The best chiller plant optimization software, and what to look for
Milvian AI teamFacilities and infrastructure intelligence
The best chiller plant optimization software measures chiller, tower and pump power against load and weather, tracks plant efficiency in kW per ton continuously, finds sequencing and setpoint faults, baselines the plant so every change is verified, reads the existing plant controls safely, and acts only within limits the owner sets, with lockouts and a full command history.
Why chiller plants are a hard case for optimization
In cooling-dominated buildings the chilled water plant is often the largest single energy user, and it is a system, not a machine. Chillers, primary and secondary pumps, condenser pumps, cooling towers and the air handlers they serve all interact. Lowering condenser water temperature helps the chiller but costs tower fan energy. Raising chilled water temperature helps the chiller but may push air handler fans harder. The best operating point moves with load and weather every hour.
Most plants run on sequences written at commissioning and adjusted by hand since. Chillers stage on conservatively, setpoints drift, a tower fan runs at full speed out of habit, and valves throttle against constant speed pumps. Nothing alarms, because the building stays cool.
Plant teams are also rightly cautious. A chiller trip in a hospital, data center or terminal is a serious event, so any software that touches the plant must earn trust slowly and prove what it did.
What the best chiller plant optimization software must do
| Requirement | Why it matters in a chiller plant | Question to ask a vendor |
|---|---|---|
| Component power metering | Plant efficiency needs chiller, pump and tower power, not just the chiller. | Do you meter every chiller, pump and tower fan? |
| Load and temperature data | kW per ton needs flow and supply and return temperatures. | Where do you get chilled water flow and temperatures, and how accurate are they? |
| Continuous plant efficiency | Efficiency drifts as tubes foul and setpoints creep. | Do you track kW per ton for the plant and each chiller over time? |
| Weather-normalized baseline | A cool month flatters a plant that changed nothing. | Is the plant baseline regressed against degree days and load? |
| Sequencing and setpoint diagnostics | Most waste is in how the plant is run, not the machines. | Can you find out of sequence staging, low chilled water setpoints and tower faults? |
| Safe integration | Plant controls keep the building cool and must not be disturbed. | Is the connection read-only until we approve specific writes? |
| Guardrails on control | Short cycling and trips damage equipment. | Do you have lockouts, limits and an audit of every command? |
| Verification | Every change should prove itself. | How do you verify savings against the plant baseline? |
The categories plant teams compare
- Chiller manufacturer plant controllersKnow their own machines very well and sequence them reliably. Some are less open about mixed fleets, pumps and towers from other makers.
- Traditional BMSRuns the plant sequence and alarms. Rarely tracks kW per ton continuously or verifies savings against weather.
- Standalone plant optimization toolsFocused on automated chiller plant control. Strong on the plant itself, often narrow beyond it.
- IoT monitoring platformsAdd sensors quickly. Usually lack the plant physics needed to explain a drop in efficiency.
- Retro-commissioning consultantsValuable for a deep one-off study and tuning. Results drift once the consultant leaves.
- Facility intelligence platformsMeasure the plant alongside the air side, water and the electrical system, verify every change, and act only within approved limits.
How to evaluate chiller plant optimization software
- Survey the plant and its instrumentsList chillers, pumps, towers, VFDs, flow meters and temperature sensors, and check which are calibrated and which are missing.
- Fill the measurement gapsAdd power meters and flow or temperature sensors where needed. Without trusted load data, kW per ton is a guess.
- Baseline before changing anythingLet the plant run as it is long enough to build a baseline against load and weather, or use historical bills where the plant is separately metered.
- Start in advisory modeHave the software recommend sequencing and setpoint changes, and let operators apply them while results are verified.
- Enable control in stagesOnly after trust is built, allow limited automatic changes on approved equipment, with lockouts and full command history.
Chiller plant optimization scorecard
| Criterion | Weight | What good looks like |
|---|---|---|
| Measurement completeness | High | Power for every chiller, pump and tower, trusted flow and temperatures |
| Efficiency tracking | High | Plant and chiller kW per ton trended continuously |
| Diagnostics | High | Sequencing, setpoint and tower faults named with causes |
| Safety and guardrails | High | Read-only by default, lockouts, limits, command history |
| Verification | Medium | Results measured against a weather and load adjusted baseline |
| Fleet neutrality | Medium | Works with mixed chiller, pump and tower makers |
| Air side context | Medium | AHU and zone data beside the plant |
| Water context | Low | Tower make-up and blowdown measured |
Integration and data in the plant room
Most plants already have a plant controller or BMS, chiller panels with Modbus or BACnet interfaces, and VFDs on pumps and tower fans. The software should read all of them, read-only to start, and add power meters where the electrical picture is incomplete. A field device network controller can then carry approved writes over BACnet or Modbus when the time comes.
Flow and temperature data deserve a hard look. A drifting temperature sensor or an uncalibrated flow meter will produce a confident but wrong kW per ton. Good software shows sensor health and flags readings that do not agree with the energy balance.
The air side belongs in the same model. Raising chilled water temperature is one of the most common plant savings, but it only works if air handlers can still meet their loads. Software that sees valve positions, supply air temperatures and zone conditions alongside the plant can tell when a reset has gone too far, before occupants or a critical space notice.
Standards and reporting
ASHRAE 90.1 sets efficiency requirements for chillers and plant design in many jurisdictions, and measurement and verification of plant savings commonly follows IPMVP principles. ISO 50001 programs often name the chiller plant as a significant energy use with its own indicators. Plant water use feeds into water audits such as AWWA M36 and certification programs such as LEED.
The software supplies the measured indicators and the verification. Engineers, auditors and certifying bodies make the judgments.
Where Aqueduct fits
aQ Core maps the plant as devices in a site model. aQ EnergySmart meters each chiller, pump and tower fan, places them in the electrical digital twin and baselines the plant against cooling degree days, so every change is verified and priced through rate configuration. aQ WaterGuard measures tower make-up and blowdown on the same data layer.
Where control is enabled, with owner approval, Smart Controls reaches chiller plants, AHUs and VAV boxes through a field device network controller over BACnet or Modbus. Control clusters let sensors and plant operate and be analyzed as one system, lockout warnings block new commands while one is still being acted on, and command history shows whether every command executed.
A first engagement usually runs in advisory mode. The plant is metered and read, the baseline forms, and Aqueduct recommends sequencing and setpoint changes that operators apply themselves. Each change is verified against the baseline. Automatic control on specific equipment comes later, only where the owner enables it, with lockouts and command history in place.
