Search

Very little work has been done on the feasibility of Machine Learning (ML) for predicting buildings energy demand right at the design stage. This feasibility, if proven, would help to avoid the construction of inefficient buildings. This paper uses dataset from 7559 buildings, and estimates their energy consumption using nine ML models. Results show that deep neural network (DNN) is the most efficient ML model with MAE, MSD and RMSE of 0.93, 1.12 and 1.06 respectively achieved in less than 7 s despite the huge data size. Its r 2 is also the highest (0.96) which means that the DNN approach manages to explain 96% of the energy consumption in buildings and only 4% remains unexplained certainly due to the limitation of independent variables. Also, this result is not affected by building clusters nor by data from a particular climate zone. As an innovation, this study proposes a model that professionals could use in the design phase of a construction project. This model will allow them to take into account all crucial aspects of the design of an energy efficient building. The model will then serve as a decision-making tool to control and optimise the project and to anticipate energy consumption even before the building is constructed. [ABSTRACT FROM AUTHOR]

• Implementation and benchmarking of five advanced control methods. • Comparison of black-, gray-, and white-box MPC with RL and approximate MPC. • Introduction of qualitative KPIs to assess the applicability. • Application in a one-year simulation to evaluate adaptability. Advanced building control strategies like model predictive control and reinforcement learning can consider forecasts for weather, occupancy, and energy prices. Combined with system and domain knowledge, this makes them a promising approach to reduce buildings' energy consumption and CO 2 emissions. For this reason, model predictive control and reinforcement learning have recently gained more popularity in the scientific literature. Nevertheless, publications often lack comparability among different control algorithms. The studies in the literature mainly focus on the comparison of an advanced algorithm with a conventional alternative. At the same time, use cases and key performance indicators vary strongly. This paper extensively evaluates six advanced control algorithms based on quantitative and qualitative key performance indicators. The considered control algorithms are a state-of-the-art model-free reinforcement learning algorithm (Soft-Actor-Critic), three model predictive controllers based on white-box, gray-box, and black-box modeling, approximate model predictive control, and a well-designed rule-based controller for fair benchmarking. The controllers are applied to an exemplary multi–input–multi–output building energy system and evaluated using a one-year simulation to cover seasonal effects. The considered building energy system is an office room supplied with heat and cold by an air handling unit and a concrete core activation. We consider the violation of air temperature constraints as thermal discomfort, the yearly energy consumption, and the computational effort as quantitative key performance indicators. Compared to the well-tuned rule-based controller, all advanced controllers decrease thermal discomfort. The black-box model predictive controller achieves the highest energy savings with 8.4 %, followed by the white-box model predictive controller with 7.4 % and the gray-box controller with 7.2 %. The reinforcement learning algorithm reduces energy consumption by 7.1 % and the approximate model predictive controller by 4.8 %. Next to these quantitative key performance indicators, we introduce qualitative criteria like adaptability, interpretability, and required know-how. Furthermore, we discuss the shortcomings and potential improvements of each controller. [ABSTRACT FROM AUTHOR]

• Implementation of a model predictive control scheme for HVAC systems is presented. • The hardware/software solution is generalisable to most building management systems. • Cloud solutions can offer low-cost, scalable control in commercial buildings. • Economic MPC reduced energy usage by 1.7% compared to baseline control. • Building thermal performance strongly affects viability of MPC schemes. Many businesses are looking for ways to improve the energy and carbon usage of their buildings, particularly through enhanced data collection and control schemes. In this context, this paper presents a case study of a food-retail building in the UK, detailing the design, installation and cost of a generalisable model predictive control (MPC) framework for its Heating, Ventilation and Air Conditioning (HVAC) system. The hardware/software solution to collect relevant data, as well as the formulation of the MPC scheme, is presented. By utilising cloud-based microservices, this approach can be applied to all modern building management systems with little upfront capital, and an ongoing monthly cost as low as $6.39/month. The MPC scheme calculates the optimal temperature setpoint required for each Air-Handling Unit (AHU) to minimise its overall cost or carbon usage, while ensuring thermal comfort of occupants. Its performance is then compared to the existing legacy controller using a simulation of the building's thermal behaviour. When simulated across two months the MPC approach performed better and achieved the same thermal comfort for a lower overall cost. The economic optimisation resulted in an energy saving of 650 kWh, with an associated cost savings of $240 (an improvement of 1.7% compared to the baseline), while the carbon optimisation gave negligible CO 2 savings due to the inability of the building to shift heating to low-carbon periods. Findings from this study indicate the potential for improving building performance via MPC strategies, however the level of impact will depend on specific building attributes. [ABSTRACT FROM AUTHOR]

Occupancy-Centric Controls (OCC) are specialized control sequences which modulate building operation as a function of the sensed occupancy status, thereby leading to increased energy savings, while maintaining thermal comfort. However, occupancy sensor nonidealities, such as bias, latency, random noise, or misdetection, can degrade the performance of OCCs. In this paper, we perform a systematic simulation-based investigation to understand the impacts of nonidealities from both occupancy counting and presence sensors on OCCs. We use a prototype medium office EnergyPlus-based building model, with a detailed zoning layout, as our evaluation testbed. Subsequently, we evaluate the effectiveness of rule-based, occupancy-centric HVAC, lighting and equipment control strategies on an annualized basis. Furthermore, we systematically characterize occupancy sensor nonidealities. Our results indicate that for counting sensors, although the effects of random noise on energy consumption and thermal comfort are minimal, sensor bias and latency can increase HVAC and whole building energy consumption by up to 3.3 % and 4.9 %, respectively. Additionally, noting that an enumeration approach to understand the sensor impact evaluation map for all considered nonidealities may be resource- and time-intensive, we propose a Bayesian Optimization (BO)-based smart sampling approach to efficiently identify the most impactful sensor nonideality sets. Our results indicate that the proposed BO-based approach is able to increase the computational efficiency by 60-fold, when compared to an exhaustive grid-search (enumeration) method. For presence sensors, we observed that higher false positive rates could have a direct impact on HVAC and whole building energy consumption. A low-grade presence sensor could lead to the increase of HVAC and whole building energy consumption by 12.5 % and 8.1 % on average. [ABSTRACT FROM AUTHOR]

[Display omitted] • The occupancy modeling approaches with a particular focus on UBEM are discussed. • Occupant-related inputs required for UBEM and the correlation between them are emphasized in the review. • Challenges in accommodating the variations in occupancy at the urban scale are reviewed. • Potential data sources for extracting building occupancy at urban scale are discussed. • UML diagram and co-simulation required to represent occupancy are discussed. Occupant-related inputs are significant parameters that influence energy simulation accuracy at both the building and urban levels. In most previous research works, fixed occupant schedules were used in urban-scale building energy modeling. The main reason is the lack of data availability to model the dynamic occupancy schedules. In recent years, urban data sets and modern estimation and detection techniques were introduced to increase the availability of occupant-related data sets. Yet, using these data to model the detailed occupancy at the urban scale is challenging and not much explored. Also, it is unclear how detailed the input regarding the occupancy should be. Addressing these research questions is the main objective of this study. This paper presents a comprehensive review of the occupant behavior (OB) modeling approaches, occupant-related input parameters with particular focus on the occupancy schedule, lighting, appliances use schedule, temperature set-point schedule, and domestic hot water usage for urban building energy modeling (UBEM). Strategies to consider the occupancy sub-models as co-simulation connecting to urban building energy simulation are discussed. Some potential datasets that could be used to derive occupant-related inputs at the urban scale are presented and highlighted. Further, the correlation between occupants' activity level, plug, and lighting loads is discussed in detail. Finally, the limitations and challenges of the occupant-related data connecting to building energy modeling are discussed, along with the research gaps and future directions of occupant-centric UBEM. [ABSTRACT FROM AUTHOR]

The objective of this work was to evaluate the interactions of users with air conditioning systems, lighting, natural ventilation, internal shading devices, and costs related to automation of a teaching building. Three behavior patterns (proactive, intermediate and careless) were created. The proactive user seeks to use natural light, uses passive solar shading and integration of natural and artificial ventilation. The intermediate user does not seek integration of natural and artificial lighting, they use hybrid ventilation as a ventilation strategy. The careless user does not use any passive strategy, predominantly uses artificial conditioning and adjusts cooling and heating setpoint to values out of the norms. In addition to the users' behaviors, the automated building was modeled. Computer simulations were performed in the EnergyPlus program, in order to measure the influence of behavior patterns on the building's energy consumption. Based on the analysis of the building's energy consumption profile, results indicate users are between the intermediate and the careless user classification, with a tendency to the intermediate user profile. The paper also presents an analysis of the payback time with the building automation in face of the user profiles. [ABSTRACT FROM AUTHOR]