A practical framework for improving cooling reliability, ventilation visibility and energy performance before adding AI analytics.

Infographic-style illustration showing a Singapore commercial facility with ACMV sensors, ventilation airflow, chilled-water piping, building controls and an analytics dashboard.

Singapore’s renewed attention on heat resilience, smart cooling and district-level energy efficiency is putting cooling operations firmly on the facilities agenda. In remarks at the Global Cooling Pledge Assembly on 17 September 2026, the Ministry of Sustainability and the Environment highlighted the role of advanced simulations, smart cooling and district-level approaches. The remarks also referenced Punggol Digital District’s centralised district cooling system as being 30% more energy efficient than conventional individual systems.

For facility managers, warehouse operators, building owners and SMEs, the practical question is not whether every building needs a complex digital platform. It is whether the building has reliable operational data to support better cooling decisions.

A sensor-led approach provides a sensible starting point. It improves visibility of ACMV performance, helps teams identify abnormal conditions earlier and creates a stronger foundation for automation or AI analytics later.

Why cooling resilience begins with data

Cooling resilience means more than keeping indoor spaces cold. It includes maintaining acceptable comfort, protecting equipment and stock, supporting indoor air quality, managing peak demand and recovering quickly when a component or control point fails.

Without dependable data, teams may rely on complaints, manual checks or isolated equipment readings. These methods can miss gradual changes such as a drifting temperature sensor, reduced airflow, fouled filters, poor chilled-water temperature difference or simultaneous heating and cooling.

Data does not replace engineering judgement. It helps engineers and facilities teams see where judgement is needed first.

1. Start with a practical sensor coverage review

Before buying more sensors, map the systems, zones and operating decisions that matter. A useful review should identify:

  • Air-conditioning zones, occupied areas and critical rooms.
  • Air handling units, fan coil units, ventilation fans and major exhaust systems.
  • Chilled-water pumps, valves, cooling coils, flow points and temperature measurements.
  • Outdoor-air intake points, return-air paths and areas with variable occupancy.
  • Existing building management system points, meters, alarms and manual logs.
  • Locations where temperature, humidity, carbon dioxide or differential pressure may support operational decisions.

Sensor placement should reflect the question being asked. A single room sensor may not represent a large warehouse, office floor or mixed-use area. Likewise, a supply-air temperature reading alone cannot confirm that occupied spaces are receiving suitable airflow.

Facilities should also check sensor condition, calibration status, communication reliability, time synchronisation and data gaps. Poor-quality data can create false alarms and reduce confidence in later analytics.

2. Make ventilation visible without compromising comfort

Ventilation monitoring should be considered alongside thermal comfort and energy performance. NEA guidance on improving ventilation and indoor air quality discusses adequate outdoor-air supply, ACMV maintenance, filtration, airflow and the consideration of sensors.

In practice, teams can begin by tracking operating status, outdoor-air damper position where available, fan status, filter condition indicators, supply and return conditions, and selected indoor-air-quality indicators appropriate to the space. Carbon dioxide trends may help identify changes in occupancy or ventilation performance, but they should not be treated as the only measure of indoor air quality.

Ventilation controls should not be adjusted solely to reduce energy use. Any change should consider occupancy, space use, humidity, pressure relationships, filtration, health requirements and the building’s operating procedures. A control strategy that saves energy but creates discomfort or inadequate ventilation is not a resilient solution.

3. Combine chilled-water and air-side data

ACMV optimisation is stronger when water-side and air-side information are reviewed together.

On the chilled-water side, useful points may include chilled-water supply and return temperatures, flow, pump status, valve positions, differential pressure and plant or system electrical consumption where metering is available. These readings can help teams understand whether cooling demand is being met efficiently or whether pumps, valves and coils are operating outside expected conditions.

On the air side, consider supply-air and return-air temperatures, fan status, fan speed, static pressure, damper position, zone conditions and operating schedules. The objective is not to collect every possible point. It is to connect equipment behaviour to the conditions experienced by occupants, processes or stored goods.

For warehouses and other operational facilities, the monitoring plan should reflect loading patterns, door openings, high-bay spaces, process heat, cold-chain requirements where applicable and areas with uneven temperature distribution.

4. Integrate controls carefully

Data becomes useful when it can be interpreted alongside schedules, setpoints, alarms and control sequences. Where a building management system is available, confirm which points are read-only and which can be adjusted. Document control ownership, override procedures and escalation responsibilities.

Integration should be staged. Begin with reliable monitoring and dashboards. Next, introduce rule-based alarms or recommendations. Only after the team has validated the data and response process should automated control changes be considered.

Examples of practical rule-based checks include:

  • A fan operating while the associated schedule indicates the zone is unoccupied.
  • A valve remaining near fully open while the zone temperature does not respond as expected.
  • Unusual differences between chilled-water supply and return temperatures.
  • Temperature or humidity drifting outside an agreed operating band.
  • Repeated alarms, overrides or communication failures on the same equipment.

These checks should support, not bypass, engineering review. Automatic changes should include limits, fallback modes and a clear method to return to a safe manual or proven control state.

5. Add comfort and operational safeguards

Optimisation should be measured against more than energy consumption. Establish operating boundaries for temperature, humidity, ventilation indicators, pressure relationships and critical process conditions. The appropriate boundaries will depend on the building, occupants, equipment and business activity.

Keep a record of complaints, hot or cold spots, equipment trips, maintenance findings and production or storage issues. These operational observations provide important context when reviewing trends.

Facilities teams should also define what happens when sensors fail. A missing value should not automatically trigger aggressive control action. The system should identify stale or implausible data, use a documented fallback and notify the responsible team.

6. Use measurement and verification to prove improvement

Before changing controls, establish a baseline. Record the period, operating hours, weather conditions where relevant, occupancy or production activity, major equipment status and energy consumption available from the site.

After implementation, compare like with like. Review energy use together with comfort, ventilation indicators, complaints, alarms, maintenance events and equipment runtime. A reduction in energy use may not represent a successful outcome if it results from reduced operating hours, lower occupancy or a degraded indoor environment.

Measurement and verification does not need to begin with a complex analytics programme. A consistent monthly review, supported by reliable sub-metering and operational logs, can already reveal whether a change is delivering the intended result.

7. Add AI only after the data foundation is ready

AI analytics can help identify patterns across many equipment points, prioritise likely faults, detect unusual energy behaviour and support predictive maintenance workflows. However, AI is not a substitute for sensor coverage, sound controls or a clear response process.

A sensible adoption path is:

  1. Stabilise the data: confirm point naming, timestamps, units, sensor reliability and communication health.
  2. Build operational visibility: create dashboards and exception reports that facilities teams can understand and act on.
  3. Validate rules: test straightforward fault and performance checks against maintenance findings.
  4. Apply analytics: use AI to identify patterns, rank priorities and support investigation.
  5. Review automation: automate only those actions with clear limits, approval paths and safe fallback conditions.

This staged approach reduces the risk of installing advanced tools before the building is ready to use them. It also helps decision-makers demonstrate value through practical outcomes such as faster fault response, improved comfort consistency, more transparent maintenance priorities and better energy visibility.

A practical next step for Singapore facilities

Cooling resilience is an engineering and operational discipline. Whether a facility uses individual ACMV systems, central plant or participates in a district cooling arrangement, the first step is to understand what is happening across the building and whether the available data can support action.

ISS can support organisations with engineering, facility management and AI automation requirements, including sensor and data reviews, ACMV performance monitoring, control-system integration planning and staged analytics adoption.

Contact ISS to discuss your facility’s cooling resilience, operational visibility and digital optimisation requirements. Visit intelligencesolutionservice.com.

Reference points