Load Balancing for BMS and Local Energy Distribution Grids

WP Leader: UCC (4C)

Participants: NEC, CARTIF


WP6 considers load balancing in buildings and local energy distribution grids, and develops tools which optimize overall energy flows with variable demands and supply. It predicts usage pattern for rooms and building zones based on information outside typical building management systems like time tables, meeting schedules, or personal agenda, complementing access control information that may already be available.
Unfortunately, neither of these information sources provides fully complete nor accurate data, we therefore combine these diverse sources to increase accuracy of our utilization forecasts.

At the same time, we also consider using more advanced time-of-use or even real-time energy pricing information to determine when to schedule certain energy intensive operations, and how to operate local energy sources and storage to minimize energy cost and emissions.
In certain situations it may be preferable to pre-heat or pre-cool a building at times of cheaper energy prices, for example using a heat-pump during the night to pre-heat a meeting room that is used during the day, and to stop heating the room during the last scheduled meeting of the day. In this way the perceived service quality for the user is not affected, while energy costs (and ultimately emissions) are reduced.

The next figure shows how the optimization solver of WP6 fits into the overall Campus21 design. A receding horizon optimizer periodically produces a solution for the next 24 hours, attempting to achieve the required comfort level while minimizing energy cost. In order to produce the schedule, we need Value Predictors which predict certain key parameters based on existing forecasts (weather), energy prices (real-time pricing), or based on our own forecasts for occupancy and utilization of rooms and building zones. These forecasting modules need current information from the BMS and historical data from the data warehouse (WP2), obtained through the middleware component (WP4).

The results of the optimization are new set points for components in the building, they are actuated in the BMS through an interface with the middleware.

In order to allow rapid deployment of our solution to new buildings, we use a declarative model to describe the optimization problem to be solved. The next figure shows the model for the swimming pool in the demonstrator site at Valladolid. We have to maintain the correct temperature in the pool, while also providing the required hot water supply to the facility, which varies significantly with utilization pattern. The utilization is predicted based on historical data, the current event schedule and background calendar information.

For heating, we can either use a solar thermal array on the roof, or we can use gas fired boilers. Storage tanks for hot water provide a buffer, which allows us to use energy from the solar array at a later time, avoiding the use of the more expensive gas boilers where possible.

From the declarative model we automatically generate the optimization model, which then integrates with the middleware to provide an updated solution every 15 minutes. In this way, we can react to short-term changes, without loosing sight of the long term objectives to minimize energy costs and emissions.


The following Deliverables document the achievements of this work package:
D 6.1: Specification of Systems Architecture and Concept for Integration and Upscaling

D 6.2: Implemented and Initially Documented Algorithms & Tools for Integrated Load Balancing

Work Package Leader:
Helmut Simonis
Cork Constraint Computation Centre
University College Cork Cork, Ireland
e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.