Data center cooling is entering a new era. Higher rack densities, accelerated computing and the growing use of liquid cooling are placing greater demands on chilled-water pumps, condenser-water pumps, cooling-tower fans, air-handling units and coolant distribution units.
AC drives, also known as variable frequency drives, or VFDs, allow these systems to adjust motor speed as cooling demand changes. When properly specified, they can improve energy efficiency, strengthen control and reduce mechanical stress. When poorly matched to the application, however, they can introduce reliability, power-quality and integration problems.
The right selection process therefore goes beyond matching a drive’s horsepower to the motor nameplate. Engineers should evaluate the complete cooling system, its operating environment and the data center’s availability requirements.
1. Start with the cooling application
First, identify what the drive will control and how that equipment behaves under real operating conditions. Common applications include:
- Chilled-water and condenser-water pumps
- Cooling-tower fans
- CRAH and air-handling-unit fans
- Dry coolers and fluid coolers
- Pumps in coolant distribution units
- Other secondary cooling-loop pumps
Centrifugal fans and pumps are especially well suited to variable-speed control. Their power requirement falls rapidly as speed is reduced. As a general illustration, the U.S. Department of Energy notes that a 10% reduction in speed can produce approximately a 30% reduction in power for centrifugal pumps, fans and compressors, depending on the system.
Actual savings depend on the system curve, static head, operating hours, control sequence and part-load profile. Engineers should use the expected load profile, not simply the design-day condition, to estimate drive size, energy performance and potential return on investment.
For pumps, confirm the required flow and pressure across every operating mode. For fans, evaluate airflow, static pressure and the minimum speed needed for stable operation. Compressor applications require additional care and should use a drive and control strategy approved for the specific compressor technology.
2. Size the drive by current and duty, not horsepower alone
Motor horsepower is a useful starting point, but output-current capability is the more important selection criterion. Compare the motor’s full-load current with the drive’s continuous output-current rating at the actual site conditions.
The specification should address:
- Motor voltage, frequency and full-load current
- Number and type of motors connected
- Variable-torque or constant-torque duty
- Required starting and accelerating torque
- Expected maximum and minimum speeds
- Acceleration and deceleration times
- Potential overloads or process transients
- Whether the motor may operate above base speed
Most centrifugal cooling fans and pumps are variable-torque loads. A drive rated for variable-torque duty can often provide the appropriate performance without the cost and footprint associated with a heavier constant-torque rating. Applications with unusually high starting torque, rapid acceleration or frequent load changes may require additional capacity.
Avoid oversizing “just in case.” A substantially oversized drive can increase cost, occupy more panel space and operate less effectively at very light loads. Where future capacity is anticipated, define that expansion explicitly and select both the motor and drive around the planned duty point.
3. Account for the installation environment
Published ratings apply under stated environmental conditions. Data center mechanical spaces can exceed those assumptions, particularly in rooftop equipment, crowded electrical rooms and enclosed pump skids.
Check:
- Maximum and minimum ambient temperature
- Installation altitude
- Humidity and condensation risk
- Dust, water and contaminant exposure
- Indoor or outdoor location
- Enclosure type and cooling method
- Required clearances and ventilation
- Heat released into the electrical room
Apply the manufacturer’s required derating for temperature, altitude, enclosure type and switching conditions. Do not treat thermal management as an afterthought: the drive itself produces heat, and that heat must be included in room and enclosure cooling calculations
Fuji Electric’s FRENIC-MEGA (G2) platform, for example, is offered across 208/230 V, 460 V and 575 V classes, with enclosure and derating requirements that vary by rating and installation. The current specification sheet should always be consulted during final selection.
4. Define the control strategy before selecting features
A drive should support the intended sequence of operations without unnecessary external hardware. Useful functions for data center cooling can include:
- PID control for pressure, temperature or flow
- Sleep and wake functions
- Minimum- and maximum-speed limits
- Controlled acceleration and deceleration
- Flying start for a motor that is already rotating
- Automatic restart after a permitted interruption
- Pump alternation or staging
- Broken-belt, loss-of-load or dry-pump detection
- Skip frequencies to avoid mechanical resonance
- Real-time scheduling
- Local and remote operating modes
The control narrative should also describe sensor failure, loss of communication and controller failure. In a mission-critical cooling system, the safest fallback may be a predefined speed rather than an immediate stop. That decision should be made through a system-level risk assessment.
Multiple sensors may be required where a single pressure or temperature reading cannot accurately represent cooling demand. Stable sensor placement and sensible PID tuning are just as important as the drive hardware; an unstable loop can cause hunting, excess energy use and premature equipment wear.
5. Integrate the drive with the building management system
Cooling equipment rarely operates in isolation. Drives should exchange meaningful information with the building management system, electrical power monitoring system or supervisory cooling controls.
Define the required communications protocol and points list during design. Typical data points include:
- Run and ready status
- Speed command and feedback
- Output frequency and current
- Power and energy
- Alarm and fault codes
- Operating hours
- Local/remote status
- Maintenance indicators
Fuji Electric’s FRENIC-MEGA (G2) includes RS-485/Modbus and supports common building-automation connectivity, including BACnet MS/TP and Metasys N2. [2] Confirm the exact protocol, interface hardware and point mapping required for each project.
Hardwired enable, run-status and critical-alarm signals may still be appropriate even when network communications are used. This gives essential functions a path that does not depend entirely on the control network.
6. Address harmonics, electromagnetic compatibility and motor protection
AC drives are nonlinear loads, so their cumulative effect on the electrical system must be evaluated. A data center may contain dozens or hundreds of drives alongside UPS systems, power supplies and other power-electronic equipment.
The design team should assess:
- Harmonic distortion at the point of common coupling
- Applicable project or utility power-quality limits
- Input reactors, DC-link reactors or harmonic filters
- Electromagnetic interference
- Grounding and cable shielding
- Motor-cable length
- Reflected-wave voltage at the motor
- Common-mode current and bearing-current risk
Long motor leads can increase voltage stress on motor insulation. Depending on cable length, motor design and switching conditions, an output reactor, dV/dt filter or sine-wave filter may be needed. Inverter-duty motors and appropriate insulation systems should be considered, especially for higher voltages, long conductors or continuous low-speed operation.
Liquid-cooling systems deserve particular attention because drives used with coolant distribution unit pumps can add to harmonic distortion. ASHRAE identifies power quality as one of the issues to consider when modernizing facilities for higher-density and liquid-cooled computing.
A project-level harmonic study is preferable to specifying mitigation devices independently for every drive. The study can identify the most effective combination of drive topology, reactors, filters and system impedance.
7. Engineer for availability and maintainability
Not every cooling motor requires the same redundancy strategy. The appropriate design depends on the facility topology, such as N, N+1, 2N or distributed redundancy.
Key questions include:
- Can another fan or pump carry the load if this unit is unavailable?
- Is a bypass required?
- Will bypass operation provide useful cooling without variable-speed control?
- Can the drive be isolated and replaced safely?
- Are critical spares available on site?
- Can one drive rating serve several standardized motor sizes?
- How quickly can trained support be reached?
A bypass is not automatically the best answer. It adds components, wiring and failure modes, while operating a fan or pump at full speed may disrupt pressure or flow control. In systems with equipment-level redundancy, a spare drive or quick-replacement strategy may provide a cleaner solution. The choice should follow a failure-mode analysis of the cooling architecture.
Specifications should also require accessible fault histories, replaceable cooling fans where applicable, clear diagnostic information and retention or backup of parameter settings.
8. Coordinate protection and emergency behavior
The drive, motor, upstream protective device and control system must be coordinated as one assembly. Define short-circuit protection, disconnecting means, branch-circuit requirements, ground-fault behavior and any safety or fire-alarm interfaces in accordance with the applicable codes and project standards.
Also document what should happen after:
- A brief voltage dip
- Complete power loss
- Generator transfer
- Loss and restoration of a control signal
- Fire-alarm activation
- Emergency power-off
- Motor or sensor failure
Automatic restart can help restore cooling quickly, but it must be applied only where restarting machinery is safe and permitted. Restart delays may need to be staggered to prevent a large block of cooling equipment from returning simultaneously after a power event.
9. Specify commissioning, not just equipment
Even a correctly selected drive can underperform if it is commissioned with default parameters. The project specification should include:
- Verification of motor nameplate data
- Direction-of-rotation checks
- Minimum- and maximum-speed settings
- Acceleration and deceleration tuning
- PID-loop tuning
- Resonance checks
- Sensor-loss and communications-loss tests
- Alarm and restart testing
- BMS point-to-point verification
- Measurement at representative load conditions
- Backup of the final parameter set
- Operator and maintenance training
Trend speed, current, power, pressure, flow and temperature during functional testing. This provides evidence that the cooling loop is stable and creates a baseline for future condition monitoring.
A system-level selection
The best AC drive is not necessarily the one with the highest power rating or longest feature list. It is the one that fits the motor, load profile, electrical system, control architecture, environment and availability strategy.
Fuji Electric offers AC drives for the fans and pumps used throughout data center cooling systems, with products spanning fractional through high-horsepower applications. The FRENIC-MEGA (G2) series is designed specifically for fan and pump control, while the broader FRENIC portfolio provides additional choices for varied performance and capacity requirements.
By involving the drive manufacturer early, and treating the drive as part of the complete cooling and power system, designers can achieve responsive control, dependable operation and better lifetime energy performance.