantified, the three highest-value interventions and the financial case in dollars. The fee is fixed and credited in full against a first-year subscription if you decide to roll out. The pilot runs on Fabrico’s manufacturing performance platform (MES, OEE, CMMS & AI), which connects machine data, OEE and loss analysis, production scheduling, SKU-level output value and maintenance in one system, so the capacity you recover is measured, planned and sustained rather than reported once.
Request a demo or start with the OEE calculator to get a first estimate of what your line could deliver.
Login
Blog
