An underground MRT station is a thermal system full of people, lights, equipment, outdoor air and train-generated heat. Comfort depends on moving heat and moisture out at the same rate the station keeps creating them.
The best HVAC system is not the coldest one. It is the one that keeps passengers comfortable with the least energy consistent with humidity control, air movement and the station’s public job.
The Tunnel Ventilation and Airflow pillar owns train-piston effects, tunnel pressure and tunnel airflow. The Train HVAC pillar owns the carriage thermal envelope.
This article owns the station thermal environment: passenger heat, equipment heat, outside-air loads, sensible and latent cooling, humidity, air distribution, HVAC energy, control setpoints and measured comfort.
LTA’s public green-station material identifies air-conditioning as one of the major energy challenges for underground MRT stations. SMRT’s Intelligent Facilities Management programme has publicly reported using predictive AI to optimise station HVAC setpoints while maintaining commuter comfort, with trial energy savings in the 5–10% range.
The RFE — What Is Station HVAC For?
remove heat and moisture from station spaces at a rate sufficient to preserve passenger comfort, equipment operability and usable airflow while avoiding cooling energy that adds no public value.
Prompt 1 — Where Does the Heat Come From?
A simplified station heat balance is:
Q̇total = Q̇people + Q̇lighting + Q̇equipment + Q̇outside-air + Q̇conduction + Q̇train/interface − Q̇cooling
If Q̇total is positive, the station warms.
Passenger heat alone is time-dependent:
Q̇people(t)=N(t) q̇person
A crowded interchange therefore creates a different cooling problem from the same station at midday.
Prompt 2 — What Is Sensible Cooling?
Sensible cooling changes air temperature.
Q̇sensible = ṁair cp (Tin−Tout)
More airflow or a larger temperature difference increases sensible cooling capacity.
But high airflow costs fan energy and can create draught discomfort.
So cooling is an optimisation, not a maximum-flow problem.
Prompt 3 — Why Does Humidity Matter?
Singapore’s outdoor air contains substantial water vapour.
Latent cooling removes moisture.
Q̇latent = ṁdry-air (ωin−ωout) hfg
ω is humidity ratio and hfg the latent heat associated with condensation.
Two stations at the same dry-bulb temperature can feel very different if humidity differs.
Thermal comfort is not a thermometer reading. It is a combined state of temperature, humidity, air speed, clothing, activity and expectation.
Prompt 4 — How Does Outdoor Air Enter the Cooling Equation?
Outdoor air is necessary for ventilation and indoor-air quality, but it carries heat and moisture.
Using air enthalpy h:
Q̇outside = ṁoa (hout−hin)
The hotter and more humid the outdoor air, the larger the conditioning load for the same ventilation flow.
Station entrances, doors and pressure effects also change infiltration.
The tunnel-airflow pillar owns the larger tunnel pressure field; this pillar owns its thermal consequence once that air enters the station environment.
Prompt 5 — How Much Electrical Power Does Cooling Need?
Coefficient of performance is:
COP = Q̇cooling/Pelectrical
So:
Pelectrical = Q̇cooling/COP
Higher COP means less electrical power for the same cooling duty.
Fan power also matters. A simplified fan relation is:
Pfan = Δp Q/ηfan
Airflow therefore has both thermal benefit and electrical cost.
Prompt 6 — Why Can a Small Setpoint Change Save So Much Energy?
If indoor target temperature is lowered, the system must usually remove more sensible heat and operate cooling equipment harder.
Conceptually:
cooling energy ≈ ∫ f(outdoor state, passenger load, equipment load, setpoint) dt
A setpoint that is unnecessarily cold multiplies across long station operating hours.
SMRT’s public IFM programme demonstrates the systems idea: predictive control adjusts HVAC setpoints to reduce energy while maintaining commuter comfort rather than holding one fixed aggressive cooling state all day.
Prompt 7 — How Does Air Distribution Affect Comfort?
Average station temperature can look acceptable while one passenger zone feels warm or draughty.
Let zone temperature be Ti.
Tavg = Σ wi Ti
A useful uniformity measure is:
σT = √[Σwi(Ti−Tavg)²]
A lower σT generally means more spatially uniform thermal conditions.
But uniformity is not sufficient if the average itself is uncomfortable.
Prompt 8 — How Does World Return Improve Station HVAC?
Useful signals include:
- zone temperature;
- humidity;
- airflow;
- equipment electrical power;
- passenger density proxy;
- outdoor weather;
- complaint or comfort feedback;
- equipment condition.
A temperature prediction residual is:
eT = Tmeasured−Tpredicted
Energy residual is:
eE = Emeasured−Epredicted
If comfort remains stable while eE falls after control optimisation, the intervention is working.
If energy falls but warm-zone complaints rise, the optimisation has deleted part of the public job.
A Fictional Station HVAC Example
Suppose a fictional station peak includes 2,000 passengers, each contributing an average 100 W of sensible heat:
Q̇people=2,000×100=200kW
Equipment and lighting add 120 kW and outside-air load adds 180 kW.
Q̇cooling≈500kW
At COP=5:
Pcooling≈100kW
If predictive control reduces required average cooling load by 7% without changing comfort:
electrical saving≈7kW during that state
Multiplied across many hours and stations, small percentages become large annual energy effects.
Deletion Tests
- Remove passenger heat: crowding has no thermal consequence.
- Remove humidity: Singapore air can be cooled without moisture removal.
- Remove outdoor air: ventilation introduces no energy burden.
- Remove COP: all cooling plants use the same electricity.
- Remove fan power: air distribution is energetically free.
- Remove spatial variation: one temperature sensor describes the whole station perfectly.
- Remove comfort feedback: energy alone defines success.
- Remove World Return: control optimisation never learns whether passengers or meters agreed.
Station-HVAC Paradoxes
- The station can save energy by cooling less without becoming hotter to passengers if control improves.
- A colder setpoint can produce worse comfort if air speed becomes unpleasant.
- More outdoor air can improve air quality while increasing cooling energy.
- A station can have the right average temperature and the wrong local thermal experience.
- The best thermal optimisation is one passengers barely notice.
The Station-HVAC Audit
- What sensible heat enters the station?
- What latent moisture load enters?
- How does passenger density change both?
- What outside-air load follows current weather?
- What airflow and supply conditions are used?
- What cooling COP and fan efficiency apply?
- How uniform are temperature and humidity across passenger zones?
- What setpoint changes reduce energy without weakening comfort?
- What equipment-condition signals change efficiency?
- What measured energy and comfort return proves the optimisation worked?
World Return — The Passenger and the Meter Must Agree
predict heat and moisture load → operate HVAC → passengers and weather change → measure temperature, humidity and energy → adjust controls → compare comfort and consumption → maintain equipment → repeat
MRT station HVAC works when heat and humidity are removed only as aggressively as the real passenger environment requires—and when energy falls without comfort disappearing.
Reader-safety note: This article does not publish station-specific ventilation layouts, emergency ventilation logic, plant-room arrangements, control thresholds or operational setpoints. The numerical example is fictional.