Case Study — Higher Education
University Campus Chiller Plant Optimization at the University of Melbourne
Fourteen chilled water plants across the Parkville and Southbank campuses, optimized building by building. Nine have completed twelve month measurement and verification, and between them they are saving 62% more energy than the models forecast.
Verified energy saved
426,444 kWh
Per year, across the nine buildings that have completed M&V
Verified carbon abated
269 tCO2
Per year, at the 2025 NEM scope 2 and 3 factor of 0.63 tCO2/MWh
Against modelled forecast
+62%
Measured savings exceeded the modelled forecast of 262,571 kWh
Buildings verified
9 of 14
The remaining five are modelled and awaiting verification
The Challenge
Fourteen plants, fourteen different problems
A university campus is not one building repeated. The University of Melbourne runs chilled water plants serving a neuroscience research facility, a rare books library, a performing arts school, a biocontainment institute and a law school, and each of those has a different load profile, a different tolerance for variation and a different set of constraints on what the plant is allowed to do.
Plant capacity across the portfolio ranges from 1,198 kWr at the Nancy Millis Building to 5,300 kWr at the Peter Doherty Institute. Chiller counts range from two to three. Some plants serve laboratories that cannot deviate. Others serve teaching space that empties for months at a time.
A single control philosophy applied across all of them would have been simpler to deliver and would have left most of the saving on the table. The university's energy team wanted each plant tuned to its own building, and wanted the results measured rather than asserted.
Every building was modelled on its own data, and every result reported here was measured against that building's own pre implementation baseline.
Portfolio Map
Where the savings are
Twelve of the fourteen buildings sit on the Parkville campus and its Carlton edge. The remaining two are at Southbank, home of the Faculty of Fine Arts and Music.
- 144Kenneth Myer, Neurosciences126,321 kWh verified
- 248Peter Doherty Institute69,961 kWh
- 401Nancy Millis50,417 kWh
- 110The Spot45,415 kWh
- 864Dance Building, Southbank42,311 kWh
- 379207 Bouverie Street29,679 kWh
- 168Doug McDonell26,333 kWh
- 880MCM Performing Arts, Southbank23,462 kWh
- 106Law Building12,545 kWh
- 177Baillieu Library43,863 kWh modelled
- 153Chemistry, West Wing26,245 kWh modelled
- 105Business & Economics19,141 kWh modelled
- 104Alan Gilbert17,960 kWh modelled
- 133Glyn Davis, Melbourne School of Design16,943 kWh modelled
Verified Results
Nine buildings, measured over twelve months
| Building | Plant | Modelled kWh | Verified kWh | Verified tCO2 | Against forecast |
|---|---|---|---|---|---|
| 144 Kenneth Myer, Neurosciences | 4,200 kWr | 46,048 | 126,321 | 80 | +174% |
| 248 Peter Doherty Institute | 5,300 kWr | 60,761 | 69,961 | 44 | +15% |
| 401 Nancy Millis | 1,198 kWr | 51,929 | 50,417 | 32 | −3% |
| 110 The Spot | 2,280 kWr | 20,839 | 45,415 | 29 | +118% |
| 864 Dance Building, Southbank | 1,900 kWr | 38,991 | 42,311 | 27 | +9% |
| 379 Building, 207 Bouverie Street | 1,240 kWr | 9,739 | 29,679 | 19 | +205% |
| 168 Doug McDonell | 2,874 kWr | 14,200 | 26,333 | 17 | +85% |
| 880 MCM Performing Arts, Southbank | 1,370 kWr | 13,431 | 23,462 | 15 | +75% |
| 106 Law Building | 3,200 kWr | 6,633 | 12,545 | 8 | +89% |
| Nine buildings verified | 23,562 kWr | 262,571 | 426,444 | 269 | +62% |
Reading the Numbers
Why the measured savings ran ahead of the model
Eight of the nine verified buildings beat their forecast, and the portfolio came in 62% above the modelled total. That gap is worth explaining, because a model that is wrong by 62% in the other direction would be a serious problem.
Exergenics models carry a deliberate factor of safety. The forecast issued to a client is the saving the plant should deliver under conservative assumptions about load, weather and how faithfully the strategy gets implemented. It is written to be a floor, not a midpoint.
Two things then move the measured result above it. Cooling demand in the reporting period is rarely identical to the baseline period, and where demand rose the optimized plant absorbed it more efficiently than the baseline plant would have. And several of these buildings had more headroom in condenser water control than the conservative model assumed.
The Nancy Millis Building is the useful counter example. It came in 3% under forecast, which says the model was close and the plant had less slack than the others. A portfolio where every building overshoots by the same margin would suggest a systematic modelling bias rather than genuine site by site variation.
In Progress
Five buildings modelled, verification scheduled
The third tranche completed modelling and implementation and has not yet been through measurement and verification. The figures below are forecasts, not measured results, and should be read as such.
| Building | Plant | Modelled kWh | Modelled tCO2 | Status |
|---|---|---|---|---|
| 177 Baillieu Library | 3,500 kWr | 43,863 | 28 | M&V pending |
| 153 Chemistry, West Wing | 2,500 kWr | 26,245 | 17 | M&V pending |
| 105 Business & Economics | 2,500 kWr | 19,141 | 12 | M&V pending |
| 104 Alan Gilbert | 1,880 kWr | 17,960 | 11 | M&V pending |
| 133 Glyn Davis, Melbourne School of Design | 1,650 kWr | 16,943 | 11 | M&V pending |
| Five buildings modelled | 12,030 kWr | 124,152 | 78 |
The Exergenics Approach
Optimization inside the university's own control system
Nothing was installed on the university network. Each strategy was delivered as setpoints and sequences for the incumbent controls contractor to code into the existing building management system.
-
Model each plant on its own data
Historical chiller, pump and cooling tower trends were drawn from each building individually. Plant that shares a campus does not share a load profile, and a model built on one building's data does not transfer to the next one along.
-
Generate a strategy the plant can actually run
Chiller staging thresholds and condenser water temperature control were generated for each plant and bounded by constraints the site team confirmed. Research and teaching buildings on this campus have hard limits on how far conditions may drift, and the strategy respects them rather than optimizing past them.
-
Implement through the incumbent contractor
The university's incumbent controls contractor coded the recommended strategies into the existing BMS. No hardware was added, no gateway was installed and no external connection to the control network was required.
-
Verify against the building's own baseline
Each verified building was regressed against measured cooling load over a twelve month reporting period and compared with its pre implementation baseline on the same basis. The portfolio figures in this case study were compiled by the university's energy consultants.
Put a number on your own campus
Exergenics optimizes chilled water plants for universities, hospitals, airports and commercial portfolios. No capital expenditure, no hardware and no disruption to operations.
Talk to an expert Download case study PDFVerified figures are drawn from twelve month measurement and verification reports and from the portfolio comparison of forecast against realised savings compiled by the University of Melbourne's energy consultants, July 2025. Carbon conversion uses 0.63 tCO2/MWh, the 2025 NEM factor for scope 2 and 3 emissions.