Closed-loop · plant controls itself

Chiller plant — escalation under a load & weather spike

You set only the triggers (demand + weather). The plant decides how to respond — pump, chiller, tower, staging — in real time.

demandBaseline load280 kW

weatherWet-bulb24 °C

Predictive (knows the spike is coming)

Plant response (computed live)

Pump speed

80%

Chillers on

Tower fans

50%

CHWS (set 7)

7.0 °C

ΔT

5.0 °C

Plant power

— kW

Chillers & capacity vs demand

Chiller 1

off

Chiller 2

off

Chiller 3

off

available —demand —

Escalation ladder

1Pump ramps up — more flow to carry the heat, holds ΔT

2Chiller loads up — compressor works harder to hold CHWS

3Tower fans ramp — reject extra heat, fight rising lift

4Stage up a chiller — capacity limit reached

5Loss of control — CHWS drifts above setpoint, comfort slips

CHWS over time (setpoint 7 °C, comfort limit 9 °C)

Hit Occupancy spike and watch the ladder climb. Toggle Predictive on, reset, and spike again — the plant pre-stages before the load arrives, so CHWS barely moves. That gap between reactive and predictive is the core value of AI optimization.