Bottleneck

A bottleneck is the resource or process step in a production/service system whose capacity is most constrained relative to demand and therefore determines total system throughput. Operationally it is diagnosed by two signals: growing queue length immediately upstream and a utilization rate exceeding ~85%. Little's Law (Little 1961) L = λW formalizes how WIP and cycle time grow ahead of the bottleneck. Goldratt's Theory of Constraints (Goldratt 1984) places the bottleneck at the center of throughput management — system output equals bottleneck rate, and any improvement to a non-bottleneck station only inflates WIP, not throughput. Hopp and Spearman (1996) in 'Factory Physics' formalize the relationship between bottleneck rate, critical WIP, and raw process time. In practice the bottleneck is not static: changes in product mix or demand pattern can shift the bottleneck to another station, which is why the TOC focusing cycle re-identifies it after every improvement. From an operations-research perspective the bottleneck corresponds to the binding constraint with the highest shadow price in an LP model; #001 job-shop scheduling makespan optimization directly targets balancing load on the bottleneck machine.