Production Capacity Planner

What this tool does

CM8-03 is a weekly rough-cut capacity planner for manufacturing. You describe each work centre — a machine or group of identical machines with a shift pattern and realistic losses — and the demand placed on it: products, weekly quantities and the run minutes per unit on each centre they visit. The tool works out the load on every centre, finds the bottleneck, shows how much of the demand mix actually fits, counts the hours you are short, and lets you test fixes — an extra shift, another machine, better OEE or a change in demand — without touching the plan itself.

It is the capacity check that sits between a sales and operations plan and detailed scheduling: does next week's (or next month's) demand fit the plant, and if not, where and by how much? Everything runs inside this single file with no account and no network request, so routings and demand figures never leave this computer.

One row per routing operation

Each row is one operation: a product on a work centre. A product that is milled, welded and assembled has three rows. Each row carries the weekly demand, the run time per unit and the setup, plus the shift pattern of the work centre it loads.

Because several rows share a centre, the centre's capacity is taken from the first row entered for that centre (matching the name, ignoring capitals and spaces). Later rows for the same centre are checked against it: the Cap. check column shows Matches or Differs — first row used, and the form warns you as you type. To change a centre's shift pattern, edit its first row. A mismatch never blocks you — it is flagged so that one mistyped row cannot quietly change a centre's capacity.

The capacity model

All figures are per week. For each work centre:

For each operation, and then summed over every operation on the centre:

An operation with no demand or no run time adds no load.

In the sample plan the mill also carries Shaft B (9.0 h) and Frame C (13.8 h), so its total is 66.8 h against 62.56 h — a load of 106.8%. It is the bottleneck.

Planned downtime and OEE

Two separate deductions keep the model honest:

• Planned downtime % — scheduled non-production time: preventive maintenance, cleaning, breaks that stop the machine, planned changeover windows. Typical planning values are 2–10%.
• OEE / efficiency % — how much of the remaining time produces good parts at rated speed. It bundles unplanned stops, slow running and scrap. About 85% is often quoted as world class for discrete manufacturing; many plants run 55–75%. If you do not measure OEE, start with a conservative 70–80%.

Do not double-count: if your OEE figure already includes planned maintenance, set planned downtime to 0. The two multiply together.

Routings and setup

• Run time per unit is the standard cycle time of one unit on that centre. For a multi-machine centre it is the time on one machine — the machine count is already in the capacity.
• Setup allowance % spreads changeover over the run. A 60-minute setup before a batch of 100 parts at 6 min/part (600 min of running) is 10%. Small batches push this up fast.
• Setup time per batch with a batch size is the explicit alternative: one setup per batch, rounded up. Use one method per row, not both, or changeover is counted twice.
• The product table shows each product's standard hours across all centres — useful for sanity-checking routings against your ERP standards.

Reading the results

• OK — load at or below the target maximum load (85% by default). Normally plannable; the slack absorbs breakdowns and rush orders.
• Near limit — above the target but not over 100%. Queues grow quickly at high utilisation; expect longer lead times and little tolerance for disruption.
• Bottleneck — over 100%. The demand does not fit. Something gives: overtime, subcontracting, late orders or lost sales.

The bottleneck tile names the centre with the highest load. The plant's output is set there: an hour lost on the bottleneck is an hour lost for the whole plant, while an hour saved on a centre with spare time saves nothing. Hours short is the sum, over every overloaded centre, of required minus available hours. The filters let you look at firm orders only, or one type of centre; load is then calculated from the filtered demand against each centre's full capacity.

Feasible mix

Feasible mix answers "how much of this week's demand actually fits?" It scales the whole product mix down in proportion until the most loaded centre reaches exactly 100%:

A feasible mix of 93.7% means you can ship 93.7% of every product's demand; the product table shows the resulting quantities. Over 100% means the whole mix fits with headroom. Changing the mix — favouring products with less bottleneck time — can beat proportional scaling; that is Theory of Constraints territory.

What-if scenarios

Two kinds of what-if never change your plan:

• Demand level in Settings scales every row's demand — 120% tests an upturn.
• Δ machines, Δ shifts per day and Δ OEE points on a centre's first row overlay its capacity. Machines and shifts cannot fall below zero and OEE is held between 0 and 100%.

When any centre has a what-if, the tiles, charts and verdict use the what-if capacity, and the work-centre table shows baseline and what-if side by side. Clear the Δ fields to return to baseline.

Closing a capacity gap

The Executive report turns the shortfall into actions, using the centre's own figures:

Overtime is stated in clock hours on the machine, because each overtime hour loses the same downtime and OEE as a normal hour. Moving volume names the product taking the most bottleneck time and, where one exists, a centre of the same type with spare time below the target — check the routing really allows it. The planning horizon setting multiplies overtime and shortfall over several identical weeks.

Accuracy and limits

This is rough-cut capacity planning. Deliberate simplifications:

• Weekly buckets, infinite queues: hours are compared as weekly totals; sequencing and queue timing within the week are not modelled.
• Identical machines: all machines in a centre are interchangeable and share one shift pattern. Model a faster or slower machine as its own centre.
• Fixed standards: demand and cycle times are single numbers. Above about 85% load, variability makes reality markedly worse than the average suggests.
• Labour is not separate: if operators rather than machines are the constraint, model the crew as the work centre.

Use the result to make capacity decisions — shifts, machines, subcontracting, demand levelling — then confirm with detailed scheduling before promising dates.

Saving your work

Operations, settings and the report header are kept in this browser's local storage as you type. Export .json is the real save: one file that Import .json restores anywhere. Export CSV gives the register for spreadsheets, and the spreadsheet template lets you import a routing file. The Executive report prints the verdict, tiles, charts, tables and your notes. Reset asks twice, then erases everything this tool has stored.