Data-driven Analytics for Sustainable Buildings and Cities

A recent surge of interest in the building energy consumption has generated a tremendous amount of energy data, which boosts the data-driven algorithms for broad application throughout industry. This chapter reviews the prevailing data-driven approaches used in building energy analysis under different archetypes and granularities including those for prediction (artificial neural networks, support vector machines, statistical regression, decision tree and genetic algorithm) and those for classification (K-mean clustering, self-organizing map and hierarchy clustering). To be specific, we introduce the fundamental concepts and major technical features of each approach, together summarizing its current R&D status and practical applications while pointing out existing challenges in their development for prediction and classification of building energy consumption. The review results demonstrate that the data-driven approaches, although they are constructed based on less physical information, have well addressed a large variety of building energy related applications, such as load forecasting and prediction, energy pattern profiling, regional energy-consumption mapping, benchmarking for building stocks, global retrofit strategies and guideline making etc. Significantly, this review refines a few key tasks for modification of the data-driven approaches in the contexts of application to building energy analysis. The conclusions drawn in this review could facilitate future micro-scale changes of energy use for a particular dwelling through appropriate retrofit in building envelop and inclusion of renewable energy technologies. They also pave an avenue to explore potential in macro-scale energy-reduction with consideration of customer demands. All these will be useful to establish a better long-term strategy for urban sustainability.

Occupancy behaviour plays an important role in energy consumption in buildings. Currently, the shallow understanding of occupancy has led to a considerable performance gap between predicted and measured energy use. This chapter presents an approach to estimate the occupancy based on blind system identification (BSI), and a prediction model of electricity consumption by an air-conditioning system is developed and reported based on an artificial neural network with the BSI estimation of the number of occupants as an input. This starts from the identification of indoor CO₂ dynamics derived from the mass-conservation law and venting levels. The unknown parameters, including the occupancy and model parameters, are estimated by using a frequentist maximum-likelihood algorithm and Bayesian estimation. The second phase is to establish the prediction model of the electricity consumption of the air-conditioning system by using a feed-forward neural network (FFNN) and extreme learning machine (ELM), as well as ensemble models. To analyse some aspects of the benchmark test for identifying the effect of structure parameters and input-selection alternatives, three studies are conducted on (1) the effect of predictor selection based on principal component analysis, (2) the effect of the estimated occupancy as the supplementary input, and (3) the effect of the neural network ensemble. The result shows that the occupancy number, as the input, is able to improve the accuracy in predicting energy consumption using a neural-network model.

In building performance simulation, occupant behaviour contributes to large uncertainties, which often lead to considerable discrepancies between actual energy consumption and simulation results. This chapter aims to extract occupant-behaviour related electricity load patterns using classical K-means clustering approach at the initial investigation stage. Smart-metering data from a case study in Shanghai, China, was used for the load pattern analysis. The electricity load patterns of occupants were examined on a daily/weekly/seasonal basis. According to their load patterns, occupants were categorized as (a) white-collar workers, (b) poor or older families and (c) rich or young families. The daily patterns indicated that electricity use was much more random and fluctuated over a wide range. Most households of the monitored communities consumed relatively-low electricity; the characteristic double peak with higher level of consumption in the morning and evening were only apparent in a relatively small subset of residents (mostly white-collar workers). The weekly analysis found that significant load shifting towards weekend days occurred in the poor or old family group. The electricity saving potential was greatest in the white-collar workers and the rich or young family groups. This study concludes with recommendations to stakeholders utilizing our load profiling results. The research provides a rare insight into the electricity-use-related occupant behaviours of Shanghai residents through the case study of two communities. The findings of the study are also presented in a meaningful way so that they can directly aid the decision-making of governments and other stakeholders interested in energy efficiency. The research results are also relevant to the building energy simulation community as they are derived from observations, and thus can have the potential to improve the efficiency and accuracy of numerical simulation results.

The lighting system accounts for 8% of the total electricity consumption in commercial buildings in the United States and 12% of the total electricity consumption in public buildings globally. This consumption share can be effectively reduced using the demand-response control. The traditional lighting system control method commonly depends on the real-time occupancy data collected using the passive infrared (PIR) sensor. However, the detection inaccuracy of the PIR sensor usually results in false-offs. To diminish the false-error frequency, the existing lighting system control simply deploys a delayed reaction period (e.g., 5–20 min), which is not sufficiently accurate for the demand-response operation. Therefore, in this research, a novel data-driven model predictive control (MPC) method that is based on the temporal sequential-based artificial neural network (TS-ANN) is proposed to overcome this challenge using an updated historical occupancy status. Using an office as case study, the proposed model is also compared with the traditional lighting system control method. In the proposed model, the occupancy data was trained to predict the occupancy pattern to improve the control. It was found that the occupancy prediction mainly correlates with the historical occupancy ratio and the time sequential feature. The simulation results indicated that the proposed method achieved higher accuracy (97.4%) and fewer false-offs (from 79.5 with traditional time delay method to 0.6 times per day) are achieved by the MPC model. The proposed TS-ANN-MPC method integrates the analysis of the occupant behaviour routine into on-site control and has the potential to further enhance the control performance practice for maximum energy conservation.

Net-zero energy building (NZEB) is widely considered as a promising solution to the current energy problem. The existing NZEBs are designed using the historical weather data (e.g. typical meteorological year-TMY). Nevertheless, due to climate change, the actual weather data during a NZEB’s lifecycle may differ considerably from the historical weather data. Consequently, the designed NZEBs using the historical weather data may not achieve the desired performances in their lifecycles. Therefore, this study investigates the climate change impacts on NZEB lifecycle performance (i.e., energy balance, thermal comfort and grid interaction) in different climate regions, and also evaluates different measures’ effectiveness in mitigating the associated impacts of climate change. In the study, the multi-year future weather data in different Chinese climate regions are firstly generated using the morphing method. Then, using the generated future weather data, the lifecycle performances of the NZEBs, designed using the TMY data, are assessed. Next, to mitigate the climate change impacts, different measures are adopted and their effectiveness is evaluated. The study results can improve understanding of the climate change impacts on NZEB lifecycle performance in different climate regions. They can also help select proper measures to mitigate the climate change impacts in the associated climate regions.

The solar energy share in Sweden will grow up significantly in next a few decades. Such transition offers not only great opportunity but also uncertainties for the emerging solar photovoltaic/thermal (PV/T) technologies. This chapter therefore aims to conduct a techno-economic evaluation of a reference solar PV/T concentrator in Sweden for building application. An analytical model is developed based on the combinations of Monte Carlo simulation techniques and multi energy-balance/financial equations, which takes into account of the integrated uncertainties and risks of various variables. In the model, 11 essential input variables, i.e. average daily solar irradiance, electrical/thermal efficiency, prices of electricity/heating, operation & management (OM) cost, PV/T capital cost, debt to equity ratio, interest rate, discount rate, and inflation rate, are considered, while the economic evaluation metrics, such as levelized cost of energy (LCOE), net present value (NPV), and payback period (PP), are primarily assessed. According to the analytical results, the mean values of LCOE, NPV and PP of the reference PV/T connector are observed at 1.27 SEK/kW h (0.127 €/kW h), 18,812.55 SEK (1881.255 €) and 10 years during its 25 years lifespan, given the project size at 10.37 m² and capital cost at 4482–5378 SEK/m² (448.2–537.8 €/m²). The positive NPV indicates that the investment on the selected PV/T concentrator will be profitable as the projected earnings exceeds the anticipated costs, depending on the NPV decision rule. The sensitivity analysis and the parametric study illustrate that the economic performance of the reference PV/T concentrator in Sweden is mostly proportional to solar irradiance, debt to equity ratio and heating price, but disproportionate to capital cost and discount rate. Together with additional market analysis of PV/T technologies in Sweden, it is expected that this chapter could clarify the economic situation of PV/T technologies in Sweden and provide a useful model for their further investment decisions, in order to achieve sustainable and low-carbon economics, with an expanded quantitative discussion of the real economic or policy scenarios that may lead to those outcomes.

Occupants often perform many types of behaviour in buildings to adjust the indoor thermal environment. In these types, opening/closing the windows, often regarded as window-opening behaviour, is more commonly observed because of its convenience. It not only improves indoor air quality to satisfy occupants’ requirement for indoor thermal comfort but also influences building energy consumption. To learn more about potential factors having effects on occupants’ window-opening behaviour, a field study was carried out in an office building within a university in Beijing. Window state (open/closed) for a total of 5 windows in 5 offices on the second floor in 285 days (9.5 months) were recorded daily. Potential factors, categorized as environmental and non-environmental ones, were subsequently identified with their impact on window-opening behaviour through logistic regression and Pearson correlation approaches. The analytical results show that occupants’ window-opening behaviour is more strongly correlated to environmental factors, such as indoor and outdoor air temperatures, wind speed, relative humidity, outdoor PM2.5 concentrations, solar radiation, sunshine hours, in which air temperatures dominate the influence. While the non-environmental factors, i.e. seasonal change, time of day and personal preference, also affects the patterns of window-opening probability. This chapter provides solid field data on occupant window opening behaviour in China, with high resolutions and demonstrates the way in analyzing and predicting the probability of window-opening behaviour. Its discussion into the potential impact factors shall be useful for further investigation of the relationship between building energy consumption and window-opening behaviour.

Classical building control systems are becoming vulnerable with increasing complexities in contemporary built environments and energy systems. Due to this, the reinforcement learning (RL) method is becoming more distinctive and applicable in control networks for buildings. This chapter, therefore, conducts a comprehensive review of RL techniques applied in control systems for occupant comfort in indoor built environments. The empirical applications of RL-based control systems are presented, depending on comfort objectives (thermal comfort, indoor air quality, and lighting) along with other objectives which invariably includes energy consumption. The class of RL algorithms and implementation details regarding how the value functions have been represented and how the policies are improved are also illustrated. This chapter shows there are limited works for which RL has been explored for controlling occupant comfort, especially in indoor air quality and lighting. Relatively few of the reviewed works incorporate occupancy patterns and/or occupant feedback into the control loop. Moreover, this chapter identifies a gap with regard to the performance of implementing cooperative multiagent RL (MARL). Based on our findings, current challenges and further opportunities are discussed. We expect to clarify the feasible theory and functions of RL for building control systems, which would promote their widespread application in built environments.

An occupant’s window opening and closing behaviour can significantly influence the level of comfort in the indoor environment. Such behaviour is, however, complex to predict and control conventionally. This chapter, therefore, proposes a novel reinforcement learning (RL) method for the advanced control of window opening and closing. The RL control aims at optimising the time point for window opening/closing through observing and learning from the environment. The theory of model-free RL control is developed with the objective of improving occupant comfort, which is applied to historical field measurement data taken from an office building in Beijing. Preliminary testing of RL control is conducted by evaluating the control method’s actions. The results show that the RL control strategy improves thermal and indoor air quality by more than 90% when compared with the actual historically observed occupant data. This methodology establishes a prototype for optimally controlling window opening and closing behaviour. It can be further extended by including more environmental parameters and more objectives such as energy consumption. The model-free characteristic of RL avoids the disadvantage of implementing inaccurate or complex models for the environment, thereby enabling a great potential in the application of intelligent control for buildings.

The Predicted Mean Vote (PMV) model is extensively used by current thermal comfort standards, such as ASHRAE 55 and ISO 7730, despite its discrepancy in predicting Thermal Sensation (TS). The implicit assumption is that PMV can be applied for predicting TS of a large population. Our statistical analysis of a subset of ASHRAE global database of thermal comfort field study shows that occupants’ expectations towards TS are affected by factors that are not accounted for in the classic PMV model, such as climate, building type, age group, season and gender. The influences of the climate and building type are more determinant. An adapted PMV (PMVa) model and an adaptation table were developed based on the selected samples to reduce this discrepancy. After adaptation, the medians of each category corresponding to the discrepancy are zero or near zero. The results also show that the adapted PMV outperforms the classic PMV in predicting TS, while increasing the overall accuracy from 36 to 39%.

PMV (Predicted Mean Vote) model is currently the most extensively used method for thermal sensation (TS) evaluation. However, the model is criticized for not being able to account for human thermal preferences and expectations. In response to these limitations, the adaptive model was developed to factor in behavioural, psychological and physiological adjustments, but it overlooks important factors such as clothing insulation, activity level and the indoor thermal environment. This chapter therefore proposes a prediction accuracy weighted voting ensemble (PAWVE) method for TS evaluation. Feature selection was used to identify the important features contributing most to TS. Data resampling was applied to improve the classification performance of the imbalanced data extracted from the ASHRAE global database. Five classifiers were eventually selected for the ensemble method based on their prediction accuracies. Results indicate that resampling is important for improving the performance of the minority classes. PAWVE with data resampling outperformed the other models, with an overall accuracy and F1-score both at 0.67. When compared with the traditional PMV model, the performance has been improved by 72%.

The particulate matters (PM10 and PM2.5) inside urban subway stations greatly influence indoor air quality and passenger comfort. This study aims to analyze and interpret the concentrations of PM10 and PM2.5, measured in several subway stations from October 9th to 22nd, 2016 in Beijing, China. The overall methodology was based on the Statistical Package for Social Science (SPSS) software while General linear model (GLM) and correlation analysis were further applied to examine the sensitivities of different variables to the particle concentrations. The data analysis showed the average overall mass ratio of PM concentrations inside subway station is about 68.7%, much lower than outdoor condition (79.6%). In the areas of the station hall and platform, the real-time PM10 and PM2.5 concentrations varied periodically. In working and operation offices, all rooms had much higher PM concentrations than the outdoor environment when its pollution level was level 3, in which the facility room reached the highest level, while the closed meeting room had the lowest. Correlation analysis results indicated that PM10 and PM2.5 concentrations were mutually correlated (average R² = 0.854), and a strong linear correlation (R² = 0.897) of the subway-station PM concentrations to the outdoor PM conditions, regardless of the outdoor atmospheric PM concentrations pollution level was. Nevertheless, the impact of passenger number and temperature and humidity on the station PM concentrations was less, when compared to the outdoor environment. This chapter is expected to provide useful information for further research and design of effective prevention measures on PM in local subway stations, towards a more sustainable and healthier built environment in the city underground.

Smart grid is triggering the transformation of traditional electricity consumers into electricity prosumers. This chapter reports a case study of transforming an existing residential cluster in Sweden into electricity prosumers. The main energy concepts include (1) click-and-go photovoltaics (PV) panels for building integration, (2) centralized exhaust air heat pump, (3) thermal energy storage for storing excess PV electricity by using heat pump, and (4) PV electricity sharing within the building cluster for thermal/electrical demand (including electric vehicles load) on a direct-current micro grid. For the coupled PV-heat pump-thermal storage-electric vehicle system, a fitness function based on genetic algorithm is established to optimize the capacity and positions of PV modules at cluster level, with the purpose of maximizing the self-consumed electricity under a non-negative net present value during the economic lifetime. Different techno-economic key performance indicators, including the optimal PV capacity, self-sufficiency, self-consumption and levelized cost of electricity, are analysed under impacts of thermal storage integration, electric vehicle penetration and electricity sharing possibility. Results indicate that the coupled system can effectively improve the district-level PV electricity self-consumption rate to about 77% in the baseline case. The research results reveal how electric vehicle penetrations, thermal storage, and energy sharing affect PV system sizing/positions and the performance indicators, and thus help promote the PV deployment.

Existing studies have developed some advanced building side controls that enable renewable energy sharing and that aim to optimize building-cluster-level performance via regulating the energy storage charging/discharging. However, the flexible demand shifting ability of electric vehicles is rarely considered. For instance, the electric vehicle charging will usually start once they are plugged into charging stations. But, in such charging period the renewable generation may be insufficient to cover the EV charging load, leading to grid electricity imports. Consequently, the building-cluster-level performance is not optimized. Therefore, this study proposes a coordinated control of building prosumers for improving the cluster-level performance, by making use of energy sharing and storage capability of electricity batteries in both buildings and EVs. An EV charging/discharging model is first developed. Then, based on the predicted future 24 h electricity demand and renewable generation data, the coordinated control first considers the whole building cluster as one ‘integrated’ building and optimizes its operation as well as the EV charging/discharging using genetic algorithm. Next, the operation of individual buildings in the future 24 h is coordinated using nonlinear programming. For validation, the developed control has been tested on a real building cluster in Ludvika, Sweden. The study results show that the developed control can increase the cluster-level daily renewable self-consumption rate by 19% and meanwhile reduce the daily electricity bills by 36% compared with the conventional controls.

Utilizing renewable energy to meet the energy demand, net-zero energy building (NZEB) is considered a promising solution to the worsening energy and environmental problems. Due to the intermittent and unstable characteristics of renewable energy (e.g. solar energy), NZEB needs to frequently exchange energy with the power grid. Such frequent energy interactions can impose negative impacts on the grid in terms of power balance and voltage stability. Existing studies demonstrated that there exist many influential parameters to NZEB grid interaction. However, the impacts of influential parameters have not been systematically compared and the key parameters with critical impacts are still unknown. Without knowing the key parameters, researchers may mistakenly optimize non-critical parameters, thereby leading to limited performance improvements; or they have to take parameters more than necessary into consideration, thereby causing unnecessarily high computation loads. Therefore, this study proposes a novel method to identify the key parameters affecting NZEB grid interactions. In the method, global sensitivity analysis is adopted to quantitatively compare the impacts of 24 influential parameters in three major performance aspects including over/under voltage, grid dependence and energy loss. Meanwhile, Monte-Carlo method is used to simulate the parameter uncertainties. The identified key parameters have been verified through comparing their performance improvements and computation loads. Providing an effective way to identify key parameters out of numerous ones, the study results can substantially reduce the unnecessary considerations of non-critical parameters in design optimizations. Also, the identified key parameters can be used for improving NZEB grid interaction with limited computing power requirement.

Solar photovoltaic (PV) is becoming one of the most significant renewable sources for positive energy district (PED) in most countries, including Sweden. The lack of innovative business models and financing mechanisms are one of the main constraints for PV’s deployment installed in local community. This chapter therefore analyses a set of peer-to-peer (P2P) business model for 48 individual building prosumers with PV installed in a Swedish community. It considers energy use behaviour, electricity/financial flows, ownerships, and trading rules in a local electricity market. Different local electricity markets are designed and studied using agent-based modelling technique, with different energy demands, cost–benefit schemes and financial hypotheses for an optimal evaluation. This chapter provides an early insight into a vast research space, i.e. the operation of an energy system through the constrained interaction of its constituting agents. The agents (48 households) show varying abilities in exploiting the common PV resource, as they achieve very heterogeneous self-sufficiency levels (from ca. 15 to 30%). The lack of demand side management suggests that social and lifestyle differences generate huge impacts on the ability to be self-sufficient with a shared, limited PV resource. Despite the differences in self-sufficiency, the sheer energy amount obtained from the shared PV correlates mainly with annual cumulative demand.

The energy shortage is one key issue for sustainable development, a potential solution of which is the integration with the renewable energy resources. However, the temporal sequential characteristic of renewable resources is different from traditional power grid. For the entire power grid, it is essential to match the energy generation side with the energy consumption side, so the load characteristic at the energy use side is crucial for renewable power integration. Better understanding of energy consumption pattern in buildings contributes to matching different source of energy generation. Under the background of integration of traditional and renewable energy, this research focuses on analysis of different household electricity consumption patterns in an urban scale. The original data is from measurement of daily energy consumption with smart meter in households. To avoid the dimension explosion phenomenon, the auto-encoder algorithm is introduced during the clustering analysis of daily electricity use data, which plays the role of principal component analysis. The clustering based on auto-encoder gives a clear insight into the urban electricity use patterns in household. During the data analysis, several feature variables are proposed, which include peak value, valley value and average value. The distinction analysis is also conducted to evaluate the analysis performance. The chapter takes households in Nanjing city, China as a case study, to conduct the clustering analysis on electricity consumption of residential buildings. The analysis results can be further applied, such as during the capacity design of district energy storage.

Urban energy mapping plays a crucial role in benchmarking the energy performance of buildings for many stakeholders. This study examined a set of buildings in the city of Borlänge, Sweden, owned by the municipality. The aim was to present a digital spatial mapping of both electricity use and district heating demand. A toolkit for top-down data processing and analysis was considered based on the energy performance database of municipality-owned buildings. The data were initially cleaned and transformed using the Feature Manipulation Engine tool (FME) and then it was geocoded using a python script with an application program interface (API) for OpenStreetMap. The dataset consists of 221 and 89 geocoded addresses for, respectively, electricity and district heating monthly consumption for the year 2018. The electricity use and heating demand in the building samples were about 24.06 kWh/m² and 190.99 kWh/m² respectively, for which great potential for saving heating energy was observed. The digital mapping revealed a spatial representation of identifiable hotspots for electricity uses in high-occupancy/density areas and for district heating needs in districts with buildings mostly constructed before 1980. These results will provide a comprehensive means of understanding the existing energy distributions to stakeholders and energy advisors. They also facilitate strategy geared towards future energy planning in the city, such as energy benchmarking policies.

Positive Energy Districts (PED) require integration of different systems and infrastructures for the optimal interactions among buildings, stakeholders, mobility, energy systems and ICT systems. Digital twin is a coupled approach for new forms of modelling and analysis based on big data and machine learning/artificial intelligence, which combines capacities of virtual model, data management, analytics, simulation, system controls, visualization and information sharing. Digital twin is regarded as a potential solution to optimize PEDs. This chapter presents a comprehensive review about digital twins for PED from aspects of concepts, working principles, tools/platform and applications, in order to address the issues of both ‘how digital PED twin is made’ and ‘how digital PED twin optimizes liveability’. Further challenges and opportunities are brought forward for discussion. The outcome of the review is expected to provide useful information for optimizing the liveability of the urban environment in line with social, economic and environmental sustainability.

Solar photovoltaic thermal (PVT) is an emerging technology, capable of producing electrical and thermal energy using a single collector. However, to achieve larger market penetration for this technology, it is imperative to have an understanding of the energetic performance for different climatic conditions and the economic performance under various financial scenarios. This chapter thus presents a techno-economic evaluation of a typical water based PVT system for electricity and domestic hot water applications in 85 locations worldwide. The simulations are performed using a validated tool with one-hour time step for output. The thermal performance of the collector is evaluated using energy utilization ratio and exergy efficiency as key performance indicators, which are further visualized by the digital mapping approach. The economic performance is assessed using net present value and payback period under two financial scenarios: (1) total system cost as a capital investment in the first year; (2) only 25% of total system cost is a capital investment and remaining 75% investment is considered with financing period with certain interest rate. The results show that such a PVT system has better energy and exergy performance for the locations with a low annual ambient temperature and vice versa. Furthermore, it is seen that the system boundaries, such as load profile, hot water storage volume, etc., can have a significant effect on the annual energy production of the system. Economic analysis indicates that the average net present values per unit collector area are 1800 € and 2200 € respectively among the 85 cities for financial model 1 and financial model 2. Nevertheless, from the payback period point of view, financial model 1 is recommended for the locations with high interest rate. The study is helpful to set an understanding of general factors influencing the techno-economic performance of PVT systems.