A Network-Based Thermal Control Strategy for Improving the Operational Sustainability of School Annex Buildings

1.1. Research Purpose

This study focuses on the crucial function of an optimized supply air control system in addressing the environmental demands of both existing and prospective school annex buildings. The core goal extends beyond simple energy efficiency to incorporate strategies for establishing and maintaining acceptable standards of indoor thermal comfort. The approach is methodological, concentrating on formulating control paradigms that achieve a simultaneous equilibrium between energy savings and user comfort. A central tenet of this research is maximizing the operational use of the school annex buildings, thereby lessening the requirement for constructing new facilities. Specifically, the investigation evaluates the performance of the suggested control system under shifting operational demands, assessing its effectiveness across various temporal and spatial conditions that involve fluctuating heating and cooling loads. Special emphasis is placed on situations that promote building flexibility, such as the efficient use of the service area, to improve resource administration and operational agility.

1.2. Research Objectives and Methods

The study is structured around several objectives aimed at enhancing the performance of existing school annex buildings:

• 1) To numerically determine the energy usage and evaluate the attainable indoor thermal comfort when a retail and a food service are integrated into existing structures, utilizing computational simulation models.
• 2) To employ simulation tools to analyze the correlation between energy consumption and thermal comfort across a spectrum of diverse operational parameters.
• 3) To examine and propose solutions for improving space utilization within these areas, explicitly supporting sustainable school management.

Following a detailed presentation of the methodology and the analysis of the simulation outcomes, this paper will feature a critical examination of its limitations and strengths. The concluding section will outline prospective avenues for future inquiry, highlighting the necessity of expanding the data collection to bolster the statistical reliability and overall robustness of the findings.

2.1. HVAC Control

Continuous studies have been dedicated to optimizing indoor thermal conditions to enhance user mental and physical comfort, simultaneously aiming to boost productivity and workability within building spaces. Given that the remodeling or reconstruction of spaces consumes significant time and material resources, unlike the manufacturing of smaller products, the efficient operation of indoor space is a critical factor throughout the building's lifecycle. Optimizing these thermal conditions is particularly important for increasing user work efficiency, leading to numerous studies focused on improving Heating, Ventilating, and Air Conditioning (HVAC) system performance. Various software solutions have been integrated into existing physical HVAC devices to effectively manage the dynamic indoor thermal quality, which shifts constantly based on user psychological and physical states as well as external climatic conditions [1~3]. Historically, the Proportional-Integral-Derivative (PID) technique provided control before the advent of modern computer technology. Subsequently, the Fuzzy Inference System (FIS) offered a methodology to effectively address the limitations inherent in deterministic control problems. The FIS was often preferred due to its operational clarity, implementing deterministic algorithms with an understandable engineering approach [4,5]. This clarity is vital because internal conditions may be manually adjusted by users based on psychological or emotional states, overriding mechanically calculated optimal settings. The FIS remains relevant across various fields through continuous algorithmic upgrades, performing efficiently in situations involving ambiguous interpretation at specific boundary conditions [6,7]. Specifically, simulation results indicate that FIS models, when supplemented with additional control methodologies and without replacing existing control facilities, are quite effective for indoor thermal control in small or older buildings [8~10].

Another approach for effective HVAC system control involves programming a schedule into the system based on anticipated external weather or occupant schedules. This method requires the collection of long-term data to ensure statistical validity. This programmable method has been favored, particularly in facilities like schools and business buildings, where occupant schedules can be predicted with relatively high probability [11,12]. Given the difficulty of finding valid regression models in traditional ways for such large datasets, many approaches utilize Artificial Neural Network (ANN) models in this phase. By collecting large amounts of data across diverse building types, the ANN can be employed to analyze the correlation between building size, room count, envelope type, and weather conditions, enabling the establishment of effective construction or operation plans. Furthermore, statistically improved data derived from various ANN models can facilitate the development of highly effective system control that responds adaptively to unexpected situations [13,14]. Research frequently focuses on modifying and supplementing the internal structure of the ANN model to optimize its application to specific architectural models. These structures are often refined by adjusting variables and parameters within optimization algorithms, contingent on available hardware resources such as processing speed and computational accuracy. A major challenge, however, is that large-scale data are essential for the effective control of power-generating mechanical facilities, but in many countries, such data are highly confidential or restricted for commercial use by third parties. To address this constraint, continuous development of various inner structures and parametric functions within the ANN is being pursued to increase statistical reliability even with limited datasets [15,16]. Confirming the precise control performance of both major and minor mechanical components necessitates a stringent time interval for both simulation and experimental investigations; consequently, numerous studies have achieved high precision, operating at the level of several seconds [16,17]. Control strategies designed to mitigate overshooting during system activation and deactivation cycles, triggered by the temperature set-point, have been validated as highly effective for enhancing energy performance. To bolster the statistical validity when training models on large historical datasets, commercial facilities are predominantly utilized to ensure data volume and reliability [18,19]. Despite significant research efforts concentrating on improving the operational efficiency of the mechanical and electrical components essential for precise control, it is posited that residual opportunities or inherent weaknesses for further energy conservation within the overall system operation still exist.

2.2. School Annex Buildings

The primary characteristics of main school buildings fundamentally dictate the design and operational needs of their annexes, particularly concerning energy consumption, operation, and architectural planning. Main school buildings typically exhibit a high, long-duration base load during the academic year. Energy use is characterized by substantial demand for HVAC (heating being dominant in temperate climates), driven by strict minimum fresh air ventilation standards mandated by codes per student density. Electrical load includes extensive task lighting and significant power for IT infrastructure [20]. The operational schedule is generally predictable and non-stop during school hours, leading to predictable, though high, daily energy profiles. Operational complexity is high, centered around safety, security, and predictable scheduling. Key operational tasks include managing daily high-volume student flow, maintaining strict access control, and providing high-density food service. Operations are generally centralized, often using a Building Management System (BMS) to control the entire facility from a single point, ensuring consistent temperature and air quality across diverse zones [21]. Main school buildings are characterized by a highly specialized, departmentalized floor plan. The architecture must accommodate large, fixed spaces (gymnasiums, cafeterias, auditoriums) alongside numerous small, repetitive spaces (classrooms). Circulation paths must handle peak-hour traffic efficiently and safely. Service areas, such as the main kitchen, boiler rooms, and administrative offices, are often large and centrally located to support the entire campus [22,23]. School annex buildings, whether newly constructed or repurposed, possess characteristics that are often a derivative or intensification of the main building’s needs, particularly focusing on flexibility and specialized use. School annex buildings, whether new construction or repurposed existing structures, possess distinct architectural, operational, and energy characteristics driven by their educational and often mixed-use functions [22,23]. Energy consumption is notably impacted by their highly discontinuous and specialized occupancy patterns, which often include evening or weekend programs distinct from the main school schedule, leading to significant HVAC loads, particularly for heating due to large internal volumes and mandatory fresh air ventilation requirements. Architecturally, the floor plan is typically defined by a need for adaptable, multi-functional spaces that can rapidly transition from classrooms to temporary retail, food service, or community facilities, often requiring greater structural flexibility and a robust, accessible service area to manage logistics, utilities, and high-density, peak-hour usage. Operationally, the primary challenge lies in achieving a dynamic balance between rigorous energy efficiency and consistent indoor thermal and air quality standards, necessitating advanced, localized supply air control strategies and high-performance building envelopes to manage transient internal gains and reduce overall operational complexity and cost.

3.1. Building Model

The U.S. Energy Information Administration’s Commercial Building Energy Consumption Survey (CBECS) has been provided as a major source for research across several sectors by documenting the energy consumption patterns of public, commercial, and residential buildings. This comprehensive report specifically quantified the Energy Use Intensity (EUI) for fourteen primary building categories. However, while providing detailed data on these established types, the survey neglects any corresponding information regarding annex buildings in several types of large-scale schools [24,25]. Therefore, to predict the energy consumption levels of the spaces, the EUIs of similar building types are properly utilized such as Food Sales, Service, and Others [24,25].

The school annex building is typically composed of office, commercial, dining hall, and restrooms. It is common to configure, by using CBECS report, retails, food sales, and services. Based on the size of the average value of service areas in three school annex buildings in New York, USA, Fig. 1. shows a newly planned example for a school annex building that consists of office, food service, and restroom. Table 1. shows the geometry information of walls, roofs, doors, and windows constituting buildings in the service area. For a simulation input for outdoor conditions, the Central Park Area in New York from the EnrgyPlus V9.5 weather data provided by the US Department of Energy was utilized. The time period was selected from 00:00 on December 24th to 24:00 on December 30th. This period requires high heating and low cooling loads in the middle of winter.

Fig. 1.

Floor plan of a school annex building

Table 1.

Geometry information of a designed temporary building

3.2. Thermal Control Method

The energy use of the model is calculated by formulating a straightforward energy balance based on thermal dynamics. This involves summing the heat input from the HVAC supply air system and subtracting the heat transfer across the envelopes of the school annex buildings. This energy exchange as the temporal change in the target indoor temperature can be described by an equation as follows [26].

where, h_out and h_in are the heat transfer coefficients (W/m·K), k is the transmission coefficient (W/m·K), A is the area (m²), D is the depth of envelope (m).

The Predicted Mean Vote (PMV) index, standardized by EN ISO 7730 and introduced by Dr. P. O. Fanger, serves as the primary metric for assessing occupant thermal comfort [27]. The simulation module utilizes six distinct variables as inputs to determine the PMV level: Dry bulb temperature; Relative humidity; Indoor air speed; Mean radiant temperature; Metabolic rate; Clothing insulation. In order to time and cost effective simulation works, the followings were assumed to reduce the calculation complexity inside: The indoor air speed is fixed at 0.1 m/s. The mean radiant temperature is approximated as the dry bulb temperature, though with a one-hour time delay. The metabolic rate is held constant at 1.2 Met (normal working activity), and the clothing insulation is set to 1.2 clo (normal clothing)[27,28].

where, M is the metabolic rate and L is the thermal load.

By combining an adaptive algorithm with the thermostat control, the possibility of improving the thermal comfort level and reducing energy use is examined. The basic operation and flow of this adaptive algorithm are shown in Fig. 2.

Fig. 2.

Adaptive control algorithm for heating air supply

For example, after the calculation of indoor temperature and PMV value under given simulation conditions, in the case of heating control, the change in the PMV value is confirmed when the indoor temperature is higher than the previous calculated result. If the PMV value is within the setting comfort range (-1<α<1) , the signal value in the previous simulation step is reduced by 10%. In this case, if the PMV value is still in a comfortable status within the range, the signal value is additionally reduced by 5% point and sent to the supply air system. If the PMV value is calculated over +1 in which people inside feel somewhat hot, the signal lowered by 10% is sent to the supply air system again to complete the adaptive process. This flow of signal control is composed of one module to operate independently for heating and cooling, respectively.

Then, the result of the adaptive algorithm is trained by the ANN for a network algorithm. This network is specifically structured as a multilayer perceptron (MLP), a type of fully connected network renowned to deal with complex non-linear mapping problems [29,30]. The typical MLP utilizes a three-layer design of Input, Hidden, and Output Layers for mapping the data. The fundamental operation involves each neuron calculating a crucial intermediate value, (n_c), defined as the sum of its inputs (x₁, ⋯, x_k) multiplied by their weights (w_ai), plus an added bias (θ_b). This value is then passed through a non-linear activation function (g_d) to produce the layer’s output [29,30]. The training of this network model was conducted using the scaled conjugate gradient algorithm over 1,000 iterations, with a setting of one epoch per iteration. The model’s performance was statistically significant, achieving a high R² value for both controlled variables: 0.99532 for mass flow rate and 0.99147 for supply air temperature. The adaptive process and the ANN algorithm for an entire simulation model were combined in the Matlab and Simulink applications as described in Fig. 3.

Fig. 3.

Simulation block model