Buildings account for approximately 40% of energy consumption in the European Union and over one-third of energy-related greenhouse gas emissions, with a significant portion attributed to heating, ventilation, and air conditioning (HVAC) systems. Effective fault detection and diagnosis (FDD) are essential for reducing energy waste and lowering maintenance costs in HVAC operations. FDD methods for HVAC systems have been extensively studied and can be broadly classified into two categories: knowledge-based and data-driven approaches. Knowledge-based approaches heavily rely on predefined rules and domain expertise and remain the most widely used in existing HVAC systems. Over the past decade, data-driven FDD approaches have gained popularity. However, data-driven FDD approaches require highquality labelled fault datasets for model training, which can be time-consuming and costly to obtain. To address this challenge, various studies have explored the use of generative adversarial networks (GANs) and other data augmentation techniques to synthesize realistic fault data and improve model performance. Despite these advancements, challenges related to generalization, scalability, and the interpretability of black-box models remain key concerns in the adoption of data-driven FDD approaches. [...]@en

Energy waste in buildings can range from 5% to 30% due to faults and inadequate controls. To effectively mitigate energy waste and reduce maintenance costs, the development of Fault Detection and Diagnosis (FDD) algorithms for building energy systems is crucial. Diagnostic Bayesian Networks (DBNs), as graphical probability models, are particularly useful in scenarios where high-quality data is not always available. While many studies have focused on single fault detection using DBNs, the occurrence of multiple simultaneous faults is common, yet the versatility of DBNs in handling such cases is rarely explored. This study adapts a DBN, initially designed for single fault diagnosis, to perform simultaneous fault diagnosis Experiments were conducted on an air handling unit (AHU) in the Netherlands, using implemented simultaneous faults to test the model. The results suggest that the DBN can detect both single and multiple faults effectively.@en

Fault detection and diagnosis (FDD) provides several interrelated benefits, including reducing energy waste, enhanced operational efficiency, and maintaining indoor comfort. The initial step in FDD is to detect deviations from normal or expected operation. However, establishing a reliable baseline can be challenging, especially when there is a lack of sufficient system documents or when complex control strategies are involved. This study investigates three feature selection methods for the baseline estimation: expert knowledge-based, correlation-based, and causality-guided, using heating coil valve control estimation as an example. These methods were tested in an office building in the Netherlands. The results show that while the correlation-based method achieved the best estimation, it may lead to false negatives due to features with reverse causality. This study aims to emphasize the necessity of causal analysis in the baseline estimation to achieve reliable FDD in buildings.@en

Automated fault detection and diagnostics (FDD) can support building energy performance and predictive maintenance by leveraging the vast amounts of data generated by modern building management systems. Diagnostic Bayesian Networks (DBN) offer a particularly promising approach due to their robustness, flexibility and scalability. However, FDD applications in whole building systems are rare, as they require the integration of different building subsystems, with their own potential faults and symptoms, which increases complexity and makes the resulting DBNs system-specific. In order to overcome these limitations, the 4S3F (four symptoms and three faults) method offers a simplified, adaptable framework for FDD implementation across building systems. In this paper, we implement the 4S3F methodology to a whole-building HVAC system in a case study office building located in the Netherlands. Our methodology uses generic, aggregated representations of individual subsystems within the building, such that FDD methods for specific subcomponents can later be incorporated where available. We first define aggregated building system groups (boiler group, chiller group, hydronic groups, ventilation groups, and end user groups) and subsequently define generic faults that can be detected with the existing sensor infrastructure. This simplified system representation is then used to define a DBN to isolate the most probable system-level faults that lead to building-level symptoms. By focusing on the whole building system, this work aims to provide the groundwork to incorporate occupant feedback and behavior in FDD.@en

This study investigates the diagnostic capabilities of a Diagnostic Bayesian Network (DBN) for air handling unit (AHU) components, particularly focusing on the heat recovery wheel (HRW) and heating coil valve (HCV). Unlike data-driven methods relying heavily on high-quality labeled data, this knowledge-based DBN is more suitable for real-world applications, where labeled faulty and normal data are hard to obtain. Notably, existing studies predominantly concentrate on developing DBN for AHU with recirculated air, neglecting thorough investigations into AHU with HRW, a prevalent system in North Europe and increasingly recommended post-COVID-19 for mitigating viral propagation. This paper presents a DBN setup with expert knowledge for an AHU with HRW, which is evaluated using experimental data from an office building in the Netherlands. The results show that the proposed DBN can successfully diagnose typical faults in HRW and HCV. @en

The energy management systems industry in the built environment is currently an important topic. Buildings use about 40% of the total global energy worldwide. Therefore, the energy management system’s sector is one of the most influential sectors to realize changes and transformation of energy use. New data science technologies used in building energy management systems might not only bring many technical challenges, but also they raise significant educational challenges for professionals who work in the field of energy management systems. Learning and educational issues are mainly due to the transformation of professional practices and networks, emerging technologies, and a big shift in how people work, communicate, and share their knowledge across the professional and academic sectors. In this study, we have investigated three different companies active in the building services sector to identify the main motivation and barriers to knowledge adoption, transfer, and exchange between different professionals in the energy management sector and explore the technologies that have been used in this field using the boundary-crossing framework. The results of our study show the importance of understanding professional learning networks in the building services sector. Additionally, the role of learning culture, incentive structure, and technologies behind the educational system of each organization are explained. Boundary-crossing helps to analyze the barriers and challenges in the educational setting and how new educational technologies can be embedded. Based on our results, future studies with a bigger sample and deeper analysis of technologies are needed to have a better understanding of current educational problems.@en

Learning and educational challenges in the field of indoor climate and building services like energy systems are mainly due to the transformation of professional practices and learning networks, a big shift in the way in which people work, communicate, and share their knowledge and the need for additional workforce, either juniors or coming from other disciplines. One of the most important factors that highly influence professional development and workplace learning is networked learning. Our goal in this study, is understanding the learning networks characteristics and patterns of interaction using Social Network Analysis techniques in three MOOCs discussion forums. The result of this study shows not only the importance of Learning networks and peer support on professionalization of learners, but also how pedagogical approach of instructors in MOOCs can foster learning networks. This novel approach in developing learning networks and communities is not only able to help connect young professionals and experienced practitioners digitally, but also it can promote professional development and innovation in the energy installation sector.@en

The average energy consumption for heating of dwellings in the Netherlands has been decreasing over the last decades as a result of increasing thermal performance of new and renovated dwellings. However, this decrease is found to be lower than energy performance models predicted. One of the possible reasons is the behaviour of the residents, which is partially determined by the thermal comfort preferences of these residents. In this paper, the relationship between thermal comfort perception, indoor climate and energy consumption is investigated using high-resolution measurement data in 93 dwellings in the Netherlands. In the OPSCHALER project, data about thermal comfort perception, indoor climate and operational energy consumption were collected in 93 dwellings in the Netherlands during periods ranging from two to twelve months over a period of two years. Comfort perception was registered using the Comfort App, an application where users record their comfort data. Indoor climate data were collected per five minutes using sensors for temperature, relative humidity and CO2 concentration in the living room, kitchen and bedrooms of the dwellings. The Comfort App asked the residents for the room they are in, their comfort perception, their activities during the last half hour and the amount of clothing they wear. Significant relationships were found between thermal sensation and thermal preference, clothing level, metabolic activity level, activities related to thermal comfort taken in the last half hour, and indoor air temperature. These data can be used to compare the comfort level registered by the residents with the comfort level predicted by the PMV model, and link this to the indoor climate and the energy consumption for heating. This information can help to understand the relations between user preferences, indoor climate and energy consumption for heating.@en

Monitoring the energy performance of very low and zero energy buildings is fundamental to evaluate the efforts made to transition into an energy neutral built environment. Post occupancy monitoring has been embedded into current practice, supported by the availability of smart meters and affordable sensor technology. However, there is still a lack of standardised monitoring guidance, which complicates the comparison between projects. In this study, we reviewed reports and publicly available documents related to the monitoring of low energy and zero energy projects in the Netherlands. A total of 12 studies reporting on 65 projects containing 4,400 dwellings were analysed. These included both new and renovated housing built in the last decade. This study aims to provide an overview of actual energy performance in energy renovation projects across the Netherlands. It also analyses the difference with predicted energy performance and analyses the perceptions of residents involved in low and zero energy renovations. It answers questions such as: What energy and behavioural data is being gathered through energy monitoring in the residential sector (related to monitoring low and zero energy buildings/dwellings)? How is the data currently being utilized? What does the data tell us about actual energy use and resident perceptions? How can monitoring be improved to help develop better energy models, and help building owners optimize their investments in energy renovation projects? The results indicate that even though monitoring building performance in the Netherlands could be considered common practice, the results are seldomly reported or communicated. Furthermore, very few projects monitor indoor conditions and occupants’ behaviour. As a consequence, the performance gaps found in these projects are not fully understood. These findings are summarised to provide an overview of the main goals for monitoring from a practical point of view. These findings are used to provide recommendations for monitoring setups according to the final goals.@en

The HVAC sector is essential to realize the energy transition and is facing numerous challenges like educating enough HVAC engineers to carry out the task and being able to integrate knowledge from the construction, energy, IT and health sectors and to cope with rapid technological changes. The availability of structured and easy-to-follow courses on HVAC and energy systems for buildings at higher education level could help to motivate (future) engineers to contribute to the HVAC sector, and to understand how challenging and high-tech it is. Such a course program would ideally also bring a basic understanding of the field to architects and building engineers, in such a way that a better common ground is created for collaboration and integrated design. It would also be useful to Machine Learning and Artificial Intelligence experts joining the HVAC sector. Last but not least, it could help bridging the gap between engineering and policy making, by here too, offering common views on primary energy, resource depletion and CO2 emissions relating to HVAC systems. The paper describes the structure and content of such an on-line course program. It was developed based on years of teaching experience with international master students of Mechanical Engineering, Civil Engineering, Architecture, Technical Management and Policy, Electrical Engineering and with professionals from housing associations, ministries and municipalities. The choices for the program structure, based on systems engineering, are underpinned and explained, as well as the choices for specific contents. Additionally, experience with the development of self-assessment tools for students, and self-paced courses is shared, as well as the feed-back from students. A first version of the course program was tested on the edX platform with more than 5000 students participating in each module and is publicly available.@en

Low Temperature Heating (LTH) of buildings is a key feature when switching to renewable energy. Even when the capacity of LTH is high enough, LTH may adversely affect indoor thermal comfort in case buildings are not suitably insulated. This paper goes deeper into methodological issues when conducting a thermal comfort assessment. Thermal comfort is either quantified by Fanger’s Predicted Mean Vote (PMV) or ranked in building comfort classes in the adaptive model. In both cases, one of the main parameters influencing comfort is the Mean Radiant Temperature (MRT). This study addresses issues with common MRT and PMV calculations in energy simulation software. The case study is TRNSYS 17. Several MRT and PMV calculation methods are compared, showing possible draw-backs and deviations from comfort standards NEN-EN ISO 7726 and 7730. For instance, in the standard heating settings in TRNBuild only the total heating capacity is specified. The radiative part is then distributed area-weighted over opaque surfaces. A more detailed option in TRNBuild is to specify the locations of radiative gains as points. In both cases, the MRT at a comfort sphere is calculated with Gebhardt-factors instead of view-factors. The standard settings may be considered too simplified for detailed comfort studies whereas the detailed model shows deviations from comfort standards NEN-EN ISO 7726 and 7730. Therefore, two additions to these models are proposed to increase accuracy. One addition is an ordinary detailed model with radiative gains as point sources in order to retrieve all surface temperatures during a desired period of time. In the second addition walls with radiators are split-up and planes are added at the locations of radiators to generate a view-factor matrix. This can be done in TRNBuild, but also in other view-factor calculation software. From model 1 all surface temperatures are retrieved. Combined with the view-factors from model 2, the MRT can be calculated.@en

In this review article, our main goal is understanding the Networked Learnings used for professional development. Networked learning can be defined as a form of learning where information and communication technology (ICT) can be used to promote connections between learners and their peers, learners and tutors and learners and learning resources. Such networks play an important role in professional development of employees in different sectors, from high tech industries to traditional businesses, and in both formal teaching and educational programs and informal learning activities. In this review, we explore how networked learning contexts, domains, and levels of scale are practiced and reported in the academic literature. And finally, we will investigate support technologies that have been used to facilitate networked learning for professional development. @en

Buildings need to be carefully operated and maintained for optimum health, comfort, energy performance, and utility costs. The increasing use of Machine Learning combined with Big Data in the building services sector has shown the potential to bring energy efficiency and cost-effectiveness. Therefore, upskilling and reskilling the current workforce is required to realize new possibilities. In addition, sharing and preserving knowledge are also required for the sustainable growth of professionals and companies. This formed the basis for the Dutch Research Council funded TransAct project. To increase access to education on the job, online learning is experiencing phenomenal growth. A study was conducted with two focus groups - professionals of a building service company and university researchers - to understand the existing challenges and the ways to improve knowledge sharing and upskilling through learning on the job. This study introduced an Enterprise Social Network platform that connects members and may facilitate knowledge sharing. As a community forum, Yammer from office 365 was used. For hosting project files, a SharePoint page was created. For online courses, the company’s online learning site was utilized. The log data from the online tools were analysed, semi-structured interviews and webinars were conducted and feedback was collected with google forms. Incentive models like social recognition and innovative project results were used to motivate the professionals for online activities. This paper distinguishes the impacts of initiatives on the behaviour of university researchers vs company employees.@en

In practice, faults in building installations are seldom noticed because automated systems to diagnose such faults are not common use, despite many proposed methods: they are cumbersome to apply and not matching the way of thinking of HVAC engineers. Additionally, fault diagnosis and energy performance diagnosis are seldom combined, while energy wastage is mostly a consequence of component, sensors or control faults. In this paper new advances on the 4S3F diagnose framework for automated diagnostic of energy waste in HVAC systems are presented. The architecture of HVAC systems can be derived from a process and instrumentation diagram (P&ID) usually set up by HVAC designers. The paper demonstrates how all possible faults and symptoms can be extracted on a very structured way from the P&ID, and classified in 4 types of symptoms (deviations from balance equations, operational states, energy performances or additional information) and 3 types of faults (component, control and model faults). Symptoms and faults are related to each other through Diagnostic Bayesian Networks (DBNs) which work as an expert system. During operation of the HVAC system the data from the BMS is converted to symptoms, which are fed to the DBN. The DBN analyses the symptoms and determines the probability of faults. Generic indicators are proposed for the 4 types of symptoms. Standard DBN models for common components, controls and models are developed and it is demonstrated how to combine them in order to represent the complete HVAC system. Both the symptom and the fault identification parts are tested on historical BMS data of an ATES system including heat pump, boiler, solar panels, and hydronic systems. The energy savings resulting from fault corrections are estimated and amount 25%. Finally, the 4S3F method is extended to hard and soft sensor faults. Sensors are the core of any FDD system and any control system. Automated diagnostic of sensor faults is therefore essential. By considering hard sensors as components and soft sensors as models, they can be integrated into the 4S3F method.@en

In practice, automated energy performance fault diagnosis systems are seldom installed in HVAC systems. The main reason is that a specific Fault Detection and Diagnosis (FDD) setup is time-consuming and expensive because the existing methods are component-specific, not aligned with HVAC design practices, and not fully automated. 4S3F (four symptoms three faults) method, based on system engineering and Diagnostic Bayesian Networks (DBN), was proposed to decrease the gap between the design of HVAC systems for buildings and energy performance diagnosis, and proofs of concepts were tested on diverse parts of the HVAC system of one specific building. In order to test the further applicability potential of the method, it is necessary to expand these tests and to study possible problems arising in practice, like the lack of sensors installed in a specific system or practical difficulties in the construction of the 4S3F Bayesian network by HVAC or control. However, due to the small number of validations carried out on the environment, parameters, and installation process of this method still need further discussion and refinements. In this paper, we investigate how to construct the DBN for the quite generic AHU (Air Handling Unit) of a, with mechanical supply and exhaust, heating and cooling coils, and heat recovery. The paper describes the possible DBN's depending on the technical design and the measurement points. The diverse Bayesians networks are compared, and it is concluded that also, with a limited number of sensors, a diagnostic network can be set up. It is also concluded that step-by-step instructions would be needed to facilitate the work of HVAC engineers when setting up the diagnosis model.@en

After the thermal renovation of a dwelling, there exists a gap between the actual and predicted energy performance. One of the reasons contributing to this gap is the poor assumptions of building thermal characteristics during the prediction stage. Nowadays, smart meters for gas and electricity, and home automation systems are becoming increasingly prominent in dwellings. Hence, there is potential to use the on-board monitored data from these sources to estimate the thermal characteristics of the actual dwellings. If it was possible to measure everything in a dwelling, then the estimation of these characteristics would become easy. However, the amount of data from the dwellings is limited. Hence with the available data, assumptions have to be made to estimate characteristics reflective of the actual dwelling. Therefore, this study investigates the impact these assumptions have on the estimated characteristics. First, a simple equation requiring minimum data is formulated to represent the heat dynamics in a building. Then, the characteristics are determined for one Dutch dwelling for the following conditions: 1. Different measurement periods, 2. Different time granularities, 3. With total (space heating + domestic hot water) and decomposed (only space heating) gas consumption data, 4. With different representations of indoor air temperature, and 5. Using electricity data to account for internal heat gains. In general, the estimated characteristics deviated for all the conditions. And thus, this study establishes the importance of well-chosen on-board monitored data.

@en

Many researchers have indicated the energy performance gap (difference between actual and predicted energy used in buildings), not only on an individual building level, but also on a building stock level. For policy makers it is important that predictions are correct on an building stock level to make them a useful tool to predict the effect of their proposed energy saving policies. Often not all input parameters for building energy simulations are known (e.g. insulation rates are often only possible to determine with destructive inspection or extensive measurements), therefore assumptions are made (e.g. assumptions for insulation rates are often made based on construction year). It is expected that a large part of the energy performance gap on building stock level are caused by incorrect assumptions of the unknown parameters in the building simulations. Previous research has shown that automated calibration of the assumptions on building stock level seems a promising method to reduce the energy performance gap and therewith make building energy simulations on building stock level a more reliable tool for policy makers. The previous research about calibration on building stock level was a proof of concept and still needs some improvements before it can be applied in practice. One of the aspects to improve the method is to determine the most suitable objective function and the most suitable optimization algorithm. In this paper we compare different objective functions (e.g. Root Mean Square Error, Mean Absolute Error, Sum of Absolute Errors). Next to that we compare different optimization algorithms (e.g. Genetic Algorithm, Particle Swarm and simulated Annealing Algorithm). For the comparison of the objective functions and the algorithms the former Dutch calculation method to determine the energy label in dwellings is used, in combination with the SHAERE database and data from the Dutch Statistics. The SHAERE database contains all input information on individual dwelling level to calculate the energy label of a dwelling of almost 2 million dwellings. The Dutch Statistics database contains the individual annual energy use of all dwelling of the Netherlands and can be linked to the SHAERE database.@en

A comparative study for a representative case district in the Netherlands with around 2500 dwellings has been executed with the purpose of identifying the best solution for a Low Temperature District Heating (LTDH) with Low Temperature Geothermal Heat (LTGH) as the main heat source. The district is presently connected to the Dutch gas network. Its heat demand and building typologies are used as departing points. The comparison is based on 3 key performances indicators (KPI’s): CO2 emissions, levelized costs of energy (LCOE) and peak electricity use. The following LTDH designs have been considered: Central heat pump and collective peak supply at 70 ⁰C / 50 ⁰C; Central heat pump and decentral peak supply at 70 ⁰C / 50 ⁰C; Decentralized heat pumps using 30 ⁰C supply temperature. Individual air source heat pumps for space heating and electric boilers for domestic hot water purposes are taken as reference. The LTDH concept with decentralized heat pumps saves 19.5 %, 16.8 % and 36 % on the CO2 emission, LCOE, and electricity use in peak load hours compared to the reference concept.@en

There are indications that energy-retrofitted buildings can create risks for indoor environmental quality (IEQ) and therefore for health and comfort of occupants. A review was conducted to identify and verify those risks, within three themes: building envelope, heating, ventilating and air conditioning (HVAC)-systems, and occupants. Publications from the last five years in major peer-reviewed journals from different fields (energy, buildings, indoor air, social sciences) were found by using a variety of keywords (health effects, occupant behaviours, energy-efficient retrofitting, etc.). For the building envelope, retrofitted buildings tend to be air-tighter and more thermally insulated. Hence, humidity problems, build-up of pollutants, and overheating may occur. Installing HVAC-systems and issues within (ducts, filters, maintenance, noise) may also compromise IEQ. Although relationships are difficult to establish, evidence shows that certain retrofits increase the risk of health problems, particularly for airways, skin, and eyes. Despite the installation of energy-retrofitting technologies, not all buildings lower their energy consumption. This is partly due to occupants (behaviours, preferences, needs, awareness) and partly due to technical issues. The studies reviewed, mainly focused on the performance gaps of energy-retrofitted homes and on energy-saving measures. “Comfort” and “health” tend to be disregarded, with both being seldom measured and only assessed by simulation. Occupant behaviours, preferences, and needs are understudied and need to be incorporated into the research and development of retrofitting measures. More interdisciplinary approaches are needed, in which buildings & HVAC-systems, occupants, health and comfort, and IEQ are investigated as interacting elements and based on an integrated approach.@en

Current methods for energy diagnosis in heating, ventilation and air conditioning (HVAC) systems are not consistent with process and instrumentation diagrams (P&IDs) as used by engineers to design and operate these systems, leading to very limited application of energy performance diagnosis in practice. In a previous paper, a generic reference architecture – hereafter referred to as the 4S3F (four symptoms and three faults) framework – was developed. Because it is closely related to the way HVAC experts diagnose problems in HVAC installations, 4S3F largely overcomes the problem of limited application. The present article addresses the fault diagnosis process using automated fault identification (AFI) based on symptoms detected with a diagnostic Bayesian network (DBN). It demonstrates that possible faults can be extracted from P&IDs at different levels and that P&IDs form the basis for setting up effective DBNs. The process was applied to real sensor data for a whole year. In a case study for a thermal energy plant, control faults were successfully isolated using balance, energy performance and operational state symptoms. Correction of the isolated faults led to annual primary energy savings of 25%. An analysis showed that the values of set probabilities in the DBN model are not outcome-sensitive.

@en