Why Data Centre PUE Can Worsen During Low IT Load Malaysia
Why Data Centre PUE Can Worsen During Low IT Load Malaysia
Content
Power Usage Effectiveness, commonly known as PUE, is widely used to evaluate data-centre energy efficiency.
A lower PUE generally indicates that a larger proportion of total facility power is reaching IT equipment instead of being consumed by cooling, electrical losses, lighting and other supporting infrastructure.
However, PUE can worsen when IT load is low—even when the data centre has efficient cooling and electrical equipment.
This is particularly important during commissioning, phased expansion and early operation of new data-centre capacity in Malaysia.
What Is PUE?
PUE is calculated as:
PUE = Total Data Centre Energy ÷ IT Equipment Energy
Total facility energy includes:
Servers and IT equipment
Cooling equipment
Pumps and fans
UPS and electrical losses
Lighting
Controls
Other supporting infrastructure
A theoretical PUE of 1.0 would mean that all incoming energy is used by IT equipment with no supporting-energy consumption.
Real facilities require cooling, power distribution and other systems, so their PUE is higher than 1.0.
Simple PUE Example
Assume a data centre uses:
Total facility power: 1,500 kW
IT equipment power: 1,000 kW
The PUE is:
1,500 ÷ 1,000 = 1.50
Now assume the IT load falls to 500 kW, but cooling and supporting systems continue consuming 500 kW.
The total power becomes 1,000 kW:
1,000 ÷ 500 = 2.00
The facility may not have developed a technical fault. The supporting load simply represents a larger proportion of the reduced IT load.
Why Supporting Equipment Does Not Fall Proportionally
Some data-centre systems have a minimum operating demand.
These may include:
Chillers
Cooling towers
Computer-room air conditioners
Computer-room air handlers
Pumps
Fans
UPS systems
Transformers
Controls
Lighting
Security and fire systems
When IT load decreases by 50%, these supporting systems may not reduce their consumption by the same percentage.
This creates poor part-load efficiency and increases PUE.
1. Cooling Equipment May Be Oversized for the Current Load
A data centre may be designed for future capacity but initially operate with only a small amount of installed IT equipment.
The cooling system may still require:
Minimum water flow
Minimum fan speed
Standby equipment
Pump circulation
Humidity control
Pressurisation
Control-system operation
As a result, cooling energy may remain relatively high while the IT denominator remains low.
2. Commissioning Loads Do Not Represent Full Operation
During commissioning, sections of the cooling and electrical infrastructure may operate before the associated server capacity is installed.
Examples include:
Cooling a partially occupied data hall
Operating pumps for testing
Running redundant equipment
Energising UPS systems at low load
Maintaining temperature and humidity in empty white space
Testing alarms and backup systems
PUE calculated during this phase should not be compared directly with a mature, fully occupied facility without explaining the operating condition.
3. UPS Efficiency May Decline at Low Load
UPS efficiency varies with load.
A UPS system designed for a high-capacity data hall may operate less efficiently when only lightly loaded.
Losses may include:
Conversion losses
Transformer losses
Battery-charging demand
Control-system consumption
Redundant UPS modules operating unnecessarily
Modular UPS architecture and load management may improve part-load performance, depending on the design and redundancy requirements.
4. Pumps and Fans May Run at Fixed Speed
A fixed-speed pump or fan may consume unnecessary energy when cooling demand is low.
Variable-speed control can reduce consumption, but only when:
Sensors are accurate
Control logic is correct
Minimum flow is respected
Valves and dampers operate properly
The system is commissioned correctly
A variable-speed drive does not guarantee efficient operation if the control setpoint is unsuitable.
5. Redundancy Carries an Energy Cost
Data centres require high availability.
Redundant systems may include:
N+1 chillers
Redundant pumps
Multiple UPS modules
Standby generators
Duplicate cooling paths
Additional fans
Some standby equipment consumes little energy, while other equipment may be operated continuously to maintain readiness or share load.
PUE improvement must not compromise the redundancy and reliability required by the facility.
6. Humidity Control May Continue at Low IT Load
When server heat output is low, temperature and humidity control may become more difficult rather than easier.
The system may use energy for:
Dehumidification
Reheating
Humidification
Preventing condensation
Maintaining room setpoints
This is especially relevant in Malaysia’s warm and humid climate.
Humidity problems can increase when outdoor air enters the building or when control sequences are not coordinated.
7. Airflow May Not Match Actual Rack Occupancy
Partially populated data halls may experience:
Empty rack positions
Missing blanking panels
Unsealed cable openings
Unbalanced supply airflow
Air recirculation
Bypass air
Excessive fan speed
Cooling equipment may move large quantities of air without effectively cooling active IT equipment.
Simple airflow-management improvements can sometimes reduce supporting energy without changing the main cooling equipment.
8. Poor Sensor Placement Can Increase Cooling Demand
Cooling controls depend on measurements such as:
Rack inlet temperature
Return-air temperature
Relative humidity
Differential pressure
Chilled-water temperature
Flow
Equipment status
A sensor placed in an unrepresentative location may cause overcooling.
Examples include:
Temperature sensor in a local hot spot
Humidity sensor near an outdoor-air leak
Differential-pressure sensor with incorrect tubing
Return-air sensor affected by bypass airflow
Uncalibrated or drifting sensor
Instrumentation should be checked before adjusting major control settings.
9. Metering Boundaries May Be Incorrect
PUE is only meaningful when the total facility and IT energy boundaries are clearly defined.
Possible errors include:
Missing cooling equipment
Including unrelated building loads
Incorrect current-transformer ratio
Double-counting meters
Estimating IT load instead of measuring it
Different time intervals
Meter drift
Communication errors
Incorrect aggregation
A worsening PUE may sometimes be caused by improved data reconciliation that reveals loads previously excluded.
10. Leakage Can Increase Water and Energy Use
Cooling-system leakage may increase:
Pump operation
Water consumption
Treatment-chemical usage
Makeup-water demand
Cooling instability
Small leaks and meter inconsistencies can remain unnoticed without regular trend review.
PUE should be assessed together with:
Water Usage Effectiveness
Makeup-water volume
Cooling load
IT load
Temperature
Flow and pressure
Equipment runtime
PUE Should Not Be Read Alone
A lower PUE does not automatically mean that every operating condition is better.
Facilities should also consider:
Reliability
Water use
Carbon intensity
Cooling resilience
IT utilisation
Rack density
Climate
Maintenance condition
Redundancy
Service availability
For example, reducing fan speed may improve energy use but create unacceptable rack-inlet temperatures if airflow is already poorly distributed.
How to Investigate High PUE at Low Load
Step 1: Confirm the Calculation
Check:
Total facility meter
IT power meter
Measurement interval
Meter calibration
Included and excluded loads
Data completeness
Step 2: Record IT Occupancy
Document:
Installed rack capacity
Active IT load
Average rack density
Data-hall occupancy
Planned future load
This helps distinguish a lightly occupied facility from a genuine efficiency decline.
Step 3: Separate Major Supporting Loads
Measure or estimate:
Chillers
Cooling towers
Pumps
CRAH or CRAC units
UPS losses
Lighting
Other facility loads
Identify which load is not reducing with IT demand.
Step 4: Review Part-Load Operation
Check whether:
Too many chillers are operating
Redundant pumps are sharing a very low load
Fans remain at fixed speed
UPS modules are lightly loaded
Empty halls are fully conditioned
Minimum-flow bypass is excessive
Step 5: Inspect Airflow
Check:
Hot- and cold-aisle separation
Blanking panels
Cable openings
Rack inlet temperature
Return-air path
Differential pressure
Fan speed
Empty rack spaces
Step 6: Verify Sensors
Compare installed sensors with portable instruments.
Check:
Temperature
Relative humidity
Differential pressure
Air velocity
Water temperature
Electrical current
A control change should not be based on an unverified sensor.
Useful Data Centre Measurements
Thermal Imaging
A thermal camera can help identify:
Hot electrical connections
Uneven rack temperatures
Cooling-distribution problems
Overheated cables
Abnormal UPS or panel conditions
Pump and motor heating
Thermal imaging shows surface-temperature patterns and should be combined with direct environmental and electrical measurements.
Temperature and Humidity Logging
A data logger can reveal:
Daily variation
Short-duration excursions
Differences between rack locations
Overcooling
Humidity instability
Changes after control adjustment
Differential Pressure
Differential-pressure measurement helps assess:
Raised-floor pressure
Hot- and cold-aisle containment
Filter restriction
Room pressurisation
Airflow imbalance
Air Velocity
Anemometers can support checks of:
Floor-tile airflow
Cooling-unit discharge
Return-air paths
Local airflow problems
Air velocity alone does not provide total airflow unless the outlet area and measurement method are considered.
Electrical Measurement
Clamp meters and electrical instruments help verify:
Current
Voltage
Load balance
UPS loading
Pump and fan demand
Circuit utilisation
Energy-efficiency analysis should use suitable power or energy measurement where required, rather than current alone.
When High PUE Is Expected
A temporarily high PUE may be understandable during:
Initial commissioning
Low occupancy
Phased server installation
Maintenance
System testing
Redundancy testing
Temporary cooling configuration
Equipment startup
The facility should document the reason and expected improvement as IT load increases or control optimisation is completed.
When High PUE Requires Investigation
Investigate when:
PUE continues rising at a stable IT load
Cooling power increases without a clear reason
Rack temperatures remain low while cooling demand rises
Water usage increases
Pumps or fans operate continuously at high speed
Sensors disagree
Humidity control becomes unstable
Leakage or meter anomalies are suspected
Efficiency does not improve after occupancy increases
Data Centre Testing Instruments from MTM Precision
MTM Precision supplies instruments for data-centre commissioning, inspection and preventive maintenance in Malaysia.
Relevant products include:
NOYAFA NF-522 thermal imaging camera
Temperature and humidity meters
Temperature and humidity data loggers
Differential-pressure meters
Airflow and anemometer instruments
Clamp meters and electrical testers
Water-quality meters for cooling systems
Leak-detection equipment
Vibration meters for pumps and cooling machinery
NOYAFA NF-8209 Pro and NF-8506 network cable testers
NOYAFA NF-IPC728HSO CCTV and network tester
These instruments can support investigation of thermal, airflow, electrical and cooling-system conditions. Formal PUE assessment also requires correctly installed energy metering and a clearly defined measurement boundary.
Contact MTM Precision
MTM Precision Sdn. Bhd. (744811-A)
Tel: 03-8080 7172
WhatsApp: +6016-660 7346
Email: mtmpre@yahoo.com / enquiry@mtmpre.com.my
Website: www.mtmpre.com.my
Visit our Puchong Showroom & Service Centre for data-centre testing instrument selection, product demonstration and technical support.
Operating hours: Monday to Friday, 9:00 AM–6:00 PM.
20 Sep 2026
