When averages hide the queue

A service counter can have enough capacity across an entire day and still build a long queue at lunchtime. The daily average hides when people arrive, how long individual requests take, and which resources are available together. Those details can determine whether a change works.

Discrete event simulation represents a system as events over time. A customer arrives, a resource becomes available, an activity finishes. Between events, the simulated clock can advance to the next change. This makes it useful for exploring processes where timing and competition for resources matter.

Start with a decision

Consider a fictional repair workshop deciding whether to add a service bay. Vehicles need both a bay and a suitably qualified mechanic. Some jobs wait for parts. A second bay may help during busy periods, but it may sit empty when mechanics or parts are the constraint. The question is which change improves completed work under realistic operating conditions.

Write that decision down before drawing a model. Choose an outcome such as completed jobs per day, time spent waiting, or the proportion finished before closing. Include practical limits: available space, working hours, and the cost of each option. These choices define what the model must represent.

Model the detail that can change the answer

For the workshop, useful inputs might include arrival patterns by time of day, a range of repair durations, resource availability, and the rules for selecting the next job. A single average duration may erase the longer jobs that block a bay. On the other hand, modeling every tool movement may add effort without changing the decision.

Build the smallest model that can distinguish the options. Ask people who know the process to review its logic. Trace a few individual jobs, test a quiet day and a deliberately overloaded day, and compare baseline behavior with observed records. A plausible animation alone cannot establish that a model represents the operation well enough.

Know when a simpler method is enough

A spreadsheet may be sufficient when the decision is a straightforward capacity calculation and timing has little effect. Simulation adds value when queues, variability, dependencies, or operating rules materially influence outcomes. It also needs data, assumptions, and time to check the results.

Before starting a study, write one sentence: we are choosing between these options, using these measures, within these constraints. If that sentence is still unclear, defining the decision is the most useful next step. The workshop is an illustration, not a reported customer result.