Safety & Risk Management in Data Centres: Emerging Fire and Electrical Challenges

Data centres quietly support almost everything people do online today. They keep cloud platforms running, store healthcare and banking records, support enterprise software, and power AI workloads that require continuous computing. Even a small disruption inside a data centre can create serious business consequences.

As data centres grow larger and more powerful, the risks inside them are also changing. These facilities now handle denser server environments, more demanding cooling systems, complex electrical infrastructure, and growing use of lithium-ion battery storage. All of this improves performance and resilience, but it also creates new safety concerns. Fire hazards, electrical faults, overheating, and battery failures are no longer rare technical issues. They have become business risks that can affect uptime, customer trust, compliance, and operational continuity.

This is why data centre safety needs to be approached in a practical and connected way — not just about installing alarms or fire extinguishing systems, but about understanding where risks come from, how they develop, and how systems and people should respond before an incident grows out of control.

Why Data Centre Safety Has Become a Core Business Need

Modern data centres run around the clock with high-density racks, nonstop electrical loads, advanced cooling equipment, backup power systems, and battery energy storage. When everything works well, the system stays stable. But if one component fails, it can affect multiple systems at once.

A loose electrical connection, an overheating cable, a UPS issue, or a battery defect may seem small at first. In a data centre, these issues do not remain isolated. Because equipment is tightly packed and operations are continuous, a minor fault can spread quickly or interrupt critical services before anyone has time to respond.

Safety is directly linked to uptime and trust

Data centre incidents are not only about physical damage. They stop business operations, affect customer-facing services, and harm a company's reputation. If a cloud platform, enterprise application, or transaction system goes offline, the impact can be immediate. Clients may lose confidence. Service-level commitments may be missed. Recovery can become expensive and complex.

Risk management in data centres therefore serves several business purposes:

• Protects uptime and service continuity
• Reduces the chance of major fire or electrical incidents
• Safeguards employees, technicians, and contractors
• Lowers financial and reputational damage after an event
• Supports compliance, and goes beyond compliance into resilience

The risk profile is changing faster than many teams realise

Older safety approaches were designed for buildings like offices or industrial warehouses. Data centres combine electrical intensity, sensitive digital assets, airflow-driven environments, and energy storage systems in one place. Traditional safety assumptions do not always translate.

Smoke may not become visible quickly because strong airflow can dilute it. Hidden spaces under raised floors or inside plenums can allow heat or smoke to spread unnoticed. High-performance computing and AI workloads push thermal stress much higher than before. Operators need a more advanced and proactive safety model than the one used in conventional facilities.

Understanding Where Fire Risk Comes From

In data centres, fire risk often begins quietly. A cable may start smouldering. A connector may develop a hotspot. Dust may accumulate in a cooling unit. A battery cell may begin to off-gas before any obvious thermal event occurs. These hidden warning signs are what make data centre fires especially dangerous.

Common sources of ignition include:

• Faults in switchgear, transformers, PDUs, and UPS systems
• Overloaded circuits and ageing insulation
• Overheating servers, storage systems, and networking hardware
• HVAC problems such as blocked filters or overheating motors
• Battery rooms and energy storage systems, especially lithium-ion batteries

These risks often overlap. A cooling issue increases equipment temperature. Increased temperature stresses electrical components. Electrical stress can trigger sparks or fire. Fire risk in data centres should never be treated as a single-issue problem.

High-density computing is making fire risk more complex

AI, analytics, and high-performance computing workloads consume more power and generate more heat. As rack density increases, the margin for error becomes smaller. If cooling, load balancing, or electrical distribution slips even slightly, the risk of overheating rises. Heat builds up faster, equipment has less tolerance for cooling inefficiencies, and more power runs through the same facility footprint.

Hidden spaces make detection harder

Underfloor spaces and overhead plenums are efficient for design but can hide developing hazards. A small fire in one of these areas may not be visible immediately. Smoke can move in unusual ways because of air circulation, and staff may notice the problem only after it has spread. Fire risk in data centres is therefore increasingly about early detection of subtle issues, not only responding to obvious emergencies.

Electrical Safety and Battery Storage Need Special Attention

Electrical infrastructure is at the heart of every data centre. Transformers, switchgear, UPS units, distribution panels, power cables, and backup systems all work continuously under heavy loads. Problems often build slowly — a loose terminal, poor cable routing, degraded insulation, or harmonic distortion can create hotspots that are not visible without proper monitoring.

Electrical safety in data centres is not just about grounding or standard inspections. It requires ongoing attention to how systems are performing under real operating conditions. To reduce electrical risk, operators need to focus on:

• Routine thermographic inspections
• Load testing of critical equipment
• Cable management that avoids overcrowding
• Proper grounding and bonding checks
• Updated one-line diagrams and fault documentation
• Clear isolation procedures for maintenance teams

Battery storage is now a critical safety concern

As more data centres use lithium-ion batteries for UPS support and energy storage, battery safety has become a major part of overall facility risk. Lithium-ion systems offer performance benefits but introduce thermal runaway risks that older battery technologies did not present in the same way. Overheating, overcharging, humidity, vibration, and physical damage can all contribute to battery failure. Gas emissions may appear before a thermal event fully develops.

Thermal runaway can escalate quickly. If a battery cell begins failing and the issue is not detected early, heat and gases can spread to nearby cells and equipment. At that point, response becomes more difficult and more urgent.

Battery safety must be planned as its own risk zone

Battery rooms should not be treated like ordinary equipment spaces. A safe battery infrastructure should include:

• Strong battery management system (BMS) monitoring
• Controlled charging and discharging limits
• Temperature monitoring and ventilation control
• Gas detection for early off-gassing signals
• Physical segregation from sensitive or high-traffic areas
• Dedicated suppression and shutdown logic where required

Battery safety is where electrical risk and fire risk converge most clearly. Failing to monitor this area properly means missing one of the most serious sources of modern data centre incidents.

Detection, Suppression, and Cooling Must Work Together

In data centres, high airflow, large rooms, hidden spaces, and early-stage smouldering conditions can delay or weaken the warning signals that ordinary detectors rely on. General standards such as NFPA 72 provide guidance but do not automatically guarantee early detection in specialised data centre conditions.

That is why many facilities now use more advanced detection strategies:

• Aspirating smoke detection that continuously samples air
• Multi-sensor systems combining smoke, heat, flame, and gas inputs
• Staged alarms that allow investigation before full emergency action
• Self-diagnostic systems that improve reliability and reduce false alarms

The purpose of detection is not just to sound an alarm. It is to give operators time to investigate, isolate the threat, and act before the incident becomes severe.

Suppression must protect both people and equipment

Fire suppression in a data centre is a delicate balance. Water-based sprinkler systems can extinguish fire effectively but may also damage sensitive IT infrastructure. Many modern facilities use clean agents, inert gas systems, water mist, or targeted rack-level suppression depending on the environment.

A layered suppression strategy typically combines:

• Early warning to identify the issue quickly
• Localised suppression to stop spread at the source
• Room-level suppression if the fire escapes containment
• Emergency shutdown logic to isolate affected zones only

The aim is to stop the hazard while protecting unaffected systems and reducing unnecessary downtime.

Cooling systems influence fire safety more than people assume

Poor airflow, blocked filters, overheating motors, and dust buildup can all become ignition sources. Airflow patterns also affect how smoke and heat travel through a facility — influencing how quickly a detector responds and how well suppression performs. Cooling and fire protection should be designed together, not in isolation.

Operators should consider how airflow may dilute or redirect smoke, whether plenum spaces need additional protection, how dust is managed in HVAC systems, and whether detector placement reflects actual heat and smoke movement.

Building a Safer and More Resilient Data Centre

Standards and codes are essential but not sufficient on their own. A compliant facility can still have significant blind spots if its actual operating conditions are not fully understood. Real safety comes from detailed risk assessment, not only from following generic rules.

A strong data centre risk programme should evaluate:

• Ignition sources in each zone
• Fuel load and material behaviour
• Ventilation and airflow patterns
• Response time and detectability
• Impact on uptime, people, and recovery costs

This analysis helps operators prioritise the most serious risks first, instead of spreading resources evenly across every possible hazard.

Operational discipline matters as much as technology

Even the best systems can fail if procedures are weak. Data centres need disciplined maintenance, housekeeping, hot-work control, contractor management, and emergency planning. Temporary cabling, rushed electrical work, and poor permit-to-work practices often create avoidable risks.

Important operational controls include:

• Task-specific risk assessments for maintenance work
• Strict hot-work permits and fire-watch procedures
• Clear emergency instructions for staff and contractors
• Regular drills that reflect real facility hazards
• Coordination with local fire responders who understand the site layout

Safety should be linked to resilience and continuity

A fire event in a data centre is also a business continuity event. The impact may extend to customers, partners, and platforms depending on what the facility supports. Safety planning should connect directly with business continuity planning — defining acceptable downtime, shutdown isolation procedures, incident response leadership, and communication protocols during an emergency.

The future will require smarter, more connected safety systems

As AI workloads grow, edge data centres expand, and battery use increases, safety management will have to become even more intelligent. Future-ready data centres will need better sensing, better analytics, better zoning, and stronger integration across fire systems, battery controls, building management, and electrical monitoring.

The most resilient facilities will be the ones that stop treating fire risk, electrical safety, cooling issues, and battery storage as separate topics. In reality, they are all part of one connected safety system. Managing them in an integrated way reduces risk more effectively and protects both infrastructure and business continuity.

C. Rajadurai (Reviewed by)

Sr. Manager – Environmental Consulting | Chola MS Risk Services

C. Rajadurai is an environmental consultant with ten years of experience in environmental engineering and sustainability consulting. His expertise spans carbon footprinting, net-zero strategy, water stewardship (AWS Certified Professional), Environmental Impact Assessments across sectors including ports, power plants, and FMCG, and geospatial studies using Remote Sensing and GIS. He is an ISO 14064 certified lead verifier for GHG accounting and a QCI/NABET approved expert for Land Use & Land Cover.