Investigation of Building Profiles for the Energy Simulation of a Factory Building: A Case Study in South Korea

Abstract

In the Republic of Korea, the 2030 Nationally Determined Contributions aim for carbon neutrality by 2050, with the building sector targeting a 32.8% reduction in carbon emissions by 2030 compared with the 2018 baseline. To achieve these goals, significant efforts are underway to improve the energy efficiency of buildings. Building energy simulation is a standard method for evaluating energy performance as it assesses the current performance and predicts the potential contributions of energy retrofitting initiatives. However, industrial factories often lack specific energy simulation profiles, posing a challenge for accurate energy performance assessment. This case study aims to bridge this gap by investigating a detailed building profile for factory building based on the extended operational data and experimental measurements within a live factory setting. Energy simulations employing these factory-specific profiles yielded R² values (coefficient of determination) of 98.2% and 94.1% for cooling and heating energy accuracy, respectively, when compared with the actual monthly consumption data. Additionally, simulations with these profiles demonstrated a 2.81% improvement in R² accuracy compared to those using conventional office building profiles, particularly enhancing the precision during the cooling season. These findings highlight the effectiveness of customized profiles in building energy simulations, ensuring more precise and reliable energy efficiency assessments.

1. Introduction

South Korea aims to reduce carbon emissions in the building sector by 32.8% from 2018 levels by 2030 [1]. To achieve this goal, considerable efforts are underway including energy remodeling and the retrofitting of existing buildings [2,3,4].

Industrial factory buildings are characterized by their significant size and high occupancy levels in addition to the substantial energy demands of their manufacturing processes. However, unlike general commercial buildings, research on energy-efficient factory buildings is limited owing to the confidential nature of pertinent information safeguarded for corporate security reasons [5,6,7,8,9]. Despite these challenges, there is an emerging commitment to enhancing energy efficiency within these buildings [10] that is driven by the impending regulatory sanctions on greenhouse gas emissions and the critical need of environmental, social, and corporate governance (ESG) management for corporate sustainability [11,12,13].

Katunsky et al. [14] attempted to devise a practical approach for conducting energy simulations in industrial buildings, aiming to identify energy conservation strategies. This comprehensive study involved an analysis of various parameters including energy consumption, surface and ambient temperatures, along with the overall energy utilization of the buildings. The resultant data facilitated the estimation of thermal energy demands, enabling the execution of dynamic simulations to assess the geometrical thermal behavior of factory buildings. Although the above study established foundational methodologies for simulating energy efficiency in industrial factory buildings, it primarily focused on assessing internal heat gains, highlighting a research void in aspects such as airtightness, indoor temperature regulation, and operational hours, with the study confined to the heating season.

In a previous study [15], a 3D model of a garment factory was constructed to analyze the thermal performance of the building envelope. The primary objective of this study is to explore the impact of various building envelope configurations on indoor thermal conditions through simulation. However, this study did not thoroughly examine other relevant factors for building simulation, relying instead on default input values.

Weeber et al. [16] conducted an energy simulation study to evaluate the energy efficiency measures in factory buildings. They developed a comprehensive assessment framework involving both energy savings and non-energy benefits such as improved thermal comfort, reduced operating costs, and decreased CO₂ emissions. Their work offers valuable insights into the framework and evaluation process required for assessing the energy efficiency of a factory. Nevertheless, it acknowledges the absence of detailed profiles for factory buildings that could enhance the precision of energy simulations.

Furthermore, efforts have been made to combine building energy modeling with manufacturing process simulations [17,18,19] to effectively optimize energy use. Duflou et al. [18] proposed coupling building energy modeling and manufacturing process simulation to optimize a manufacturing process line that considers the building energy. Garwood et al. [19] reviewed various modeling approaches and simulation tools, underscoring the challenges associated with employing energy modeling in manufacturing settings. Although these efforts on combining manufacturing process and building energy modeling have reached a “proof of concept” stage, they highlight the need for further research to enable practical applications of building energy modeling.

In summary, while previous studies have employed simulations incorporating various factors to enhance the energy performance of industrial buildings, no research has specifically examined the building characteristics that directly affect energy performance. This gap is largely attributed to security restrictions, which severely limit the direct inspections of factory buildings [8].

Consequently, the present case study aims to investigate a detailed factory building profile suitable for practical energy simulation applications. This study provides rare and valuable information about the building profiles required for building energy simulation because it is a long-term, real-world investigation of an industrial factory building. This profile was derived from a year-long investigation of operational data and field measurements within a factory building in the Republic of Korea. The items in the building profile were investigated based on inputs from ECO2: the Korean government’s building energy performance evaluation program based on ISO 52016 [20]. This was to enable the results of this study to be directly applied to simulations for the energy efficiency of Korean industrial factory buildings.

After establishing the industrial building profile, the energy performance of the factory building was assessed and contrasted with actual energy usage data to validate the efficacy of these profiles. Additionally, the present study examined the distinct characteristics of the factory building profile by comparing it with a standard office building profile, thereby contributing to a more nuanced understanding of energy simulation requirements for industrial settings.

The structure of this paper is organized as follows. Section 2 describes the factory building in South Korea considered for this case study. The investigation of building profiles for the factory building through measured field data and collected operation data is presented with each result investigated in Section 3. In Section 4, the building profiles of the investigated factories are first consolidated and characterized in comparison to typical office buildings, and then the results of the energy simulations performed in ECO2 based on the investigated factory building profiles are presented and discussed. The main conclusions of this study are summarized in Section 5.

2. Target Factory Building

The target industrial building was a manufacturing factory in the Republic of Korea. As depicted in Figure 1, the facility features a rectangular structure with a pitched roof constructed with a steel framework and dimensions spanning 300 m (L) $\times$ 260 m (W) $\times$ 13.5 m (H). This single-story building has a total floor area (A) of 78,000 m², equipped with 28 gates for the ingress and egress of automotive parts and materials, and it is complemented by windows along its ground-level and top perimeter. The interior space is almost a single zone, consisting of a subzone for the temporary storage of materials on the perimeter and a main zone in the center (Figure 2). The main zone is a completely open plan space without any walls. Photographs of the actual building could not be used owing to the security policy of the company operating the factory building.

Figure 3 illustrates the annual weather data for the region where the building is located. The temperatures ranged from approximately −18 °C to 33.8 °C, and the absolute humidity ranged from 0.76 to 34.06 g/kg_a. Thus, it is a seasonal climate with considerably large fluctuations in both temperature and humidity.

The heating and cooling of the entire zone are provided by 22 air handling units (AHUs), which utilize chilled water and steam supplied from a central plant. The total cooling and heating capacities of the AHUs are 2414 kW and 5212 kW, respectively. The locations of the AHUs are illustrated in Figure 2.

3. Building Profiles of the Factory Building

For energy simulation of the factory building, the building profiles included variables such as the operation schedule, internal heat gains, hot water demand, air changes per hour, and room air temperatures, which are essential for most energy simulation software. As detailed in Section 3.1, the profiles for the air change rates and zone air temperatures were derived from experimental data. The profiles for the operation schedule, internal heat gain, and hot water demand compiled from the building operation data are presented in Section 3.2.

3.1. Field Data-Based Building Profiles

3.1.1. Air Change Rate

The air change rate of a building typically involves a blower door test to assess the building performance [21]. However, the vast volume of industrial factories, such as the one under study with 1,053,000 m³, complicates the creation of a significant pressure difference when using fans. Instead, the air change rates were measured by employing the tracer gas decay method [22] using the carbon dioxide (CO₂) generated by the occupants [23]. In particular, the CO₂ concentration was measured between 23 February 2022 and 21 March 2022 for the heating period and 23 July 2022 and 21 September 2022 for the cooling period.

• Measurement setup

CO₂ concentrations were monitored using a HOBO MX 1102A sensor (Onset, MA, USA) operating based on the principle of non-dispersive infrared (NDIR) absorption. This instrument can accurately measure concentrations ranging from 0 to 5000 ppm, with a precision of ±50 ppm. The measurements were recorded at 5 min intervals to balance the responsiveness of gas concentration readings with the battery life of the instrument.

The distribution of the sensor points is projected in Figure 2, with five allocated for the heating season measurements and two for the cooling season. The sensors were installed in the facility manager’s walkway, out of sight of the workers. This ensured that the sensors remained safe during long-term measurements; further, the measurement points were selected such that an even distribution among external and internal zones was realized.

The reduction in sensor points during the cooling period was a precaution to protect the measurement equipment in the operational factory setting. Despite the reduced number of sensors, the measurement accuracy was not compromised because of the effective mixing of indoor air, facilitating the successful application of the tracer gas decay method. The indoor air mixing level can be determined from the results measured during the heating season using the coefficient of variation (CV) values in Equations (1)–(3) [24]. The uniformity of CO₂ concentrations across five measurement locations during the heating season was confirmed by an average CV of 7.7%, demonstrating consistent CO₂ levels across different points [25].

$$\overline{C_{RA}\left(\tau \right)}={\displaystyle \frac{{\sum }_{p=1}^n{C}_p\left(\tau \right)}{n}},$$

(1)

$$\mathsf{\delta }{C}_p\left(\tau \right)=\sqrt{{\displaystyle \frac{{\sum }_{p=1}^n{(C_p(\tau )-\overline{C_{RA}\left(\tau \right)})}^2}{n}}}$$

(2)

$$CV={\displaystyle \frac{\mathsf{\delta }{C}_p\left(\tau \right)}{\overline{C_{RA}\left(\tau \right)}}}.$$

(3)

2.

Analysis method

In total, 7483 and 17,450 data points were collected for the heating and cooling seasons, respectively. Given the time-series nature of the CO₂ concentration data, preprocessing steps were performed to minimize noise. MATLAB was utilized for all data processing and analysis tasks. The preprocessing phase included the application of a Hampel filter to eliminate spark-type outliers, with a window size of 30 min (six samples) and a three-sigma threshold for outlier detection [26]. Following the removal of spark-type outliers, a moving-average filter was employed to further smoothen the data series by setting a 15 min window (three samples) to smoothen the data fluctuations rather than excluding the outliers.

After preprocessing, the analysis equations of the tracer gas decay method [22,24,27] were applied to calculate the air change rate per hour (ACH) utilizing the mass conservation equation in Equation (4). Upon establishing a sufficient concentration of tracer gas within the zone, G = 0, the equation was simplified, as expressed in Equation (5). The external air CO₂ concentration (COA) was acquired from the Korea Meteorological Administration [28], ensuring the accuracy of environmental conditions in the calculations.

$$V\left({\displaystyle \frac{d{C}_{RA}}{dt}}\right)=G+Q(C_{OA}-C_{RA}),$$

(4)

$$V\left({\displaystyle \frac{d{C}_{RA}}{dt}}\right)+Q\left(C_{RA}-C_{OA}\right)=0.$$

(5)

Equation (5) can be integrated into Equation (6), and the two-point decay method can be established using Equation (7). This equation represents the slope of $\ln (C_{RA}-C_{OA})$ over time, which is expected to exhibit a linear relationship for accurate evaluation of the ACH. To achieve this accuracy, we employed the segmented linear regression method proposed in previous research [23]. The criterion of the determination coefficient R² was set to >0.99, and the data points in which the indoor–outdoor CO₂ concentration differential decreased below the equipment uncertainty were excluded from the analysis, as depicted in Figure 4.

$$\left(C_{RA,t}-C_{OA}\right)=(C_{RA,0}-C_{OA}){·e}^{-{\displaystyle \frac{Q}{V}}t}=(C_{RA,0}-C_{OA})·e^{-ACH·t},$$

(6)

$$ACH={\displaystyle \frac{\ln \left(\frac{C_{RA,t}-C_{OA}}{C_{RA,0}-C_{OA}}\right)}{∆t}}={\displaystyle \frac{\ln \left(C_{RA,t}-C_{OA}\right)-\ln (C_{RA,0}-C_{OA})}{∆t}}.$$

(7)

3.

Air changes per hour

The analysis of the ACH, as shown in Figure 5, utilized 12,961 and 7200 data points for the cooling and heating seasons, respectively. This study aimed to establish a representative ACH value for the building, and it was calculated as an average from the individual ACH values derived. To calculate this average, we excluded outliers beyond the upper and lower 5% confidence intervals. The resultant ACH for the factory was determined to be 0.3672.

3.1.2. Zone Air Temperature

The zone air temperature is one of the most critical profiles in building energy simulations, particularly for factory buildings with large open spaces. Therefore, the simulation zones need to be determined based on the temperature distribution characteristics as different areas within a single, undivided space may exhibit distinct thermal behaviors depending on air conditioning and air intake locations. To accurately simulate the zone of the large factory building, we conducted long-term vertical and horizontal temperature distribution measurements, providing essential data on the thermal characteristics of these zones.

• Vertical zone air temperature distribution

Vertical zone air temperatures were measured during two periods: from 21 August 2022 to 18 September 2022 for the cooling season, and from 20 November 2022 to 4 December 2022 for the heating season. K-type thermocouples were secured to a metal wire, which, in turn, was affixed to the roof truss centrally located near Air Handling Unit (AHU)-11, as indicated in Figure 2. Four sensors were vertically positioned at 1.5 m intervals, as illustrated in Figure 6. Data collection occurred every minute using a TESTO 176 T4 logger (TESTO, Titisee-Neustadt, Germany), which can measure temperatures ranging from −200 °C to 1000 °C with an accuracy of ±0.5 °C.

An analysis of the vertical temperature differences over time for both the cooling and heating seasons is shown in Figure 7. Additionally, the vertical temperature distributions for representative days of each season are plotted in Figure 8. During the cooling season, the temperature differences started to increase around 10 am and decreased from 4 pm onward (Figure 7a). Typically, industrial factory buildings feature skylights or top-side windows, which allow for greater solar exposure at higher elevations. Despite the expectations from natural convection laws that predict hot air rises, we assumed that the air at the top would be cooler. However, it can be observed from the measurement results that the temperature increased as the height decreased (Figure 8a). This anomaly was attributed to the operational dynamics within the factory, where the internal heat sources were predominantly located at lower levels, the air conditioning ducts were positioned at mid-height, and natural ventilation occurred through skylight openings during the cooling season.

In the heating season, the temperature differentials exhibited no significant trends, maintaining a consistent range between 1.0 °C and 2.0 °C. Figure 8b shows that natural stratification occurred when the AHU was deactivated, with the mid-height areas warming upon AHU activation owing to the positioning of the supply air ducts. Additionally, the temperatures at the top levels increased once again when the solar altitude was sufficiently high such that the top side received more insolation.

2.

Horizontal-zone air temperature distribution

The horizontal-zone air temperature data were collected from the sensor outputs of the AHUs located in the return air ducts. The temperature sensor was a T-type thermocouple, whose measurable range is between −20 °C to 100 °C with an accuracy of $\pm$0.5 °C. Eleven AHUs (3, 4, 6, 9, 11, 12, 13, 16, 17, 18, and 19, as depicted in Figure 2) were selected for the analysis of horizontal air temperature distribution, contributing to the definition of the building profile. The exclusion of the remaining 11 AHUs was because of their placement in areas lacking air conditioning. The temperature data were recorded on an hourly basis throughout the year 2022, with values aggregated into monthly averages for comparative analysis across locations and to establish representative building profiles. Figure 9 depicts these monthly temperature values across different horizontal locations using three-dimensional graphs, with uniformity assessed via standard deviation measures. The standard deviations for horizontal temperatures varied between 0.4 and 1.0, with maximum temperature disparities across the AHUs being in the range of 1.4–1.8 °C.

The findings in Section 3.1.2 suggest that a factory with evenly distributed air conditioning, as was observed in this study, may be thermodynamically treated as a single zone. Figure 10 shows a summary of the monthly zone air temperatures for the factory, indicating a trend similar to that observed in conventional buildings. Notably, a reduced temperature was recorded in April, possibly because of the deactivation of climate control systems during the transitional season, which was coupled with unexpectedly low external temperatures.

3.2. Operation Data-Based Building Profiles

3.2.1. Operation Schedule

The operation schedule was composed of two categories (both of which are tailored to the operational regulations of the company, as listed in

Table 1. The operation schedule of the factory building.

	Start Time	End Time
HVAC Schedule	6:00 am	1:00 am (+day1)
Occupant Schedule	7:00 am	0:00 am (+day1)

): heating, ventilation, and air conditioning (HVAC), and occupant activity schedules. Unlike the typical differentiation between weekdays and weekends, the factory adhered to a unified schedule across both, spanning three shifts without operational breaks on holidays.

Although the operation schedule is specific to this Korean factory and not universally applicable, it serves as a comparative benchmark and highlights the distinct operational characteristics of industrial versus non-industrial buildings. In addition, this survey information could facilitate the preparation of estimates even when the operation schedule for a factory is unavailable.

3.2.2. Internal Heat Gain

The internal heat gains within the factory building originate from three primary sources: the occupants, lighting, and manufacturing equipment. The occupant-related heat gain can be calculated using Equation (8). The heat emission per person (${\dot{Q}}_{occupant,person}$) for a light bench work at the factory was defined as 220 W/person, in accordance with the ASRHAE Handbook [29]. There were 789 workers (N) in a 78,000 m² area (A) working simultaneously. In addition, the total working time for a day was 17 h. Consequently, the calculated occupant heat gain per area was 38 Wh/(day∙m²).

$${\dot{Q}}_{occupant,area}={\displaystyle \frac{N\times {\dot{Q}}_{occupant,person}\times Workingtime}{A}}.$$

(8)

Meanwhile, the internal heat gains from the lighting and equipment were derived from the facility management data and technical specifications from the manufacturers. The total number of lighting sources and the efficiency of each source according to the type of lighting were determined to calculate the heat gains from them. In addition, the heat emissions from the equipment were calculated by the facility managers on the basis of the electricity consumption for the intermediate seasons and the number and characteristics of the manufacturing equipment. In the intermediate season, the air conditioning system was not operated; therefore, most of the electricity consumption was attributed to the lighting and manufacturing equipment. This formed the basis for estimating the internal heat gains.

However, the specifications of the manufacturing equipment were not provided during the collaborative research owing to strong security reasons. Therefore, the description was not detailed, and only the concept of the calculation was provided with the results of internal heat gain from the lighting and equipment as 414 Wh/(day∙m²). Assessing the internal heat gains from the manufacturing activities was extremely challenging and could have a significant impact on the building energy; moreover, the operating schedule can fluctuate depending on the production demand and economic cycles [30].

Therefore, the obtained internal heat gain values can be considered as an investigated reference case. They are not the focus of this study as internal heat gain is more related to the characteristics of the industrial sector than to the inherent characteristics of the factory building. To appropriately evaluate this aspect, a combination of using building modeling and manufacturing process modeling would be required, as has been noted in previous studies [11,19,21,22]. As this case study, the inherent performance of the building was focused on, and we utilized the investigated internal heat gains of 414 Wh/(day∙m²).

3.2.3. Hot Water Demand

The hot water demand of the factory building (Q_hw) was quantified at 1.47 Wh/(day·m²), and it was derived using Equation (9). This calculation was based on a reported average hot water (V_hw) usage of 4.16 L/(person·day) [31], with hot water and tap water temperatures assumed to be 45 °C and 15 °C, respectively. The calculation considered the total number of occupants (N) and the total floor area of the building (A), excluding the hot water utilized in the manufacturing processes, as described in Section 2 and Section 3.2.2.

$$Q_{hw}={\displaystyle \frac{V_{hw}N{c}_{p,w}(T_{hw}-T_{tap})}{A}}.$$

(9)

4. Results and Discussion

4.1. Comparison of the Profiles Between the Office and the Factory

Table 2. The building profiles of the office and factory for the energy simulation.

Building Profiles		Factory	Office (>30 m2)
Operation schedule	Occupant	7:00–24:00	9:00–18:00
	HVAC	6:00–1:00	7:00–18:00
ACH [h−1]		0.367	0.3
Minimum outdoor air flow rate per area [m3/(h∙m2)]		5.13	6
Required hot water per area [Wh/(day∙m2)]		1.47	30
Internal heat gain [Wh/(day∙m2)]	Occupant	38	55.8
	Equipment	414	126
Set point [°C]	Heating	19.8	20
	Cooling	25.0	26
Monthly day of use [days]	January	24	22
	February	20	19
	March	26	21
	April	25	22
	May	23	22
	June	26	20
	July	26	22
	August	19	21
	September	20	18
	October	26	21
	November	26	21
	December	26	21

presents a comparison of the final building profiles of the factory and those of the office, which are defined in the “Building Energy Efficiency Rating System Operating Regulations” of the Republic of Korea [32]. The building profile of the office defined in this regulation was developed by surveying the actual operation of many of the office buildings in Korea.

Except for the minimum outdoor airflow rate per area, all of the metrics were derived from the results in Section 3. According to Korean ventilation standards [33], the required minimum ventilation rate for factory workers is 35 m³/h, allowing for the calculation of the minimum outdoor airflow rate per area based on number of occupants and the total factory area. This is in accordance with the national standards and legislation, and it is essentially a value that changes with the occupancy density of the building. The higher values of the internal heat gain from occupants and the minimum outdoor air flow rate of office building indicate that the office had a higher occupant density than a factory. Meanwhile, the hot water demand per unit area in the factory was substantially lower than that in office settings.

Contrary to initial expectations, the ACH and temperature setpoints in the factory did not significantly deviate from the office norms. Pre-analysis assumptions suggested a higher ACH in the factory owing to its structural characteristics; however, the surface-to-volume ratio (S/V) in the factory, which was markedly lower than in typical office buildings owing to its vast open spaces, contributed to this parity, as discussed in Section 2.

The building profiles were developed based on the assumption that the statistical averages derived from the data investigated were representative. However, these profiles were subject to uncertainties and may deviate from the actual characteristics of real buildings. Furthermore, even buildings with similar usage can exhibit significantly different attributes. Despite these limitations, the establishment of building profiles and their application as representative values in building energy simulations hold significant value. They provide a standardized set of parameters for consistent evaluation, and they are independent of the need to precisely fit actual building data.

4.2. Energy Simulation Using the Investigated Building Profiles

Energy simulations were performed to validate the reliability of the developed building profiles for industrial factories by comparing the actual monthly energy consumption of the factory with the simulation outcomes obtained from applying these profiles.

4.2.1. Energy Simulation Method

The simulations utilized ECO2, which is the official building energy evaluation software endorsed by the Korean government and is based on the DIN V 18599-1 [34] and ISO 52016-1 [20] standards. In South Korea, this program is used to certify zero-energy buildings at the permitting stage of new construction.

This study employed ECO2 for the evaluation over other commercially available programs such as EnergyPlus and TRNSYS. This is because the building profile was investigated using ECO2 to assess the impact of the results of this study on the energy performance evaluation of factories in South Korea.

The objective of ECO2 is to appraise building energy performance across various parameters, including thermal load and primary energy consumption for heating, cooling, lighting, hot water, and ventilation. Within the ECO2 software, 20 distinct building profiles are available, catering to a wide range of spaces such as residential areas, offices, conference rooms, auditoriums, cafeterias, restrooms, occupied and accessory spaces, warehouses, computer rooms, kitchens, hospital and guest rooms, classrooms, lecture rooms, stores, exhibition halls, reading rooms, and sports facilities [32]. The current version of the program does not have a building profile for a factory, and one goal of this study is to address this drawback so that energy performance assessments for factories can be performed in the future.

ECO2 follows the calculation approach of ISO13790 [35], the predecessor standard to ISO 52016 [20], as shown in Figure 11. This calculation procedure follows the typical building energy simulation program, and ECO2 uses monthly results based on the quasi-steady state calculation method. Detailed energy consumption calculations were programmed according to DIN V 18599-5 [36], ISO 52016-1 [20], DIN V 18599-3 [37], DIN V 18599-8 [38], and DIN V 18599-4 [39] for the five elements, namely heating, cooling, air conditioning, domestic hot water, and lighting, respectively.

4.2.2. Performing Simulation and Results

The factory building profile (Table 2) was incorporated into the developer mode of the program for simulation purposes. The simulation assessed the energy requirements for cooling and heating based on the detailed building information provided in Section 2. The component characteristics are summarized in

Table 3. The overall heat transfer coefficient of the building components.

Component	Materials	U-Value [W/m2·K]	G-Value
External wall with insulation		0.48	–
External wall without insulation		2.40	–
Floor		3.16	–
Roof		0.50	–
Single-glazed window		6.60	0.688
Door		0.48	–

, although specific details on mechanical equipment, including the fans, pumps, gas boilers, chillers, and air conditioning units, have been omitted owing to industrial confidentiality.

The main zone of the factory was air-conditioned with air handling units consisting of gas-fired boilers and centrally supplied chillers, while some sub-zones and small break areas were equipped with electric heat pumps. According to the facility manager, information regarding the building’s envelope was similar to that of a typical factory; therefore, there was no security issue. However, equipment information could not be disclosed because many of the pieces of information related to product production could be estimated, particularly with regard to capacity.

The simulated and actual values for the monthly primary energy consumption was compared to validate the building energy simulation, as shown in Figure 12. The actual cooling and heating energy consumptions were collected from the factory energy monitoring system (FEMS). The domestic hot water energy was excluded from the comparative assessment as it was not separately metered in the FEMS.

For quantifying the accuracy of the building energy simulation, statistical performance parameters were used such as the root-mean-squared error (RMSE) and the coefficient of determination (R²), which have generally been used in the literature [40,41,42,43]. The comparison results indicated the RMSEs for the cooling and heating energy consumptions at 3.9 kWh/m² and 7.4 kWh/m², respectively. Additionally, the R² values for cooling and heating energy consumptions were recorded at 98.2% and 94.1%, respectively. Despite the certain monthly discrepancies observed (notably in March and August for heating and cooling), the simulations demonstrated R² values exceeding 90%, affirming their effectiveness in estimating the monthly energy consumption of a factory building.

4.3. Energy Simulation of the Factory Based on Office Building Profiles

As described in Section 4.2.1, an additional energy simulation was performed using the same models by substituting the factory building profiles with those of office buildings. This effort aimed to assess the impact of applying the devised building profiles on the accuracy of simulating the energy consumption of the factory buildings.

Figure 13 reveals that the monthly heating energy consumption yielded comparable results for the office and factory profiles. Nevertheless, the monthly cooling energy consumption was lower when employing the office building profile, resulting in a decrease in the average R² value by 2.81% for simulations using the office profile.

From the investigation results in this study, given that factory buildings operate over more extended periods and more days monthly than office buildings, both the heating and cooling energy consumption were expected to be higher in the factory settings. However, the factory had 3.29 times more internal heat per unit area than the office, and the water heating requirement was significantly higher in the office at 20.4 times the factory value. Therefore, the heating energy consumption of the office building was evaluated to be similar to that of the factory.

Moreover, the summer cooling energy for offices was relatively lower than that for factories because of the extended operational hours of the factories, as well as due to the internal heat gain from the equipment in the factories. The variances in internal heat gains from the occupants, ACH, and set temperatures further contributed to the differences that can observed in Figure 13. Nonetheless, this section highlights only the factors deemed to have the most substantial impact on the differences in the energy consumption figures.

5. Conclusions

This research explored the energy simulation profile for factory buildings based on empirical operational data and comprehensive measurements from a manufacturing facility in the Republic of Korea. The annual operation period of the factory was notably extended: 198.2% longer than that of an office setting. The internal heat gain attributed to the occupants in the factory constituted merely 68.1% of that in the office environments, whereas the heat gains from equipment surged to 328.6% more than that in the offices. This difference arose from the lower occupancy density in the factories and the significant heat output from the industrial machinery. Moreover, the hot water demand in the factory was significantly lower than that of the office, showing an 83.7% reduction. This reduction reflects the lower occupant density per unit area in the factory compared to the office.

The room air temperatures during both cooling and heating periods were similar to those in the office buildings, indicating that factories are managed to maintain thermal comfort for workers. This study documented consistent horizontal and vertical temperature distributions within the factory, achieving uniformity within a 2 °C range. Air tightness assessments through prolonged measurements indicated the ACH of the factory to be 22.3% higher than that observed in the office environments.

Employing the specifically formulated factory building profiles for energy simulations demonstrated their accuracy by aligning closely with the actual energy usage and exhibiting an improved R² value by 2.81% over the office building profiles, particularly during the cooling season.

As the push for carbon neutrality in the building sector intensifies, optimizing energy efficiency in industrial factory buildings is increasingly aligned with ESG management principles. Therefore, the factory building profiles developed in this study hold significant value for industry stakeholders.

Furthermore, the process of the long-term investigation of the characteristics of large factory buildings aids practitioners in many companies who are tasked with enhancing the energy efficiency in their factories. In addition, the investigation process can provide insights into methodologies for investigating the building types that currently lack information for building energy simulation, such as other types of factory buildings, transportation buildings, military buildings, special purpose buildings (resource recycling, power generation, broadcasting and telecommunications, funerals, etc.). The information thus gained would help these practitioners initiate their engineering approaches.

This case study focuses on a specific factory building, which limits its general applicability. Therefore, the application of this factory profile as a universal benchmark or the case study was constrained. Nevertheless, it holds value as it investigated the building profile for energy simulation in a factory that is characterized by its large size and high security, for which minimal information is available when compared to typical buildings.

Future research should focus on regularly updating factory building profiles by exploring a diverse range of manufacturing facilities, following the methodology outlined in this research.