Services / How it works

The software behind every service

The Calibrated Digital Twin Behind Every Exergenics Service

Exergenics builds a calibrated digital twin of a facility’s central chilled water plant and simulates it under every operating condition to find better ways to run it, upgrade it, and design what replaces it.

CALIBRATED TWINCT-1CT-2CT-3CONDENSER WATERChiller 1Chiller 2Chiller 3kWr, kWkWr, kWkWr, kWCHILLED WATERFIELD DEMAND, kWr1Physics engine2Machine learning3Predictive control4M&V baselineWET BULB, °C
5–35%
less chilled water plant energy. Published engagements sit between 6.8% and 33.4%, each measured to IPMVP Option B.
10–15%
peak demand reduction, from staging on field demand rather than on chiller loading.
~18 mo
average payback. The full range across published sites is 6–24 months.
None
new hardware, new technicians or new software on site. Your incumbent controls contractor implements.
Architecture
API-first, with pull and push capability
Deployment
Remote and air-gapped from plant control
Calibration
ASHRAE Guideline 14, or calibration fees refunded
Deployed
Healthcare, education, office, retail, infrastructure, hotel and casino assets in Australia and the US

Why central plants need digitisation

Chillers, pumps and cooling towers account for 25–50% of a commercial building’s electricity. Each machine has its own efficiency profile, and every profile depends on what the others are doing. The setpoint that is right for one chiller at one load and one wet-bulb temperature is wrong an hour later.

The plant obeys thermodynamics, so there is an optimal way to run it. Finding that optimum across millions of combinations is a simulation problem, not a rule-of-thumb problem. Simulation and machine learning find those setpoints from day one.

Four layers, one calibrated digital twin

Each layer feeds the next. The output is a control strategy in industry-standard format, with its savings forecast before anything changes on site.

1

Physics engine

Simulates the fluid mechanics and thermodynamics of the plant: mass and energy flows through chillers, pumps, towers and the loops that join them.

2

Machine learning layer

Fits the operational efficiency profile of every chiller, pump and cooling tower to the plant’s own trend data. Thousands of measured operating points per machine, not four rating points.

3

Model predictive control layer

Genetic optimisation across every load, weather and stage the plant will see. Produces the setpoints, staging and sequencing, and the retrofit and design recommendations.

4

Measurement and verification layer

An energy baseline model to IPMVP Option B, Retrofit Isolation. Every forecast is tested against it after twelve months.

Calibration guarantee. The twin is proven against the plant’s own data to ASHRAE Guideline 14, or calibration fees are refunded in full.

Air-gapped. All modelling, tuning and verification is done remotely. Nothing connects to plant control.

API-first. Data arrives from any BMS, historian or data lake. Results are available through the client portal and API.

How the software works

Historical telemetry is collected from the BMS, historian, fault-detection platform or data lake already on site, normalised through the Exergenics ETL process, and used to train equipment efficiency models and to learn the site’s load profile and weather patterns. A new model is built for each building or district cooling system and the central plant that serves it.

Once calibration is complete, a multi-stage optimisation loop solves the plant as one system: the load balancer sets the split between machines, the staging optimiser sets when they come on and off, and the condenser water loop is solved for lowest total power rather than lowest chiller power.

The output is a functional description in industry-standard format, implemented by your incumbent controls contractor in the existing BMS. Recommendations, reports, charts and data are available through the API and the client portal, which can be white-labelled.

After the strategy is implemented, the model keeps ingesting plant data, so the strategy stays current as the plant and its use change.

What data you need

Share historical data through API integration into your data warehouse, historian or fault-detection platform, or as a flat-file export from the BMS. Six months minimum, twelve ideal, at 15-minute intervals or better. Chiller power is the most valuable point in the room. A plant with no trending can still start.

Stage
Data available
What the model is built from
Target
1
Site walk
One structured survey of the plant room. A like-for-like model matched from thousands of chillers already on the platform.
5–10%
2
Documentation
Functional descriptions, O&M manuals, as-builts and OEM part-load data. The model matches your actual machines.
10–15%
3
Load profile
Thermal side trended: temperatures and flow or chiller load.
15–20%
4
Full trends
Thermal and electrical sides trended for six months or more. A calibrated twin, an energy baseline, and verified savings.
20–25%

Targets are cumulative, for total chilled water plant energy. Earlier stages are simulation-based estimates; only the final stage is independently verified.

Frequently asked questions

Does the software connect to our BMS or plant controls?

No. Exergenics does not connect to plant control and does not write to the BMS. All work is remote and air-gapped.

Can you build a twin with no trend data?

Yes. At the site-walk stage, a like-for-like model is matched from thousands of chillers already on the platform.

What if the twin does not match our plant?

The twin is proven against the plant’s own data to ASHRAE Guideline 14, or calibration fees are refunded in full.

Services built on the twin

Start with a Feasibility Study

A desktop forecast from your plant’s own BMS data: data gap analysis, a plant health check and a Plant Potential business case. Credited in full against any option purchased within 90 days.

Priced per kilowatt of refrigeration (kWr) of installed plant capacity. Request a price list