Optimising energy performance and comfort in office spaces

Context

This case study applies to building zones equipped with sensors that measure temperature and activity, a controllable Heating, Ventilation, and Air Conditioning (HVAC) system, and, optionally, a local PV installation.

Objective

Buildings are responsible for almost one-third of the global energy consumption and CO2 emissions. Hence, their role in mitigating climate change is crucial. HVAC systems alone account for approximately 40% of the total energy consumption of a building.

In this case study, we show that it is possible to reduce HVAC systems consumption by finding the optimal trade-off between minimising energy consumption, guaranteeing a good level of comfort, reducing energy costs, and maximising renewable energy usage.

Solution

Energis developed and deployed a solution consisting of a rule-based AI system, capable of controlling HVAC devices in a smart and autonomous manner, during the Innoviris TeamUp research program. 

The solution was installed on our edge device Raspicy and tested in a scenario comprising an office zone equipped with temperature, humidity, CO2, movement sensors, and a local PV installation. The communication between the smart controller and the HVAC system occurred through ModBus protocol, although several other alternatives can be considered, such as BMS and an already-existing HTTP API.

The rule-based AI controller has an Inference Engine that applies condition-action rules stored in the Knowledge Base to make decisions and deduce new facts. 

As a first step, a Knowledge Engineer gathers the (building) domain knowledge with the help of domain experts and then converts that knowledge into rules that can be added to the system Knowledge Base.  

After that initial phase, the system works autonomously, interacting with the environment to perceive its state (what my world looks like now?) and applying the rules stored in the Knowledge Base to decide the best next action to take (what should I do now based on what I know?). 

Besides using sensor data, the rule-based controller leverages weather and occupancy forecasts to adapt proactively to the predicted conditions of the environment. For example, preheat the zone if occupancy and low external temperature are expected in the coming hours.

Outcome

As demonstrated during our research program, our rule-based controller was able to improve the level of comfort in the zone, reducing at the same time the energy consumption of the HVAC system without any human intervention. The optimisation algorithm also took the RES production into account, taking maximum advantage of solar irradiation whenever available. 

Moreover, being able to forecast weather and occupancy conditions was a significant factor to increase the energy efficiency, e.g. leveraging the thermal inertia of the building at the end of the working day.

Monitoring of the Controller

The two panels of the controller’s dashboard below were used to monitor the indoor temperature and the actions taken. The controller preheats the office if the zone is cold and will be occupied in the coming hours. It made the best use of the PV production whenever the irradiation was good enough and stopped heating when it was time to leverage the thermal inertia of the building until the end of the working day.

The comfort panels allowed us to monitor the comfort in depth, e.g. follow the evolution of the Predicted Percentage of Dissatisfied (PPD) and Predicted Mean Vote (PMV) indexes used by ASHRAE¹ to assess thermal comfort. PPD, indicated with a green line in the comfort panel, estimates the percentage of people that are dissatisfied with the current level of comfort (cooling or heating) and must be lower than 20% during working hours (grey area in the dashboards).

¹ The American Society of Heating, Refrigerating and Air-Conditioning Engineers. www.ashrae.org 

Monitoring of KPIs

A global view of the comfort conditions was also necessary. We created the below dashboards that monitor comfort scores e.g. temperature, humidity. These scores indicate the percentage of time the indoor condition was in the comfort range during working hours.