|Abstract (english)|| |
Energy and resource efficiency in combination with renewable energy sources represent the backbone of future sustainable development in any sector. In this context, reduction of energy consumption in buildings is a vital element in the long-term transition towards low-carbon society. The European Union has identified buildings as being the most promising target for improving energy efficiency and has quantified a significant energy-saving potential associated with infrastructure and equipment investments. Although the theory often cites the so-called universally applicable solutions, practical experiences confirm that it is not possible to expect successful implementation of the initially defined energy efficiency programs without the proper understanding of the implementation environment. According to many studies, the majority of contemporary energy efficiency related policies and instruments are mainly oriented towards technical systems (building envelope, heating systems, electrical appliances), while behavioural changes are addressed only by the improved availability of information through awareness campaigns. The presented dissertation explores the theoretical background of energy management systems in buildings and defines the key contextual parameters necessary for proper evaluation of energy efficiency measures. A comprehensive literature review clearly confirmed that there is a need to connect energy consumption with the human factor in a systematic manner through a system of metering, monitoring and evaluation of energy and environmental performance. Also, it is vitally important to properly appreciate the internal and external contextual factors that influence the energy performance. Related to the energy performance in buildings the notion of context refers to the characteristics of the building’s operation, installed devices and systems, presence of a building occupant and their behaviour. The conventional performance monitoring methods and tools do not include prediction engines which would allow early warnings in the case of potential future abnormal situations. This lack of awareness about possible future abnormal situations limits the scope of today’s energy management systems and it has been a trigger for the presented research work. The main reason for introducing contextual parameters in the analysis of energy efficiency in buildings is to enable a proper understanding of energy consumption by building users and to encourage them to choose energy-efficient solutions. The research work described in this dissertation has been inspired by the recommendations found in different studies and papers during the literature review, where it was stated that the future research should focus on developing an efficient and effective performance monitoring system for promotion of energy awareness in buildings to enable building users to become aware of the energy performance in real-time, facilitating more effective business decisions based on accurate and timely information. The research question that arises in this context is: “How to sustainably empower building users who operate various systems and work in buildings to achieve enduring performance improvements?” The dissertation starts with the introduction and a comprehensive literature review regarding the framework for the energy management systems in buildings. It is followed by the second chapter which explains the general context and concept of energy efficiency in buildings, focusing on the legislative framework in Slovenia and the European Union, nearly zero energy buildings and smart grids. Additionally, it explores the main barriers for investing in energy-efficiency technologies and outlines the most important targets and future development challenges. The third chapter provides the reference architecture of the proposed model of context-sensitive energy management system in buildings. It also recognises the need for a systematic approach to identify the significant energy saving potentials and key contextual factors affecting energy performance. It emphasises the importance of the key performance indicators and gives an overview of prediction models based on the regression analysis, artificial neural networks and neuro-fuzzy modelling approach. This chapter also deals with the modern scientific approaches for supporting energy awareness including early warning, optimisation and multi-criteria decision making. The fourth chapter presents the most important results of the performed research work which is organised in four different case studies where the selected functionalities of the proposed model have been critically reviewed. This chapter illustrates the power of introducing dynamic and context-sensitive indicator called Energy Performance Coefficient in three different contextual situations: a complex of buildings for educational and research purposes, a multi-dwelling building and a group of elementary schools and kindergartens. The final chapter gives a summary of the key findings of the conducted research work and it is concluded by a proposal of possible future research work and addressed challenges. One of the basic assumptions of the concept proposed in this dissertation is that the monitored energy consumption, enriched with information about its context, can be the basis for the identification of energy profiles as well as energy efficiency improvements in buildings. The reference architecture of the proposed model of context-sensitive energy management system in buildings has evolved from the traditional management structure, plan-do-check-act model. The traditional plan-do-check-act model has been upgraded with the integration of a prediction engine, knowledge repository and different approaches for decision support. The energy manager and his team are placed in the centre of the proposed model. In this sense the energy manager is a person who evaluates energy use, designs and then supervises the implementation of energy efficiency measures. Additionally, the energy managers are responsible for closing the gap between theory and practice and for providing the necessary background and help for decision makers when they are dealing with energy-related issues. One of the assumptions of this research work is that even the most sophisticated information about energy consumption remains as inert data unless it is acted on by skilled and motivated employees. Additionally, the reference architecture of the proposed model of context-sensitive energy management system in buildings includes the following subsystems: • Subsystem for data acquisition and contextualisation, • Knowledge repository and • Subsystem for performance monitoring and process optimisation including prediction engine, consumption modelling, calculation of dynamic and context sensitive key performance indicators and various approaches for decision support. The concept of three subsystems has evolved from the idea to enrich the existing solutions with the awareness about the possible future abnormal situations and to empower energy managers with the appropriate knowledge to perform preventive actions. In this sense, context variables are variables that affect the energy consumption but are kept steady for periods that go beyond the process time constants, for example number of building occupants (employees), time of the day, day of the week, outside temperature, inside temperature or even energy prices. The key objective of monitoring and modelling energy profiles in buildings is to provide the necessary background of energy consumption and to allow informed decisions to be made. Energy consumption patterns are essential not only for a better understanding of past behaviours, but also for forecasting the future building energy demand and particularly for estimating the energy requirements of alternative consumption strategies. In the proposed model of context-sensitive energy management system in buildings, the modelling starts with the integration of energy within the activity flow charts, which is the basis for decisions on setting up the structure of energy cost centres (ECC). Similarly to industrial applications, the ECCs are the core elements of an entire energy model of a building or a complex of buildings. During the preformed research work it was recognised that the energy consumption modelling and prediction was a necessary step, but only the first one in the process of the performance improvement. The purpose of each defined key performance indicator is to enable the energy manager to react appropriately to changes in the building operation related with the context of energy use. The benchmarking values of all key performance indicators relevant for the energy managers and maintenance staff must be context-specific. Additionally, it has been revealed that the information on the key performance indicator alone was not solving the problem and the intention of the proposed model was to incorporate the required knowledge on how to correctly interpret different values of key performance indicators. That was the reason why the dynamic and context-sensitive indicator called Energy Performance Coefficient, which is the ratio of actual to predicted or benchmark energy consumption, has been introduced. The purpose of the Energy Performance Coefficient is to early identify or predict changes in energy consumption or context and to present that information to the building energy manager in a simple and straightforward way. During the research work, three different prediction models based on the regression analysis, artificial neural networks and neuro-fuzzy modelling approach were compared. For the initial testing the data samples were collected on an hourly interval over a period of one year, which resulted in 8759 samples of each input signal. The necessary input data for all compared prediction models consist of actual (hourly average) values of electricity consumption, building occupancy, type of working day, outside temperature and global solar irradiance at the time t. The output of the prediction engine is the predicted electricity consumption at the time t+1. During the next 12 months, additional 1440 samples were collected and the developed models were compared. The testing results confirmed that all the developed models are able to give a relatively accurate prediction of the ECC energy consumption if the model is well calibrated and the reliable history data is available. However, the analysis revealed that the model based on the adaptive neuro-fuzzy inference system (ANFIS) was capable of providing slightly better prediction results than the other two models. Also, in this case the hybrid learning algorithm proved to be more suitable than the back propagation learning algorithm. The objective of the first case study was to analyse the strengths and weaknesses of the customized educational solution called EUREM (European Energy Manager) for the training of energy managers in Slovenia and to outline the most important targets and challenges for the future development. The data presented in this case study consist of the data obtained through an evaluation of all the submitted energy concepts, a regular evaluation of individual trainers and lectures (which normally takes place immediately after each training module), additional interviews with selected participants and a survey among former participants some months or years after their participation in a EUREM training program. All the data collection activities included quantitative and qualitative approaches with the goal being to evaluate the different dimensions of the strengths and weaknesses experienced during the implementation of the EUREM education and training program in Slovenia. The presented case study has clearly confirmed that the integration of the postulates of energy efficiency into daily operational practice is a continuous process that requires additional skills and knowledge. Additionally, the conducted research work confirms that the successful implementation of energy efficiency projects requires interdisciplinary knowledge related to energy management, renewable energy sources, energy auditing, building and facility management, energy trading, economics, financing, planning and maintenance. The objective of the second case study was to test the capabilities of the ECC based modelling of energy consumption in educational and research buildings by using different modelling approaches and to test different data visualisation approaches. At the beginning of the modelling process a comprehensive analysis of the energy and water consumption at the selected location was preformed and important physical components and activities that characterise both energy and water consumption were recognised. The guiding principle for ECCs setup was to follow the activities, organisation structure and energy flows at the location. The presented modelling concept provides a framework for performance monitoring and targeting at each designated responsibility centre and directly connects people with tasks in each ECC. To cover energy and water consumption in all the identified ECCs, it was necessary to buy and install thirteen electricity meters, install a meter for extra light fuel oil and upgrade the existing utility meters for water and electricity together with the new communication equipment for data acquisition. During the testing period, this approach required continuous refinement of the initially defined procedures for responding to abnormal situations and implementation of corrective measures based on real-time measurements. The initial testing results based on the provided feedback, visualisation of electricity consumption and personal involvement of nominated employees at the level of each ECC, confirmed that significant electricity savings can be achieved. Visualisation of electricity consumption through the heat map analysis has been upgraded by using the Energy Performance Coefficient. Additionally, it was recognized that in their daily work, the energy managers must focus on working with the maintenance staff, because the technical expertise and even the most efficient equipment will fail to produce positive results unless they are properly motivated and committed to performance improvements. In the third case study, the capabilities of the ECC based modelling of heat consumption in a multi-dwelling building were evaluated. The biggest challenge during the conducted research work was how to make the consumption and billing data understandable to the occupants in the selected building. The obtained results clearly confirmed the applicability of the ECC based modelling for providing systematic framework and energy consumption analysis in the comprehensive environment of the multi-dwelling building. The dwelling occupants were informed about their performance relative to the performance of other contextually similar dwellings in the same building. One dwelling owner was acting as energy manager for the entire building and that has proved to be an initial change of context and a crucial moment in the transfer of sustainable ideas. Additionally, in this case study, the applicability of the Multi-Attribute Utility Theory (MAUT) method as a decision support tool for the selection of optimal energy-renovation scenario of the single family house was tested. Due to the dependence on many input and output parameters, which can be very complex, the energy efficiency projects represent a typical multi-criteria decision problem. In the conducted research work the MAUT method was used to compare five different energy-renovation scenarios of the selected single family house. Also, strengths and weaknesses of the selected multi-criteria method and its role in the decision process (selection of the optimal energy-renovation scenario) were critically reviewed. It was noticed that the construction of the utility function for each criterion requires a lot of effort and experience by the energy manager. The objective of the fourth case study was to evaluate the capabilities of the data envelopment analysis in combination with ECC based modelling for the electricity consumption analysis in the selected group of elementary schools and kindergartens. The electricity consumption data was obtained through the cooperation with the local energy agency which was acting as an energy manager for the selected group of buildings. The obtained results confirmed the potential of combining the data envelopment analysis and the ECC based modelling in order to discover the schools with the capacity for energy efficiency improvements. Additionally, the combination of results obtained through the data envelopment analysis with the contextual parameters hidden in the structure of the ECCs, provides additional and valuable information for the energy manager and helps him with the proper benchmarking of analysed energy management practices. However, it has to be noticed that the complexity of the proposed model arises from the many variables that provide contextual information, including the end-user behaviour during the usual daily operations. The other limitations of the proposed model are related with the requirements for expert knowledge during the training period and the definition of the initial set of contextualised benchmark values. Based on the experiences gained through different case studies it is clear that the additional research work must include experts from social sciences that will deal with people’s values and attitudes toward the energy efficient behaviour. Also, the additional research and analysis must be carried out for a larger group of similar sites to obtain reliable proof of the proposed model, especially having in mind new trends in modelling the energy consumption and prediction due to high penetration rate of renewable energy sources in the buildings sector. The conducted research work has clearly revealed that the implementation of the context-sensitive energy management system in buildings will only be effective if those who need to act are motivated and trained to become involved and to make energy-efficient actions. In this context, the proposed model based on the ECC modelling allows the individualisation of responsibilities within each ECC, which is a necessary precondition to achieve enduring energy performance improvements.