Ammonia Drill Failure and Fifteen-Minute Accountability

The failures are layered, not a single siren problem. The primary notification failed and the backup took four minutes, which consumes nearly a third of the accountability objective. The control room did not announce wind direction, so the assembly-area decision was left to employees. Contractors were absent from the roster, and the mutual-aid fire department could not obtain an occupancy list at 12 minutes. The 18-month drill interval and open corrective actions show a broken corrective-action system, not just a training gap.

Immediate actions: test the siren and backup activation weekly; add a redundant notification path such as mass text, radio, and handheld air horns; script and rehearse the wind-direction announcement; integrate contractor sign-in with the site occupancy list; and provide the fire department with a printed and electronic occupancy summary at the gate. Open a corrective-action tracker with owners and due dates, and close the previous drill's actions before scheduling the next exercise. Set a 90-day full-scale drill and a monthly notification test. These actions are inexpensive and address the failures that most directly threaten the 15-minute objective.

Move from one paper roster to zone-based accountability. Divide the site into four or five mustering zones, each with a marshal, a backup, a roster, and a physical checkpoint. On evacuation, employees report to their zone marshal rather than to a single crowd; contractors report to the contractor coordinator. Use a buddy system and a sweep team for restrooms, break rooms, and maintenance areas. The control room should maintain a live occupancy count from badge-in data, contractor sign-in, and visitor logs, but keep a printed fallback because the drill showed electronic dependency is risky.

At the assembly area, marshals report zone counts to the incident commander at 5, 10, and 15 minutes. Missing names are passed to a search adviser, not read aloud to the crowd. The fire department receives the same dashboard at the gate. Rehearse the system with unannounced drills, including one with the siren disabled and one with a contractor surge. Set a target of 95% accountability at 10 minutes and 100% at 15 minutes, and measure the time to produce the occupancy list separately. The design must work without network access.

Coordination should be pre-incident, not at the gate. Meet with the fire department to agree on a unified command structure, a liaison role, the occupancy-list format, and the staging area. Provide a pre-incident plan with the ammonia tank location, water requirements, evacuation zones, and the site's wind-direction procedure. Invite the fire department to the next drill and have them request the occupancy list as a test. The site should also designate a public-information lead who can draft and release a community notification within the regulatory window.

For the 30-minute notification requirement, prepare templates in advance for shelter-in-place and evacuation, with pre-approved contacts and a call tree. Test the notification system monthly and log the time from decision to release. During an incident, the incident commander should authorize notification early rather than waiting for complete information; the message can be updated. Assign a scribe to record decision times. After the drill, measure notification time as a key performance indicator alongside accountability. The fire department and community notification processes should be exercised together, because the drill showed that information gaps at the gate delay both.