Building Performance Data Research Articles

Empirical hurricane fragility assessment of elevated and slab-on-grade residential houses

The widespread hurricane-induced damage to residential houses in the United States leads to severe social and economic impacts. Recent community-wide efforts by researchers within the natural hazards community to collect and curate post-hazard damage data have advanced the potential to assess structural vulnerability during extreme wind events. These efforts have led to data re-use and further analysis for detailed evaluation, allowing for the comparison of the performance of elevated versus slab-on-grade single-family residential houses during hurricanes. The empirically derived fragility functions from these damage assessment data offer a unique opportunity to evaluate the vulnerability of constructed buildings. This paper presents the development of empirical fragility functions for elevated and slab-on-grade residential houses, along with their building envelope components, when subjected to hurricane winds. It assesses the post-hurricane occupiability and the annual probability of failure for both elevated and slab-on-grade houses. Utilizing post-hurricane building performance data collected from 618 elevated and 1188 slab-on-grade single-family residential houses impacted by Hurricanes Irma and Harvey in 2017, we analyzed the relationship of damage to the estimated peak 3-sec gust wind speed at each house. The analysis yields the assignment of each house to one of five damage state categories, from which we developed the fragility functions of buildings and building components in accordance with FEMA's HAZUS Hurricane model. Results showed that the developed empirical fragility functions for building envelope components of elevated houses have a higher probability of failure compared to those installed on slab-on-grade houses. Additionally, the likelihood of elevated houses being unoccupiable after a hurricane is significantly higher than that of slab-on-grade houses in both Florida and Texas. For residential buildings constructed in the state of Florida, this study finds that elevated houses have a 39 %–74 % higher annual probability of failure due to hurricane wind compared to slab-on-grade houses for the different considered damage states.

Transfer learning integrating similarity analysis for short-term and long-term building energy consumption prediction

Currently, building energy consumption prediction models are usually based on a large amount of historical operational data in high demands of building operating hours and monitoring systems. However, many buildings may lack operational data due to relatively limited monitoring systems, causing the failure to use data-driven methods to characterize the energy profile. In this context, transfer learning is a promising method to establish the knowledge transfer between many high-quality building operation datasets and a small amount of target building data, and to help predict energy consumption in the target building. This paper studies the possibility of employing transfer learning to achieve both short and long-term building energy consumption prediction. Firstly, a similarity analysis method, based on variable modal decomposition and dynamic time warping, is proposed for identifying the source buildings with the most similar energy features to the target building. Then, transfer learning for long-term prediction air-conditioning energy consumption is developed with weather parameters generated by the Morphing method as inputs. For the short-term single-step prediction, the proposed model CV-RMSE improves 81.3% (AEC) and 77.4% (EEC), respectively, compared to the prediction model that does not implement the transfer learning strategy and directly uses the target BEC data. As for the short-term multi-step prediction, the proposed model CV-RMSE improves 62.0% (AEC) and 65.5% (EEC), respectively. For the long-term prediction, the average CV-RMSE for the whole year is 6.62% and 11.15% for the proposed and directly target domain-based model, respectively. The proposed method explores the practicality of transfer learning in building energy forecasting, contributing to the use of existing building operation data for energy management at different timespan.

Application of high performance computing based on big data in architectural interior design

In response to the problems of insufficient data management and low construction quality in architectural interior design, high-performance computing technology was proposed to be applied to optimize the overall effect of architectural interior design. This article proposed some solutions to the problems in interior design of buildings, and applied high-performance computing technology to building design materials, building operation data processing, etc. Furthermore, combined with multi-objective genetic algorithm, relevant experimental tests were conducted on the optimization of interior space environment design in buildings. The experimental results showed that under the algorithm proposed in this paper, the average health standard compliance of each scheme reached 92.17%, and the average emotional requirement satisfaction reached 88.42%. The average interface detail attention reaches 84.65%, and the average environmental comfort reaches 92.87%. Under the genetic algorithm, the average compliance of health standards for each scheme was 88.14%, and the average satisfaction of emotional needs was 84.17%. The average attention to interface details was 81.18%, and the average environmental comfort was 90.30%. From the above data, it can be seen that the algorithm proposed in this paper can achieve good optimization results in the design of indoor space environment in buildings.

Performance Analysis of Application of BIM Technology and Two-Way Progressive Optimization Algorithm in Energy-Saving Design of Residential Buildings Based on BIM Technology

The current residential energy-saving design is lagging behind, and lack of systematic design methods and scientific technical means seriously restricts process of improving quality of residential buildings in China. With development and application of digital information technology in construction field, BIM technology has brought new opportunities construction industry, and building performance data has become information basis for scientific promotion of energy conservation. This paper uses BIM technology as a platform to establish an information base of common building materials and components in Chongqing through literature research, field studies, summarization, and application practice, and to help architects make wise decisions at early design stage by using "one-click" scheme comparison and design evaluation to avoid a large number of potential energy consumption problems at a later stage, It greatly improves energy efficiency and design quality of buildings.

Assessing the Impact of Regulatory Frameworks on the Success of Green Building Projects in Singapore

This paper examines the influence of regulatory frameworks on the success of green building projects in Singapore. By analyzing the Building and Construction Authority’s (BCA) Green Mark Scheme and other related regulations, the study highlights how these frameworks have shaped the environmental sustainability practices within the construction industry. Through a mix of qualitative interviews with industry professionals and quantitative analysis of building performance data, the research demonstrates that regulatory frameworks not only enhance environmental outcomes but also drive innovation in building technologies and practices. The findings suggest that while these regulations have led to significant improvements in energy efficiency and resource management, challenges such as high compliance costs and the need for continuous professional development remain. The paper concludes with recommendations for evolving these frameworks to better support sustainable development in Singapore’s urban landscape.

Smart operational technologies programming for whole-building decarbonisation

Have you ever finished designing a new building and wondered what technologies are specified for the building? When using delegated design for building technologies, is the scope clearly defined or is it left to the vendor to decide? If you do not know the answers to these questions — or worse, you say to yourself, ‘my vendor has that’ — this paper is for you. Regardless of project delivery, if building owners want to democratise their building data, take advantage of digital twin technologies and advanced data layer capabilities and operate their buildings at the highest possible levels of performance, they must take control of their building’s operational technologies (OTs). Delegated design of OTs is typically the root cause of building owners not controlling or owning their building performance data. This paper is intended to provide building owners with the basic framework for OT programming, which is the only way to ensure ownership and control over building performance data. Our previous Corporate Real Estate Journal paper from Vol. 12, No. 3, Spring 2023 titled ‘Integration of building science and data science to de-risk an affordable strategy for building decarbonisation’ provides a higher-level view of the benefits from the integration of OTs. The convergence of global challenges requires building owners to address building performance as performance is now expected by owners, investors, occupants and governments. Merging building science and data science is the most cost-effective way to deliver expected building performance over the life of a building. Smart building infrastructure technology democratises building data and is the key to cost-effectively controlling building performance. Data is the new gold and smart operational technology programming is the new shovel.

Automated data mining framework for building energy conservation aided by generative pre-trained transformers (GPT)

Data mining technologies have showed promising capabilities in extracting building operation patterns from massive amounts of building operational data for energy conservation. However, the number of the extracted operation patterns is always large. It is time-consuming and tedious for users to find valuable operation patterns among them. To overcome this barrier, this study proposes an automated data mining framework based on maximal frequent itemset mining and generative pre-trained transformers (GPT). An improved maximal frequent itemset mining-based data mining method is developed to extract non-redundant operation patterns from numerous building operational data for reducing the number of the extracted operation patterns. A template-based prompt generation method is proposed to transform the extracted operation patterns into prompts. The prompts are inputted into GPT to determine whether there are energy waste patterns hidden in the extracted operation patterns. It liberates humans from the tedious work on analyzing the extracted operation patterns. The framework is applied to analyze the one-year operational data from a real-world building chiller plant system for verifying its performance. Most of the energy waste patterns in this system are detected successfully, such as valve faults, low chilled water outlet temperature, small chilled and cooling water temperature differences, and improper coordinated control among devices. The detection accuracy of GPT is 89.17 % for energy waste patterns and 99.48 % for normal operation patterns. The response time and cost of GPT are 6747.60 s and $17.68, respectively.

Evaluating Indoor Air Quality Monitoring Devices for Healthy Homes

In light of COVID-19, people are increasingly anxious about indoor air quality data in places where they live and work. Access to this data using a consumer-grade air quality monitor has become a way of giving agency to building users so that they can understand the ventilation effectiveness of the spaces where they spend their time. Methods: Fourteen low-cost, air quality devices marketed to consumers were tested (seven types, two of each product): AirBird, Airthings View Plus, Aranet4 Home, Awair Omni, Eve Room, Laser Egg + CO2, and Purple Air PA-1. The study focus was accuracy and useability using three methods: a low-cost laboratory setting to test accuracy for CO2; a comparison to a calibrated, research grade meter for particulate matter (PM2.5), temperature, and relative humidity; and short-term field testing in a residential environment to understand the quality of feedback given to users. Results: Relating to accuracy, all devices were within acceptable ranges for temperature, relative humidity, and CO2, and only one brand’s results met the accuracy threshold with the research grade monitor when testing PM2.5. In terms of usability, a significant variation in response time and data visualization was found on the devices or in the smartphone applications. Conclusions: While accuracy in IAQ data is important, in low-cost air quality devices marketed to consumers it is just as important that the data be presented in a way that can be used to empower people to make decisions and modify their indoor environment. We concluded that response time, user-interface, data sharing, and visualization are important parameters that may be overlooked if a study just focuses on accuracy. The design of the device, including its appearance, size, portability, screen brightness, and sound or light warning, must also be considered. The act of measuring is important, and more studies should focus on how users interpret and react to building performance data.

Integrating active learning and semi-supervised learning for improved data-driven HVAC fault diagnosis performance

Data-driven methods have drawn increasing interests in HVAC fault diagnosis tasks due to their intrinsic advantages in making real-time automated decisions. To ensure the reliability of data-driven models, it is essential to prepare sufficient labeled data for predictive modeling. In practice, it can be very time-consuming and labor-intensive to determine the actual operating condition or label of each data sample (e.g., Normal or Faulty), making it highly challenging to develop robust data-driven solutions through conventional supervised learning methods. To tackle such challenges, this study proposes a data analytic framework to integrate active learning and semi-supervised learning to utilize massive unlabeled data for improved fault diagnosis performance. More specifically, five active learning methods have been tested to quantify their effectiveness in discovering valuable unlabeled data for expert labeling. Semi-supervised data-driven models have been developed to enable autonomous knowledge discovery from unlabeled building operational data through self-training protocols. Data experiments have been conducted to explore the separated and integrated values of active and semi-supervised learning. The results show that active learning can effectively identify valuable data samples for fault diagnosis and thereby, reducing approximately 50% labeling costs. Cost-effective combinatorial strategies have been derived to integrate active learning and semi-supervised learning for practical applications. The research outcomes are valuable for developing advanced data-driven solutions with substantial decreases in manual costs.

Comparison of supervised algorithms for automated data analysis in existing buildings

Non-residential buildings are equipped with building automation systems (BAS). BAS provide a vast amount of building operation data. However, the analysis of this data is laborious and requires a scalable and automated process. The authors developed the BUDO schema for a standardized, object-oriented description of monitoring metadata. The translation process employs artificial intelligence to translate the properties of data streams (labels, time series) from BAS into the BUDO schema. The translated data is then analyzed using various applications to detect inefficiencies, such as incorrect placement of outdoor temperature sensors or a weekend shutdown that has not been properly implemented. The paper presents a process that automatically translates, structures and analyzes operational data to identify potential for energy savings. The OOM4ABDO project monitors 138 buildings to analyze and improve building operations, whose subset of data we are using here. In addition, we use an open source data set of about 70 buildings. We compare several feature extraction methods and classifiers on their ability to support each of the stages (label translation, time series translation, application). The algorithms used reach F1 scores of up to 98% for label translation, 90% for time series translation and 92% for applications.

Synconn_build: A python based synthetic dataset generator for testing and validating control-oriented neural networks for building dynamics prediction

Applying model-based predictive control in buildings requires a control-oriented model capable of learning how various control actions influence building dynamics, such as indoor air temperature and energy use. However, there is currently a shortage of empirical or synthetic datasets with the appropriate features, variability, quality and volume to properly benchmark these control-oriented models. Addressing this need, a flexible, open-source, Python-based tool, synconn_build, capable of generating synthetic building operation data using EnergyPlus as the main building energy simulation engine is introduced. The uniqueness of synconn_build lies in its capability to automate multiple aspects of the simulation process, guided by user inputs drawn from a text-based configuration file. It generates various kinds of unique random signals for control inputs, performs co-simulation to create unique occupancy schedules, and acquires weather data. Additionally, it simplifies the typically tedious and complex task of configuring EnergyPlus files with all user inputs. Unlike other synthetic datasets for building operations, synconn_build offers a user-friendly generator that selectively creates data based on user inputs, preventing overwhelming data overproduction. Instead of emulating the operational schedules of real buildings, synconn_build generates test signals with more frequent variation to cover a broader range of operating conditions.•Synconn_build is an open-source tool designed to address the lack of datasets for benchmarking control-oriented building dynamics prediction models.•The tool automates simulations, data acquisition, and EnergyPlus configuration, guided by user inputs.•Synconn_build prevents data overproduction by selectively creating data, offering a user-friendly approach to dataset generation.

Data-driven online energy management framework for HVAC systems: An experimental study

The optimization of the operation of heating, ventilation, and air conditioning (HVAC) systems has gained considerable attention due to its energy conservation potential. Collecting various building operation data presents an opportunity to improve the efficiency of energy management and reduce consumption. In this paper, a data-driven online energy management framework is proposed for enhancing the effectiveness of HVAC systems and minimizing energy consumption. To streamline the modeling process and improve the accuracy of temperature estimation, our proposed framework adopts a symbolic regression approach to estimate building thermodynamics based on the collected data. It also formulates a model predictive control (MPC)-based HVAC scheduling strategy to optimize HVAC operations while maximizing thermal comfort and reducing energy consumption and peak power demand. We conducted on-site experiments to evaluate the proposed framework and demonstrated a notable reduction in total energy consumption in both HVAC cooling and heating operations: average reductions of 4.9% and 30.2%, and 49.3% and 73.9% for best cases in the on-site experiment under certain conditions. In the heating operation, the proposed data-driven HVAC management framework efficiently utilized indoor/outdoor heat generation to maintain comfortable temperature levels without activating the HVAC. Such effective utilization of heat sources shortened the HVAC runtimes and further reduced the total energy consumption. With the proposed framework, the average HVAC peak power demand fell by 25.8% and 35.1% during cooling and heating operations in the on-site experiment with certain scenarios. On-site experiment results demonstrated the proposed framework can efficiently suppress peak power demand, reduce energy consumption in both cooling and heating operation under certain conditions. The framework’s data-driven approach provides an effective means of optimizing HVAC operations, reducing energy consumption, and improving the thermal comfort of the building’s occupants.

Similarity learning-based fault detection and diagnosis in building HVAC systems with limited labeled data

Machine learning has been widely adopted for fault detection and diagnosis (FDD) in heating, ventilation and air conditioning (HVAC) systems over the past decade due to the ever-increasing availability of massive building operational data. Machine learning-based FDD is flexible and accurate but heavily relies on the availability of sufficient labeled data to develop supervised or unsupervised models. However, collecting labeled data is usually labor-intensive for various types of faulty conditions, significantly limiting the practical implementation of machine learning-based FDD. Therefore, this study proposes a similarity learning-based method using Siamese networks to improve the performance of machine learning-based FDD in applications with limited labeled data. Unlike the conventional supervised approach, the proposed Siamese networks contain two identical long short-term memory subnetworks which take a pair of multivariate time-series samples from the building energy management system as input. The number of training samples can be significantly augmented by generating pairs randomly. In this way, the generalization ability of the machine learning-based FDD is significantly improved in practical applications. Two case studies were designed and conducted using experimental data when labeled data were limited and imbalanced to validate the proposed similarity learning-based method. In case 1, the proposed method improves the fault diagnostic accuracy by at most 45.7% compared with the baseline model when the number of labeled data is limited. In case 2, the proposed method demonstrated better generalization ability when the labeled data is imbalanced.

Graph neural network-based spatio-temporal indoor environment prediction and optimal control for central air-conditioning systems

Model-based optimal control has proven its effectiveness in optimizing the performance of central air-conditioning systems in terms of thermal comfort and energy efficiency. It was often assumed that temperature distribution in the entire air-conditioned space is uniform and can be represented by a single or averaged measurement in optimization. However, actual distribution in the air-conditioned space is usually uneven, which can affect thermal comfort and indoor air quality. The dynamics of air conditioning systems and their interactions with the built environment exist in both time and space domains. Conventional data-driven approaches typically concentrate on the temporal correlations only among building operation data, while neglecting the spatial correlations. This study proposed a spatio-temporal data-driven methodology for optimal control of central air conditioning systems, which aims to address the challenging issue of uneven spatial distributions of environmental parameters in air-conditioned space. The methodology consists of graph-based multi-source data integration, graph neural network-based indoor environment modeling and spatio-temporal model-based online optimal control. The proposed methodology is tested on real air handling unit-variable air volume (AHU-VAV) system in a high-rise office building through a cloud-based platform. Spatio-temporal data from building automation system (BAS), internet of things (IoT) devices and building information modeling (BIM) are integrated and organized as graph structure. Graph neural network models are developed for predicting the evolution of indoor air temperature distribution under different control settings. The developed graph neural network-recurrent neural network (GNN-RNN) model architectures show enhanced accuracy than conventional deep learning models (i.e., convolutional neural network-recurrent neural network (CNN-RNN), Dense-RNN). And the cloud-based online test demonstrates that the proposed model-based control strategy improves the percentage of time achieving Grade I thermal comfort from 36.5% (i.e., existing control strategy) to 81.3%.

SIREN – smart ventilation for infection risk mitigation and HVAC energy efficiency: a case study amid the COVID-19 pandemic

The COVID-19 pandemic has underscored the need for effective ventilation control in public buildings. This study develops and evaluates a smart ventilation control algorithm (SIREN) that dynamically adjusts zone and system-level HVAC operation to maintain an acceptable COVID-19 infection risk and HVAC energy efficiency. SIREN uses real-time building operation data and Trim & Respond control logic to determine zone primary and system outdoor airflow rates. An EnergyPlus and CONTAM co-simulation framework was developed to assess its performance across various control scenarios and US climate zones. Results show that SIREN can flexibly control infection risk within a customized threshold (e.g. 3%) for every zone, while traditional controls cannot. At the building level, SIREN’s HVAC energy consumption is comparable to a fixed 70% outdoor airflow fraction scenario, while its infection risk is lower than the 100% outdoor airflow scenario, illustrating its potential for safe and energy-efficient HVAC operation during pandemics.

Interpretable machine learning for building energy management: A state-of-the-art review

Machine learning has been widely adopted for improving building energy efficiency and flexibility in the past decade owing to the ever-increasing availability of massive building operational data. However, it is challenging for end-users to understand and trust machine learning models because of their black-box nature. To this end, the interpretability of machine learning models has attracted increasing attention in recent studies because it helps users understand the decisions made by these models. This article reviews previous studies that adopted interpretable machine learning techniques for building energy management to analyze how model interpretability is improved. First, the studies are categorized according to the application stages of interpretable machine learning techniques: ante-hoc and post-hoc approaches. Then, the studies are analyzed in detail according to specific techniques with critical comparisons. Through the review, we find that the broad application of interpretable machine learning in building energy management faces the following significant challenges: (1) different terminologies are used to describe model interpretability which could cause confusion, (2) performance of interpretable ML in different tasks is difficult to compare, and (3) current prevalent techniques such as SHAP and LIME can only provide limited interpretability. Finally, we discuss the future R&D needs for improving the interpretability of black-box models that could be significant to accelerate the application of machine learning for building energy management.

An outlier management framework for building performance data and its application to the power consumption data of building energy systems in non-residential buildings

The existence of outliers and noise is one of the major causes of poor quality of the building performance data. Methods to address such abnormal data have received significant attention from several research interests including data-driven fault detection, diagnosis, and estimation (FDD&E), performance optimization, predictive modeling, pattern recognition of building power consumption, etc. However, there are many factors that may eventually lead to unreliable FDD&E results or data cleaning effects. For instance, the misconception of the statistical definition of outliers, the confusion about the concepts of outliers and noise, and the neglect of the potential impact of the outlier processing way on the study results, etc. This study gives an interpretation to the statistical definition of outliers using the domain knowledge in building energy systems. The outliers in the building performance data are classified into three categories. Based on such classification, an outlier management framework is developed to achieve reliable outlier detection results and accurate outlier estimates. In Case study 1, under the proposed outlier management framework, about 81.2%–88.6% of the power consumption observations can be identified as outliers or normal observations. In terms of the outlier estimation accuracy, the RMSE of the chiller/chilled water pumps/cooling water pumps power outlier estimation results can be limited to less than 5.2, 0.5 and 0.6 kWh. The results of Case study 1 show that the proposed outlier management framework is generic. The framework can help reducing unnecessary monitoring cost of the HVAC system. In Case study 2, proper outlier management method helps improve the chiller power prediction results. The RMSE of the Basic group, Method 1 group and Method 2 group power prediction results are 0.55, 0.54 and 0.2 kWh. The results of Case study 2 show that the proposed outlier management framework can contribute to improve the results reliability of building performance analysis related studies. In this study, an outlier management framework is developed for the building performance data. The proposed framework aims to improve the quality of the data cleaning work. In the following research, we would dedicate to improve the efficiency of the data cleaning work. In other word, to develop specific online outlier management algorithms for specific studies under the proposed framework.

A recommendation system for energy saving and user engagement in existing buildings

Ranked as the major energy-consuming sector, the building industry continuously introduces enhanced energy-management systems towards more energy efficiency. Despite technological progress, high amounts of resources are still wasted worldwide due to the wrong usage of energy systems in existing building stock. As a response, this paper proposes a recommendation system that is based on information and communication technologies and that provides building users with hands-on executable recommendations to reduce their energy consumption. Indeed, because of their lack of energy awareness and knowledge, people may need guidance in their daily usage of a building to change their energy behaviour and to save energy. For that purpose, the recommendation system analyses building operation data in real time to identify energy wastes as well as suitable energy conservation measures that shall help in saving energy. One main challenge is to provide a highly scalable solution that can be easily applied in a wider range of buildings, thus enabling resource saving at a large scale or even worldwide. For that purpose, the system relies on a model-based expert system for easy adaptation in any building regardless of its type, size and usage.

Query-adaptive training data recommendation for cross-building predictive modeling

Predictive modeling in buildings is a key task for the optimal management of building energy. Relevant building operational data are a prerequisite for such task, notably when deep learning is used. However, building operational data are not always available, such is the case in newly built, newly renovated, or even not yet built buildings. To address this problem, we propose a deep similarity learning approach to recommend relevant training data to a target building solely by using a minimal contextual description on it. Contextual descriptions are modeled as user queries. We further propose to ensemble most used machine learning algorithms in the context of predictive modeling. This contributes to the genericity of the proposed methodology. Experimental evaluations show that our methodology offers a generic methodology for cross-building predictive modeling and achieves good generalization performance.

A short-term building energy consumption prediction and diagnosis using deep learning algorithms

Short-term energy consumption prediction of buildings is crucial for developing model-based predictive control, fault detection, and diagnosis methods. This study takes a university library in Xi’an as the research object. First, a time-by-time energy consumption prediction model is established under the supervised learning approach, which uses a long short-term memory (LSTM) network and a Multi-Input Multi-Output (MIMO) strategy. The experimental results validate the model’s validity, which is close enough to physical reality for engineering purposes. Second, the potential of the people flows factor in energy consumption prediction models is explored. The results show that people flow has great potential in predicting building energy consumption and can effectively improve the prediction model performance. Third, a diagnostic method, which can recognize abnormal energy consumption data is used to diagnose the unreasonable use of the building during each hour of operation. The method is based on differences between actual and predicted energy consumption data derived from a short-term energy consumption prediction model. Based on actual building operation data, this work is enlightening and can serve as a reference for building energy efficiency management and operation.