Embedded controller · chillers
Chiller optimization, in stages.
Start with the setpoints your building system already writes: no change to the chiller, no OEM needed. Then go inside the machine with its maker, where one decision sets the compressor, pumps and fans together.
35%less energy per ton-hour than a variable-speed chiller, modeled at the AHRI 550/590 part-load pointsThe results
The stages
Four stages. Only the last two need the chiller's maker.
- Stage 1
Read-only assessment
Your trend data and a physics model of the plant, its decisions set against your history. No command sent.
No OEM needed - Stage 2
Setpoint optimization
Condenser water, chilled water, tower fans and plant pumps, written by your building system. Nothing inside the chiller changes.
No OEM needed - Stage 3
Inside the chiller
Compressor, expansion valve and the chiller's own pumps and fans as one decision, through an interface the maker approves.
With the OEM or its integrator - Stage 4
Built in at the factory
The optimizer ships inside new chillers, commissioned from a model of the machine, licensed per unit.
The OEM's product
In every stage, your plant controls keep staging, towers and targets, and the chiller's own interlocks and pressure limits act whatever the controller does. Each stage goes ahead only with your approval.
Stage 2 · no OEM needed
Colder condenser water is not free.
It saves compressor power, but the towers and pumps spend power to make it. Stage 2 is designed to find the setpoint where their total is lowest, again as load and weather move.
- Condenser water: the entering setpoint, often still set by hand.
- Chilled-water supply: raised when the load allows it.
- Towers and plant pumps: set with the chillers, not apart from them.
Stage 3 · with the OEM · modeled
Against a modern variable-speed chiller: 35% less energy per ton-hour.
The same 30-ton R32 chiller and plant, two controls. Legacy B runs its variable-speed compressor and chilled-water pump on PI loops and switches whole fixed-speed fans. AReCoS sets the compressor, both pumps and all seven dry-cooler fans as one decision.
- 35%less energy per ton-hour, AHRI 550/590 part-load weights: 1.02 to 0.66 kW/ton.
- 7–19%less per ton-hour on two Atlanta office days: 7% on the design day, 19% on a July weekday.
- 5.7 → 1.7 kWmean fan power: seven fans slowed together with the compressor, not switched one by one.
- $1,675a year for one chiller in an Atlanta office, 39% less energy: an estimate at 11.53 ¢/kWh.
Against a staged chiller · two scrolls switched on and off
125 → 1 compressor starts on the design day · 19.7 → 0 an hour at quarter load
AReCoS turns one variable-speed scroll down instead of cycling a tandem pair, holding the leaving water without the start-stop wear. Energy per ton-hour: 33% less on the same weights.
| Load | Outdoor air | Legacy B, kW/ton | AReCoS, kW/ton | Energy per ton-hour |
|---|---|---|---|---|
| 100% · 30 tons | 95 °F | 1.36 | 1.35 | 1% less |
| 75% · 22.5 tons | 80 °F | 1.07 | 0.92 | 14% less |
| 50% · 15 tons | 65 °F | 0.93 | 0.60 | 35% less |
| 25% · 7.5 tons | 55 °F | 1.21 | 0.40 | 67% less |
| AHRI-weighted | 1.02 | 0.66 | 35% less |
Physics-based model results on a 30-ton R32 reference chiller built from public data, both controls from the same start; not a certified IPLV. Office loads: the DOE ASHRAE 90.1-2019 Medium Office prototype in Atlanta (EnergyPlus). Stages 1 and 2 are measured on your own plant.
Stage 1 · the first step
One plant, read-only. A short data sample is enough to begin.
We ask for
Trends every 1 to 5 minutes from one or two chillers: power, load, water temperatures and flows, tower and pump power, wet-bulb. No source code, no network passwords.
You get
Energy per ton-hour across chillers, towers and pumps, against the optimization you already run, and which stage pays first.
Who takes part
AReCoS and your facilities team for stages 1 and 2. The chiller's maker or its integrator joins at stage 3.
Let's talk.
Tell us about your plant, or the chillers you build.
Grand Rapids, Michigan 49546