Capacity & Bottlenecks

Measure what each step can do against what is asked of it, find the step that actually limits output, and choose the cheapest way to get more through it. You leave with one named constraint and one dated action, not a list of general improvements.

Theory of Constraints has a large applied literature but its status as a theory is disputed: mixed research. Little's law is a mathematical result for stable systems, not an empirical finding. The utilization bands on this page are a rule of thumb. The numbers in the example are illustrative.

Use it when:

• Work queues up at some point and you do not know whether to add people, tools or hours, or where.
• Output is stuck although every team says it is busy.
• You must decide where to spend an improvement budget, in a workshop, an office process or a small business.

Skip it when:

• You do not yet know the steps or who does them: first draw 8.1 Process Map (Current State).
• The issue is quality or a recurring defect, not volume: use 8.7 Root Cause & Waste Notes.

Theory of Constraints claims that at any moment one constraint sets the throughput of the whole system, so improving anywhere else is largely wasted until the constraint is found and either exploited or elevated (Goldratt & Cox, 1984; Goldratt, 1990). The evidence is mixed: many favorable applied cases, but the framework's standing as a theory is questioned. One part stands on firmer ground: the link between work in progress, time in the system and throughput is a theorem for stable systems (Little, 1961), which is why pushing more work into a loaded system lengthens queues instead of raising output.

So this page finds the constraint from data and physical evidence (where work piles up), not from complaints, because the loudest step is often not the binding one. It exploits before it elevates, since recovering lost capacity usually costs less than buying more. The limit: the constraint moves with the mix of work, so treat the result as a measurement to repeat, not a permanent answer.

Use observed output, not job descriptions. Use the same unit for every step (items per hour or per day). Utilization is demand divided by capacity.

Strong answer: "Finishing booth: capacity 16 items per day, demand 18 per day, utilization 112 percent. Cutting: capacity 45, demand 18, utilization 40 percent." (illustrative)

Process (see 8.1):

Current output per day or week, and the target:

Rule of thumb for reading utilization (adjust to how variable your work is): under about 70 percent means spare capacity, 70 to 90 means adequate, over 90 means at risk.

Illustrative example: the workshop and all numbers are made up.

A small furniture workshop in Finland ships late. Every team says it is overloaded. The owner measures five steps: cutting, sanding, assembly, finishing and packing.

Section 1: the finishing booth manages 16 items a day against demand of 18; every other step has spare capacity.

Section 2: finished-but-unpainted frames stack up in front of the booth, and packing is idle in the afternoon.

Section 3: exploit by running the booth through lunch with staggered breaks and batching same-color jobs; subordinate by releasing orders at the booth's pace. Elevate (a second booth) is postponed.

Section 4: the booth gets 6 productive hours of 8. The lost two are changeovers, waiting for sanded frames and breaks.

Section 5: WIP is cut from 60 to 32 items. Throughput stays near 16 a day but the cycle time falls from almost four days to two. Next likely constraint: sanding.