Key takeaways
  • A drill that goes perfectly usually tested the plan people already knew, at a time they expected.
  • Vary the scenario, the time and the blocked assumptions: a blocked exit, an absent warden, a failed alarm point.
  • The debrief and the actions from it are the deliverable. The drill itself is just how you generate them.
The drill that teaches nothing
Announced, same scenario, same time of day

Evacuating a building at 10am on a Tuesday, announced a week ahead, confirms that people can walk outside. It does not test whether the plan works during a shift change, in the dark, with a visitor in the building.

Planning a drill worth running

  • Choose a scenario from the actual risk assessment, not from habit.
  • Vary time of day, shift and day of week across the year.
  • Introduce a realistic complication: a blocked route, an unavailable warden, a failed call point, a person requiring assistance.
  • Decide in advance what you are testing and what a failure would look like.
  • Brief observers on what to watch and where to stand, and give them something to record on.

What to observe

  • Time to alarm, time to full evacuation, time to accounting for everyone.
  • Whether the alarm was audible and visible everywhere, including plant rooms and toilets.
  • Whether people went to the correct assembly point and stayed there.
  • How visitors, contractors and anyone requiring assistance were handled.
  • Whether processes and equipment were made safe, or simply abandoned.
  • Whether the roll call actually reconciled, and how long it took.

Debriefing

Do it immediately while it is fresh, with the observers and a cross-section of participants. Ask what was confusing, not just what went wrong. Confusion is the leading indicator of failure in a real event, when stress is higher and information is worse.

Recording it

Scenario, date and time, participants and numbers, timings, observations, issues found, actions with owners and dates. Regulators and auditors ask for drill records, but the more important reason is that the action list is the only durable output. A drill with a record and no actions is a record that nothing was learned.

Special cases worth drilling

Not only fire. Spill response, confined space rescue, medical emergency in a remote area, loss of power, and evacuation during a night shift with minimal staffing. The scenarios with fewest people available are the ones where the plan most often turns out to depend on someone who is not there.