What is throughput?

Throughput is the amount a process completes per unit of time, such as units produced per hour or patients treated per day. It measures what is actually delivered, as opposed to what could be delivered in theory. Raising throughput is often the aim of process improvement, but it is limited by the slowest step.

How it works

In a process with several steps in sequence, throughput is set by the slowest step, the bottleneck. Speeding up any other step only builds up work in front of the slow one. Throughput also relates to the other flow measures through Little's law: work in progress equals throughput times the time an item spends in the system, so for the same work in progress, higher throughput means shorter times.

Theoretical capacity and throughput are different. Breakdowns, waiting for inputs and variation all pull real throughput below capacity. Pushing utilization to the limit also makes queues grow quickly, so a process is often kept below its full capacity to keep waiting times acceptable.

Example

Say a bakery line has three steps: mixing at 120 loaves an hour, baking at 80 and packing at 100. The line's throughput is 80 loaves an hour, set by the oven. Adding a second packing station raises packing to 200 but changes nothing, since the oven still limits the result. Adding a second oven raises baking to 160, and then mixing at 120 becomes the new limit, so throughput rises to 120. The numbers are an illustration.

How to model it in Machinations

Chain Pools or Queues for each step, connected by Converters whose rate is the step's capacity. The final Drain or Pool collects finished goods, and a Register can compute completed units per step of time. Make each step's capacity a Lever. Play and chart the completions to see how throughput follows the slowest step.