This article describes an experiment and field test conducted to improve the energy efficiency and effectiveness of an HVAC system of a factory building in the automotive industry in Września, Poland, using indoor air quality parameters. The factory building was built in 2014 and is divided into two sections. The northern section ( Figure 3, right) of the hall is the main assembly line area (Montage); it stretches over 475 m in length, whereas the southern section of the hall, which has a length of 375 m, is reserved for the finishing area ( Figure 3, left). In the finishing area the manufactured cars are processed, washed and inspected thoroughly before the final test drive, this area of the hall is significantly narrower than the assembly line area.

The assembly area (Montage; M01 to M16) is served by 16 HVAC systems, whereas the finishing area is served by six HVAC systems (F01 to F06). However, with regard to the production process, the HVAC in the finish area is not only used to condition the building area, but also to supply quality control and test drive, for example, roller test rig. This limits the design of the experiment; only the main assembly line area (M01–M16) can be used for experiments with. The 16 HVAC supply systems installed in the assembly hall are presented in Table 1. The HVAC systems in the finish line were outside the scope of this study, and remained unchanged throughout the experiment. Each of the 16 HVAC systems was equipped with two ventilator resps. fans on the supply side, and two fans on the exhaust side. The supply air fans of M01-M04 have a nominal electric power of 18.5 kW, while all other fans have a power of 22 kW. Exhaust air fans operate at nominal power of 18.5 kW, 22 kW resp. 30 kW, depending on the system. They can deliver air flow rates of up to 63,000 m ³/h or 75,000 m ³/h at the nominal power. In total, 64 fans in the assembly area required 676 kW of fan power on the supply air side and 740 kW of fan power on the exhaust air side.

The floor area and nominal supply air volume flow rates of dedicated HVAC systems for the complete hall are summarized in Table 2.

Based on data given, the floor related supply air volume flow rate of the HVAC system for the assembly area for this case study is calculated at 15.69 m ³/hm ². However, owing to the narrow construction of the building and higher demands on air quality as a result of the exhaust gases released, the finishing area has a much higher specific volumetric flow rate than the assembly line area. In view of the very high air quality recorded in the hall so far, the question arises as to whether it’s necessary to deliver volumetric flow rates of air of up to 22.70 m ³/hm ². These are the design state conditions with typical flow rates for assembling halls between 20 m ³/hm ² and 30 m ³/hm ² ( VDI 3802, 2014). A reduction in the volume flow rate leads to a significant reduction in the fan electric power consumption for air transport and distribution.

The operation of fans is governed by a consistent set of laws dictated by speed, power, and pressure. If the speed (RPM) of a fan is altered, it will reliably result in a corresponding change in the power required to run it at the new RPM, as well as the pressure rise it generates. The volumetric flow rate of the fans was calculated based on the relationship between the flow power of the fan and its speed ( VDI 6014, 2016). N 1 N 2 = V ̇ 1 V ̇ 2   and  P 1 P 2 = ( N 1 N 2 ) 3 where: P 1 , P 2 = Flow Power in [ W ],

V ̇ 1 , V ̇ 2 = Volumetric Flow Rate [ m 3 h ]

N 2 , N 2 = Fan Speed [RPM].

Speed control by frequency control flow power is not very different from the electrical power of the fan; therefore, these relations are often used to estimate electrical power consumption in a very simplified way ( Bureau of Energy Efficiency, 2023) and can be treated in the same way here or by ( VDI 6014, 2016). Because the air flow is proportional to the speed of the fan, slowing down the fan also means a linear reduction in the volumetric flow rate, so that a 10 % reduction in volumetric flow rate will lower the electrical power demand by approximately 30%.

The HVAC systems listed in Table 1 are shown on the right side of Figure 3, which shows the outline of the supply air ventilation zone according to the distribution of the supply air duct system in the hall. The heating demand of the building was covered by using direct gas heaters and heat recovery in the HVAC system. They do not have humidifying, dehumidifying, or cooling capabilities. Air is supplied close to the occupied zone and workplaces by displacement, such as diffusers (for low momentum stratified flow) at a height of approximately 3 m. In this way, the ventilation system generates a uniform velocity and temperature field in the complete hall. During the heating period, the HVAC system was set to establish a 20.5 °C of room air temperature under all conditions in this study, measured at two different positions at approximately 3 m height for each system. Figure 4 shows a schematic of the parts used in the HVAC system installed in the factory hall, which applies to each of the 16 HVAC systems.

As mentioned above, each ventilation system was equipped with temperature and CO ₂ sensors close to the East and West walls of the building at a height of approximately 3 m. This is very helpful, but not enough, to monitor thermal comfort and indoor air quality close to the workplace. Therefore, a network of approximately 120 sensors for air temperature and humidity, and some additional sensors for operative temperature and CO ₂ were installed. However, this was not sufficient to obtain the details of the indoor air climate of the complete hall. Thus, four zones were recognized as representative based on the production process, and worker occupancy was defined in the hall. The locations of these zones are shown in Figure 5. Of these zones, Zone 4 was selected for the reduction of the air flow rates and for this study. In this area, there are several windshield gluing machines and workers performing manual work, as well as a break area for workers. Zone 4 was supplied by the ventilation systems M05–M08.

Figure 6 shows an overview of the installations used to measure temperature, relative humidity, and CO ₂ in zone 4 as an example. The setup consists of a vertical air temperature and humidity profile within the reach of the server transmitter area with nine–ten sensors. The reason for this is to measure the temperature gradient from floor to ceiling, so that we can ensure that there are no large temperature differences that could cause unpleasant draughts. The vertical distance between the sensors was 1 m, and the uppermost sensor was fixed to the bottom of the roof frame. In addition, there were two to three “smaller” vertical temperature profiles with four to five sensors spaced 2 m apart. Operative temperature and CO ₂ sensors were installed at a height of approximately 1.50-1.60 m, close to the worker’s area.

Figure 7 shows the detailed positions of the sensors in relation to the relevant nearby facilities, such as the break room, windscreen bonding system, manipulators, and screws. Sensor S ₁ was placed in proximity to the air supply duct of the HVAC system. Five Sensors were installed on S ₄ at regular distances starting from 1.5 m height to 3 m, 5 m, 7 m, and 9 m, respectively, to monitor the room temperature at the working level and to monitor the vertical temperature gradient in the hall. In Figure 8, examples of the installed routers, gateways, and sensors used in this study are shown in Figure 8.

The developed system was built following the approach by the authors ( Arendt et al., 2018), and in-depth knowledge of the structure and structural elements of the building was not required. The approach to hardware development was expandable and exchangeable with other possibilities of data inclusion.

The sensors used to measure temperature and humidity are from Technoline (TX29 DTH-IT) and are capable of transmitting the measured data at a frequency of 868 MHz. All sensors were calibrated accordingly prior to use, and the error tolerances were approximately +-0.3 K and +-5 %, respectively. To upload the measured data to the web servers of TU Dresden, the appropriate receiver (JeeLink) must be configured on a Raspberry Pi (RPi) server. Each RPi Server has a coverage of 50 m –100 m and can manage 30 air temperature/humidity sensors, as well as two to three operative temperature sensors and one CO ₂ sensor. The respective FHEM OpenSource-based server for data management also operates on this server. The wireless sensors collect data and send them through RPi servers and a gateway to the database server at TU Dresden. The data are visualized to observe the experiments online and, if necessary, to immediately react to overshooting. Figure 9 shows the design of the communication structure of the measurement system.

The sensor nodes remained in operation in 2019 and 2020. The analysis presented here focused on the dataset from the winter months of both 2019 and 2020. Three conditions were chosen from the dataset: March 2019 (Condition 1 or C1), April/May 2019 (Condition 2 or C2), and February/March 2020 (Condition 3 or C3). For convenience, the three conditions are mostly referred to as C1, C2, and C3, respectively. The outside temperature was between -5 °C and 20 °C in March 2019 and between -3 °C and 28 °C in April 2019. In the winter period from February to March 2020, the outdoor temperature was between -17 °C and 20 °C ( Weather Spark, 2019, 2022). The three time slots are shown in Figure 10.

The objective is to investigate the potential for energy savings for the HVAC system owing to volume flow reduction while maintaining acceptable air quality and thermal comfort levels. This is achieved by varying the airflow rates (C1, C2, and C3) and observing how these changes impact the indoor climate by conducting measurements. The method of varying the air flow rates may not be systematic, but some special conditions stopped the systematic investigations in autumn 2019, especially in 2020. Therefore, we switched back to higher airflow rates after a stepwise reduction in the spring of 2019. Nevertheless, there were sufficient data for analysis.

Table 3 provides information on the test conditions conducted in accordance with the time of year, power requirement, and volumetric flow rate. The “Nominal Condition” column shows the standard power requirement and volumetric flow rate of 22 kW and 2 × 37,500 m ³/h respectively. Condition C1 (from 1st March–31st March 2019) had a lower calculated power requirement of 12 kW at a volumetric flow rate of 2 × 31,000 m ³/h. This was the starting point of our investigation because the operator became aware of the data of the building management system that it is not necessary to run the HVAC system at maximum capacity. Condition C2 (from 18th April to 18th May2019) had the lowest calculated power requirement of 3.5 kW at a volumetric flow rate of 2 × 20,000 m ³/h.

Similarly, condition C3 (from 20th February to 20th March 2020) had a calculated power requirement of 8.5 kW at a volumetric flow rate of 2 × 27,000 m ³/h. On condition 3 (C3), the fans were reduced to approximately 8.5 kW or 40% of the electrical power and on condition 2 (C2) to about 3.5 kW or about 13.75% of the nominal power.

The power requirements data were calculated using VDI 6014 (2016) and compared to the data delivered by the building management system. In the case of the supply system, the data coincide well, but for the exhaust side, the power requirement is always lower because the short-duct system generates lower pressure losses than the supply system.

It should also be noted that further reductions can be achieved by further reducing the air flow rate or by switching off the ventilation at night or when there is no human presence, thus leading to further energy savings. Systems for monitoring emissions, temperature, and relative humidity are recommended so that the HVAC system can be controlled to achieve the defined thresholds.

Figure 12 shows the air temperature directly in front of an HVAC supply air outlet ( Figure 7, S ₁), whereas Figure 13 shows the room air temperature in the nearby working area ( Figure 7 and S ₄). The mean difference between C1 and C3 was approximately 1 K when comparing the room air temperature values under all three conditions. The air temperatures during C1 and C3 also followed the weekday–weekend heating profile of the production site. C2 showed higher room temperatures in general, owing to the higher outdoor air temperature between April and May 2019 compared to February and March. In contrast to the supply air temperature at C1 and C3, the supply air temperature of C2 approximately represents the outdoor air temperature because the heating module was shut off during the warmer season transition time (April to May). This can also be observed in Figure 13, where the room air temperature moves above the desired 20.5 °C threshold and does not follow the weekday-weekend heating profile.

Figure 14 (a), (b), and (c) show the vertical temperature profiles in the hall for all three test conditions. The temperature at 1,5 m height is often higher than that at 3 m because the supply openings are at 3 m.

Because of internal heat gains, the supply air temperature is often lower than the room air temperature. The temperature is highest at 7 m. At 9 m, the temperature drops again because the sensor is directly below the ceiling and close to or almost adjacent to the cold outside air. This vertical temperature difference can be clearly observed in Figure 14 (a) and Figure 14 (b). During the seasonal transition C2, as shown in Figure 14 (c), the measured air temperatures were closer to each other along the height.

Overall, there is no difference in the thermal and air flow situation in the hall when comparing conditions C1, C2, and C3. This means that the indoor airflow and thermal stratification seem to be independent of the airflow rate within the range between C1 and C2. The already good situation in the indoor climate of the hall is not affected by reducing the air volume rates within the limits discussed above. For a comprehensive assessment, humidity and CO ₂ levels should be evaluated in the same manner.

The relative humidity measurement data for all three conditions ( Figure 15 (a), (b), and (c)) show little to no differences along the height.

Figure 16 shows the CO ₂ measurement data for all the three test conditions. In general, the measured CO ₂ concentration was in the 400–700 ppm band. The CO ₂ concentrations during the C1 and C3 test periods were higher than the CO ₂ concentration measured during the C2 test period. The higher concentration can be explained by the use of direct gas heating in the HVAC system during the test periods C1 and C3, which represent the 2019 and 2020 heating periods, respectively. In both cases, the difference was quite small, indicating that the CO ₂ concentration was well below the Pettenkofer limit of 1,000 ppm. This knowledge is relevant for worker safety because it shows that the direct gas heater does not cause any significant increase in CO ₂ emissions. Overall, this also highlights that the change or reduction in the flow rate under the chosen conditions does not negatively affect the air conditions, as can also be seen from the temperature and relative humidity results. According to the trends in globally averaged CO ₂ determined from NOAA Global Monitoring Laboratory measurements, the average CO ₂ concentration of outdoor air between 2019 and 2020 lies between 409 ppm and 411 ppm ( Lan et al., 2023).

Finally, Table 4 displays the volumetric flow rate and the measured energy consumption of the three different conditions (C1, C2, and C3) during field testing. The first column lists the condition numbers, and the second column indicates the date and time for each condition. The third column shows the volumetric flow rate in cubic meters per hour (m ³/h) under each condition. The fourth and fifth columns indicate the energy consumption in kilowatts (kWh) for the supply and exhaust fans, respectively. Energy consumption was calculated using the data delivered by the building management system for the selected time period.

Overall, this table provides an overview of the volumetric flow rates and electrical energy consumption under the three conditions in this study. We only discuss electrical energy, although there are also savings in gas consumption, which is a subject for another publication. It is clear that energy consumption varies significantly among the three conditions. Condition C1 had a volumetric flow rate of 2 × 31,000 m ³/h, resulting in the highest energy consumption of 16,151 kWh (10,132 kWh for supply fans and 6,019 kWh for exhaust fans).

In contrast, condition C3, with a volumetric flow rate of 2 × 27,000 m ³/h, consumed 14,965 kWh (9,351 kWh for the outlet fans and 5,614 kWh for the exhaust fans). Finally, condition C2 had the lowest energy consumption of 4,540 kWh (2,798 kWh for the supply fans and 1,742 kWh for the exhaust fans), with a volumetric flow rate of 2 × 20,000 m ³/h.

Comparing the energy consumption between the three conditions, it is clear that condition C3 had lower energy consumption than condition C1, with a significant energy saving of 1,186 kWh compared to condition C1. Furthermore, condition C2, which had the lowest volumetric flow rate compared to the other two conditions, resulted in a noteworthy reduction in the energy consumption of 11,611 kWh when compared to the energy consumed in condition C1.

Overall, significant energy savings can be achieved by optimizing the volumetric flow rate and operating conditions, taking into account the current processes, and by running a network of monitoring sensors, as demonstrated by the lower energy consumption in conditions C2 and C3 compared to condition C1.