Bottleneck Analyser

What this tool does

CM8-288 finds the constraint in a chain of process steps and sizes what fixing it is worth. You enter one row per step or resource — the rate it runs at, the hours it is available, how much of that time it really runs, and how much of what it makes comes out good the first time. The tool works out the effective capacity of every step, names the one with the least, shows how much of each step's capacity demand is already consuming, and puts a figure on the units the chain is failing to produce.

The bottleneck is simply the step with the lowest effective capacity. Everything upstream of it can only usefully produce what it can take; everything downstream can only receive what it lets through. The output of the whole chain is the output of that one step, and no amount of improvement anywhere else changes that number by a single unit.

Everything runs inside this single file. There is no account, no upload and no network request of any kind, so your rates, downtime, scrap and margins never leave the computer you are using.

Effective capacity, not nameplate

This is the point of the tool, so it comes first. The nameplate rate of a step tells you almost nothing about what it can produce. A machine sold as 40 units an hour, scheduled for 40 hours a week, does not make 1,600 units a week. It breaks down. It gets changed over. It waits for an operator, a fixture, a crane, a decision. And some of what it does make comes back as scrap or rework, consuming capacity twice. What is left after all of that is the effective capacity, and it is the only capacity figure worth planning with.

Three multipliers, each of which can quietly halve a step. In the sample chain the turning cell has a nameplate of 36 units an hour over 40 hours — 1,440 a week — but 82% uptime and 96% first-pass yield reduce that to about 1,134. The saw next to it looks similar on a specification sheet and produces three times as much. If you planned that chain from nameplate rates you would promise a customer 1,440 units and be short every single week, with no idea why.

The second chart splits every step's nameplate capacity into three: what it actually delivers, what downtime takes, and what quality takes. Two steps can end up with the same effective capacity for completely different reasons, and the reason determines the fix. A slow step needs a faster process or more hours. An unreliable step needs maintenance, setup reduction, or whatever it is waiting for. A wasteful step needs a quality fix — and a quality fix at the constraint is worth far more than the same fix anywhere else, because every scrapped unit there was made with time the chain can never get back.

An hour lost at the bottleneck

If you take one idea from this tool, take this one. An hour lost at the bottleneck is an hour lost for the entire chain. An hour lost at a non-bottleneck costs nothing at all.

The constraint sets the output. If it stops for an hour, the chain produces an hour less, permanently — there is no way to make it up, because the constraint was already running flat out before and after. If a step with 60% utilisation stops for an hour, it simply uses some of the spare capacity it has been carrying all along, and the chain produces exactly the same as it would have done.

This changes what you do on the floor. The constraint gets the best operator, the first maintenance slot, the spare tooling on the shelf, cover through breaks and shift changes, and material staged in front of it so it never waits. Meanwhile, a non-constraint standing idle is not a problem to be solved. It is spare capacity doing its job — absorbing the day-to-day variation that would otherwise reach the constraint and stop it. Managers who chase utilisation on every machine convert that protection into stock and make the chain slower, not faster.

The five focusing steps

The theory of constraints reduces to five steps, and they are worth stating plainly:

1. Identify the constraint. Find the step with the least effective capacity — the red bar on the capacity profile. Not the one people complain about, not the noisiest, not the most expensive: the one with the smallest number.
2. Exploit it. Get more out of it without spending money. Stop it breaking, stop it waiting, stop it running scrap, stop it doing work that could be done elsewhere.
3. Subordinate everything else to it. Run the rest of the chain at the constraint's pace rather than at its own maximum. Schedule to it, buffer in front of it, and accept idle time elsewhere.
4. Elevate it. If demand still exceeds capacity after all that, buy more — a second machine, another shift, a subcontractor, an outsourced operation.
5. Repeat. Go back to step one, because the constraint has moved.

The last step is the one people skip, and it is why the tool counts the steps above 90% utilisation. In the sample chain, lifting turning by 12% does not give you 12% more output — it gives you a few per cent before milling, sitting at about 93% utilisation, takes over as the constraint. Knowing that in advance stops you promising a customer a volume that a second, hidden constraint will not support, and it tells you what to work on next. The "Constraint and next constraints" table is that queue: every step ranked by effective capacity, least first, in the order they will become the problem.

Exploit before you elevate

The instinct on finding a bottleneck is to buy another one. It is almost always the wrong first move, and it is certainly the expensive one. Before spending anything, work through the cheap questions:

• Stop breaking it. What does the uptime figure actually consist of? Breakdowns, changeovers and waiting are three different problems with three different fixes.
• Stop starving it. Does it ever stand idle waiting for material, an operator, a programme, an inspection result or a decision? A small buffer in front of the constraint is one of the few stocks worth holding.
• Stop putting scrap through it. Inspect before the constraint, not after. Time spent machining a unit that was already defective is capacity destroyed for the entire chain.
• Stop giving it work it does not need to do. Can any operation move to a step with spare capacity? Can a variant be subcontracted? Can the constraint run through breaks and shift changes?

Exploitation routinely finds 10–20% on a constraint for the cost of some attention, which is why it comes before capital in the sequence.

Why balancing every step to equal capacity fails

A tidy-looking chain where every step has the same capacity is a fragile one. Real processes vary: a breakdown here, a slow batch there, an absence, a bad delivery. In a perfectly balanced chain there is nowhere for that variation to be absorbed, so every disruption anywhere propagates straight to the output. Worse, the constraint jumps around week to week, and nothing can be protected because everything is the constraint some of the time.

Deliberate imbalance is the better design: one known constraint, protected, with genuine spare capacity either side of it. That spare capacity is not waste — it is what stops normal variation from reaching the step that matters. Aim to be well-managed rather than fully utilised, and use the utilisation chart to check that the non-constraints really do have room, rather than sitting at 97% and pretending to be safe.

Labour constraints behave differently

A machine constraint is fixed until you buy something. A labour or skill constraint is not: capacity moves with overtime, an extra shift, cross-training or a temporary hire, often within a week and usually without capital. That flexibility is real, but it comes with its own limits — fatigue, competence, the number of people who genuinely hold the skill, and the fact that a second shift needs a supervisor and a maintenance window as well as an operator. Record labour steps with the hours of everyone doing the work added together, and read the capacity-per-operator column as the honest measure of what one more person would add. If your constraint is a skill rather than a machine, the first question is usually not "who else can we hire" but "who else can we train, and what is stopping them being trained".

What the gap is worth — honestly

Enter a contribution per unit — selling price less the truly variable cost of one unit — and the tool multiplies the weekly shortfall at the constraint by that figure and by your working weeks per year. In the sample chain that is about 46 units a week the chain cannot make, at 18.50 contribution, over 48 weeks.

Read the number with care. It is only realisable if you can actually sell the extra units. If demand was optimistic, or the order book is seasonal, or the customer will not take more than they already order, then closing the capacity gap produces stock rather than income and the figure is fiction. Contribution is also not profit: fixed costs are not in it, so do not present it as a bottom-line improvement. Use it for what it is good for — a defensible order-of-magnitude answer to "what is this bottleneck costing us", which is usually enough to justify a week of setup-reduction work and often enough to justify a machine.

Note that the tool takes one contribution figure for the whole chain rather than adding one per step, because the units are the same units flowing through every step. Enter it identically on every row, or leave it blank and the tile will say so.

The spreadsheet workflow

If the rates, hours and downtime already live in a spreadsheet, you do not have to retype them.

• Spreadsheet template in the toolbar saves a CSV whose headings are exactly this tool's column names — value stream, step order, process step or resource, resource type, units per hour at full speed, available hours per week, uptime percent, quality yield percent, operators required, demand per week — with a guidance row underneath showing what each column expects, including the accepted values for the resource-type list and the unit for each number.
• Open it in any spreadsheet, fill in one row per process step, and delete the guidance row before saving. Keep the file as CSV.
• Import spreadsheet reads it back. Columns are matched by heading, so their order does not matter and extra columns of your own are ignored; numbers and list values are converted for you. Rows missing a required column, or failing a validation check, are skipped and reported by row number.

The file is read by this page in your browser: nothing is uploaded and the original spreadsheet is not changed. Importing adds to what is already here rather than replacing it.

FAQ

Where do I get the uptime figure? From a downtime record, not from memory. If you already measure overall equipment effectiveness, use its availability component and put the performance losses into the rate. If you have nothing, time the step for a week — the exercise itself is usually more revealing than the number.

Should the rate include the operator's breaks? No. The rate is what the step does while it is running. Breaks and everything else that stops it belong in the uptime percentage, which keeps the two causes separate and visible on the loss chart.

What if demand differs along the chain? Enter the real figure for each step. If a step scraps 5%, the steps before it must produce more than the steps after it, and utilisation should be measured against what each one actually has to make. The tiles use the highest demand in the chain so the constraint is judged against the hardest requirement it faces.

Is the bottleneck always a machine? No. It is often a person, a skill, an inspection, a subcontractor, a decision or a supplier. The resource-type list exists so those show up honestly alongside the plant.

How does this relate to mapping and balancing? The Value Stream Map tool answers "where does the time go" and the Line Balancing tool answers "how do I share the work evenly between stations"; this one answers "which step limits output, and what is that costing". They are complementary — the map finds the queues, and a queue that never clears is usually sitting in front of the constraint this tool names.

Can one file hold more than one chain? Yes. Every step carries a value stream name, and steps sharing a name form one chain. The tiles, charts and tables report the chain with the most steps, since capacities from different product lines cannot sensibly be combined.

Saving your work

Steps, settings and the report header are written to this browser's local storage as you type, and the toolbar shows the time of the last save. That storage belongs to one browser on one computer: another browser, a private window, a second machine or a clean-up tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you the register for spreadsheet work. Reset asks twice, then erases everything this tool has stored. There is no undo.

Accuracy & disclaimer

The arithmetic here is deliberately simple — three multiplications for capacity and one division for utilisation — and the tool does it faithfully. Everything that matters sits underneath: whether the rate was measured or remembered, whether the uptime figure comes from a record or from optimism, and whether the demand is an order book or a hope.

Every capacity here is built from averages, and averages hide variability. A step with 15% spare capacity on paper can still starve the constraint on a bad Tuesday, because what protects a bottleneck is not average spare capacity but spare capacity available at the moment it is needed. Use this analysis to decide where to look and what to protect, verify on the floor, and re-run it after every change — the constraint moves, and a bottleneck analysis that has not been refreshed since the last improvement is describing a chain that no longer exists.