Configuring Sapphire HVAC Shutdown Before Discharge

A Sapphire suppression sequence must treat HVAC shutdown and agent release as separate timed paths. In the resolved configuration, the releasing sequence gave the CRAC units a 20 s pre-discharge interval; measured full shutdown took 8 s, leaving a nominal margin. Triggering an EPO only when discharge begins provides no pre-discharge margin.

Which shutdown approach fits the timing requirement?

Use pre-discharge shutdown when the design objective is to have the CRAC units stationary before agent release. The cited interpretations of NFPA 2001 differ on whether equipment should stop or may remain operating when it only recirculates protected-room air. Resolve that point against the adopted edition, the Sapphire system installation instructions, and the authority having jurisdiction. For information-technology facilities, also review NFPA 75, Section 8.4; for telecommunications facilities, review NFPA 76.

Where does the shutdown request travel?

Follow the signal from detection to final fan motion. The recommended path starts with the initiating devices, passes through the releasing control logic, branches to the HVAC stop interface, and later reaches the agent-release actuator. The shutdown path is not complete merely because a panel relay changes state.

If an EPO operates through additional relays, controllers, or power equipment, every hop adds a possible open circuit, logic mismatch, or delay. A controller showing “off” does not prove that rotating equipment has stopped.

What physical checks come before sequence changes?

Layer one first. Identify the exact releasing-panel output, the intermediate shutdown relay, the CRAC input, and the circuit power source. Compare contact form and normal state at both ends. Confirm that the fire-system output can operate the receiving circuit without exceeding either device rating; obtain those ratings from the panel and CRAC documentation rather than inferring them.

Check continuity through the complete shutdown circuit and test each intermediate relay. Look for a maintained-versus-momentary command mismatch, shared EPO loads, controller input filtering, local/remote mode selection, and restart logic that can override the fire command. Verify whether loss of circuit power produces a stop, a run, or an indeterminate state.

Inspect the air path separately. Classify each unit as closed recirculation, outside-air supply, exhaust, or a combination. Recirculation can aid agent mixing, while supply or exhaust paths can move agent across the enclosure boundary. Damper travel and fan coast-down are mechanical processes and require direct timing measurements.

How should pre-discharge shutdown be configured?

1. Document the cause-and-effect sequence from each qualifying detector state through HVAC shutdown, warning functions, the discharge delay, and release.
2. Use a dedicated, supervised interface where required by the approved system design. Do not presume that the building EPO implements the needed order; it did not in this installation.
3. Start the HVAC stop command early enough that every affected CRAC unit reaches a complete stop before release. The resolved sequence used a 20 s delay.
4. Measure each CRAC unit from the actual stop command to zero fan motion. The recorded full-stop time here was 8 s.
5. Calculate the observed nominal margin as M = Tdelay - Tstop. For the recorded values, . This calculation assumes the 8 s measurement starts at the shutdown command and ends at complete mechanical stop.
6. Review the sequence with the suppression-system manufacturer, HVAC manufacturer, authority having jurisdiction, and any party responsible for NFPA 75 or NFPA 76 compliance.

A separate FM-200 arrangement cited a built-in 45 s delay while sending an HVAC shutdown signal. That value demonstrates another sequence architecture, not a transferable Sapphire setting. Use only the delay approved for the installed releasing system and protected hazard.

How is the timing margin verified?

Use a controlled functional test that records common reference times. Capture detection qualification, shutdown-output transition, CRAC input recognition, fan stop, delay expiration, and simulated release. Repeat the test for every CRAC unit because controller behavior, drive deceleration, belts, dampers, and rotating inertia can produce different stop times.

Test abnormal states as permitted by the approved commissioning plan: loss of interface power, an open shutdown circuit, a CRAC in local mode, and failure of an intermediate relay. Confirm the panel indication and response for each condition rather than relying on an undocumented assumption about fail-safe behavior.

Which sequence mistakes recur?

The most common mistake is equating an electrical command with mechanical completion. Issuing shutdown and discharge simultaneously leaves the entire coast-down interval inside the agent-release event. Another mistake is adding a delay without identifying its purpose. An evacuation delay, shutdown interval, and release delay may occupy the same timeline, but each serves a different function and must appear clearly in the cause-and-effect documentation.

Do not treat all air handlers alike. A CRAC that recirculates room air presents a different agent-retention problem from a unit that introduces outside air or exhausts to another space. Also avoid copying the 45 s delay from the FM-200 example or relying on the initial belief that Sapphire discharge occurs within 10 s. Read the configured delay from the installed releasing control and confirm it by test.

FAQ