Process flow, bottlenecks, capacity and throughput

Reciprocals answer different questions

In our original deterministic model, preparation takes 2 min/kit, assembly 3 and checking 1. Each has its own dedicated resource. Sixty available minutes divided by activity time gives nominal capacities 30, 20 and 60 kits/hour. These are rates under stated conditions, not the number guaranteed in a particular first hour. Startup, downtime, setups and transfer rules are omitted.

A single kit beginning in an empty line needs at least 2+3+1=6 minutes for those sequential activities, before any waiting. That six-minute flow time differs from the three-minute spacing that the slowest activity can support between successive completions after a suitable pipeline is established. Adding activity times and taking a bottleneck rate answer different questions.

A series route cannot outrun its required activities

With every kit requiring every activity, no rework, adequate buffers and separate resources, the steady serial rate cannot exceed the smallest capacity: 20 kits/hour at assembly. Demand may be lower, and actual throughput can be lower because of missing inputs or other losses. The bottleneck calculation is an upper-bound model under these assumptions.

Reduce assembly time to 1.5 min/kit. Its nominal capacity becomes 40/hour, but preparation remains 30/hour. The serial bound moves to 30/hour. The change can help, yet it does not make the whole route capable of 40. If one person performs several activities, combine that person’s required work per kit before calculating their resource capacity.

Measure use without making busyness the objective

At a steady demand of 12 kits/hour and no losses, preparation needs 24 minutes/hour, assembly 36 and checking 12. Their nominal time utilisations are 40%, 60% and 20%. An idle checker is compatible with the route meeting demand. Producing excess unchecked inventory solely to keep everyone busy can increase waiting without providing more completed service.

Real capacity analysis preserves product mix, quality, setups, maintenance, breaks and shared tools. A resource may be starved for input or blocked by a full downstream buffer. A different mix can move the constraint because products use resources differently. Measure completed acceptable output and the conditions behind the rate rather than treating the fastest observed cycle as a sustainable whole-system capacity.

Check a proposed improvement against demand

If demand is 24 kits/hour, the original 20/hour upper bound is insufficient under the model. Faster checking alone leaves assembly at 20. Faster assembly raises the bound to 30, so the simplified rate test no longer rules out meeting 24.

That result is a reason to investigate the proposal, not a forecast of 24 successful deliveries every hour. Check startup, variability, supply, quality and human work. If a single resource instead performs all three original activities, its 6 min/kit workload limits it to 10/hour regardless of the separate box labels.

Try the reasoning

1. What becomes limiting after assembly drops to 1.5 minutes?

   Show the answer and reasoning

   Preparation at 2 minutes per kit, or 30 kits/hour, under the dedicated-resource assumptions.

2. Does six minutes of processing imply ten kits/hour for three separate resources?

   Show the answer and reasoning

   No. Six minutes is one kit’s sequential processing time. Separate resources can work on different kits simultaneously.

3. Why can actual throughput be lower than the nominal bottleneck capacity?

   Show the answer and reasoning

   Demand, availability, variability, starvation, blocking, setup, rework and other omitted constraints can reduce completed output.

What to carry forward

Find the limiting resource under explicit assumptions, and keep flow time, capacity, utilisation and throughput distinct.