Kaiyu Sun
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Kaiyu Sun has a total of 72 co-authors. Here are the top co-authors by number of shared publications:
- Tianzhen Hong
Building Technology and Urban Systems Division, Lawrence Berkeley National Laboratory, Berkeley, USA
19 shared publications Wanni Zhang
5 shared publicationsDa Yan
4 shared publicationsTianzhen Hong
4 shared publicationsMary Ann Piette
3 shared publicationsMax Wei
3 shared publications
Latest Publications
- Open Access
- 28 March 2026
- Published by Elsevier in Sustainable Cities and Society
- Vol. 143
District-Scale Analysis of Electricity Load and Strategies to Improve Energy Reliability Using Prototype District Models
- 28 March 2026
- Published by Elsevier in Sustainable Cities and Society
- Vol. 143
Open AccessDistrict-Scale Analysis of Electricity Load and Strategies to Improve Energy Reliability Using Prototype District Models
- Open Access
- 12 March 2026
- Published by IOP Publishing in Environmental Research Communications
- Vol. 8 (4)
A Rapid Embodied Carbon Assessment Tool for Priority Materials
Embodied carbon limits within building materials are a driving factor in global trade, generating new research and analysis tools in industry. These product assessments, which require utilizing life-cycle assessment (LCA) across broad supply chains, can be expensive, time and data-intensive, and subject to significant variations. Existing methods and tools, such as specific environmental product declarations, typically do not capture these variations and dynamics in supply and manufacturing. Moreover, models and tools must enable stakeholders to assess customized supply chains and future scenarios. In this study, we present the Rapid Embodied Carbon Assessment and Target-setting for Emissions-intensive Materials (REDuCE) tool for building materials. We developed a tool that allows users to select production technologies, transportation mode and distances, concrete carbonation, fuel sources, and regional electricity mixes to supply customization for cement and concrete produced and consumed in California. We generated and integrated a material demand model using residential building stock projections. We provide the user with a wide range of mitigation alternatives through low-carbon production pathways, material use efficiency, transportation modes, and projected electricity grid mixes. In a case study application through four mitigation scenarios, we find emission savings up to 80% by maximizing user mitigation alternatives, primarily driven by reductions in material use intensities. This work represents a foundation for expanding LCA and embodied carbon tools to better enable stakeholders to rapidly and accurately assess customized supply chains while meeting trade requirements.
- 12 March 2026
- Published by IOP Publishing in Environmental Research Communications
- Vol. 8 (4)
Open AccessA Rapid Embodied Carbon Assessment Tool for Priority Materials
Embodied carbon limits within building materials are a driving factor in global trade, generating new research and analysis tools in industry. These product assessments, which require utilizing life-cycle assessment (LCA) across broad supply chains, can be expensive, time and data-intensive, and subject to significant variations. Existing methods and tools, such as specific environmental product declarations, typically do not capture these variations and dynamics in supply and manufacturing. Moreover, models and tools must enable stakeholders to assess customized supply chains and future scenarios. In this study, we present the Rapid Embodied Carbon Assessment and Target-setting for Emissions-intensive Materials (REDuCE) tool for building materials. We developed a tool that allows users to select production technologies, transportation mode and distances, concrete carbonation, fuel sources, and regional electricity mixes to supply customization for cement and concrete produced and consumed in California. We generated and integrated a material demand model using residential building stock projections. We provide the user with a wide range of mitigation alternatives through low-carbon production pathways, material use efficiency, transportation modes, and projected electricity grid mixes. In a case study application through four mitigation scenarios, we find emission savings up to 80% by maximizing user mitigation alternatives, primarily driven by reductions in material use intensities. This work represents a foundation for expanding LCA and embodied carbon tools to better enable stakeholders to rapidly and accurately assess customized supply chains while meeting trade requirements.
- 18 July 2025
- Published by Elsevier in Building and Environment
- Vol. 284
Ten questions concerning calibrating building energy simulation models
- 18 July 2025
- Published by Elsevier in Building and Environment
- Vol. 284
Ten questions concerning calibrating building energy simulation models
- Open Access
- 1 May 2025
- Published by KEAI Publishing in Energy and Built Environment
- Vol. 7 (5) , 1166-1185
High-performance windows improve thermal survivability of occupants during cold snaps
- 1 May 2025
- Published by KEAI Publishing in Energy and Built Environment
- Vol. 7 (5) , 1166-1185
Open AccessHigh-performance windows improve thermal survivability of occupants during cold snaps
- 22 November 2024
- Published by Elsevier in Energy and Buildings
- Vol. 327
Framework to select robust energy retrofit measures for residential communities
- 22 November 2024
- Published by Elsevier in Energy and Buildings
- Vol. 327
Framework to select robust energy retrofit measures for residential communities
- 3 February 2022
- Published by Elsevier in Sustainable Energy Technologies and Assessments
- Vol. 52
Large scale energy analysis and renovation strategies for social housing in the historic city of Venice
- 3 February 2022
- Published by Elsevier in Sustainable Energy Technologies and Assessments
- Vol. 52
Large scale energy analysis and renovation strategies for social housing in the historic city of Venice
- 1 February 2022
- Published by Elsevier in Building and Environment
- Vol. 210
Investigation of pre-cooling as a recommended measure to improve residential buildings’ thermal resilience during heat waves
- 1 February 2022
- Published by Elsevier in Building and Environment
- Vol. 210
Investigation of pre-cooling as a recommended measure to improve residential buildings’ thermal resilience during heat waves
- 1 December 2021
- Published by Elsevier in Energy and Buildings
- Vol. 252
Passive cooling designs to improve heat resilience of homes in underserved and vulnerable communities
- 1 December 2021
- Published by Elsevier in Energy and Buildings
- Vol. 252
Passive cooling designs to improve heat resilience of homes in underserved and vulnerable communities
- 26 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 486-491
Multi-Scale Feature Distillation for Anomaly Detection
Due to the diversity and scarcity of defective samples, anomaly detection is usually taken as one of the main solutions of defect detection in intelligent manufacturing industry. Unsupervised anomaly detection usually faces two main challenges: (1) Without effective knowledge transferring, diversity and scarcity of defect samples led to model's training difficulties, especially for deep CNN model; (2) Large scale variations of defects led to the lack of generalization ability for typical CNN models with single or a few scale represented features. In this work, we propose a simple yet powerful unsupervised anomaly detection method by: i) The teacher and student networks are introduced to train deep CNN-based detection networks. With a powerful deep pre- trained neural network as teacher, and one proposed combined- distance based loss function, a student network with the same architecture is developed to learn teacher's knowledge, and the defective regions are recognized by the larger feature distances compared with the normal regions; ii) Multi-scale feature distillation is developed to against the scale variations. Unlike most knowledge distillation methods that only distill feature output by the last layer, we distill the features on different levels of feature maps output by the intermediate layers, and develop one combined-distance and Gaussian Kernel based receptive field up-sampling mechanism, to generate high resolution anomaly heatmap. Evaluation of the widely used MVTec anomaly detection dataset shows that our method is competitive to the state-of-the-art methods with over 10 times inference speed increase: from 2.8fps@256 × 256 to 33.2fps; The experiments on other two anomaly detection datasets, ShanghaiTech Campus dataset (STC) and Magnetic Tile Defects dataset (MTD), are also given to verify the generalization of the proposed work.
- 26 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 486-491
Multi-Scale Feature Distillation for Anomaly Detection
Due to the diversity and scarcity of defective samples, anomaly detection is usually taken as one of the main solutions of defect detection in intelligent manufacturing industry. Unsupervised anomaly detection usually faces two main challenges: (1) Without effective knowledge transferring, diversity and scarcity of defect samples led to model's training difficulties, especially for deep CNN model; (2) Large scale variations of defects led to the lack of generalization ability for typical CNN models with single or a few scale represented features. In this work, we propose a simple yet powerful unsupervised anomaly detection method by: i) The teacher and student networks are introduced to train deep CNN-based detection networks. With a powerful deep pre- trained neural network as teacher, and one proposed combined- distance based loss function, a student network with the same architecture is developed to learn teacher's knowledge, and the defective regions are recognized by the larger feature distances compared with the normal regions; ii) Multi-scale feature distillation is developed to against the scale variations. Unlike most knowledge distillation methods that only distill feature output by the last layer, we distill the features on different levels of feature maps output by the intermediate layers, and develop one combined-distance and Gaussian Kernel based receptive field up-sampling mechanism, to generate high resolution anomaly heatmap. Evaluation of the widely used MVTec anomaly detection dataset shows that our method is competitive to the state-of-the-art methods with over 10 times inference speed increase: from 2.8fps@256 × 256 to 33.2fps; The experiments on other two anomaly detection datasets, ShanghaiTech Campus dataset (STC) and Magnetic Tile Defects dataset (MTD), are also given to verify the generalization of the proposed work.