HVAC systems account for 40-60% of total building energy consumption, making them the most promising area for energy-saving breakthroughs. However, traditional HVAC design often focuses on peak load conditions on the "design day," while overlooking the fact that buildings operate at partial load for the vast majority of their 8,760 annual operating hours. The core value of smart HVAC lies in using Building Management Systems (BMS) for real-time monitoring and automated control to drive HVAC systems toward optimal efficiency at every moment of operation[1]. From temperature and humidity sensors to cloud computing platforms, from PID control loops to machine learning algorithms, smart HVAC represents the deep convergence of HVAC engineering and information technology -- this is not merely a technical upgrade, but a fundamental transformation of the HVAC engineer's role.

1. Overview of BMS Building Management Systems

A Building Management System (BMS), also known as a Building Automation System (BAS), is a centralized platform integrating hardware and software for monitoring and managing mechanical and electrical equipment within buildings. The ISO 16484 series of standards formally defines it as Building Automation and Control Systems (BACS), encompassing the complete lifecycle from project specification and system implementation to performance verification[1].

The core functions of BMS can be summarized in four aspects:

  • Monitoring: Real-time acquisition of data from hundreds to thousands of sensing points within the building, including temperature, humidity, pressure, flow rates, power consumption, and equipment operating status, providing complete visibility of building operations
  • Control: Automatic adjustment of chillers, pumps, fans, valves, dampers, and other end actuators based on preset logic or algorithms to maintain specified environmental conditions
  • Optimization: Minimizing energy consumption while meeting comfort requirements through scheduling management, load forecasting, equipment rotation, and coordinated control strategies
  • Reporting: Recording historical trend data, generating energy consumption reports, and issuing anomaly alerts to support facility management team decision-making

A typical BMS employs a three-tier architecture. The bottom layer is the Field Level, containing various sensors (temperature, humidity, CO2, differential pressure, etc.) and actuators (motorized valves, variable frequency drives, damper actuators, etc.). The middle layer is the Automation Level, where Direct Digital Controllers (DDC) execute control logic -- DDCs can operate independently, maintaining basic control functions even when the upper-level network is disrupted. The top layer is the Management Level, providing graphical user interfaces (GUI), databases, trend analysis, and alarm management functions[1].

It is worth noting that BMS does not manage only HVAC systems. A complete BMS typically integrates HVAC, lighting, power, fire protection, access control, and elevator subsystems. However, in practice, HVAC systems remain the most critical target of BMS control strategies due to their highly dynamic nature and significant energy-saving potential.

2. HVAC Automatic Control Strategies

The goal of HVAC automatic control is to maximize system efficiency while maintaining indoor environmental quality. Modern BMS can execute control strategies spanning every component from chiller plants to terminal spaces.

Chiller Plant Optimization

The chiller is the single largest energy-consuming component in HVAC systems, and its control strategy has the most significant impact on overall energy savings. Chiller Sequencing automatically starts and stops chillers based on system load, preventing chillers from operating in the inefficient low-load range. Advanced strategies include Condenser Water Reset -- lowering the condenser water supply temperature when outdoor wet-bulb temperature decreases can improve chiller COP by approximately 2-3% per 1°C reduction[2]. However, this must be balanced against the increased energy consumption of cooling tower fans, requiring optimization based on overall chilled water plant efficiency (kW/RT).

AHU Control

Air Handling Unit (AHU) control strategies directly affect supply air quality and energy efficiency. Supply Air Temperature Reset is the most fundamental energy-saving measure: raising supply air temperature at partial load (e.g., from 13°C to 16°C) reduces chilled water consumption and lowers reheat energy use. Economizer control increases outdoor air intake or enables full outdoor air operation when outdoor air enthalpy is lower than return air enthalpy, achieving "free cooling." ASHRAE Guideline 36 provides standardized high-performance operating sequences for these control sequences[2].

VAV Variable Air Volume Terminal Control

Variable Air Volume (VAV) systems are the mainstream HVAC type for modern commercial buildings. Each VAV box adjusts airflow based on zone temperature demand -- when load decreases, dampers close and airflow reduces, causing the supply fan to reduce speed via the variable frequency drive. According to fan affinity laws, when airflow drops to 80%, fan power is only 51% of the original (power is proportional to the cube of speed), yielding extremely significant energy savings.

Demand-Controlled Ventilation

ASHRAE Standard 62.1 allows the use of CO2 sensors as proxy indicators for occupant density to dynamically adjust outdoor air intake[7]. In spaces with highly variable occupancy such as conference rooms and lecture halls, Demand-Controlled Ventilation (DCV) can maintain indoor air quality while avoiding the energy waste of introducing excessive outdoor air during low occupancy. Typical DCV strategies maintain CO2 concentrations between 800-1,000 ppm, saving 20-30% of outdoor air conditioning energy compared to fixed outdoor air volume designs.

Optimal Start/Stop Control

Traditional fixed schedules often start HVAC systems too early to ensure room temperature reaches the setpoint before occupied hours. BMS optimal start/stop control calculates the latest possible start time by learning building thermal mass characteristics and predicting outdoor air conditions, ensuring timely setpoint achievement while avoiding unnecessary early operation. Similarly, optimal stop control can shut down the chiller before occupied hours end, utilizing the building's thermal storage effect to maintain temperature until the end of the workday. These two strategies combined can save 10-15% of daily operating hours.

3. Communication Protocols and System Integration

The value of BMS depends on its integration capability -- whether it can connect equipment from different manufacturers, different eras, and different functions into a coordinated operating whole. Communication protocols are the critical foundation for achieving this goal.

BACnet

BACnet (Building Automation and Control Networks) is an open communication protocol defined by ASHRAE Standard 135[3] and is currently the most widely adopted building automation protocol globally. BACnet defines standardized Object Models and Services, enabling devices from different manufacturers to communicate with each other. It supports multiple network layer technologies, including BACnet/IP (Ethernet-based), BACnet MS/TP (RS-485-based), and BACnet/SC (Secure Connect, providing TLS encryption). BACnet's greatest advantage is interoperability -- building owners are not locked into a single vendor's ecosystem.

Modbus and LON

Modbus is a simple and long-established serial communication protocol still widely used at the equipment level for chillers, boilers, and variable frequency drives. Its advantages are simplicity and reliability, but it lacks BACnet's object-oriented architecture and self-describing capabilities. The LonWorks (LON) protocol once held higher market share in European markets, with its distributed architecture allowing direct peer-to-peer communication between field controllers, but it has gradually been replaced by BACnet/IP in recent years.

Integration Challenges and Trends

Although the proliferation of open protocols has significantly improved the system integration landscape, practical challenges remain: inconsistent object naming across different brands of BACnet devices, inaccessible proprietary extension points, the need for gateways for protocol conversion of legacy systems, and more. Furthermore, with the maturation of IoT technology, IT-domain communication methods such as MQTT and RESTful APIs are rapidly entering the building automation field. The rise of cloud-based BMS platforms has further broken the boundaries of traditional architectures -- edge computing handles real-time control on-site, while data analytics and machine learning are executed in the cloud, forming a hybrid architecture.

4. AI and Machine Learning Applications in HVAC Energy Savings

Traditional BMS control strategies are primarily rule-based -- for example, "enable the economizer when outdoor air temperature falls below 15°C." While effective, these rules cannot address the numerous nonlinear, time-varying, and uncertain factors in building operations. The introduction of AI and machine learning is transforming HVAC control from "reactive response" to "proactive prediction"[4].

Model Predictive Control (MPC)

Model Predictive Control (MPC) is currently the most actively researched AI HVAC control method[8]. MPC builds mathematical models of building thermal dynamics and, combined with weather forecasts, electricity price information, and occupancy schedules for the coming hours, solves for the optimal control trajectory. For example, pre-cooling the building during off-peak electricity rate periods, leveraging the thermal storage capacity of the building structure to reduce chiller operation during peak rate periods. Research shows that MPC can save 15-30% of HVAC energy compared to traditional PID control while maintaining or even improving indoor comfort[4].

Occupancy Detection and Adaptive Control

Traditional HVAC scheduling assumes fixed occupancy patterns, but actual building usage patterns are highly dynamic. By integrating multi-source data from Wi-Fi connected device counts, infrared sensors, access control systems, and even image recognition, AI algorithms can estimate the actual number of occupants in each zone in real time and dynamically adjust HVAC and ventilation volume. In the post-pandemic era where hybrid work has become the new norm, the energy-saving potential of occupancy-based HVAC control is even more significant -- unoccupied floors do not need to run at full load.

Fault Detection and Diagnostics (FDD)

ASHRAE Guideline 36 includes automated Fault Detection and Diagnostics (FDD) as standard equipment in high-performance control sequences[2]. FDD continuously analyzes sensor data and equipment operating parameters to automatically identify abnormal conditions[5]. Common HVAC faults include stuck valves, sensor drift, damper linkage detachment, and insufficient refrigerant. Machine learning methods -- particularly anomaly detection algorithms in unsupervised learning -- can build baseline models from historical normal operating data and automatically flag operating states that deviate from the baseline, without needing expert rules written for each fault mode. Research indicates that HVAC systems in buildings commonly exhibit 5-20% of "hidden energy waste" from undetected equipment faults or control sequence malfunctions[5].

Digital Twin

A Digital Twin is a high-fidelity virtual model of a building that synchronizes data with the physical building in real time. In the HVAC domain, Digital Twins integrate building energy simulation engines (such as EnergyPlus), BMS real-time data, and machine learning models, enabling: offline testing and validation of control strategies (avoiding "experiments" on physical buildings), simulation analysis of future scenarios (such as the impact of climate change on HVAC loads), and continuous energy performance baseline comparisons. Digital Twins represent the ultimate vision for smart buildings -- optimize in the virtual world, execute in the real world.

5. Development of Intelligent Buildings in Taiwan

Taiwan has promoted the "Intelligent Building Label" (IBL) certification system since 2004, administered by the Architecture and Building Research Institute of the Ministry of the Interior[6]. The IBL evaluation covers eight major indicators: information and communication, safety and disaster prevention, health and comfort, equipment energy efficiency, structured cabling, system integration, facility management, and smart innovation. The "Equipment Energy Efficiency" and "System Integration" indicators are directly related to BMS implementation.

Integration with Green Building Certification

Taiwan's EEWH Green Building Label focuses on energy-efficient design of the building's physical environment (such as envelope insulation and lighting efficiency), while the Intelligent Building Label emphasizes improving building operation and maintenance efficiency through ICT technology. The two complement each other: good building envelope design reduces the base load, while BMS smart control ensures the HVAC system maintains high efficiency during the operational phase. In recent years, policies have encouraged new public buildings to obtain both certifications.

BMS Energy-Saving Achievements

Based on domestic and international case studies, comprehensive BMS implementation and commissioning can achieve 15-30% HVAC energy savings for existing buildings. The energy savings do not come from a single technology but from the combined effect of multiple control strategies: optimal start/stop scheduling reduces unnecessary operating hours, chiller plant optimization improves partial-load efficiency, supply air temperature reset reduces chilled water consumption, demand-controlled ventilation reduces outdoor air conditioning energy, and FDD eliminates waste caused by hidden faults. It is worth noting that BMS energy savings require ongoing maintenance -- without regular system recommissioning and calibration, control quality will degrade over time, a phenomenon known as "Performance Degradation."

Challenges for Existing Buildings

The vast majority of existing buildings in Taiwan were not equipped with comprehensive BMS at the time of construction, or use outdated proprietary systems. These buildings face smart-upgrade challenges including: insufficient sensor coverage resulting in data gaps, legacy equipment not supporting open communication protocols requiring gateway installation, outdated controller firmware unable to support advanced strategies, and a shortage of qualified system integrators for commissioning. Government subsidy programs for existing building energy improvement provide partial financial support for building owners, but technical barriers and uncertainty in investment payback periods remain the primary obstacles to smart renovation. Taiwan's HVAC engineers play a critical role in this process -- requiring not only traditional mechanical and electrical engineering expertise but also the ability to bridge OT (Operational Technology) and IT (Information Technology) integration.

Conclusion

The development of smart HVAC and BMS is essentially the evolution of HVAC engineering from the static mindset of "install and forget" to the operational mindset of "continuous monitoring and dynamic optimization." When AI algorithms can predict tomorrow's HVAC loads and preemptively adjust strategies, when digital twins can verify the impact of every control parameter in the virtual world, when cloud platforms can aggregate operational data from hundreds of buildings for cross-building benchmarking -- the HVAC engineer's role evolves from equipment selector to system optimizer. Under the global pressure of net-zero carbon emissions, smart HVAC is no longer an "icing on the cake" value-add option, but a core strategy for building energy efficiency.