Singapore Haze-Ready Buildings: Using AI Indoor-Air-Quality Monitoring to Automate ACMV Response

A practical framework for turning air-quality data into faster, consistent and auditable ACMV and facility-management actions.

Why haze readiness needs more than manual checks

For Singapore facility managers, building owners, warehouse operators and SMEs, haze readiness is an operational issue as well as an occupant-protection responsibility. When outdoor air quality deteriorates, teams may need to review outdoor-air intake, verify ACMV operation, monitor indoor conditions, manage filters and communicate with occupants quickly.

Manual checks can help, but they may be inconsistent across shifts, buildings and operating teams. A connected indoor-air-quality monitoring system can provide a clearer operating picture and trigger defined actions when conditions change. With suitable automation, the facility does not simply display sensor readings; it creates alerts, routes tasks, records decisions and supports a repeatable response.

Singapore’s recent haze-risk planning and updated guidance on ventilation and indoor air quality reinforce the value of preparedness. Building owners and facilities managers should consider how their ACMV systems, monitoring arrangements and response procedures work together before poor outdoor air quality occurs.

What a haze-ready monitoring system should measure

A practical system should combine several data sources instead of relying on a single indoor sensor:

• Indoor PM2.5: Helps identify particulate conditions in occupied areas, loading zones, offices, production areas or other sensitive spaces.
• CO2: Provides a useful indicator of occupancy-related ventilation conditions. It should be interpreted with the building’s use, ventilation design and operating schedule in mind.
• Outdoor-air-quality data: External readings can provide context for haze conditions and support decisions about outdoor-air intake and operating modes. Outdoor readings are not a direct substitute for indoor measurements.
• ACMV data: Fan status, operating mode, damper position, pressure or differential-pressure readings, filter condition indicators and alarms can help determine whether the system is responding as intended.
• Occupancy and schedule data: Work hours, tenant schedules, warehouse shifts and room use can help reduce false alarms and make responses more relevant.

Measurements should be reviewed for sensor placement, calibration, maintenance and data quality. A poorly located or neglected sensor can produce a misleading picture, even when the dashboard looks technically advanced.

How AI and automation improve the response

AI should be applied as a decision-support and workflow layer around sound engineering controls. It should not be treated as a replacement for qualified ACMV assessment or site-specific operating procedures.

For example, an AI-enabled platform can identify patterns such as rising indoor PM2.5 after a change in outdoor conditions, a repeated increase in particulate levels near a loading area, or a mismatch between a commanded ACMV mode and actual fan or damper status. It can then recommend or initiate the next approved workflow, depending on the control permissions defined by the facility.

Useful automation actions may include:

1. Detect: Compare indoor PM2.5, CO2, outdoor-air-quality inputs and ACMV signals against approved site thresholds and operating rules.
2. Alert: Notify the facility manager, duty engineer or operations team through the agreed channel.
3. Recommend: Present the relevant response playbook, such as checking doors, reviewing outdoor-air intake, inspecting filters or verifying equipment status.
4. Act: Where the system has been engineered and authorised to do so, change a predefined ACMV operating mode or issue a command through the building-management system.
5. Escalate: Route unresolved or worsening conditions to the responsible person, with time-based reminders.
6. Record: Store the reading, alert, acknowledgement, action, person responsible and resolution status for later review.

For safety and reliability, automated control should use permissions, interlocks, manual override options and clear fallback procedures. A facility may choose to automate alerts and work orders first, then introduce limited operating-mode changes after testing.

Example haze-response workflow

A simple workflow can be structured around verified conditions rather than one generic alarm. Consider an office, warehouse or mixed-use facility with indoor PM2.5 sensors, CO2 sensors and ACMV integration.

When outdoor conditions worsen, the system checks whether indoor PM2.5 is also rising. If indoor readings remain normal, the system may issue an advisory and continue monitoring rather than creating unnecessary disruption. If indoor PM2.5 exceeds the site’s approved action level, the system can notify the duty team, display the affected zones and create an inspection task.

The task may require the team to verify doors and openings, review outdoor-air supply, check the selected ACMV mode, inspect filter differential pressure and confirm that equipment alarms are clear. If the site procedure permits a mode change, the system may apply a predefined setting or request approval before doing so.

If CO2 then rises because outdoor-air supply has been reduced, the system should not treat this as a simple success or failure. It should show the trade-off for review by the responsible engineering or facilities team. Haze response must protect indoor-air quality while maintaining suitable ventilation and occupancy conditions. The correct action depends on the building, system design, occupants and approved procedures.

Haze response is different from general energy optimisation

Energy optimisation normally focuses on reducing consumption while maintaining comfort and operating requirements. Haze response has a different priority: maintaining a controlled indoor environment and ensuring that the facility can respond consistently when outdoor air quality is poor.

These objectives can overlap, but they should not be combined into one uncontrolled automation rule. For example, reducing outdoor-air intake may affect energy use, but the decision must consider ventilation, indoor CO2, space usage, equipment limitations and applicable guidance. Similarly, increasing fan operation may improve filtration performance in some systems but could create other engineering or maintenance considerations.

A haze-ready strategy therefore needs clearly separated operating modes, such as normal operation, heightened monitoring, haze response, recovery and manual intervention. Each mode should define who can activate it, what equipment changes are permitted, which alarms are generated and when the facility returns to normal operation.

Filter maintenance and evidence matter

Monitoring is only useful if the facility can act on the findings. Automated alerts should connect to maintenance workflows rather than remain as dashboard notifications.

When particulate readings rise or filter differential pressure changes, the system can create a work order for inspection. The workflow can capture the location, equipment reference, sensor readings, technician acknowledgement, inspection result, filter replacement or cleaning action and closure time. Photo evidence or service notes may also be included where appropriate.

This creates a practical record for internal review and supports more consistent handovers between facility teams, contractors and management. It also helps identify recurring issues, such as a particular zone with infiltration, an ACMV unit that repeatedly alarms or a maintenance interval that needs review.

A practical implementation path for SMEs and larger facilities

Facilities do not need to begin with a complex, fully integrated platform. A staged approach can reduce risk and demonstrate value:

• Stage 1 — Assess: Map critical zones, existing sensors, ACMV equipment, outdoor-air intakes, occupancy patterns and current haze procedures.
• Stage 2 — Pilot: Install or connect monitoring in selected representative areas, such as an office floor, warehouse zone or sensitive room.
• Stage 3 — Define playbooks: Agree alert levels, responsible persons, escalation times, approved ACMV actions and manual fallback procedures.
• Stage 4 — Integrate: Connect the monitoring layer with dashboards, work-order systems, messaging tools or building-management systems where suitable.
• Stage 5 — Test and improve: Run drills, review false alarms, verify sensor performance and update procedures after actual events or maintenance findings.

Useful measures include alert acknowledgement time, response completion time, percentage of actions closed with evidence, sensor availability, recurring alarm locations and the number of unresolved ACMV or filter-related issues. These indicators help turn haze readiness into a measurable Smart FM activity.

Prepare before the next haze event

A connected monitoring system cannot compensate for inadequate ACMV maintenance, poor sensor placement or unclear responsibilities. It works best when engineering controls, operating procedures, trained personnel and digital workflows are designed together.

For Singapore businesses, a practical first step is to review how indoor-air-quality data is currently collected, who receives an alert, what action follows and how the outcome is recorded. ISS can help organisations discuss engineering, facility management and AI automation requirements, from pilot monitoring and dashboards to workflow integration and controlled ACMV response.

Contact ISS to discuss how your building, warehouse or SME facility can become more haze-ready with measurable and auditable indoor-air-quality response processes.