Kingman’s Queue Approximation, explained.
Kingman’s approximation relates mean queueing delay in a stable single-server system to utilisation, mean service time and variability of arrivals and service.
Why it happens
Mean service time is two minutes. Approximate waiting is ((ca²+cs²)/2)×ρ/(1−ρ)×2; arrival rate is ρ/2, allowing an approximate queue length through Little’s relation.
Kingman’s approximation relates mean queueing delay in a stable single-server system to utilisation, mean service time and variability of arrivals and service.
Read the result
Read all outputs as approximations from one stationary single-server model. A higher utilisation denominator amplifies the effect of supplied variability.
A worked example
Keep utilisation while changing variability
At utilisation seventy percent and both coefficients of variation one, inspect approximate queue wait about 4.67 minutes.
Reduce both variability coefficients to one-half while leaving mean service and utilisation fixed.
Compare wait about 1.17 minutes, then raise utilisation to inspect why spare capacity matters to delay.
OPTIONAL DEEPER DETAILGo deeper: inside the model
Inside this model
Mean service time is two minutes. Approximate waiting is ((ca²+cs²)/2)×ρ/(1−ρ)×2; arrival rate is ρ/2, allowing an approximate queue length through Little’s relation.
Where this idea is useful
A practical use
Explain why matching average capacity to average demand can leave a highly variable operation with long queues.
A common misconception
“Equal average demand and capacity summaries determine queueing delay completely.”
Variability and utilisation interact, so systems with similar averages can have very different waiting under appropriate queueing assumptions.
What this explanation leaves out
- Appropriate stationary renewal arrivals, service assumptions, finite moments and utilisation below one are required. This is an approximation, not an exact multi-server, scheduled-arrival or bursty-system forecast.
Is the zero-variability result an exact prediction for every scheduled service?
No. The formula is a model approximation, and real schedules, dependence or multiple servers may violate its conditions.
Which source of variability is being hidden by your average workload figure?
Associated thinkers
Associations marked provisional are awaiting source review.
Further reading
Explore the original research or the teaching reference behind this experiment.