Capacity Planning: Key Checks Before Release

Capacity planning is the work of matching the work available to the work the shop can actually complete, and its central idea is that the shop is limited by one step rather than by its average. That step sets the throughput of the whole plant, and everything that happens at the other steps only changes how much work is waiting in front of it. A shop that plans from its average capacity will promise more than it can deliver during the months when the mix is unfavourable, and the failure will look like an execution problem when it is really an arithmetic one.

The Bottleneck Sets the Rate

The bottleneck is the step whose capacity is fully used while the other steps have spare time. It might be the drilling area, the lamination press, the plating line or the test area depending on the product mix, and it can change from month to month. The flow rate of the plant is the flow rate of that step, and an improvement anywhere else produces extra work in progress rather than extra output.

Identifying the bottleneck should be done from data rather than from opinion. The simplest method is to compare the load placed on each step by the current order book with the hours available at that step, taking account of the yield and the rework. The step whose ratio is highest is the constraint, and the plan should be built around it until the mix or the capacity changes.

Load Levelling

Load levelling is the practice of moving work in time so that the bottleneck is neither starved nor overloaded. Starving it wastes the most expensive resource in the plant, while overloading it creates a queue that grows faster than the extra input justifies. The tool is the release of orders into the flow, not the pushing of more work into the front of it, since work released early simply waits somewhere further down.

The release decision should follow the bottleneck, so that a job is released when the constraint will be ready for it and not before. This is uncomfortable in a shop that measures activity by how busy each step looks, because levelling deliberately leaves some steps with spare time. The measure that matters is the throughput of the constraint, and a shop that keeps every step equally busy will always have more work in progress than it needs.

Utilisation and Its Limits

Utilisation is the share of available time that a resource is producing, and it is a useful measure and a dangerous target. A step loaded to one hundred percent has no capacity to absorb any variation, so the queue in front of it grows without limit whenever something goes slightly wrong. In practice, a plant that runs its constraint at eighty to eighty five percent of its theoretical capacity will usually deliver more than one that tries to run it at a hundred.

The relationship between utilisation and lead time is not linear. Small increases in load at high utilisation produce very large increases in queue, which is why a shop can be only slightly busier and much slower at the same time. This is the arithmetic that most delivery failures hide behind, and it is worth showing once to the people who set the delivery promises.

Measuring Real Capacity

Real capacity is a measured figure rather than a nominal one. It has to account for the shift pattern, the planned maintenance, the changeover time, the yield and the rework, and the time lost to the ordinary disruptions of a working shop. A nominal figure from the equipment specification is a starting point, and the difference between it and the measured figure is usually twenty to thirty percent, which is exactly the gap that makes a plan fail.

The measurement should be repeated periodically, because capacity changes with the mix, the tooling and the workforce. A press that can produce a certain number of panels a shift at one panel size will produce a different number at another, and a plan that assumes one figure for all products will be wrong in both directions. Recording the actual hours and the actual output by step, with the fabrication notes for the batches, gives the data without any additional measurement effort.

Planning the Order Book

Planning the order book means comparing the load it represents against the capacity of the constraint, week by week, and deciding what to do with any excess. The options are to extend the delivery date, to subcontract a portion, to add a shift, or to decline the work. Each has a cost and a consequence, and the decision should be made when the order is quoted rather than when it is already in the shop.

A load profile that is visible for the next few weeks is enough for most shops. The profile should be expressed in hours at the constraint rather than in orders or in money, because hours are what the plan consumes, and it should be updated weekly from the confirmed order book rather than from the enquiry pipeline. Mixing confirmed orders and hopes in the same chart destroys the value of the chart.

Throughput and Improvement

Throughput improvement should be aimed at the constraint, and only at the constraint, until it moves. Adding capacity upstream of the bottleneck produces more waiting panels, and adding capacity downstream produces idle equipment, and both add cost without output. Once the constraint has been relieved, the constraint moves to another step and the analysis has to be repeated, which is why capacity work is continuous rather than a project.

The cheapest throughput gains are usually in the constraint’s own losses. Changeover time, unplanned stops, small stoppages and quality losses at the bottleneck are all recoverable hours, and they cost attention rather than capital. A process flow that records stops and changeovers at the constraint will usually find several hours a week that no equipment purchase could have recovered as cheaply.

Practical Rules

Find the constraint from data, plan the release of work around it, and keep its load below the point where queues become unstable. Measure real capacity rather than nominal capacity, express the order book in hours at the constraint, and recheck the measurement after any change in mix, tooling or shift pattern, because a capacity figure that is a year old is a guess with a number attached.

Improve the constraint before anything else, and repeat the analysis when the constraint moves. Review the load profile weekly with the quality and delivery data beside it, since a plan that ignores scrap and rework will overstate the capacity it thinks it has and the shortfall will appear as a missed delivery rather than as a planning error.

FAQ

What is a bottleneck? The step whose capacity is fully used while other steps have spare time. It sets the output rate of the whole plant.

Should utilisation be maximised? No. Very high utilisation leaves no room for variation and makes queues unstable. Eighty to eighty five percent at the constraint is often more productive.

How is capacity measured? From actual output and hours at each step, adjusted for mix, maintenance, changeovers and yield, rather than from equipment specifications.