VESDA Aspirating Fire Detection for Irish Data Centres

Fire detection in a data centre is fundamentally different from fire detection in any other building type. The unique combination of high-velocity airflow, inaccessible voids, physical containment structures and the absolute intolerance of false alarms makes conventional point detection inadequate for the task. Aspirating Smoke Detection (ASD) — and specifically the VESDA (Very Early Smoke Detection Apparatus) product family from Xtralis — has become the standard approach for Irish data centres that take life safety and business continuity seriously. This guide explains the engineering rationale, IS 3218 compliance framework, product range, design principles and suppression integration that every Irish data centre designer and facility manager needs to understand.

Why Conventional Point Detectors Fail in Data Centres

A conventional optical or ionisation point detector relies on smoke physically rising to the detector location and entering the sensing chamber. In a standard commercial building, this works adequately. In a data centre, four specific conditions undermine this mechanism:

• High-velocity airflow: CRAC and CRAH units create strong directional airflows — typically 0.5 to 2.5 m/s across the data hall floor. Smoke generated by an overheating component is rapidly diluted and dispersed by this airflow before it can accumulate to a detectable concentration at any fixed point detector. The same cooling air that protects IT equipment from thermal failure is the mechanism that makes early conventional smoke detection unreliable.
• Inaccessible voids: Raised floor plenums (typically 600–900 mm deep) and above-ceiling cable voids are primary locations for cable faults and electrical ignition. Conventional point detectors are difficult to install, maintain and test in these environments, and access for routine servicing requires disruption to the live data hall.
• Hot aisle and cold aisle containment: Physical containment structures — hard ceilings over hot aisles, containment doors at aisle ends — create isolated zones that a single data hall detector cannot adequately survey. Smoke from a rack fire within a contained hot aisle may not reach any detector outside the containment for a significant period.
• Uptime imperative: Data centre operators typically require 99.999% availability (Tier III/IV). A false alarm that triggers suppression discharge or building evacuation causes an outage event. The consequence of a false alarm can be more operationally damaging than a brief real fire event caught at an early stage. Conventional detectors have far higher nuisance alarm rates than ASD systems.

ASD and VESDA Fundamentals

An Aspirating Smoke Detection system actively draws air samples from the protected space through a network of sampling pipes and delivers them to a high-sensitivity laser detection chamber housed in a remote detector unit. This is the fundamental distinction from passive point detectors: the ASD system brings the air to the detector, rather than waiting for smoke to drift to a fixed sensor.

The VESDA detection cycle operates as follows:

1. An internal fan draws air continuously through a network of sampling pipes. Pipe sampling holes are positioned at intervals defined by the ASD zone design — typically every 1–2 metres in high-risk areas or every 2–4 metres in lower-risk coverage zones.
2. The sampled air passes through a filter that removes particulates (dust, fibres) which could cause nuisance alarms while allowing smoke particles to pass through to the detection chamber.
3. A high-intensity laser beam passes through the filtered air sample in the detection chamber. Smoke particles scatter the laser light in a pattern that the detector's photodiode array and signal processor analyse to determine smoke concentration in percentage obscuration per metre (%/m).
4. The signal processor compares the measured concentration against four programmable alarm thresholds: Alert (lowest — early warning, investigate), Action (prepare response), Fire 1 (confirmed fire — initiate suppression protocol) and Fire 2 (second confirmation or higher concentration — immediate response).

The sensitivity of VESDA detection — as low as 0.005%/m obscuration — is approximately 30 times more sensitive than conventional optical point detectors. This allows detection of incipient fire signatures (the invisible aerosols produced by overheating insulation or components before visible smoke is produced) typically 15–30 minutes before a detectable fire event, providing a substantial investigation window before any suppression action is required.

Xtralis VESDA Product Range

Xtralis (part of Honeywell since 2016) manufactures the VESDA product family. For Irish data centre applications, four models are most commonly encountered:

IS 3218 Compliance Framework for ASD in Irish Data Centres

IS 3218:2013 (plus its National Annex, updated by S.R. 51:2017) is the Irish Standard for fire detection and alarm systems in buildings, and it is the primary compliance framework for Irish data centre fire detection design. The standard is derived from EN 54 series standards but includes Irish-specific amendments and occupancy category definitions that differ from the UK BS 5839 framework.

Key IS 3218 provisions relevant to ASD in Irish data centres:

• Occupancy category for data centres: Data centres are typically classified as Category P2 (property protection of the whole building) or Category L2/L3 (life protection of defined areas where personnel are present). The category determines the detection coverage requirements and the required response time performance of the ASD system.
• ASD as an alternative to point detection: IS 3218 explicitly permits ASD systems as an alternative to conventional point detectors under Clause 15, provided the ASD system complies with EN 54-20 (Aspirating Smoke Detectors) and the design is carried out in accordance with the manufacturer's design tools and either AS 4428.3 or EN 54-20 Annex B pipe network calculation methodology.
• Design documentation: The IS 3218 responsible designer (typically an ETCI-registered engineer or a certified fire detection system designer) must document: zone layout drawings showing all sampling pipe routes and hole positions; pipe network hydraulic calculations confirming adequate flow at all sampling points; alarm threshold settings with engineering justification; and the commissioning test protocol including response time verification by zone.
• ETCI compliance: All Irish data centre fire alarm installations must comply with the ETCI National Rules for Electrical Installations (ET 101) and be carried out by RECI/ETCI registered contractors. The ASD system's fire alarm control panel (FACP or FACIE) must carry CE marking and comply with EN 54-2 (Control and Indicating Equipment).

Sampling Pipe Design Principles

The sampling pipe network is the most critical design element of an ASD system and the area where design errors most commonly compromise performance. The fundamental design principles for Irish data centre ASD pipe networks are:

• Pipe material and diameter: Standard VESDA-E pipe is 25 mm OD rigid CPVC or ABS. Flexible 25 mm convoluted conduit is permitted for short sections (under 2 m) at detector connections and at sampling hole positions in constrained locations. Pipe diameter must not be reduced below 25 mm in the sampling network — reduction creates flow imbalance that invalidates the hydraulic calculation.
• Maximum pipe length: Per EN 54-20 and VESDA design rules, total equivalent pipe length (including fittings expressed as equivalent lengths) must not exceed the detector model's rated maximum. Exceedance of maximum pipe length reduces sensitivity by increasing transit time and dilution of the sample.
• Sampling hole sizing: Hole diameters are calculated using the VESDA design software (VESDAnet Designer or ASPIRE2) to achieve balanced flow at all sampling points. A typical 25 mm pipe with 40 sampling holes may use hole diameters ranging from 1.2 mm at the distal end to 3.0 mm near the detector, ensuring approximately equal air volume is drawn from each position.
• Transit time: The time for a smoke sample to travel from the most remote sampling hole to the detector chamber must not exceed 90 seconds per EN 54-20. Transit time is calculated as part of the pipe network hydraulic design and must be verified during commissioning.
• End cap purge: All pipe runs must terminate in an end cap with a purge hole (typically 4 mm) that allows a small flow of clean air to prevent pipe blockage at the terminus. Purge holes must not be blanked during installation.

ASD Zone Coverage Design for Data Centres

A comprehensive ASD zone design for an Irish data centre typically addresses four distinct detection zones, each with different sampling pipe configurations and alarm threshold settings:

Zone 1 — Under-Floor Plenum

The raised floor plenum is a primary fire risk zone due to the concentration of power distribution cables, copper and fibre data cabling, and PDU connections. Sampling pipes are routed within the plenum at low level, with sampling holes positioned at regular intervals to detect incipient overheating from cable faults or PDU connection issues. Alarm thresholds are typically set at the most sensitive levels (Alert at 0.02%/m, Fire 1 at 0.1%/m) given the high risk of smouldering cable fires that produce low-concentration smoke before flaming combustion begins. Floor tile lifting for maintenance must include a permit-to-work procedure to avoid accidental damage to ASD pipe runs.

Zone 2 — Hot Aisle Containment

Within hard-ceiling hot aisle containment enclosures, sampling pipes are routed along the containment ceiling with sampling holes directed downward into the server exhaust air stream. Because hot aisle air temperatures can reach 45–55°C, the ASD detector and pipe must be rated for the maximum operating temperature (VESDA-E VEP: rated to 60°C ambient). Alarm thresholds in hot aisle zones may be set slightly higher than general area thresholds to account for any residual thermal aging effects on the detection chamber at elevated temperatures.

Zone 3 — Above False Ceiling

The above-ceiling cable management void carries power and data cabling feeding from the main distribution frame and overhead cable trays to the rack containment system. This void is typically inaccessible during normal operation and represents a cable fire risk. ASD sampling pipes are routed above the false ceiling tiles, avoiding conflict with cable trays and fire suppression pipework.

Zone 4 — General Data Hall

A general area ASD zone covers the main data hall volume above the containment structures and any uncovered areas. This zone provides a second layer of detection supplementing the targeted containment zone coverage and catches smoke from sources that may not be within a contained zone — aisle end areas, perimeter walks, equipment staging areas.

Integration with Clean Agent Suppression Systems

The integration between VESDA detection and clean agent gaseous suppression is the most safety-critical interface in the data centre fire strategy. Irish data centres use three primary clean agent suppression agents:

• FM-200 (HFC-227ea): The most widely installed clean agent in Irish data centres built between 2000 and 2020. Effective at 6.25–9% concentration by volume; safe for occupied areas. No ozone depletion potential but has a high global warming potential (GWP = 3,220). Subject to EU F-Gas Regulation restrictions; new installations are increasingly moving to alternatives.
• Novec 1230 (FK-5-1-12): The preferred alternative for new Irish data centre installations. Effective at 4.2–5.8% concentration; zero ozone depletion; very low GWP (1). Longer hold time than FM-200 in many enclosures. Slightly higher agent cost offset by lower quantity required.
• CO2: Used in unoccupied areas such as cable basement voids, generator enclosures and unattended UPS rooms. CO2 is not permitted in occupied data halls without engineered personnel safety procedures and timed discharge delays that are impractical in staffed areas.

The VESDA-to-suppression integration signal chain operates through the Fire Alarm Control and Indicating Equipment (FACIE). VESDA outputs its four alarm levels as volt-free contacts or addressable loop signals to the FACIE. The FACIE maps these to suppression control actions:

1. Alert signal received: FACIE triggers a local audible warning tone and sends a notification to the BMS and operations monitoring system. No suppression action. A 60-second investigation window allows operations personnel to inspect the zone before any further escalation.
2. Action signal received: FACIE arms the suppression system and initiates the pre-discharge evacuation sequence — continuous alarm sounders, strobe beacons at zone entry points, and automated HVAC shutdown for the affected zone (to prevent dilution of the suppression agent). Operations personnel evacuation commences.
3. Fire 1 signal received: FACIE initiates the suppression discharge countdown, typically 30 seconds (allowing completion of evacuation), then triggers gas discharge. Abort switches within the zone perimeter allow override of the countdown if personnel confirm the alarm is spurious.
4. Fire 2 signal received: FACIE may trigger immediate discharge (countdown bypassed) where the fire strategy has determined that the risk of rapid fire spread justifies elimination of the delay. This is typically only used in unoccupied zones.

AI-Enhanced Detection and Particle Discrimination

The most significant advance in ASD technology over the past five years is the integration of machine learning algorithms for particle discrimination — the ability to distinguish between smoke particles and other airborne particulates that can cause nuisance alarms: dust, condensation droplets, aerosol cleaning products and steam from cooling systems.

Traditional ASD systems use a single-wavelength laser and a photodiode array to measure light scattering. AI-enhanced detection platforms (including the latest Xtralis VESDA-E firmware updates and competing products from Hochiki and System Sensor) use multi-wavelength laser sources combined with ML models trained on thousands of particle scatter signatures. The result is a significant reduction in nuisance alarm rates from non-smoke particulates, without reducing the system's sensitivity to genuine smoke particles.

For Irish data centres, particle discrimination is particularly valuable in environments with elevated dust loading during construction phases (commissioning during fit-out), in battery rooms where off-gassing from VRLA battery banks can contain aerosol electrolyte particles, and in generator rooms where diesel exhaust from generator testing can reach low concentrations in adjacent protected zones.

BMS and FACIE Output Signals

Beyond the suppression integration, the VESDA system provides multiple output signals to the broader building management and monitoring infrastructure:

• BMS integration: VESDA fault signals (low airflow, filter saturation, detector contamination, power supply fault) are typically integrated into the BMS as Modbus TCP or BACnet inputs, ensuring facilities teams receive maintenance alerts through the same operational dashboard used for HVAC and electrical monitoring.
• Addressable loop output: In facilities using an addressable fire alarm panel as the FACIE (compliant with EN 54-2), VESDA can interface via an addressable protocol module (VESDA-E APIC card supporting Apollo, Hochiki or Notifier protocols) enabling zone-level alarm information to appear on the FACIE graphic display.
• Remote VESDA monitoring: VESDAnet allows up to 20 VESDA-E detector units to be networked together and monitored from a single graphical workstation. For large Irish data centres with multiple ASD zones, a central VESDAnet workstation provides a single view of all zone alarm states, flow rates, detector contamination levels and maintenance due dates.
• DCIM integration: Some Irish operators integrate VESDA alert and fault signals into their DCIM platform alongside power and thermal data, creating a single operations interface that correlates thermal events (detected by DCIM) with smoke events (detected by VESDA) for root cause analysis.

Irish Standards and Regulatory Context

The complete regulatory and standards framework for VESDA and ASD systems in Irish data centres encompasses:

• IS 3218:2013 + S.R. 51:2017: Primary Irish Standard for fire detection system design. Compliance is a statutory requirement for new-build data centres and major refurbishment works in Ireland.
• EN 54-20:2006: Harmonised European Standard for Aspirating Smoke Detectors. All ASD units installed in Ireland must carry CE marking demonstrating EN 54-20 conformity. Pipe network design calculations must follow EN 54-20 Annex B or AS 4428.3 methodology.
• EN 54-2:2012: Fire Alarm Control and Indicating Equipment standard. The FACIE must comply with EN 54-2 for the fire detection and alarm system to be IS 3218-compliant.
• EN 54-13:2005: Assessment of compatibility of components of fire detection and fire alarm systems. Requires documented evidence that the combination of ASD detector, FACIE, sounders, strobes and suppression interface is a compatible, tested system.
• ETCI ET 101: National Rules for Electrical Installations — all fire alarm wiring, containment and connections must comply. ASD sampling pipe is not classified as electrical containment but must be installed in accordance with the ETCI-registered contractor's design.
• S.I. 58 of 2000 (Fire Services Acts): Irish statutory requirement for fire safety systems in certain building categories. Data centres above a threshold size are subject to fire safety certificate requirements from the relevant local authority, which will typically include review of the fire detection system design against IS 3218.

Future Directions in Data Centre ASD

Three trends are shaping the next generation of ASD design for Irish data centres. First, liquid cooling integration: as direct liquid cooling (DLC) and immersion cooling deployments increase in Irish hyperscale facilities, the fire risk profile changes — there is less cabling density but new risks from dielectric fluid leaks and electrical arcing in liquid-cooled manifolds. ASD zone designs will need to adapt to cover liquid cooling manifold areas and cold plate distribution zones. Second, very high density AI compute: GPU and AI accelerator racks exceeding 100 kW per rack create extreme thermal conditions that make the hot aisle zone a more critical ASD zone than ever, and may require dedicated per-rack ASD micro-sampling from within the rack enclosure itself. Third, wireless ASD: battery-powered ASD sampling heads with wireless connectivity to a central detector unit are emerging as a solution for retrofit installations where pipe routing is impractical in an existing live data hall — not yet mainstream in Ireland but likely to appear in refurbishment specifications within the next three to five years.