Factory Mutual Data Center Design Beyond Building-Code Compliance

Code Compliance vs Factory Mutual (FM) vs Owner / Insurer Criteria

Four layers must remain explicit: adopted building and fire code with Authority Having Jurisdiction (AHJ) review; Factory Mutual engineering-based property-loss-prevention guidance; project-specific owner and insurer criteria; and the final requirements established by the actual project team. Factory Mutual recommendations are not automatically universal mandates. Applicability, priorities, and acceptance must be confirmed for the project.

CODE-COMPLIANT DOES NOT AUTOMATICALLY MEAN LOSS-PREVENTION OPTIMIZED.

Start With the Site, Not the Sprinkler System

Loss-prevention thinking begins before a building exists. Screen flood exposure, elevation, drainage, wind, hail, seismic and geotechnical conditions, wildfire and smoke where relevant, emergency-response access, utility resilience, water availability, surrounding property exposures, and vulnerable adjacent uses. A site with cheaper land can become the more expensive site when mitigation consumes land, capital, redundancy, schedule, or operating flexibility. Flood planning must use current project-specific hazard data and applicable guidance. Do not assume a universal elevation or freeboard. Coordinate flood elevation, freeboard, surface and roof drainage, critical-equipment placement, below-grade vulnerability, access during an event, and recovery. Keep critical electrical and continuity infrastructure out of vulnerable locations where feasible; exact criteria require project-specific confirmation.

The Building Envelope Is a Loss-Prevention System

The envelope is not aesthetic wrapping. For the roof, coordinate membrane, insulation, deck, attachment, wind uplift, hail exposure, drainage, ponding, equipment curbs, penetrations, safe access, and maintainability. For walls, examine combustibility, insulation, cladding, moisture durability, impact resistance, fire resistance, and interfaces with doors and louvers. For penetrations, coordinate electrical, mechanical, plumbing, fire-protection, cable, structural, and future work. Risk often sits at a transition: roof-to-wall, wall-to-foundation, panel joint, door opening, louver, cable or pipe penetration, equipment curb, or expansion joint. Details must preserve weather, fire-resistance, thermal, air, and water-control continuity. Confirm that specified products are suitable within the tested or approved assembly; a listed component does not automatically make an unreviewed combination acceptable. Roof replacement, inspection, drain cleaning, curb access, and equipment removal must remain practical throughout facility life.

1. 1

   Roof

   Verify the complete assembly, attachment, drainage paths, exposed equipment interfaces, penetrations, access, and repair strategy against project criteria.

2. 2

   Walls

   Resolve material combustibility and durability, then detail cladding, insulation, openings, joints, and fire-resistance continuity as one system.

3. 3

   Penetrations

   Create owned details and change control for every service, future opening, curb, louver, joint, and interface.

Fire Detection: Find the Event Before It Becomes the Event

Data centers combine high asset concentration, continuous electrical load, strong airflow, concealed cable pathways, raised-floor or overhead distribution, and multiple potential ignition sources. Very Early Warning Fire Detection (VEWFD) uses high-sensitivity detection that can provide earlier warning than conventional arrangements, depending on the design. It is not a universal answer. Detection design must follow actual airflow: hot-aisle and cold-aisle containment, sampling locations, ceiling configuration, return-air paths, and underfloor or overhead conditions can all affect transport of smoke. Commissioning should include representative smoke testing, verified sampling and alarm thresholds, documented alarm-response sequences, interfaces with suppression and air systems, and recurring inspection. The project fire protection engineer, insurer, manufacturer, and Authority Having Jurisdiction must confirm the appropriate technology and sequence.

Why Water-Based Fire Protection Still Matters

“Water and servers do not mix, therefore data centers should avoid sprinklers” is incomplete. Localized water damage must be compared with uncontrolled fire, heat, smoke, corrosive combustion products, cascading failures, and total room loss. The goal is not to put water everywhere; it is to provide a reliable, engineered way to control a local fire before it becomes a catastrophic facility event. Wet and preaction systems involve different operating, detection, accidental-discharge, and maintenance considerations. Coordinate zoning, isolation, drainage, protected-area boundaries, testing, and impairment management so protection remains available without creating uncontrolled secondary exposure. Clean-agent systems can play an important role, but they do not universally replace water-based protection. Agent selection, enclosure integrity, occupant safety, fire scenario, reignition potential, and water-based backup belong to one integrated fire strategy. Final system choice requires project-specific confirmation.

Uninterruptible Power Supply (UPS), Batteries, and Energy Storage

The Uninterruptible Power Supply (UPS) strategy is architectural as well as electrical. Review redundancy topology, physical segregation, room location, fault isolation, maintenance bypass, ventilation, fire strategy, equipment movement, and replacement access. A bypass that crosses the same vulnerable space as the primary path may preserve a diagram but not real resilience. Lithium-ion batteries introduce thermal-event, propagation, gas and off-gas, detection, ventilation, separation, emergency-response, maintenance, and access questions. Chemistry, cell and module construction, rack configuration, enclosure, controls, tested performance, and system scale all matter. Battery Energy Storage System (BESS) installations located outdoors or centrally can create different spatial, exposure, drainage, security, and fire-protection questions than rack-level or UPS batteries. Battery-fire strategy is chemistry-, configuration-, enclosure-, testing-, and project-specific. It requires fire protection engineering and review of applicable Factory Mutual guidance, National Fire Protection Association (NFPA) standards, manufacturer information, insurer criteria, and Authority Having Jurisdiction requirements. Do not assume water is always prohibited or always sufficient. Separation, ventilation, suppression, drainage, emergency access, and response tactics require project-specific confirmation rather than an invented threshold.

Cooling, Water, and Leak Management

High-density computing changes the property-loss profile. Heating, Ventilation, and Air Conditioning (HVAC) design must address cooling redundancy, airflow, equipment zoning, control dependencies, and failure isolation—not efficiency alone. For liquid cooling, coordinate direct-to-chip circuits, each Coolant Distribution Unit (CDU), distribution piping, couplings, valves, leak detection, containment, drainage, and serviceability. Water systems need reliable sources, appropriate quality and treatment, management of scaling, corrosion, and biological growth where relevant, plus leak isolation, overflow control, and drainage. Heat rejection brings its own dependencies: cooling towers, dry coolers, chillers, pumps, heat exchangers, structural support, weather exposure, and replacement access. A leak, pump failure, control fault, or maintenance event should not disable a disproportionate amount of Information Technology (IT) capacity. Map what each valve isolates, where leaked fluid travels, which alarm acts first, how drainage behaves, what equipment sits below pipes, and whether service can occur without crossing live critical zones. Coordinate controls and physical containment so a single fault does not become a shared failure.

Electrical Rooms, Generators, and Common-Mode Failure

Coordinate utility service, transformers, switchgear, electrical-room separation, busway, UPS equipment, batteries, emergency generators, fuel systems, maintenance, and equipment replacement as connected risk systems. Redundancy is not useful when supposedly independent systems share the same vulnerable room, flood zone, overhead pipe exposure, fire compartment, control dependency, or maintenance pathway. Concrete examples include two redundant electrical paths through one vulnerable room; two cooling systems dependent on one water header; redundant generators served by one vulnerable fuel-transfer arrangement; and independent IT halls served through a shared, unprotected electrical or mechanical bottleneck. Architects should overlay routes, rooms, fire compartments, water paths, access zones, controls, and replacement routes—not merely compare separate discipline diagrams. Ask what single event can defeat both sides and who owns the corrective action.

Factory Mutual (FM) Data Center Risk-Control Layers

Use these eleven layers as a concise multidisciplinary agenda. Each layer needs an owner, evidence, unresolved-risk statement, and project-specific acceptance path.

1. 1

   Site and flood exposure

   Screen natural and surrounding hazards, access, elevation, drainage, and recovery.

2. 2

   Roof and building envelope

   Coordinate complete assemblies, attachment, weather resistance, and maintainability.

3. 3

   Penetrations and combustible materials

   Control interfaces, continuity, substitutions, and future change.

4. 4

   Very Early Warning Fire Detection (VEWFD)

   Match detection and response to actual airflow and fire scenarios.

5. 5

   Water-based fire suppression

   Integrate reliable control with zoning, isolation, drainage, and impairment planning.

6. 6

   Data hall and aisle containment

   Coordinate airflow boundaries with detection, suppression, egress, and access.

7. 7

   Uninterruptible Power Supply (UPS) and lithium-ion batteries

   Resolve chemistry, configuration, separation, protection, ventilation, and replacement.

8. 8

   Electrical rooms and utility resilience

   Separate paths and remove shared spatial, environmental, and control vulnerabilities.

9. 9

   Generators and fuel systems

   Test fuel delivery, transfer, containment, weather, access, and maintenance dependencies.

10. 10

    Heating, Ventilation, and Air Conditioning (HVAC), cooling, and water

    Plan leak control, isolation, drainage, redundancy, treatment, and heat rejection.

11. 11

    Inspection, testing, and business continuity

    Keep protection functional through turnover, impairment, change, and recovery.

The Architect's Factory Mutual (FM) Review Checklist

Six clearly owned checklist groups keep risk decisions visible across planning, detailing, coordination, commissioning, and operations.

1. 1

   A. Site and hazards

   Flood exposure; wind; hail; seismic and geotechnical risk; wildfire and smoke where relevant; emergency access; adjacent property exposure.

2. 2

   B. Building envelope and roof

   Roof assembly and attachment; drainage; exterior walls; insulation; penetrations; louvers; doors; weather continuity; fire-resistance continuity.

3. 3

   C. Fire protection

   Detection; Very Early Warning Fire Detection (VEWFD) where appropriate; sprinkler or preaction strategy; clean agent; compartmentation; fire pump; water supply; impairment strategy.

4. 4

   D. Electrical and batteries

   Utility redundancy; transformers; switchgear; Uninterruptible Power Supply (UPS); batteries; generators; fuel; separation; replacement access.

5. 5

   E. Cooling and water

   Heating, Ventilation, and Air Conditioning (HVAC); liquid cooling; Coolant Distribution Unit (CDU); leak detection; drainage; treatment; heat rejection; maintenance access.

6. 6

   F. Operations and continuity

   Inspection; testing; maintenance; alarm response; impairment procedures; emergency response; training; recovery; change management.

When Factory Mutual (FM) Should Enter the Project

Bring loss-prevention criteria into every decision gate. The workflow is iterative: later testing can reveal a planning assumption that must be corrected, and operational changes must return to the same risk logic.

1. Step 1

   Site screening

   Evaluate hazards before acquisition.

2. Step 2

   Concept design

   Identify owner, insurer, and Factory Mutual criteria before the layout is frozen.

3. Step 3

   Envelope + fire strategy

   Coordinate materials, compartmentation, detection, suppression, drainage, and access.

4. Step 4

   Power + battery + cooling coordination

   Test whether supposedly redundant systems are genuinely independent.

5. Step 5

   Detailed risk review + insurer alignment

   Resolve identified loss-prevention issues before construction.

6. Step 6

   Testing + operations + continuity

   Verify that protection systems remain functional after turnover.

A Practical Design Review Matrix

The matrix below identifies typical coordination roles, not universal assignments. The project responsibility matrix, contracts, applicable law, insurer engagement, and Authority Having Jurisdiction process establish who actually decides, designs, reviews, and accepts each item.

What Developers Should Ask Before Design Freeze

Turn unresolved risk into direct questions with named owners and dates. Answers should cite current project evidence rather than a prior project's assumptions.

1. 1

   Criteria

   Is the insurer known? Is Factory Mutual criteria applicable? Has the team received current project-specific insurer requirements? Who owns final insurer coordination?

2. 2

   Site and envelope

   Have natural hazards and flood exposure been reviewed? Is the roof assembly compatible with loss-prevention criteria? Are wall and insulation combustibility decisions resolved?

3. 3

   Fire and batteries

   Is the fire strategy integrated with aisle containment and airflow? Are battery chemistry and configuration known? What still requires Authority Having Jurisdiction confirmation?

4. 4

   Power and cooling

   Are critical electrical systems protected from common-mode failure? Is liquid cooling included? Where can leaked water or coolant go?

5. 5

   Independence and recovery

   Can major equipment be replaced without taking unrelated systems offline? Can redundant systems actually fail independently?

Common Design Mistakes

These mistakes usually arise when one discipline's apparently complete answer is not tested against the whole-facility loss scenario.

1. 1

   Mistake 1

   Treating Factory Mutual (FM) as a late-stage specification check.

2. 2

   Mistake 2

   Assuming code compliance equals insurer acceptance.

3. 3

   Mistake 3

   Selecting roofing and wall assemblies without understanding loss-prevention criteria.

4. 4

   Mistake 4

   Treating clean-agent suppression as a universal replacement for water-based fire protection.

5. 5

   Mistake 5

   Adding battery systems after room geometry and ventilation strategy are fixed.

6. 6

   Mistake 6

   Designing liquid cooling without leak containment and failure isolation.

7. 7

   Mistake 7

   Calling a system redundant when both paths share a common vulnerability.

8. 8

   Mistake 8

   Ignoring maintainability and equipment replacement.

9. 9

   Mistake 9

   Failing to carry Factory Mutual lessons into repeatable reference designs.

DCFR Design Principle

Factory Mutual (FM) should be treated as an early feasibility and design-control lens. DCFR would flag a concept that appears code-compliant but retains unresolved property-loss exposure involving flood, roof, envelope, combustible assemblies, penetrations, fire protection, batteries, electrical systems, generator and fuel systems, cooling, water, common-mode failure, or operational continuity.

CODE COMPLIANCE ALONE DOES NOT AUTOMATICALLY MAKE A DATA CENTER LOSS-PREVENTION OPTIMIZED, RESILIENT, OR INSURABILITY-READY.