Chilled water plants account for a significant share of the energy used in buildings.
This session will present Machine Learning techniques to develop an optimal controls strategy to reduce energy use while maintaining the required chilled water production.
The strategy identifies in real-time the optimal number of chillers, cooling load distribution amongst chillers and condenser water flow setpoints that minimize power usage of the plant.
The presenter, Michael Berger, head of R&D at Conserve It, will show how he built a Digital Twin of the plant which updates automatically based on operating data.
Berger said the data relied on data pre-processing and expert-knowledge-driven constraints, to capture equipment performance variations over time.
Computing efficiency was a core requirement, leveraging the improvements in computing capabilities of Edge devices in recent years, to allow the solution to be fully deployable on site, without the need for components in the cloud.
This eliminated hurdles such as on-going fees and security or stability issues that may arise with constant internet connections requirements.
Results from deployment at several sites will be presented to demonstrate significant energy savings.
As the Head of R&D at Conserve It, Berger leads research, prototypes, and trial machine learning algorithms for analytics and real-time optimization of live equipment in the built environment.
Applications range from innovative chiller plant optimization & predictive maintenance to advanced controls solutions.
Berger graduated with a Master of Engineering in France before gaining experience at a leading research centre.
He gained experience at the Environmentally Sustainable Design consultancy in Singapore. He joined Conserve It in 2014, where he relies on years of experience in the fields of Energy Efficiency, Chiller Plant Optimization and Machine Learning.