For the fourth and most recent instalment of the Sim4Blocks webinars, partners HES-SO led a presentation on optimisation strategies.

HES-SO is the largest University of Applied Sciences in Switzerland and as a part of it the Institute of Systems Engineering (ISI) works extensively in the field of collecting energy consumption data, bringing to the Sim4Blocks project laboratory infrastructure and researching control and monitoring equipment within buildings.

Ramanunni Menon represented HES-SO and led the presentation which looked at a small part of the broad topic: optimisation strategy. At HES-SO they have been studying and collecting data from a small municipality in Switzerland called Naters with the goal to use optimisation algorithms so the buildings and system devices can be used for demand response (DR) purposes.

The project is looking at two different cases:

  • Building/local area level optimisation
  • Aggregator/market level optimisation

Ramanunni focused on building level optimisation and the two main constraints applicable to Naters from this perspective. Firstly, one limitation is how the market (ancillary market/reserve regulations and national/local grid regulations) regulates how DR can be applied.

The second issue, and one that is more important to Naters, is the system constraints – the presence or absence of pre-existing sensors or controllers for heat-pumps and storage devices. These pre-existing sensors will directly impose on how much control can be taken of the buildings and devices and therefore, how DR can be applied using optimisation algorithms.

An example of how system constraints are a major issue for municipalities like Naters is that if you take the circulation of the heating network in the area it is on all of the time and cannot be switched on or off. Without the control or flexibility to turn it on and off a successful DR model is harder to implement.

At the building level optimisation, HES-SO aim to use model predictive control (MPC) for the algorithms as compared to conventional control (CC), which sees a pre-defined transfer function input information into the device to create the output data. MPC on the other hand replaces the pre-defined transfer function with an optimiser instead, and this is where the algorithms are kept.

This is an advantage because not only are changes in objectives and constraints for a heating grid taken into account, it is possible to maximise the use of external factors and provide an input into the plant that suits a DR system.

These MPC systems can be distributed separately and can have a different set of objectives and constraints programmed to represent one building or a set of four for example, making them ideal for implementing at building and district level. Each MPC is then programmed to match its residency such as a single person house (SPH) or multi-person house (MPH).