Back to search

FORSKSYSTEMET-FORSKSYSTEMET

Evaluation of scalable methods of MPC on building HVAC systems in virtual and real environments

Awarded: NOK 4.6 mill.

Buildings account for a large share of global energy consumption. In Norway, over 60% of energy use in buildings is for heating purposes, and a significant portion of this is directly powered by electricity. There is great potential to reduce energy use in buildings through better control. At the same time, buildings represent a major flexibility resource in the energy system. There is significant pressure on the power grid, and improved building operations, both by using less energy and by shifting energy consumption over time, can free up capacity for other purposes, such as industrial electrification, new industries, or electric vehicle charging. This can reduce the need for investments in the power grid. Since the majority of buildings that will be in use over the next 25 years are already built, it is important to target measures at the existing building stock. This project explores how data-driven control algorithms can be used to manage heating, cooling, and ventilation systems in buildings more optimally, enabling more flexible consumption control. The main focus is on Model Predictive Control (MPC). This is a method based on a simplified model of the building and energy system, optimizing energy use over time. It can use future predictions, such as weather forecasts and price signals, to ensure optimal operation. Constraints can also be included, such as desired room temperature or component capacities in the system. Although MPC has been used in industrial processes for many years and its potential has been demonstrated in many scientific articles, it is still rarely applied in real buildings. There are several reasons for this: • Buildings are different: All buildings and their users are unique, making it difficult to generalize across buildings. • Insufficient and low-quality data: Instrumentation levels vary greatly between buildings, and even well-equipped buildings often have poor-quality or mislabeled data. • Non-standardized communication: There are many different communication and connection protocols, requiring various adaptations for each implementation. To increase the likelihood of large-scale implementation, the project focuses on finding MPC concepts that are scalable across different buildings and can operate with limited data. To test and evaluate the various MPC approaches, both virtual test environments and real buildings are used. Virtual tests are conducted in BOPTEST (Building Optimization Performance Test)—an open simulation platform that allows standardized comparison of control strategies. This provides a safe and cost-effective arena for development and prototyping. Additionally, experiments are conducted in actual buildings owned by SINTEF, including the modern ZEB laboratory. This provides access to real-time data and the opportunity for controlled experiments. Part of the project involves comparing tests in virtual and real environments. Based on this, the virtual test environments can be further developed to provide realistic comparisons of different control algorithms.

Funding scheme:

FORSKSYSTEMET-FORSKSYSTEMET

Funding Sources