ORNL/TM-2026/4715 Generalizable Web User Interface for Scalable and Streamlined Deployment of Building Energy Management Systems in Small and Medium-Sized Commercial Buildings Jihoon Chung Sen Huang Boming Liu Jamie Lian Jin Dong et al. September 2026 Approved for public release. Distribution is unlimited. ORNL is managed by UT-Battelle LLC for the US Department of Energy DOCUMENT AVAILABILITY Online Access: US Department of Energy (DOE) reports produced after 1991 and a growing number of pre-1991 documents are available free via https://www.osti.gov/. The public may also search the National Technical Information Service’s National Technical Reports Library (NTRL) for reports not available in digital format. DOE and DOE contractors should contact DOE’s Office of Scientific and Technical Information (OSTI) for reports not currently available in digital format: US Department of Energy Office of Scientific and Technical Information PO Box 62 Oak Ridge, TN 37831-0062 Telephone: (865) 576-8401 Fax: (865) 576-5728 Email: reports@osti.gov Website: https://www.osti.gov/ This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. ORNL/TM-2026/4715 Electrification and Energy Infrastructures Division GENERALIZABLE WEB USER INTERFACE FOR SCALABLE AND STREAMLINED DEPLOYMENT OF BUILDING ENERGY MANAGEMENT SYSTEMS IN SMALL AND MEDIUM-SIZED COMMERCIAL BUILDINGS Jihoon Chung Sen Huang Boming Liu Jamie Lian Jin Dong Srinivas Katipamula Robert Lutes September 2026 Prepared by OAK RIDGE NATIONAL LABORATORY Oak Ridge, TN 37831 managed by UT-BATTELLE LLC for the US DEPARTMENT OF ENERGY under contract DE-AC05-00OR22725 CONTENTS LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v LIST OF ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 RELATED WORKS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 SUMMARY AND CONTRIBUTIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2. GENERALIZABLE WEB USER INTERFACE FRAMEWORK . . . . . . . . . . . . . . . . . . 6 2.1 AUTOMATED CONFIGURATION FOR STREAMLINED DEPLOYMENT . . . . . . . . 6 2.1.1 Unified YAML Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.1.2 Automated Configuration Conversion . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.1.3 Semantic Building Model Import . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2 INTEROPERABILITY AND ADAPTABILITY ACROSS HETEROGENEOUS SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2.1 Configuration-Driven Web User Interface . . . . . . . . . . . . . . . . . . . . . . . 8 2.2.2 Open-Source, Scalable System Architecture . . . . . . . . . . . . . . . . . . . . . . 11 2.3 INTEGRATED DEPLOYMENT WORKFLOW . . . . . . . . . . . . . . . . . . . . . . . . 14 2.4 SUMMARY OF THE MAIN BENEFITS . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3. VALIDATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 3.1 SIMULATION-BASED DEPLOYMENT TEST . . . . . . . . . . . . . . . . . . . . . . . . 17 3.1.1 Test Case Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 3.1.2 YAML Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.1.3 Deployment Results and Interface Adaptation . . . . . . . . . . . . . . . . . . . . . 20 3.2 SEMANTIC BUILDING MODEL CONFIGURATION TEST . . . . . . . . . . . . . . . . 20 3.2.1 Studied Semantic Building Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.2.2 Validation Process and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 4. FIELD DEMONSTRATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 4.1 STUDIED BUILDING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 4.2 YAML CONFIGURATION AND DEPLOYMENT . . . . . . . . . . . . . . . . . . . . . . 24 4.3 COMPARATIVE MONITORING RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.4 REQUIRED EFFORT FOR THE DEPLOYMENT PROCESS . . . . . . . . . . . . . . . . 26 5. DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 6. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 6.1 LIMITATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 6.2 FUTURE WORK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 7. REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 APPENDIX A. SUPPLEMENTARY FIGURES AND TABLES . . . . . . . . . . . . . . . . . . . . A-1 iii LIST OF FIGURES Figure 1. Configuration converter workflow and automated transformation process. . . . . . . 7 Figure 2. Detailed examples of converting a YAML configuration file to system-specific configuration files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Figure 3. Example of a semantic building model in Brick Schema (left) and a converted building configuration file (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Figure 4. Dashboard layouts for scalable BEMS deployment. . . . . . . . . . . . . . . . . . . 10 Figure 5. Building configurations displayed in grid-tile (left) and list (right) layouts in the deployed web user interface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Figure 6. YAML configurations for data visualization (left) and their implementation in web applications (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Figure 7. System architecture of the proposed scalable BEMS. . . . . . . . . . . . . . . . . . 13 Figure 8. Overview of the proposed system and research workflow. . . . . . . . . . . . . . . . 14 Figure 9. Schematic diagram of HVAC systems in the BESTEST Air and BESTEST Hydronic test cases. Adapted from (Blum and Helsen 2023). . . . . . . . . . . . . . . . . . . 18 Figure 10. Configuration creation and conversion process for the BESTEST Hydronic test case in the simulation-based deployment test. . . . . . . . . . . . . . . . . . . . . . . . . 19 Figure 11. Building unit configuration in the prepared YAML file for the BESTEST Air and BESTEST Hydronic test cases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Figure 12. Deployed web application using the BESTEST Air and BESTEST Hydronic configuration files. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Figure 13. Converted building unit configuration file for zone RM050 from the studied Brick model (left) and corresponding deployed web interface (right). . . . . . . . . . . . . 23 Figure 14. Exterior view of the small commercial office building used for the field demonstration. 25 Figure 15. Deployed web application for monitoring Room 125 with setback control (a) and constant set point (b) in the studied building. . . . . . . . . . . . . . . . . . . . . . 26 Figure A-1. Concept diagram of the proposed generalizable web user interface. . . . . . . . . . . A-2 Figure A-2. Schematic diagram of the proposed deployment workflow. . . . . . . . . . . . . . . A-2 Figure A-3. Thermostat controller in BESTEST Air (top) and BESTEST Hydronic (bottom) test cases. PI: proportional-integral controller; Hys: hysteresis controller. Adapted from (Blum and Helsen 2023). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-3 Figure A-4. VOLTTRON agent configuration in the prepared YAML Ain’t Markup Language (YAML) file for the simulation-based deployment test. . . . . . . . . . . . . . . . . A-5 Figure A-5. Configuration creation and conversion process in the field demonstration. . . . . . . A-7 Figure A-6. Deployed web application for monitoring Room 112 (a) and Room 119 (b) with setback control (left side) and constant set point (right side) in the studied building. . A-9 iv LIST OF TABLES Table 1. Comparative analysis of existing building energy management solutions. . . . . . . 4 Table A-1. Building unit configurations for the simulation-based deployment test. . . . . . . . . A-3 Table A-2. Configuration settings and data points for the simulation-based deployment test. . . . A-4 Table A-3. Building unit configurations in the field demonstration. . . . . . . . . . . . . . . . . A-4 Table A-4. VOLTTRON configuration settings and data points for ecobee and Modbus driver agents in the field demonstration. . . . . . . . . . . . . . . . . . . . . . . . . . . . A-6 Table A-5. Detailed information of open-source code and libraries. . . . . . . . . . . . . . . . . A-8 v LIST OF ABBREVIATIONS AEMS Autonomous Energy Management Software AHU air handling unit API application programming interface ASHRAE American Society of Heating, Refrigerating and Air-Conditioning Engineers BAS building automation system BEMS building energy management system BMS building management system BOPTEST Building Optimization Testing Framework DNP3 Distributed Network Protocol 3 DOE US Department of Energy HTTP Hypertext Transfer Protocol HTTPS Hypertext Transfer Protocol Secure HVAC heating, ventilation, and air conditioning IoT Internet of Things JSON JavaScript Object Notation MQTT Message Queuing Telemetry Transport ORNL Oak Ridge National Laboratory PI proportional-integral PNNL Pacific Northwest National Laboratory PTAC packaged terminal air conditioner RDF Resource Description Framework REST Representational State Transfer RTU rooftop unit SMCB small and medium-sized commercial building VAV variable air volume YAML YAML Ain’t Markup Language vi ABSTRACT Small and medium-sized commercial buildings (SMCBs) comprise 94% of US commercial buildings yet face significant barriers to implementing building energy management systems despite advances in smart device technology. Existing solutions present critical limitations: cloud-based API solutions simplify deployment but create vendor lock-in constraints; commercial integrated software solutions ensure compatibility via standardized protocols but require substantial cost and technical expertise; open-source IoT platforms offer cost-effective vendor independence but provide insufficient standardized protocol support for commercial building automation. This research presents a generalizable web user interface framework that bridges the gap between evolving smart device capabilities and lagging software infrastructure for SMCBs. The proposed system integrates VOLTTRON open-source middleware with an automated configuration converter that transforms unified specifications written in YAML, a human-readable data-serialization format, into system-specific files, streamlining manual setup processes. The vendor-agnostic architecture supports industry-standard protocols (BACnet and Modbus) and semantic building models while providing adaptive web interfaces that dynamically adjust to various building configurations. Demonstrations through simulation-based testing and a field deployment show automatic interface adaptation across heterogeneous HVAC systems and multizone monitoring. The automated configuration converter also substantially reduces labor-intensive setup. 1. INTRODUCTION 1.1 BACKGROUND Small and medium-sized commercial buildings (SMCBs), defined as commercial buildings of 50,000 ft2 (4,645 m2 ) or less, comprise 94% of commercial buildings in the United States while consuming approximately 47% of the total commercial building energy use, excluding malls (Kim, Katipamula, and Lutes 2019; U.S. Energy Information Administration 2021; Luo et al. 2017; Ye, Zuo, and Wang 2019). Targeting SMCBs for energy efficiency improvements could achieve as much as a 17% reduction in total energy use, demonstrating their significant effect on national energy consumption (Barnes and Parrish 2016; Huppert et al. 2013). This substantial energy consumption underscores the critical need for effective energy management strategies within the SMCB sector. A building energy management system (BEMS)—also known as building automation system (BAS) or building management system (BMS)—coordinates various building equipment, including HVAC, lighting, and energy meters, through real-time monitoring, automated controls, and intelligent scheduling to improve performance and reduce energy consumption (Bhatt and Verma 2015; Domingues et al. 2016). Effective BEMS deployment fundamentally depends on seamless integration of hardware and software components: hardware elements, including sensors and controllers, interface with mechanical equipment, whereas software platforms coordinate their operation, analyze data, and enable advanced control strategies. However, approximately 84% of SMCBs do not have adequate BEMS infrastructure, having received limited attention because of the prohibitively high costs of retrofitting these buildings (U.S. Energy Information Administration 2021; Guo et al. 2025). Recent advances in IoT technology have driven increasing adoption of smart sensors and controllers in SMCBs, creating significant opportunities for sophisticated energy management through enhanced monitoring and control capabilities (Kim and Braun 2018). These smart devices include thermostats with real-time temperature displays and remote controls, wireless environmental sensors enabling scalable multizone monitoring, occupancy sensors for automated lighting and HVAC control, smart plugs for real-time energy tracking, and CO2 sensors integrated into smart ventilation systems (Stopps and 1 Touchie 2021; Jornet-Monteverde and Galiana-Merino 2020; Chaudhari et al. 2024; Chiesa et al. 2019). Collectively, these technologies can reduce energy consumption by automatically adjusting set points based on occupancy detection (Tamas, O’Brien, and Quintero 2021), enable precise zone-level control that eliminates waste in unoccupied areas (Wang et al. 2018), and allow real-time monitoring to identify unnecessary equipment operation during peak demand periods (Suryadevara and Biswal 2019). However, in practice, realizing these benefits requires comprehensive software interfaces that provide essential functionality, such as real-time data visualization, unified monitoring, remote control, and data management for advanced control strategies, while minimizing manual configuration efforts. Despite these rapid hardware advances, software development has not kept pace, creating a critical gap between device capabilities and accessible building automation. Commercial platforms such as ecobee’s SmartBuildings and Google Nest’s Devices Access rely on vendor-specific, cloud-based ecosystems that constrain broader implementation across various building environments. More critically, these API-based approaches require programming expertise to develop custom integration code within a centralized platform, resulting in significant technical barriers because most SMCB operators have limited specialized knowledge. The need for specialized labor directly translates into substantial implementation costs (Trenbath et al. 2022), creating a particularly acute barrier for SMCBs, which typically operate with limited budgets and lack the dedicated facility management staff found in larger commercial buildings (Ham et al. 2024; Guo et al. 2025). Beyond technical complexity, existing software solutions lack a centralized interoperability hub that enables scalable communication among heterogeneous devices, preventing unified integration and coordinated control of systems across different vendors and protocols. Current approaches require labor-intensive setup processes involving extensive manual configuration, custom programming, and system integration work by specialized technicians, with programming and testing alone accounting for up to 25% of total system costs (Paul et al. 2025; Trenbath et al. 2022). A case study demonstrated that labor hours reached 65 h (equivalent to $7,500) just for system setup and configuration in a 1,035 m2 SMCB, highlighting the prohibitive initial cost that prevents widespread adoption (Guo et al. 2025). This absence of a streamlined deployment process forces building operators to repeat a separate configuration procedure for each building system, preventing scalability across multiple SMCBs. Furthermore, many existing software platforms remain tightly coupled with specific devices or ecosystems and provide limited support for standardized communication protocols such as BACnet and Modbus (Essa et al. 2023; Wu and Liu 2019; Malkawi et al. 2023; Mataloto, Ferreira, and Cruz 2019). These limitations may constrain the scalability and flexibility of existing software platforms across various building types and limit integration capabilities with the heterogeneous device configurations commonly used in the building management industry. Recognizing these critical barriers, the US Department of Energy’s Building Technologies Office has initiated development of low-cost, interoperable building controller solutions specifically targeting SMCBs through multilaboratory research led by Oak Ridge National Laboratory and Pacific Northwest National Laboratory (PNNL) to address both implementation costs and standardization requirements for widespread deployment (U.S. Department of Energy, Building Technologies Office 2024). 1.2 RELATED WORKS Although various interface solutions have been developed to integrate smart devices with building monitoring and control systems, existing approaches can be categorized into two primary groups, each presenting unique advantages and limitations that collectively highlight the need for more comprehensive integration strategies in SMCB environments (Table 1). The first group leverages cloud-based APIs for device–software integration, offering streamlined setup processes that have contributed to their 2 adoption in approximately 6.5% of SMCBs (Neukomm, Nubbe, and Fares 2019; U.S. Energy Information Administration 2021). Representative solutions include Google Nest, ecobee Smart Thermostat, Amazon Smart Thermostat, and Honeywell T-Series thermostats, which provide user-friendly mobile applications and reduced upfront infrastructure requirements through Wi-Fi connectivity and vendor-specific cloud platforms (Google, n.d.-c; Ecobee, n.d.-c; Honeywell, n.d.; Amazon, n.d.; Ayan and Turkay 2018; Tamas, O’Brien, and Quintero 2021). However, these vendor-specific ecosystems lack generalized interfaces capable of accommodating heterogeneous devices across different vendors, requiring labor-intensive custom integration work when combining devices from multiple vendors and limiting interoperability and long-term system scalability across various building configurations (Trenbath et al. 2022; Gaidon and Poplawski 2017). The second group encompasses integrated software solutions that use industry-standard communication protocols, including BACnet, Modbus, KNX, and LonWorks, which have been used in the building automation industry for decades (Domingues et al. 2016; Modbus IDA 2004; International Organization for Standardization 2004; ISO/IEC 2006, 2012). These standardized approaches ensure compatibility across different manufacturers and enable comprehensive, vendor-agnostic building automation solutions that are particularly beneficial for scalable implementations (Bhatt and Verma 2015). Representative commercial platforms include Normal Framework, Facilio, KODE OS, and WEBCTRL, which provide robust BACnet/Modbus integration and vendor-agnostic device support (Normal Software 2025; Facilio, n.d.; KODE OS, n.d.; WebCTRL®, n.d.). However, these solutions typically impose substantial licensing costs and require specialized technical expertise for deployment and configuration, contributing to the labor-intensive deployment processes that create substantial implementation barriers in the SMCB sector (Trenbath et al. 2022; Ham et al. 2024; Kim and Braun 2018). In contrast, open-source software platforms provide cost-effective, community-driven middleware solutions. Representative platforms include Node-RED, Home Assistant, ThingsBoard, and OpenHAB, which offer flexible integration frameworks for connecting heterogeneous IoT devices without proprietary lock-in constraints through various IoT network protocols, such as MQTT, Zigbee, Z-Wave, and LoRaWAN (Node-RED, n.d.; Home Assistant, n.d.; MQTT, n.d.; Zigbee, n.d.; Z-Wave, n.d.; LoRaWAN, n.d.). These open-source solutions provide substantial cost reductions, extensive customization flexibility, and visual programming tools for customized algorithms through intuitive interfaces (Baig et al. 2021; Chen et al. 2024; Mataloto, Ferreira, and Cruz 2019; Korodi et al. 2024). However, they primarily focus on IoT sensor integration and residential smart home applications, lacking native support for industry-standard protocols such as BACnet and Modbus that are essential for commercial building automation. This absence of protocol support necessitates additional custom development and integration efforts to interface with commercial BASs. This analysis reveals that although each category addresses specific aspects of smart devices, none provides a comprehensive solution for SMCB environments. Two fundamental limitations persist across the existing approaches: (1) labor-intensive deployment processes requiring extensive manual configuration and specialized technical expertise and (2) a lack of generalized interfaces capable of accommodating heterogeneous devices across various building systems. API-based solutions, while offering streamlined initial setup, require labor-intensive custom integration work because of their lock-in constraints that limit interoperability and scalability. Commercial integrated software solutions ensure robust compatibility through standardized protocols but impose labor-intensive deployment processes that require specialized technical expertise for system configuration, resulting in substantial implementation cost barriers. Open-source IoT platforms offer vendor independence and customization advantages but necessitate additional custom development and integration efforts to interface with commercial BASs due to limited native support for industry-standard protocols. These limitations collectively underscore the critical need for an integrated approach that combines streamlined deployment workflows with generalized 3 Table 1. Comparative analysis of existing building energy management solutions. Protocol Support Interoperability Implementation Hardware Semantic Simplified Open No HTTP/ Category Tool BACnet Modbus -vendor model config. -source licensing Reference HTTPS agnostic integration approach availability cost Google Nest ✓ ✓ (Google, n.d.-c) API-based ecobee ✓ ✓ (Ecobee, n.d.-c) solution Amazon Smart ✓ ✓ (Amazon, n.d.) Thermostat Honeywell T-Series ✓ ✓ (Honeywell, n.d.) Normal Framework ✓ ✓ ✓ ✓ ✓2 (Normal Software Integrated 2025) software Facilio ✓ ✓ ✓ ✓ (Facilio, n.d.) 4 solution KODE OS ✓ ✓ ✓ ✓ (KODE OS, n.d.) WEBCTRL ✓ ✓ ✓ ✓ (WebCTRL®, n.d.) Node-RED ✓ ✓1 ✓1 ✓ ✓ ✓ (Node-RED, n.d.) Home Assistant ✓ ✓1 ✓ ✓3 (Home Assistant, n.d.) ThingsBoard ✓ ✓1 ✓1 ✓ ✓3 (ThingsBoard, n.d.) OpenHAB ✓ ✓1 ✓ ✓2 ✓ ✓ (openHAB, n.d.) PNNL AEMS ✓ ✓ ✓ ✓ ✓ ✓ (Lutes, Allwardt, Bleeker, et al. 2025) This work Proposed framework ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Notes: 1 Protocol support is available only via add-on or plugin extensions and is not natively integrated into the core platform. 2 This solution offers semantic model generation only; existing model import is not yet supported. 3 The community edition is available as open source; professional/enterprise editions require licensing fees. interfaces adaptable to various device configurations, thereby minimizing manual configuration and technical expertise requirements while enabling comprehensive building automation for SMCB operators. To address these identified limitations in existing approaches, this research builds upon recent advances in open-source technologies that provide essential foundations for scalable, streamlined BEMS deployment. PNNL’s Autonomous Energy Management Software (AEMS) provides a comprehensive web application framework built on the VOLTTRON open-source platform, offering building control capabilities with particular focus on rooftop units (RTUs) and heat pumps (Katipamula, Lutes, and Herr 2023; Katipamula 2022). The underlying VOLTTRON platform provides a robust middleware framework for integrating distributed energy resources through industry-standard protocols, including BACnet and Modbus, offering a vendor-agnostic foundation for building automation (Katipamula et al. 2016; Haack et al. 2016). Brick Schema has emerged as a standardized semantic building model that enables consistent representation of building systems and equipment relationships through the Resource Description Framework (RDF), facilitating robust interoperability across various building management applications (Fierro et al. 2020; Balaji et al. 2016). Furthermore, React provides a component-based JavaScript library that enables the creation of responsive, dynamic web applications that adapt to varying data structures through reusable interface components and efficient state management (Meta Platforms, Inc., n.d.). Although these technologies address specific aspects of BEMS—AEMS providing a VOLTTRON-compatible web framework, Brick Schema enabling semantic interoperability, and React facilitating dynamic web interfaces—their integration into a unified, generalizable framework for streamlined BEMS deployment with minimal configuration effort remains unexplored for the SMCB sector. 1.3 SUMMARY AND CONTRIBUTIONS Current smart device integration approaches reveal significant limitations across three primary categories: cloud-based API solutions offer simplified deployment but create fragmented systems that require significant integration effort when combining devices from multiple vendors; commercial integrated software solutions ensure robust compatibility via industry-standard protocols but impose substantial licensing costs and require specialized technical expertise; and open-source IoT platforms provide cost advantages but primarily focus on residential smart home applications, lacking native support for industry-standard protocols essential for commercial building automation. To address these identified limitations, this research presents a framework for developing generalizable web user interfaces for BEMS that bridges the critical gap between evolving smart device capabilities and accessible software infrastructure in SMCBs. The main contributions of this work are: • A generalizable web user interface framework that provides essential BEMS functionality—including real-time data visualization, unified monitoring, remote control, and configuration-driven interfaces that dynamically adapt to heterogeneous building systems, significantly reducing deployment complexity across various system configurations. • An automated configuration conversion algorithm that transforms a unified, natural-language-friendly YAML (YAML Ain’t Markup Language) specification into multiple heterogeneous system-specific configuration files, with the integrated capability to parse semantic building models (e.g., Brick Schema) for automated configuration generation, reducing labor-intensive manual setup processes. • A scalable deployment architecture integrating VOLTTRON open-source middleware to provide vendor-agnostic device connectivity through industry-standard protocols (BACnet and Modbus), eliminating proprietary ecosystem dependencies while ensuring compatibility with existing building automation infrastructure and IoT devices. 5 2. GENERALIZABLE WEB USER INTERFACE FRAMEWORK Existing BEMS deployment approaches for SMCBs face two fundamental challenges that impede widespread adoption: (1) labor-intensive deployment processes that require extensive manual configuration across multiple system components and (2) the lack of generalized interfaces capable of accommodating heterogeneous building devices and HVAC configurations. This research proposes a generalizable web user interface framework, a single interface that adapts to heterogeneous HVAC systems at the campus, building, and system levels (Figure A-1), to address these challenges through two complementary approaches. First, an automated configuration converter transforms a single, unified YAML specification into multiple system-specific configuration files, eliminating fragmented manual processes while ensuring data consistency across components, as described in Section 2.1. Second, a configuration-driven web interface dynamically adapts navigation hierarchies, controller displays, and visualization parameters to heterogeneous HVAC configurations without requiring separate interface development, as described in Section 2.2. Moreover, the proposed framework leverages a scalable, open-source VOLTTRON middleware architecture, enabling vendor-agnostic interoperability across heterogeneous building devices through industry-standard protocols (BACnet and Modbus). These approaches are intended to reduce deployment complexity and technical expertise requirements, eliminate vendor lock-in constraints, and improve scalability across diverse SMCB environments. 2.1 AUTOMATED CONFIGURATION FOR STREAMLINED DEPLOYMENT Conventional BEMS deployment requires manual creation of multiple, disparate configuration files across system components, contributing to labor-intensive processes and technical expertise barriers. This subsection presents the proposed framework’s approach to minimizing human involvement through unified YAML configuration specification, automatic semantic model parsing, and automated conversion algorithms that eliminate fragmented manual setup while ensuring data consistency across components. 2.1.1 Unified YAML Configuration The proposed configuration converter aims to address the critical challenge of managing multiple, disparate configuration files required by different system components in building energy management deployments. Existing open-source system implementations typically require building operators to manually create and maintain numerous configuration files. These files are distributed across different directories and formatted according to specific software requirements for web servers, clients, middleware, and device managers. This fragmented configuration approach contributes significantly to the labor-intensive setup processes and technical expertise requirements that constitute one of the primary barriers to SMCB automation adoption. The complexity is further compounded by the need to maintain data consistency across multiple configuration files, where discrepancies in port numbers, device identifiers, or system parameters can result in deployment failures requiring specialized technical knowledge to diagnose and resolve. To address these challenges, the proposed configuration converter consolidates all required system configurations into a single, unified YAML file while automatically generating platform-specific configuration files, thereby eliminating manual configuration errors and reducing deployment complexity (Predoaia et al. 2024). YAML is selected for nontechnical users, providing natural-language-like key-value syntax without requiring extensive quotation marks or the complex bracket and tag structures typical of JavaScript Object Notation (JSON) and XML. Moreover, YAML offers several critical advantages over alternative formats: (1) it natively supports various data types including booleans, strings, numbers, nested arrays 6 with minimal syntax requirements, (2) it enables structured hierarchical data organization through flexible indentation, and (3) it allows inline comments for documentation (Ben-Kiki, Evans, and Net 2009). This format enables seamless automated conversion to multiple system-specific configuration files, ensuring data consistency across components while maintaining human readability. As illustrated in Figure 1, the automated conversion process employs a systematic algorithm that parses and transforms the unified YAML input file into four distinct output configurations tailored to specific system components. The converter generates separate environment variable files (.ENV) for both the web client (containing client port, login information, API URLs, and application settings) and web server (specifying server port, administrative credentials, database configurations, and API endpoints). Simultaneously, the converter produces JSON configuration files for building unit specifications (including building and system identifiers, controller variables, schedules, and user interface preference settings) and VOLTTRON middleware configurations (encompassing VOLTTRON port assignments, agent identifiers, data update frequencies, and data point specifications). The unified YAML file does not include all configuration details but contains the essential information needed by nonexpert users to set up each system component and deploy the entire system. The systematic transformation process ensures that all interdependent system components initialize with consistent configuration parameters while maintaining the specific formatting requirements necessary for seamless integration with the AEMS web client, AEMS web server, and VOLTTRON middleware platforms. Figure 1. Configuration converter workflow and automated transformation process. 2.1.2 Automated Configuration Conversion The configuration converter implements several streamlined deployment features that enhance accessibility for nontechnical building operators while handling potential errors throughout the conversion process. Figure 2 illustrates the detailed examples of converting a unified YAML file into system-specific configuration files required for system deployment. Natural-language-friendly configuration keys (e.g., application title) 7 specified in the YAML file are automatically translated into machine-readable keys required by specific system components (e.g., REACT_APP_TITLE), enabling building operators to configure systems using intuitive terminology. Moreover, these natural-language key names can be customized according to user preferences or operational conventions, providing flexibility for various operational environments. The converter incorporates comprehensive error handling through optional value processing, where omitted configuration parameters are automatically replaced with predefined default values to ensure successful deployment even with incomplete user input. Additionally, although output files are generated using relative file paths within the repository for simplified deployment, the system supports user-specified absolute file paths through the configuration file path parameters, facilitating integration with existing organizational file management and deployment workflows. 2.1.3 Semantic Building Model Import The integration of semantic building models represents a critical advancement for scalable BEMS, enabling standardized building representation and system interoperability across various building configurations (Pritoni et al. 2021). In particular, Brick Schema has been widely recognized in research studies and increasingly adopted by building applications (Pritoni et al. 2021). It represents building systems in RDF format using a set of relationships of concepts for locations (e.g., rooms), equipment (e.g., HVAC), control point (e.g., set points), and measurement (e.g., temperature). To allow advanced users to seamlessly import multiple building configurations from existing building management applications, the proposed configuration converter leverages Brick Schema models by automatically parsing semantic descriptions to generate building unit configuration files. The converter recognizes hierarchical building elements such as “campus,” “building,” and “system” values through location classes (e.g., brick:Site, brick:Building, and brick:HVAC_Zone) and their relationships (e.g., brick:hasPart). Control points can be identified using equipment classes (e.g., brick:AHU and brick:VAV) and their association with point classes (e.g., brick:hasPoint). The automatically generated files include essential information required for system deployment, including campus, building, and system identifiers and control data points, thereby eliminating the need for manual configuration document creation, as illustrated in Figure 3. 2.2 INTEROPERABILITY AND ADAPTABILITY ACROSS HETEROGENEOUS SYSTEMS Existing BEMS interfaces lack interoperability and adaptability across heterogeneous building devices and HVAC configurations, requiring separate development for different system types. To address this challenge, this subsection presents configuration-driven web interfaces that dynamically adapt to various system configurations and an open-source architecture that supports industry-standard communication protocols for vendor-agnostic device integration. 2.2.1 Configuration-Driven Web User Interface The web user interface employs a hierarchical, configuration-driven layout that dynamically adapts to various building configurations while providing scalable navigation across campus, building, and zone levels, as depicted in Figure 4. The interface automatically generates multilevel navigation layouts based on the building unit configuration JSON files, rendering campus tabs, building lists, and zone panels that reflect the actual building hierarchy specified in the unified YAML input. Furthermore, as shown in Figure 5, the system provides flexible display options to accommodate different operational preferences and building portfolio sizes, including both grid-tile layouts that provide visual building images for intuitive navigation, and list layouts that organize the same information into sortable, filterable tabular formats with comprehensive building metadata, including campus, building, building type, system, operator, and time zone. This adaptive layout supports navigation from individual-zone monitoring to campus-wide management while maintaining consistent user experience across different SMCB environments and building portfolio sizes. 8 Figure 2. Detailed examples of converting a YAML configuration file to system-specific configuration files. The visualization configuration system provides comprehensive customization capabilities through a simplified configuration protocol that enables building operators to create tailored data visualizations without requiring specialized programming knowledge. As illustrated in Figure 6, the chart configuration allows users to specify detailed visualization parameters, including chart types (e.g., line, box, and scatter), 9 Figure 3. Example of a semantic building model in Brick Schema (left) and a converted building configuration file (right). Figure 4. Dashboard layouts for scalable BEMS deployment. selected variables for multiparameter plotting, and comprehensive styling options encompassing line color, width, dash pattern, marker size, and symbol type to enhance data clarity. The system automatically translates these human-readable configuration specifications into interactive charts that support temporal trend analysis through line charts, statistical distribution visualization through box plots, and correlation assessment through scatterplots. These visualization configurations are dynamically applied to real-time building data streams, enabling building operators to create personalized monitoring dashboards that align 10 Figure 5. Building configurations displayed in grid-tile (left) and list (right) layouts in the deployed web user interface. with operational requirements while eliminating dependence on external visualization tools or specialized data analysis expertise. 2.2.2 Open-Source, Scalable System Architecture To enable vendor-agnostic interoperability across heterogeneous building devices, the proposed framework employs a four-layer system architecture that integrates open-source VOLTTRON middleware with industry-standard communication protocols. As illustrated in Figure 7, the system architecture of the proposed framework consists of four layers: an application layer, a data management layer, a network layer, and a data source layer. Data Source Layer: The foundation layer encompasses three primary data source categories that provide comprehensive building system coverage. Installed control and sensing devices (smart thermostats, IoT environmental sensors, and energy meters) provide real-time operational data, and a building control emulator enables testing and validation using advanced, high-fidelity virtual building simulations (Blum et al. 2021). External data sources, including electricity pricing services, carbon-intensity data, and weather APIs provide contextual information necessary for advanced energy management and demand-response strategies (Ma et al. 2021; OpenWeather, n.d.; National Renewable Energy Laboratory (NREL), n.d.; National Energy System Operator (NESO), n.d.). Network Layer: The communication layer manages data transmission between the data source layer and the middleware through multiple standardized protocols to ensure broad device compatibility across different manufacturers and system types. This layer leverages widely adopted industry communication standards, with BACnet and Modbus protocols providing robust connectivity with building automation equipment commonly found in commercial buildings (International Organization for Standardization 2004; Modbus IDA 2004). Additionally, the layer supports Message Queueing Telemetry Transport (MQTT) messaging protocol for wireless integration with IoT devices and smart building components, whereas Hypertext 11 Figure 6. YAML configurations for data visualization (left) and their implementation in web applications (right). Transfer Protocol (HTTP) and Hypertext Transfer Protocol Secure (HTTPS) enable secure web-based communication with Representational State Transfer (REST) API-based external services and smart devices accessible via Internet connections (MQTT, n.d.; World Wide Web Consortium (W3C), n.d.). Additionally, VOLTTRON middleware allows integration with Distributed Network Protocol 3 (DNP3), an open, standardized communication protocol originally developed for electric utility applications, thereby providing reliable communication with utility-grade devices and advanced metering infrastructure (DNP Users Group, n.d.; Njova, Ogudo, and Umenne 2020). By using industry-standard communication protocols, this approach eliminates vendor lock-in constraints while ensuring compatibility with both existing building automation equipment and emerging smart-device technologies. Data Management Layer: The data management layer centers on VOLTTRON, an open-source, agent-based control and sensing platform developed at PNNL specifically for integrating distributed energy resources (Katipamula et al. 2016). VOLTTRON serves as the primary middleware platform, coordinating communication among multiple driver agents that interface with heterogeneous building devices and systems through a 12 Figure 7. System architecture of the proposed scalable BEMS. unified message bus architecture (Haack et al. 2016). The architecture incorporates multiple configurable driver agents, each designed to handle specific communication protocols and device types, providing the extensibility necessary to accommodate various smart devices commonly deployed in SMCB environments. The layer includes InfluxDB, an open-source time-series database for managing long-term temporal information, including sensor reading and control records. InfluxDB supports the high-frequency, high-volume, real-time data collection requirements typical of building monitoring applications, making it particularly suitable for scalable building operations (InfluxData, n.d.-a; Giacobbe et al. 2018; Li and Hong 2025). Application Layer: The top layer provides user-friendly, customizable graphical user interfaces that allow users to control and monitor various building systems across campuses. These web clients are connected to the centralized web server, facilitating concurrent multiuser access for building operators, owners, and system managers (Lutes, Allwardt, Bleeker, et al. 2025). The application layer also supports integration with external dashboard platforms, including Grafana and Chronograf, extending visualization capabilities beyond the native interface to accommodate various user preferences and existing organizational dashboard systems (Grafana Labs, n.d.; InfluxData, n.d.-b). This flexibility provides operators with options for selecting visualization tools that best align with their operational preferences and existing operational workflows. 13 2.3 INTEGRATED DEPLOYMENT WORKFLOW As presented in Figure 8, the proposed framework integrates the unified YAML configuration (Section 2.1.1), automated configuration converter (Section 2.1.2), automatic semantic model parsing (Section 2.1.3), configuration-driven web interfaces (Section 2.2.1), and VOLTTRON-based scalable system architecture (Section 2.2.2) into a streamlined BEMS deployment workflow that systematically addresses both labor-intensive deployment processes and interoperability limitations. As illustrated in Figure A-2, the deployment workflow consists of four sequential phases that transform a single, unified configuration specification into a fully operational BEMS capable of accommodating heterogeneous HVAC configurations and building equipment without manual customization. Figure 8. Overview of the proposed system and research workflow. Phase 1: General Configuration File Creation: The workflow begins with the creation of a single, human-readable YAML configuration file that serves as the centralized specification for the entire BEMS. This approach streamlines extensive manual configuration across multiple software components and interfaces required for BEMS deployment. The YAML format was selected because of its intuitive, natural-language-like syntax, enabling building operators and system integrators to define comprehensive system configurations without requiring specialized programming knowledge (Ben-Kiki, Evans, and Net 2009; Schiera et al. 2021). This configuration file enables building operators to specify web client settings (application titles and user interface preferences), web server configurations (administrative credentials and port assignments), building unit descriptions (system types, default set points, and schedules), and middleware platform settings (VOLTTRON port configurations and driver agent information). This unified configuration approach eliminates the fragmented setup procedures of existing solutions while providing a single, comprehensive file that documents essential system settings for simplified maintenance and future modification. Phase 2: Configuration Input File Conversion: The second phase employs an automated configuration converter that transforms the YAML file into multiple platform-specific configuration files required by different system components. This conversion process generates environment variable files (.ENV), JSON configuration files, and other format-specific inputs necessary for seamless integration with the underlying middleware platform and web application frameworks. The automated conversion mechanism addresses one of the primary cost drivers identified in traditional BEMS implementations by eliminating 14 the manual configuration processes that typically require specialized technical expertise and contribute to the substantial deployment labor costs (Trenbath et al. 2022; Neukomm, Nubbe, and Fares 2019). By automating this conversion process, the proposed approach significantly reduces both manual configuration errors and deployment time, ensuring consistency across different system components. The detailed conversion process and converter implementation are described in Section 2.1.2. Phase 3: Application Installation and Execution: The third phase focuses on system deployment through a Docker containerized architecture (Docker, n.d.) that encapsulates the middleware, web server, and database components in portable, isolated environments for consistent deployment across various hardware configurations. The web server container seamlessly communicates with multiple client applications through RESTful APIs, enabling real-time data exchange and concurrent user access, whereas the VOLTTRON middleware container establishes direct communication channels with building devices via standardized BACnet and Modbus protocols (Katipamula et al. 2016). This containerized approach eliminates platform-specific dependencies and enables rapid automated deployment while supporting scalable multiclient connections and real-time data synchronization. The standardized container deployment process minimizes manual configuration requirements and reduces technical expertise barriers, allowing building operators to launch comprehensive monitoring systems through simplified, automated deployment scripts. Phase 4: Building System Monitoring and Control: The final phase provides building operators with comprehensive web-based interfaces that dynamically adapt to user preferences specified in the YAML configuration file. The responsive interface automatically displays relevant controllers based on the configured building systems while enabling visualization customization through selectable chart types (line charts, scatterplots, and box plots), configurable variables for data plotting, and adjustable visual parameters, including line types, colors, and widths. The interactive dashboards present customizable charts for energy consumption tracking, indoor environment monitoring with configured set points, and equipment status displays with flexible scheduling interfaces that support daily schedules, national holidays, and customized schedule and holiday configurations. The web interface responds to configuration settings specified in the YAML file, enabling building operators to modify system parameters and operational schedules through customized graphical user interfaces that eliminate the need for specialized programming expertise. Detailed implementation of these configuration options and interface customization capabilities is described in Section 2.2.1. 2.4 SUMMARY OF THE MAIN BENEFITS The proposed generalizable web user interface framework delivers four primary benefits that address the critical barriers impeding widespread BEMS adoption in the SMCB sector: • Minimized manual configuration efforts: The unified YAML configuration and automated conversion mechanism consolidate fragmented manual processes into streamlined workflows, enabling building operators to deploy BEMS without specialized programming expertise or labor-intensive setup procedures. • Automated semantic model integration: The automated Brick Schema parsing capability facilitates integration with existing semantic building models, reducing the human effort required to create configurations and enabling the systematic reuse of standardized building representations across multiple facilities. • Dynamic adaptability to heterogeneous systems: The configuration-driven web user interface dynamically adapts to various HVAC configurations without requiring separate interface development for each system type and is designed to scale from individual buildings to campus-wide deployment. 15 • Vendor-agnostic interoperability: Building on open-source VOLTTRON middleware and its support for industry-standard protocols, the framework is designed to enable building operators to leverage existing building automation infrastructure and emerging smart-device technologies without proprietary ecosystem dependencies. Collectively, these benefits transform previously inaccessible building automation capabilities into practical deployment opportunities for the underserved SMCB sector by minimizing human effort and enhancing BEMS interoperability, thereby establishing a foundation for widespread implementation of sophisticated energy management strategies. 16 3. VALIDATION The validation process evaluates two critical capabilities of the proposed generalizable web user interface across two distinct usage scenarios: (1) manual YAML-based configuration for buildings without existing semantic models (Section 3.1) and (2) automated conversion from existing semantic building models (Section 3.2). The first scenario addresses new deployments in which building operators must create system configurations from scratch, whereas the second scenario targets facilities with established semantic building models that seek to leverage existing model files for rapid deployment. This approach demonstrates both the platform’s flexibility in accommodating various system types through a unified YAML configuration and its capability to eliminate manual setup by automatically parsing semantic building models. 3.1 SIMULATION-BASED DEPLOYMENT TEST The simulation-based deployment test validates the primary usage scenario in which building operators create system configurations from scratch to deploy the BEMS in their facilities. This scenario requires the manual creation of YAML configuration files to specify system parameters, control points, and visualization preferences. Through this unified YAML configuration approach, the validation demonstrates the web user interface’s capability to dynamically adapt to different HVAC system and visualization configurations without interface modifications. This testing leverages the BOPTEST, which provides high-fidelity virtual building control test beds for systematic evaluation under controlled conditions (Blum et al. 2021). To evaluate the platform’s flexibility across different HVAC system types, the test employs a single YAML configuration file that encompasses two distinct BOPTEST test cases: BESTEST Air representing a fan-coil-based cooling and heating system, and BESTEST Hydronic representing a radiator-based heating system. This unified configuration file specifies both system configurations within a hierarchical structure, defining separate building units with their respective control points, visualization parameters, and operational schedules. The configuration converter processes this comprehensive YAML file to generate two system-specific JSON files that enable the web application to visualize and seamlessly switch between the two HVAC configurations without requiring separate deployments or interface modifications. The validation process involves three sequential phases. First, the unified YAML configuration file is created to define both HVAC systems, including their distinct control variables (fan speeds and air temperatures for the fan-coil-based system; pump speeds and water temperatures for the radiator-based system), visualization preferences, and system-specific parameters. Second, the configuration converter automatically transforms this single YAML file into multiple system-specific configuration files, including environment variable files for web client and server components, building unit specifications for both test cases, and configuration files for VOLTTRON middleware and its driver agent for each HVAC system type. Third, the web application is deployed using these generated configuration files to demonstrate its ability to dynamically render appropriate control interfaces and monitoring dashboards for each system configuration based solely on the configuration specifications. 3.1.1 Test Case Description The simulation testing employs two well-established BOPTEST test cases that represent different HVAC systems within standardized single-zone building environments. Both test cases use the BESTEST Case 900 model specification, featuring identical building envelopes with a 48 m2 floor area (6 m × 8 m), a 2.7 m ceiling height, and 12 m2 of two south-facing windows, but implement distinct heating and cooling 17 technologies to validate the interface’s adaptability (Judkoff and Neymark 1995). The BESTEST Air test case represents a commercial office environment located in Denver, Colorado (ASHRAE climate zone 5B, characterized by a cool-dry climate), employing a four-pipe fan-coil system with a variable-speed fan, chilled-water cooling coil, and hot-water heating coil and providing comprehensive air-side HVAC control through modulated supply air temperature, zone air heating and cooling temperature set points, and fan speed (ASHRAE 2020; Blum and Helsen 2023). In contrast, the BESTEST Hydronic test case represents a building environment located in Brussels, Belgium (ASHRAE climate zone 4A, characterized by a mixed-humid climate), using a radiator-based hydronic heating system with a circulation pump and gas-fired boiler. The system modulates the supply water temperature through PI control to maintain the zone operative temperature (Blum and Helsen 2023). Detailed HVAC system specifications are illustrated in Figure 9 and Figure A-3. These contrasting system configurations enable a systematic demonstration of the proposed interface’s capability to automatically adapt controller displays, data visualization, and monitoring dashboards through the unified YAML configuration approach. Figure 9. Schematic diagram of HVAC systems in the BESTEST Air and BESTEST Hydronic test cases. Adapted from (Blum and Helsen 2023). 3.1.2 YAML Configuration The YAML configuration specification consolidates essential information from data sources to enable automated deployment of both test cases. As summarized in Table A-1 and Table A-2, the building unit specification includes hierarchical spatial information (campus, building, and system identifiers), control points with default values, and operational characteristics (local time zone, building type, operator, and occupancy schedules), whereas the BOPTEST driver agent configuration includes an API endpoint (http://localhost:5000) and the mapping between the platform’s control point names (e.g., SupplyHeaterSetpoint) and BOPTEST-specific API parameter names (e.g., oveTSetSup_u). Figure 10 illustrates how these tabulated specifications are organized into the unified YAML file, which the configuration converter automatically translates into system-specific configuration files while maintaining data consistency across the deployment workflow. As shown in Figure 11 for the building unit and visualization parameters and in Figure A-4 for the VOLTTRON middleware, the configuration file employs a hierarchical structure that defines both test cases as separate building units within the same deployment. For the BESTEST Air configuration, the 18 Figure 10. Configuration creation and conversion process for the BESTEST Hydronic test case in the simulation-based deployment test. YAML specification defines control variables specific to air-based HVAC systems, including the supply air temperature set point (SupplyAirSetpoint), supply fan speed control (SupplyFanSpeed), and zone air cooling and heating set points (ZoneAirCoolingSetpoint, ZoneAirHeatingSetpoint). The visualization parameters specify line charts for tracking zone air temperature (ZoneAirTemperature) and cooling power consumption (CoolingPowerConsumption) against cooling and heating set points, with distinct color coding (blue for cooling and red for heating) and a box plot for statistical analysis of zone air temperature distribution. In contrast, the BESTEST Hydronic configuration specifies control variables tailored to radiator-based heating systems, including circulation pump control (ControlStagePump), the supply water temperature set point (SupplyHeaterSetpoint), and zone operative heating and cooling set points (ZoneOperativeCoolingSetpoint and ZoneOperativeHeatingSetpoint). The visualization configuration employs a line chart for zone operative temperature and a scatterplot correlating CO2 concentration (ZoneCo2Concentration) with heating power consumption (HeatingPowerConsumption). Both configurations share common structural elements, including VOLTTRON port assignments (tcp://127.0.0.1:23921), driver agent specifications with API endpoints, and data point definitions with labels and units. The configuration converter processes these specifications, generating separate configuration files for each building unit while preserving their interconnected deployment structure. 19 3.1.3 Deployment Results and Interface Adaptation The deployed web application demonstrates dynamic interface adaptation based on the processed YAML configurations, automatically rendering system-appropriate control panels and monitoring dashboards without manual interface customization. As depicted in Figure 12, the platform successfully deploys both HVAC configurations within a unified interface framework, with distinct visualization and control elements that reflect each system’s operational characteristics. For the BESTEST Air system, the interface automatically generates a control panel with sliders for the configured control points: the supply air temperature set point, fan speed, and the zone cooling and heating set points. The monitoring dashboard displays real-time zone air temperature overlaid with dynamic cooling and heating set point indicators, using the blue–red color scheme specified in the YAML configuration. The dashboard also plots cooling power consumption reported by the emulator on a secondary axis, enabling users to monitor the zone’s thermal response and its energy use simultaneously. For the BESTEST Hydronic test case, the interface adapts to display heating-specific controls based on the YAML specification. The control panel presents pump stage, the supply water temperature set point, and the zone operative heating and cooling set points. The line chart displays the zone operative temperature overlaid with its heating and cooling set points, which take distinct values according to the configured occupancy schedules. The other scatterplot represents the relationship between heating power consumption and zone CO2 concentration using data from both occupied and unoccupied periods provided by the emulator. Both interfaces maintain consistent navigation structures and responsive layouts while adapting their content to system-specific requirements. The platform’s automatic generation of appropriate control elements, visualization parameters, and operational indicators validates the effectiveness of the configuration-driven approach, demonstrating that building operators can deploy various HVAC system interfaces through YAML specifications without requiring programming expertise or manual interface development. This result illustrates the platform’s capability to bridge the gap between heterogeneous HVAC architectures and unified building management interfaces, which is a prerequisite for scalable deployment across various SMCB environments. 3.2 SEMANTIC BUILDING MODEL CONFIGURATION TEST The semantic model test demonstrates the second usage scenario, in which facilities possess Brick Schema documentation of their building systems. This scenario enables building operators to automatically generate building unit configuration files by directly importing semantic building models while still specifying other system parameters (web client, web server, and middleware settings) in a YAML document. The test employs an industrial Brick model representing a complex commercial building environment to evaluate the configuration converter’s capability to automatically extract hierarchical building structures, identify control points, and generate deployment-ready configuration files from existing semantic descriptions without manual intervention. This automation significantly reduces the configuration effort required of operators with established semantic building models. 3.2.1 Studied Semantic Building Model The validation uses the reference Brick model bldg27 obtained from the Brick Ontology website, which represents a validated semantic description of an anonymized real building in Turtle data format (Brick, n.d.; Fierro et al. 2019). This model represents a real commercial building with a 20,605 ft2 (1,913 m2 ) footprint and a multizone HVAC system, providing an appropriate test bed for demonstrating the platform’s capability to process real-world building complexity in semantic data format. The studied model features a representative variable air volume (VAV)-based HVAC system configuration commonly found in small 20 commercial buildings. The system consists of one air handling unit (AHU) (AC07) serving 17 VAV boxes distributed across two floors. Each VAV box feeds a dedicated HVAC zone corresponding to individual rooms, with the first floor containing seven zones and the second floor containing 10 zones. The semantic model includes comprehensive point definitions for each component: the AHU contains nine control and sensor points such as supply air temperature set point, mixed air temperature, and return air temperature Each VAV box incorporates three control points, including zone temperature set points, a damper position set point, and heating mode. Together with additional zone-level sensor points, this yields approximately 100 distinct control and sensor points across the building, providing a robust test case for evaluating the converter’s ability to parse complex hierarchical relationships and generate accurate configuration files from a real, multizone semantic model. 3.2.2 Validation Process and Results The conversion process begins with importing a comprehensive Brick model that includes hierarchical levels (building and zones), various equipment types (AHUs, VAV boxes, and dampers), and extensive control point definitions (set points, sensors, and HVAC mode). The configuration converter automatically parses the RDF-formatted Brick model, identifying building elements through their semantic relationships and class definitions. The converter recognizes spatial hierarchies through location classes (e.g., brick:Site, brick:Building, and brick:HVAC_Zone) and their brick:hasPart relationships, whereas equipment and control points are identified through equipment classes (e.g., brick:AHU and brick:VAV) and their associated point classes via brick:hasPoint relationships. Because the studied Brick model lacks campus information and default control values required for BEMS deployment, the converter automatically supplements these missing parameters with predefined default values that users can easily adjust according to their operational preferences. Upon successful parsing, the converter generates a structured building configuration file that preserves the semantic model’s hierarchical organization while translating it into the JSON format required by the web application, as shown in Figure 13. The deployed web interface dynamically renders the imported building structure, automatically generating navigation hierarchies that reflect the campus–building–zone relationships defined in the original Brick model. This automated translation eliminates the manual configuration processes typically required when implementing building management systems for existing facilities with established semantic models, demonstrating the platform’s capability to leverage standardized building information models for rapid deployment. 21 Figure 11. Building unit configuration in the prepared YAML file for the BESTEST Air and BESTEST Hydronic test cases. 22 Figure 12. Deployed web application using the BESTEST Air and BESTEST Hydronic configuration files. (a) (b) Figure 13. Converted building unit configuration file for zone RM050 from the studied Brick model (left) and corresponding deployed web interface (right). 23 4. FIELD DEMONSTRATION Field testing was conducted in an occupied commercial building to demonstrate the feasibility of the platform’s deployment workflow under real operating conditions. Unlike the controlled simulation environment, this real-world testing demonstrates the platform using smart thermostats and energy meters across different zones within the same building through the unified YAML configuration approach. It specifically covers the complete deployment workflow, from YAML configuration creation to automated conversion, system deployment, and multizone monitoring. By implementing configured thermostat set points and schedules across multiple zones, the demonstration assesses the platform’s practical applicability for scalable building energy management in SMCB environments. 4.1 STUDIED BUILDING The studied building is a two-story commercial office building located in Oak Ridge, Tennessee, within ASHRAE climate zone 4A, characterized by mixed-humid climate (ASHRAE 2020), as shown in Figure 14. This facility, originally constructed in 1988, underwent comprehensive envelope improvements in 2023, representing a typical SMCB environment with enhanced thermal performance characteristics. The building encompasses 1,035 m2 of total floor area organized through perimeter-core zoning, where the central core accommodates common areas including conference rooms and common spaces, whereas the perimeter zones comprise individual office spaces. The building accommodates approximately 60 occupants following typical commercial office schedules, with operational hours from 7:00 a.m. to 5:00 p.m. on weekdays. The facility employs an all-electric HVAC infrastructure comprising heterogeneous system types: RTUs provide centralized climate control for core zones and corridors, whereas individual package terminal air conditioning units (PTACs) serve each perimeter office space. Each HVAC unit operates through dedicated ecobee smart thermostat controls that enable Wi-Fi connectivity and vendor API integration for comprehensive building management capabilities. To enable real-time energy monitoring, the building incorporates a power-metering infrastructure in which individual current transformers monitor each HVAC circuit dedicated to a specific HVAC unit. These transformers communicate with a centralized meter that connects to a computer via Modbus Ethernet protocol, enabling periodic power data collection at 1-minute intervals. The collected energy data are subsequently forwarded to the VOLTTRON middleware platform through an API. This smart thermostat and energy-metering infrastructure provides real-time access to zone temperatures, power consumption, and remote set point adjustment functionality, creating a representative test bed for demonstrating the proposed platform’s capability to manage multiple HVAC system configurations and monitor energy use within a real building environment. 4.2 YAML CONFIGURATION AND DEPLOYMENT To demonstrate the practical deployment workflow, a unified YAML configuration file was created for the studied building, incorporating specifications for representative perimeter office spaces. The YAML file incorporated essential building data defining Room 112, Room 119, and Room 125 within the studied building; integration settings for the smart thermostats and energy meters (API endpoints and polling intervals); and visualization preferences (chart types, variable selections, and display parameters). The configuration specified two operating strategies to illustrate energy-saving potential. On June 13, a baseline strategy maintained constant set points throughout the monitoring period, holding a 70°F (21.1°C) heating set point and a 74°F (23.3°C) cooling set point. On June 14, a schedule-based setback strategy applied the same 70°F heating and 74°F cooling set points during occupied periods but raised the cooling set point to 85°F (29.4°C) and lowered the heating set point to 60°F (15.6°C) during unoccupied periods to reduce 24 Figure 14. Exterior view of the small commercial office building used for the field demonstration. cooling energy use in the cooling-dominated test period. The automated configuration converter processed this unified YAML specification to generate four system-specific configuration files: environment variables for web client and server components, building unit specifications in JSON format, and VOLTTRON middleware configurations, as illustrated in Figure A-5. The deployed web interface automatically rendered the zone monitoring dashboard based on these configuration files, demonstrating the platform’s capability to accommodate various operational strategies without manual interface customization. 4.3 COMPARATIVE MONITORING RESULTS A comparative 24-hour controlled monitoring test was conducted on two consecutive weekend days, June 13 and June 14, 2026, while the building was unoccupied to eliminate confounding factors from actual occupancy, focusing on cooling system performance and energy consumption under a predefined occupancy schedule. Outdoor temperatures over the two test days ranged from approximately 72°F to 87°F (22.2°C–30.6°C), providing comparable thermal conditions for the two control strategies. Figure 15 and Figure A-6 present the deployed web application monitoring Room 112, Room 119, and Room 125 across both test days. On June 14, the web application implemented the setback control strategy under a predefined occupancy schedule: occupied set points (70°F heating, 74°F cooling) from 5:00 a.m. to 6:00 p.m. and setback set points (60°F heating, 85°F cooling) from 6:00 p.m. to 5:00 a.m. In contrast, June 13 maintained the constant 70°F heating and 74°F cooling set points throughout the entire period, serving as the baseline for energy comparison. The line charts display zone air temperature trajectories (orange solid lines) closely tracking their respective cooling set points (red dashed lines). The chart on June 14 shows the set point transitions, and the chart on June 13 shows the constant set point baseline. The purple solid lines represent power consumption, with the setback control on June 14 exhibiting less cooling power during the setback period than the constant baseline on June 13. This demonstration confirms that 25 the platform can deploy, run, and visualize two distinct set point strategies and compare their energy use through a single unified interface. (a) (b) Figure 15. Deployed web application for monitoring Room 125 with setback control (a) and constant set point (b) in the studied building. 4.4 REQUIRED EFFORT FOR THE DEPLOYMENT PROCESS To quantitatively evaluate the required effort for deployment process, a test was conducted with a junior researcher with a Ph.D. in electrical engineering who was not involved in this application development. First, a 40-minute training session was provided to introduce the general concept of this project and deployment process, including writing YAML configuration files and installing Docker containers, as described in Section 2.3 and illustrated in Figure A-2. The researcher then independently implemented the deployment process while the required time was measured. The complete deployment process for Room 125 required 1 hour and 27 minutes of labor, encompassing downloading and extracting repository files (8 minutes), YAML configuration file creation and automated conversion (10 minutes), and Docker container installation with system initialization (9 minutes). The researcher spent approximately 1 hour debugging to correct YAML syntax errors during configuration conversion and address Docker installation issues caused by software version compatibility. The measured deployment effort provides empirical evidence for evaluating the proposed framework’s practical applicability and labor requirement reduction compared with conventional BEMS deployment approaches, as discussed in Section 5. 26 5. DISCUSSION The simulation-based testing and field deployment demonstrate that the proposed generalizable web user interface framework addresses key technical barriers impeding wider adoption of BEMS within the SMCB sector. Specifically, the framework adapts a single configuration-driven interface to heterogeneous building systems, automates the conversion of a unified specification into system-specific configuration files to reduce manual setup, and supports vendor-agnostic device integration through industry-standard protocol support. The following discussion synthesizes these findings into the key contributions and their implications for scalable building energy management deployment in SMCB environments. Configuration-driven adaptability to heterogeneous building systems: Simulation-based testing demonstrated the framework’s adaptability across heterogeneous HVAC configurations, deploying both a fan-coil-based system with supply air temperature and fan-speed controls (BESTEST Air) and a radiator-based system with pump and water temperature controls (BESTEST Hydronic) through a single, unified YAML specification. As demonstrated in Section 3.1.3, the web interface automatically adapted system-appropriate control panels, visualization charts (box plots for BESTEST Air versus scatterplots for BESTEST Hydronic), and operational schedules without manual interface customization. The field demonstration in Section 4.3 further validated this adaptability through real-time monitoring of three office zones with smart thermostat and energy meter configurations while implementing two distinct operational strategies, a schedule-based setback control versus a constant set point baseline, within the unified web user interface. These results indicate that building operators can deploy monitoring dashboards across various building systems without developing separate interfaces for each configuration, reducing deployment complexity for comprehensive automation in the SMCB sector. Automated configuration conversion eliminating manual setup processes: The automated configuration converter validated seamless transformation from unified YAML specifications to system-specific configuration files across multiple test scenarios, mitigating the labor-intensive setup processes identified as primary cost barriers. The simulation-based deployment test in Section 3.1.2 demonstrated the conversion of unified, human-readable YAML specifications into four system-specific configuration files (web client, web server, building unit, and VOLTTRON middleware) for both test cases, maintaining data consistency across interdependent system components without manual intervention. The semantic building model test in Section 3.2 validated seamless transformation from the industrial Brick model containing 17 VAV zones into deployment-ready building unit configurations without manual intervention. The automated converter successfully parsed hierarchical relationships (brick:hasPart), identified equipment classes (brick:AHU and brick:VAV), and extracted control points (brick:hasPoint) while automatically supplementing missing default values with user-adjustable defaults. Furthermore, the deployment-effort evaluation in Section 4.4 showed that configuring and deploying a single-zone monitoring interface for the studied building required 1 hour and 27 minutes of labor for one operator, an approximately 98% reduction compared with the 65 hours that Guo et al. (2025) reported for the software installation and configuration of a conventional BEMS in a similarly sized SMCB. This reduction indicates that the automated configuration process can substantially mitigate the technical expertise and labor requirements for BEMS deployment. Vendor-agnostic interoperability through standardized protocol integration: In the field demonstration described in Section 4.3, the platform integrated ecobee smart thermostats through their vendor-specific API and Schneider power meters over the Modbus protocol, interfacing with both simultaneously without device-specific integration programming. This vendor-agnostic interoperability eliminates proprietary ecosystem dependencies that constrain existing API-based solutions, enabling building operators to 27 leverage various building devices without vendor lock-in constraints or the substantial licensing costs associated with commercial integrated software. 28 6. CONCLUSION This research addresses the critical gap between advancing smart device capabilities and accessible software infrastructure in the SMCB sector by developing a generalizable web user interface framework. The proposed platform mitigates deployment complexity through automated configuration conversion, industry-standard protocol integration, and configuration-driven user interfaces. By consolidating fragmented setup processes into unified YAML specifications and leveraging open-source VOLTTRON middleware, the framework reduces labor-intensive configuration requirements that constitute substantial implementation cost barriers, enabling building operators to deploy comprehensive monitoring systems without specialized programming expertise or proprietary ecosystem dependencies. Demonstrations through simulation-based testing and a field deployment illustrate the platform’s adaptability across heterogeneous HVAC configurations, its automated semantic-model parsing, and its practical operational feasibility. The simulation-based testing showed interface adaptation between fan-coil-based and radiator-based systems without manual customization, whereas Brick Schema integration enabled automated conversion of a real multizone semantic building model into a deployment-ready configuration. The field demonstration confirmed practical deployment in an occupied office building, including real-time multizone monitoring and the implementation and visualization of two contrasting set point strategies. Consequently, these results indicate that the platform can accommodate heterogeneous building systems through streamlined deployment workflows that substantially reduce software installation and configuration effort. The developed framework establishes a foundation for scalable building energy management in the underserved SMCB sector, transforming previously inaccessible automation capabilities into practical deployment opportunities. By mitigating substantial upfront investments, vendor lock-in constraints, and specialized technical expertise requirements, this research offers opportunities for implementing sophisticated energy optimization strategies. The platform’s open-source architecture and standardized-protocol support provide a foundation for future integration of advanced control algorithms, grid-interactive capabilities, and campus-wide building energy management, contributing to significant energy performance enhancement in the commercial building sector. 6.1 LIMITATIONS Although the demonstration results show the feasibility and effectiveness of the proposed generalizable web user interface framework, several limitations still remain. First, the field demonstration was constrained to a single commercial office building with ecobee smart thermostats and PTAC/RTU systems. Although the simulation-based testing demonstrated adaptability across fan-coil and hydronic radiator configurations, comprehensive validation requires deployment across various SMCB environments with varying HVAC types, building uses, and geographic locations, as well as simultaneous deployment across multiple buildings on a campus to assess scalability. Second, although the semantic model integration was validated using a single industrial Brick model representing a 17-zone VAV system, testing various Brick models with varying HVAC types and hierarchical structures would ensure robust conversion capabilities. Additionally, compatibility with other widely adopted semantic model frameworks remains unexplored, potentially limiting interoperability with industrial semantic building models. Third, although the field demonstration acquired power-meter data over Modbus, this data reached the platform through an intermediate API rather than the native VOLTTRON fieldbus drivers; end-to-end validation of the native drivers for BACnet, Modbus, KNX, LonWorks, and MQTT remains future work. 29 Fourth, the field demonstration employed only a basic schedule-based setback strategy over two test days. Extended deployment incorporating sensor-driven occupancy, seasonal variations, and advanced control algorithms is necessary to quantify long-term performance benefits and system reliability under various operational and climate conditions. Fifth, although the deployment-effort measurement in Section 4.4 provides a single-subject, single-zone data point, a quantitative cost–benefit analysis comparing total implementation costs, including labor hours across multiple nonexpert operators, with those of conventional BEMS applications was not conducted. Such economic validation would strengthen the value proposition for SMCB operators considering platform adoption. 6.2 FUTURE WORK Future research directions can advance the platform’s capabilities and practical applicability across broader SMCB contexts. First, multisite deployment validation across SMCBs with varying HVAC configurations, building types, and climates would enable a comprehensive scalability evaluation of the proposed platform. Large-scale campus deployment studies could assess concurrent multibuilding management capabilities, network scalability, and centralized monitoring performance under real-world operational conditions with varying system complexities and deployment scenarios. Second, validating semantic model integration across various industry Brick models with varied equipment types, control hierarchies, and point definitions would ensure robust conversion capabilities across a range of real-world building configurations. Furthermore, expanding integration to the Project Haystack or ASHRAE 223P ontology frameworks would enhance interoperability with various BEMS applications (Project Haystack, n.d.; Fierro, n.d.). Future work could explore automated translation or semantic alignment enhancement, facilitating seamless conversion and bidirectional data exchange with industrial building management applications employing various semantic representations. Third, integrating advanced control strategies would unlock sophisticated energy-optimization capabilities, including model predictive control, reinforcement learning, and ASHRAE Guideline 36 standardized sequences (Blum et al. 2022; Faulkner et al. 2023; Wang and Hong 2020). Future work could develop modular control frameworks within the open-source architecture, enabling building operators to deploy various control strategies through the proposed web interfaces while incorporating external data sources such as weather forecasts, occupancy predictions, and dynamic electricity price signals. This integration would enable BEMS platforms to autonomously optimize energy performance while maintaining occupant comfort, reducing reliance on manual interventions and unlocking grid-interactive capabilities for flexible demand response. Fourth, a comprehensive cost–benefit analysis comparing implementation costs, operational savings, and return on investment with those of conventional BEMS solutions would provide quantitative evidence supporting adoption decisions for the proposed platform. Comparative studies employing extensive surveys and focus group interviews with SMCB operators could validate practical advantages while identifying deployment barriers. Furthermore, such research could provide a foundation for developing standardized cost–benefit frameworks accounting for hardware investments, installation labor, software licensing, maintenance requirements, and realized energy savings, enabling SMCB owners to make informed investment decisions based on building-specific deployment scenarios. 30 ACKNOWLEDGMENTS This material is based upon work supported by the U.S. Department of Energy, Office of Critical Minerals and Energy Innovation (CMEI), specifically the Building Technology Office (BTO) under contract number DE-AC05-00OR22725. This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of its employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. 31 7. REFERENCES Amazon. n.d. “Amazon Smart Thermostat.” Accessed June 8, 2025. https://a.co/d/iKiuV5X. ASHRAE. 2020. ANSI/ASHRAE Addendum a to Standard 169-2013: Climatic Data for Building Design Standards. Technical report. American Society of Heating, Refrigerating and Air-Conditioning Engineers. Ayan, Onur, and Belgin Turkay. 2018. “Smart thermostats for home automation systems and energy savings from smart thermostats.” In 2018 6th International Conference on Control Engineering & Information Technology (CEIT), 1–6. IEEE. Baig, Mirza Jabbar Aziz, M Tariq Iqbal, Mohsin Jamil, and Jahangir Khan. 2021. “Design and implementation of an open-Source IoT and blockchain-based peer-to-peer energy trading platform using ESP32-S2, Node-Red and, MQTT protocol.” Energy reports 7:5733–5746. Balaji, Bharathan, Arka Bhattacharya, Gabriel Fierro, Jingkun Gao, Joshua Gluck, Dezhi Hong, Aslak Johansen, Jason Koh, Joern Ploennigs, Yuvraj Agarwal, et al. 2016. “Brick: Towards a unified metadata schema for buildings.” In Proceedings of the 3rd ACM International Conference on Systems for Energy-Efficient Built Environments, 41–50. Barnes, Elizabeth, and Kristen Parrish. 2016. “Small buildings, big impacts: The role of small commercial building energy efficiency case studies in 2030 Districts.” Sustainable Cities and Society 27:210–221. Ben-Kiki, O., C. Evans, and I. döt Net. 2009. “YAML Ain’t Markup Language (YAML™) Version 1.2.” Accessed June 8, 2025. https://yaml.org/spec/1.2.1/. Bhatt, Jignesh, and HK Verma. 2015. “Design and development of wired building automation systems.” Energy and Buildings 103:396–413. Blum, D., and L. Helsen. 2023. “Test Cases: Ready-made building emulators.” International Building Performance Simulation Association. Accessed June 8, 2025. https : / / ibpsa . github . io / project1 - boptest/testcases/. Blum, David, Javier Arroyo, Sen Huang, Ján Drgoňa, Filip Jorissen, Harald Taxt Walnum, Yan Chen, Kyle Benne, Draguna Vrabie, Michael Wetter, et al. 2021. “Building optimization testing framework (BOPTEST) for simulation-based benchmarking of control strategies in buildings.” Journal of Building Performance Simulation 14 (5): 586–610. Blum, David, Zhe Wang, Chris Weyandt, Donghun Kim, Michael Wetter, Tianzhen Hong, and Mary Ann Piette. 2022. “Field demonstration and implementation analysis of model predictive control in an office HVAC system.” Applied energy 318 (119104): 10–1016. Brick. n.d. “Brick: A Uniform Metadata Schema for Buildings.” Accessed September 1, 2025. https : / / brickschema.org/. Chaudhari, Pratiksha, Yang Xiao, Mark Ming-Cheng Cheng, and Tieshan Li. 2024. “Fundamentals, algorithms, and technologies of occupancy detection for smart buildings using IOT sensors.” Sensors 24 (7): 2123. 32 Chen, Cheng-Yi, Sheng-Han Wu, Bo-Wun Huang, Chao-Hung Huang, and Cheng-Fu Yang. 2024. “Web-based Internet of Things on environmental and lighting control and monitoring system using node-RED, MQTT and Modbus communications within embedded Linux platform.” Internet of Things 27:101305. Chiesa, Giacomo, Silvia Cesari, Miguel Garcia, Mohammad Issa, and Shuyang Li. 2019. “Multisensor IoT platform for optimising IAQ levels in buildings through a smart ventilation system.” Sustainability 11 (20): 5777. DNP Users Group. n.d. “Innovative Technology for Reliable and Secure OT Communications.” Accessed June 8, 2025. https://www.dnp.org/. Docker. n.d. “Develop faster.” Docker Inc. Accessed June 8, 2025. https://www.docker.com/. Domingues, Pedro, Paulo Carreira, Renato Vieira, and Wolfgang Kastner. 2016. “Building automation systems: Concepts and technology review.” Computer Standards & Interfaces 45:1–12. Ecobee. n.d.-c. “Smart Thermostats.” Accessed June 8, 2025. https : / / www . ecobee . com / en - us / smart - thermostats/. Essa, Mohamed El-Sayed M, Ahmed M El-shafeey, Amna Hassan Omar, Adel Essa Fathi, Ahmed Sabry Abo El Maref, Joseph Victor W Lotfy, and Mohamed Saleh El-Sayed. 2023. “Reliable integration of neural network and internet of things for forecasting, controlling, and monitoring of experimental building management system.” Sustainability 15 (3): 2168. Facilio. n.d. “Facilio: Connected building operations & IoT-led property platform.” Accessed Sep. 1, 2025. https://facilio.com/. Faulkner, Cary A, Robert Lutes, Sen Huang, Wangda Zuo, and Draguna Vrabie. 2023. “Simulation-based assessment of ASHRAE Guideline 36, considering energy performance, indoor air quality, and control stability.” Building and Environment 240:110371. Fierro, Gabe. n.d. “Digitizing the Delivery of Controls and Analytics through the Proposed ASHRAE Standard 223P.” Accessed Sep. 1, 2025. https://gtf.fyi/files/2024-ASHRAE-223P-Fierro.pdf. Fierro, Gabe, Jason Koh, Shreyas Nagare, Xiaolin Zang, Yuvraj Agarwal, Rajesh K Gupta, and David E Culler. 2020. “Formalizing tag-based metadata with the brick ontology.” Frontiers in Built Environment 6:558034. Fierro, Gabe, Marco Pritoni, Moustafa AbdelBaky, Daniel Lengyel, John Leyden, Anand Prakash, Pranav Gupta, Paul Raftery, Therese Peffer, Greg Thomson, et al. 2019. “Mortar: an open testbed for portable building analytics.” ACM Transactions on Sensor Networks (TOSN) 16 (1): 1–31. Gaidon, C., and M. Poplawski. 2017. Connected Lighting System Interoperability Study Part 1: Application Programming Interfaces. Technical report PNNL-26946. Richland, WA (United States): Pacific Northwest National Laboratory (PNNL). Giacobbe, Maurizio, Chakib Chaouch, Marco Scarpa, and Antonio Puliafito. 2018. “An implementation of influxdb for monitoring and analytics in distributed iot environments.” In International conference on the Sciences of Electronics, Technologies of Information and Telecommunications, 155–162. Springer. Google. n.d.-c. “Nest Thermostats Help You Stay Comfortable & Save Energy.” Accessed June 8, 2025. https://store.google.com/category/nest_thermostats. 33 Grafana Labs. n.d. “Dashboard HTTP API.” Grafana Documentation. Accessed June 8, 2025. https : / / grafana.com/docs/grafana/latest/developers/http_api/dashboard/. Guo, Mingyue, Boming Liu, Sen Huang, Jin Dong, Jianming Lian, and Zheng O’Neill. 2025. “Deploying advanced supervisory control strategies to small-and medium-sized commercial buildings: Case study and lessons learned.” Energy and Buildings 338:115710. Haack, Jereme, Bora Akyol, Craig Allwardt, Srinivas Katipamula, Zachary Beech, Robert Lutes, Joseph Chapman, and Kyle Monson. 2016. “VOLTTRON™: Using distributed control and sensing to integrate buildings and the grid.” In 2016 IEEE 3rd World Forum on Internet of Things (WF-IoT), 228–232. IEEE. Ham, Sang-woo, Lazlo Paul, Donghun Kim, Marco Pritoni, Richard Brown, and Jingjuan Dove Feng. 2024. “Decarbonization of heat pump dual fuel systems using a practical model predictive control: Field demonstration in a small commercial building.” Applied Energy 361:122935. Home Assistant. n.d. “Awaken Your Home.” Open Home Foundation. Accessed June 8, 2025. https : / / www.home-assistant.io/. Honeywell. n.d. “Thermostats — Shop Smart Models for Any Room.” Honeywell Home. Accessed June 8, 2025. https://www.honeywellhome.com/us/en/products/air/thermostats/. Huppert, M, R Cochrane, P Frey, J Wiser, S Denniston, M Frankel, and D Hewitt. 2013. “Realizing the Energy Efficiency Potential of Small Buildings.” National Trust for Historic Preservation Preservation Green Lab, Seattle, WA (June 2013). InfluxData. n.d.-b. “Chronograf.” InfluxData Inc. Accessed June 8, 2025. https://www.influxdata.com/ time-series-platform/chronograf/. . n.d.-a. “InfluxDB Database.” InfluxData Inc. Accessed June 8, 2025. http://influxdata.com/lp/ influxdb-database/. International Organization for Standardization. 2004. ISO 16484-5: Building Automation and Control Systems, Part 5: Data Communication Protocol. Technical report. Suomen Standardisoimisliitto. ISO/IEC. 2006. ISO/IEC 14543-3-1: Information Technology — Home Electronic Systems (HES) Architecture — Part 3-1: Communication Layers — Application Layer for Network Based Control of HES Class 1. Technical report. International Organization for Standardization (ISO) and International Electrotechnical Commission (IEC). . 2012. ISO/IEC 14908-1: Information Technology — Control Network Protocol — Part 1. Technical report. International Organization for Standardization (ISO) and International Electrotechnical Commission (IEC). Jornet-Monteverde, Julio Antonio, and Juan José Galiana-Merino. 2020. “Low-cost conversion of single-zone HVAC systems to multi-zone control systems using low-power wireless sensor networks.” Sensors 20 (13): 3611. Judkoff, R., and J. Neymark. 1995. International Energy Agency Building Energy Simulation Test (BESTEST) and Diagnostic Method. Technical report NREL/TP-472-6231. Golden, CO (United States): National Renewable Energy Laboratory (NREL). 34 Katipamula, S. 2022. Autonomous Energy Management Software System for Small Commercial Buildings in Support of Decarbonization. Technical report PNNL-32889; 569 (78802). Richland, WA (United States); Norristown, PA (United States): Pacific Northwest National Laboratory (PNNL) and Intellimation LLC. Katipamula, S., R. G. Lutes, and T. Herr. 2023. Design Requirements and Software Specification for the Autonomous Energy Management Software System for Small Commercial Buildings. Technical report PNNL-33134. Richland, WA (United States): Pacific Northwest National Laboratory (PNNL). Katipamula, Srinivas, Jereme Haack, George Hernandez, Bora Akyol, and Joseph Hagerman. 2016. “VOLTTRON: An open-source software platform of the future.” IEEE Electrification Magazine 4 (4): 15–22. Kim, Donghun, and James E Braun. 2018. “Development, implementation and performance of a model predictive controller for packaged air conditioners in small and medium-sized commercial building applications.” Energy and Buildings 178:49–60. Kim, Woohyun, Srinivas Katipamula, and Robert Lutes. 2019. “Development and evaluation of HVAC operation schedule detection algorithm.” Energy and Buildings 202:109350. KODE OS. n.d. “KODE OS: Centralised building data & operations platform.” Accessed Sep. 1, 2025. https://kodelabs.com/platform/. Korodi, Adrian, Ioana-Victoria Nit, ulescu, Adriana-Anamaria Fülöp, Vlad-Cristian Vesa, Petru Demian, Robert-Adelin Braneci, and Daniel Popescu. 2024. “Integration of Legacy Industrial Equipment in a Building-Management System Industry 5.0 Scenario.” Electronics 13 (16): 3229. Li, Han, and Tianzhen Hong. 2025. “A digital twin platform for building performance monitoring and optimization: Performance simulation and case studies.” In Building Simulation, 1–19. Springer. LoRaWAN. n.d. “LoRaWAN transforms businesses by connecting wireless IoT sensors simply and affordably.” LoRa Alliance. Accessed June 8, 2025. https://lora-alliance.org/. Luo, Xuan, Tianzhen Hong, Yixing Chen, and Mary Ann Piette. 2017. “Electric load shape benchmarking for small-and medium-sized commercial buildings.” Applied energy 204:715–725. Lutes, R., C. Allwardt, A. Bleeker, and N. Tenney. 2025. “Volttron PNNL AEMS.” [Source code]. GitHub. https://github.com/VOLTTRON/volttron-pnnl-aems. Lutes, R., C. Allwardt, S. Niddodi, and D. Raker. 2025. “Volttron PNNL Applications.” [Source code]. GitHub. Ma, Zheng, Anders Clausen, Yun Lin, and Bo Nørregaard Jørgensen. 2021. “An overview of digitalization for the building-to-grid ecosystem.” Energy Informatics 4 (Suppl 2): 36. Malkawi, Ali, Stephen Ervin, Xu Han, Elence Xinzhu Chen, Sunghwan Lim, Spyridon Ampanavos, and Peter Howard. 2023. “Design and applications of an IoT architecture for data-driven smart building operations and experimentation.” Energy and Buildings 295:113291. Mataloto, Bruno, Joao C Ferreira, and Nuno Cruz. 2019. “LoBEMS—IoT for building and energy management systems.” Electronics 8 (7): 763. Meta Platforms, Inc. n.d. “React: The library for web and native user interfaces.” Accessed June 8, 2025. https://react.dev/. 35 Modbus IDA. 2004. “Modbus Application Protocol Specification v1.1a.” 51 pp. http://www.modbus.org/ specs.php. MQTT. n.d. “MQTT: The Standard for IoT Messaging.” Accessed June 8, 2025. https://mqtt.org/. National Energy System Operator (NESO). n.d. “Carbon Intensity API.” Accessed June 8, 2025. https : //carbonintensity.org.uk/. National Renewable Energy Laboratory (NREL). n.d. “Utility Rates API.” Accessed June 8, 2025. https: //developer.nrel.gov/docs/electricity/utility%E2%80%90rates%E2%80%90v3/. Neukomm, M., V. Nubbe, and R. Fares. 2019. Grid-Interactive Efficient Buildings Technical Report Series: Overview of Research Challenges and Gaps. Technical report. U.S. Department of Energy. Njova, Dion, Kingsley Ogudo, and Patrice Umenne. 2020. “Packet Analysis of DNP3 protocol over TCP/IP at an Electrical Substation Grid modelled in OPNET.” In 2020 IEEE PES/IAS PowerAfrica, 1–5. IEEE. Node-RED. n.d. “Low-code Programming for Event-Driven Applications.” OpenJS Foundation. Accessed June 8, 2025. https://nodered.org/. Normal Software. 2025. “Normal Framework (NF): Software Framework for Building and IoT Systems.” Accessed November 8, 2025. https://www.normal.dev/. openHAB. n.d. “Empowering the Smart Home.” openHAB Foundation. Accessed June 8, 2025. https : //www.openhab.org/. OpenWeather. n.d. “Weather forecasts, nowcasts and history in a fast and elegant way.” OpenWeather Foundation. Accessed June 8, 2025. https://openweathermap.org/. Palantir Technologies. n.d. “Blueprint.js: A React-based UI toolkit for the web.” Accessed June 8, 2025. https://blueprintjs.com/. Pallets. n.d. “Flask: A lightweight WSGI web application framework.” Accessed June 8, 2025. https : / / flask.palletsprojects.com/en/stable/. Paul, Lazlo, Flavia De Andrade Pereira, Anand Krishnan Prakash, Sang Woo Ham, Jingjuan Dove Feng, Rich Brown, and Marco Pritoni. 2025. “Open building operating system: a grid-responsive semantics-driven control platform for buildings.” Science and Technology for the Built Environment 31 (3): 294–311. PostgreSQL. n.d. “PostgreSQL: The world’s most advanced open source relational database.” Accessed June 8, 2025. https://www.postgresql.org/. Predoaia, Ionut, Dimitris Kolovos, Antonio Garcia-Dominguez, Matthias Lenk, Wolfram Ebel, and Jan Burkl. 2024. “Towards Processing YAML Documents with Model Management Languages.” In Proceedings of the ACM/IEEE 27th International Conference on Model Driven Engineering Languages and Systems, 970–979. Prisma Data, Inc. n.d. “Prisma: Instant Postgres plus an ORM for simpler db workflows.” Accessed June 8, 2025. https://www.prisma.io/. Pritoni, Marco, Drew Paine, Gabriel Fierro, Cory Mosiman, Michael Poplawski, Avijit Saha, Joel Bender, and Jessica Granderson. 2021. “Metadata schemas and ontologies for building energy applications: A critical review and use case analysis.” Energies 14 (7): 2024. 36 Project Haystack. n.d. “Project Haystack: An Open-Source Initiative to Streamline Working with IoT Data.” Accessed September 1, 2025. https://www.project-haystack.org/. PyYAML. n.d. “PyYAML: YAML parser and emitter for Python.” Accessed June 8, 2025. https://pyyaml. org/. RDFLib Team. n.d. “RDFLib documentation.” Accessed September 1, 2025. https://rdflib.readthedocs.io/ en/stable/. Schiera, Daniele Salvatore, Luca Barbierato, Andrea Lanzini, Romano Borchiellini, Enrico Pons, Ettore Bompard, Edoardo Patti, Enrico Macii, and Lorenzo Bottaccioli. 2021. “A distributed multimodel platform to cosimulate multienergy systems in smart buildings.” IEEE Transactions on Industry Applications 57 (5): 4428–4440. Stopps, Helen, and Marianne F Touchie. 2021. “Residential smart thermostat use: An exploration of thermostat programming, environmental attitudes, and the influence of smart controls on energy savings.” Energy and Buildings 238:110834. Suryadevara, Nagender Kumar, and Gyan Ranjan Biswal. 2019. “Smart plugs: Paradigms and applications in the smart city-and-smart grid.” Energies 12 (10): 1957. Tamas, Ruth, William O’Brien, and Mario Santana Quintero. 2021. “Residential thermostat usability: Comparing manual, programmable, and smart devices.” Building and Environment 203:108104. ThingsBoard. n.d. “ThingsBoard Open-source IoT Platform.” Accessed June 8, 2025. https://thingsboard. io/. Trenbath, Kim, Ryan Meyer, Korbaga Woldekidan, Kristi Maisha, and Morgan Harris. 2022. Commercial Building Sensors and Controls Systems-Barriers, Drivers, and Costs. Technical report. National Renewable Energy Laboratory (NREL), Golden, CO (United States). U.S. Department of Energy, Building Technologies Office. 2024. “Building Technologies Office Peer Review Program 2024 [Conference program].” U.S. Department of Energy. Accessed June 8, 2025. https://www.energy.gov/sites/default/files/2024-11/BTOPeerReview2024_Program.pdf. U.S. Energy Information Administration. 2021. “Commercial Buildings Energy Consumption Survey (CBECS).” Accessed June 8, 2025. https://www.eia.gov/consumption/commercial/data/2018/. Wang, Wei, Junqi Wang, Jiayu Chen, Gongsheng Huang, and Xiaoyang Guo. 2018. “Multi-zone outdoor air coordination through Wi-Fi probe-based occupancy sensing.” Energy and Buildings 159:495–507. Wang, Zhe, and Tianzhen Hong. 2020. “Reinforcement learning for building controls: The opportunities and challenges.” Applied Energy 269:115036. WebCTRL®. n.d. “WebCTRL® Building Automation System.” Accessed Sep. 1, 2025. https : / / www . automatedlogic.com/en/products/webctrl-building-automation-system/. World Wide Web Consortium (W3C). n.d. “REST.” World Wide Web Consortium. Accessed June 8, 2025. https://www.w3.org/2001/sw/wiki/REST. Wu, I-Chen, and Chi-Chang Liu. 2019. “A visual and persuasive energy conservation system based on BIM and IoT technology.” Sensors 20 (1): 139. Ye, Yunyang, Wangda Zuo, and Gang Wang. 2019. “A comprehensive review of energy-related data for US commercial buildings.” Energy and Buildings 186:126–137. 37 Z-Wave. n.d. “The Smart Home is Powered by Z-Wave.” Z-Wave Alliance. Accessed June 8, 2025. https: //z-wavealliance.org/. Zigbee. n.d. “Building the Foundation and Future of the IoT.” Connectivity Standards Alliance. Accessed June 8, 2025. https://csa-iot.org/. 38 APPENDIX A. SUPPLEMENTARY FIGURES AND TABLES A-1 APPENDIX A. SUPPLEMENTARY FIGURES AND TABLES Figure A-1. Concept diagram of the proposed generalizable web user interface. Figure A-2. Schematic diagram of the proposed deployment workflow. A-2 Figure A-3. Thermostat controller in BESTEST Air (top) and BESTEST Hydronic (bottom) test cases. PI: proportional-integral controller; Hys: hysteresis controller. Adapted from (Blum and Helsen 2023). Table A-1. Building unit configurations for the simulation-based deployment test. BESTEST Air BESTEST Hydronic Campus BOPTEST Building BESTEST-Air BESTEST-Hydronic System bestest_air bestest_hydronic Local time zone UTC Building type Small office Operator Operator A Control points Supply air temperature set point Supply set point of the heater (parameter name, (SupplyAirSetpoint, 74) (SupplyHeaterSetpoint, 74) default value) Supply fan speed (SupplyFanSpeed, 0.9) Control signal to control pump stage (ControlStagePump, 1) Zone temperature cooling set point Zone temperature cooling set point (ZoneAirCoolingSetpoint, 75) (ZoneOperativeCoolingSetpoint, 75) Zone temperature heating set point Zone temperature heating set point (ZoneAirHeatingSetpoint, 70) (ZoneOperativeHeatingSetpoint, 70) Unoccupied zone temperature cooling set point (UnoccupiedCoolingSetPoint, 84) Unoccupied zone temperature heating set point (UnoccupiedHeatingSetPoint, 62) Operating schedule Monday–Friday: 8 a.m. – 5 p.m. Saturday–Sunday: Always off A-3 Table A-2. Configuration settings and data points for the simulation-based deployment test. BESTEST Air BESTEST Hydronic Application http://localhost:5000 programming interface (API) endpoint Control step 5 minutes (300) Time period Peak heat day (peak_heat_day) Energy pricing Highly dynamic electricity price (highly_dynamic) Control points Supply air temperature set point Supply set point of the heater (parameter name, (fcu_oveTSup_u, SupplyAirSetpoint) (oveTSetSup_u, SupplyHeaterSetpoint) corresponding point) Supply fan speed (fcu_oveFan_u, Control signal to control pump stage SupplyFanSpeed) (ovePum_u, ControlStagePump) Zone temperature cooling set point Zone temperature cooling set point (con_oveTSetCoo_u, (oveTSetCoo_u, ZoneAirCoolingSetpoint) ZoneOperativeCoolingSetpoint) Zone temperature heating set point Zone temperature heating set point (con_oveTSetHea_u, (oveTSetHea_u, ZoneAirHeatingSetpoint) ZoneOperativeHeatingSetpoint) Sensor points Supply air mass flow rate Zone CO2 concentration (parameter name) (fcu_reaFloSup_y) (reaCO2RooAir_y) Zone air CO2 concentration Operative zone temperature (reaTRoo_y) (zon_reaCO2RooAir_y) Zone air temperature (zon_reaTRooAir_y) Pump power consumption (reaPPum_y) Supply fan power consumption Heating power consumption (reaQHea_y) (fcu_reaPFan_y) Cooling power consumption (fcu_reaPCoo_y) Heating power consumption (fcu_reaPHea_y) Table A-3. Building unit configurations in the field demonstration. Parameter Value Campus Case study campus Building Studied building System Room 125 Local time zone US/Eastern Building type Medium office Operator Building manager A Control points (parameter name, Zone temperature cooling set point (ZoneAirCoolingSetpoint, 74) default value) Zone temperature heating set point (ZoneAirHeatingSetpoint, 70) Heating, ventilation, and air conditioning (HVAC) operating mode (HVACMode, “Auto”) Unoccupied zone temperature cooling set point (UnoccupiedCoolingSetPoint, 85) Unoccupied zone temperature heating set point (UnoccupiedHeatingSetPoint, 60) Operating schedule, setback control Monday–Sunday: 5 a.m.–6 p.m. Operating schedule, constant set point Monday–Sunday: Always on A-4 Figure A-4. VOLTTRON agent configuration in the prepared YAML Ain’t Markup Language (YAML) file for the simulation-based deployment test. A-5 Table A-4. VOLTTRON configuration settings and data points for ecobee and Modbus driver agents in the field demonstration. ecobee driver agent Modbus driver agent API endpoint http://localhost:5005 http://localhost:5008 Heartbeat period 1 minute (60) Control points Zone temperature cooling set point (parameter name, (desiredCool, ZoneAirCoolingSetpoint) — corresponding point) Zone temperature heating set point (desiredHeat, ZoneAirHeatingSetpoint) HVAC operating mode (hvacMode, HVACMode) Sensor points Zone air temperature (zoneTemperature) Electric power of the equipment (power) (parameter name) Electric voltage of the equipment (voltage) Electric current of the equipment (current) Power factor of the equipment (power_factor) A-6 Figure A-5. Configuration creation and conversion process in the field demonstration. A-7 Table A-5. Detailed information of open-source code and libraries. Category Repository Description Ref. / Library VOLTTRON Agent-based middleware platform that facilitates (Lutes, Middleware communication between distributed data sources and Allwardt, building automation systems through standardized Niddodi, messaging protocols. et al. 2025) Python Flask Lightweight web framework enabling rapid (Pallets, development of RESTful APIs and web services. n.d.) PNNL AEMS Autonomous energy management software (AEMS) (Lutes, from Pacific Northwest National Laboratory (PNNL) on Allwardt, Web application VOLTTRON, providing a web-app skeleton with Bleeker, foundational components for building control interfaces. et al. 2025) React JavaScript library for component-based UIs with (Meta dynamic state management and content rendering. Platforms, Inc., n.d.) Prisma Database toolkit that generates type-safe client libraries (Prisma and manages schema migrations for relational Data, Inc., databases. n.d.) Blueprint.js TypeScript user interface (UI) toolkit with prebuilt (Palantir interface elements (e.g., charts, navigation) for Technologies, consistent design. n.d.) InfluxDB Time-series database optimized for high-frequency (InfluxData, Database sensor data storage and real-time queries. n.d.-a) PostgreSQL Relational database with advanced SQL features for (PostgreSQL, structured data storage. n.d.) Configuration RDFLib Python library for parsing, querying, and manipulating (RDFLib converter Resource Description Framework (RDF) data. Team, n.d.) PyYAML Python parser for reading/writing YAML with support (PyYAML, for complex structures. n.d.) Simulation-based BOPTEST Control testing framework with standardized virtual (Blum building control building models to evaluate control algorithms and et al. 2021) emulator energy optimization strategies. A-8 (a) (b) Figure A-6. Deployed web application for monitoring Room 112 (a) and Room 119 (b) with setback control (left side) and constant set point (right side) in the studied building. A-9