Resource schedules, breaks, failures, and maintenance
Model when capacity is actually available and define what happens to work when shifts, breaks, failures, or maintenance remove a resource.
A nominal headcount of six does not mean six servers are available all day. Shifts, breaks, training, travel, planned maintenance, and unplanned failures change effective capacity over time.
If demand also varies by time of day, align the availability schedule with the demand profile. A daily average can hide the exact period where queues form.
When a shift ends or a break begins, a busy resource might finish the current task, stop immediately and resume later, hand the work to another resource, or restart the task. Those choices can materially change waiting, overtime, and completion times.
Write the rule down for each important resource. Do not let the simulation package's default decide a labor or machine policy that the real system handles differently.
For an unreliable machine or service, model both how long it operates before failure and how long recovery takes. If failures depend on age, workload, or elapsed operating time, a memoryless failure assumption can give the wrong downtime pattern.
Planned maintenance is not the same as a random breakdown: its timing is scheduled and operators may be able to drain work before it starts.
A resource busy for six hours of an eight-hour available shift has 75% utilization for that shift. Dividing by a 24-hour calendar day would answer a different question.
When comparing alternatives, state whether breaks, failures, maintenance, and off-shift time count as unavailable, idle, or busy. Otherwise two utilization numbers may not be comparable.
If the availability pattern is wrong, downstream waiting and utilization can look plausible while answering the wrong staffing or reliability question.
- Compare simulated available headcount by time period with the roster or machine schedule.
- Check the number and duration of failures against maintenance records when they exist.
- Inspect queues around shift changes and downtime events; averages can hide handover spikes.
- Check what happens to work already in service when capacity disappears.