Comparison of Multi-Control Strategies for the Control of Indoor Air Temperature and CO 2 with OpenModelica Modeling

: As most residents spend more than 90% of their time in buildings, acceptable and reasonable control of both indoor thermal comfort and air quality is imperative to ensure occupants’ health status and work productivity. However, current control strategies generally take either thermal comfort or indoor air quality as a single loop, rather than the concurrent control of two. To analyze their mutual inﬂuence, this study investigated the performance of three multi-control approaches, i.e., proportional integral derivative (PID) control of thermal comfort and a ﬁxed outdoor air ratio, PID control of thermal comfort and design outdoor air rate, and PID control of thermal comfort and occupancy-based demand-controlled ventilation. As a pilot study, three typical control methods were implemented to a multi-zone building via OpenModelica modeling. The results indicate that indoor air temperature can be well-maintained under three control methods, however, the CO 2 concentration under the ﬁxed outdoor air ratio was over 1000 ppm, leading to poor indoor air quality. The control strategy with the design outdoor air rate could not properly ensure the CO 2 concentration, due to the over-ventilated or under-ventilated phenomena, subsequently resulting in unnecessary energy waste. The occupancy-based demand controlled ventilation could maintain the CO 2 concentration under the set-point with an intermediate power energy utilization.


Introduction
Among building energy consumers, heat, ventilation, and air conditioning (HVAC) systems dominate more than 40% of power consumption, to maintain acceptable indoor air quality (IAQ) and thermal comfort (TC) status [1,2]. These two objectives are consistently urgent and essential issues for human health, since most people spend more than 90% of their time inside buildings [3,4]. A poor IAQ or uncomfortable indoor environment could cause health problems and diseases, as well as achieve the proper balance between IAQ and energy saving. In [25], a dynamic demanded-controlled ventilation strategy of CO 2 control was applied to an hourly scheduled sports training center, by setting a base ventilation rate for unoccupied periods and calculating the ventilation rate dynamically at each occupied period. The results showed that the strategy could save 34% of energy related to ventilation air as compared to proportional control. Except for adjusting ventilation rate, an alternative control strategy of supply air CO 2 concentration could also achieve CO 2 control. In [26], a supply CO 2 -based, demand-controlled ventilation strategy was proposed by determining the CO 2 concentration of supply air to meet the IAQ requirement of the critical zone, which showed that such a control strategy could maintain good IAQ with 3.1% annual energy saving. While the supply CO 2 -based, demand-controlled method was based on the design occupancy profile rather than the real occupancy, when analyzing the CO 2 levels, researchers assumed the supply air rate was constant, without considering the changes of indoor air temperature, which was not in accordance with the actual situation. In the authors' previous work [27], the occupancy-based DCV (demand-controlled ventilation) strategy was proposed and applied to a multi-zone building, to optimize indoor air quality. However, when coordinating the ventilation rate for multi zones, the supply air rate and indoor air temperature were predetermined without consideration of the variation of indoor air temperature.
Through the above statement, we note that the current control strategies of indoor air temperature and CO 2 are independently performed, without considering the mutual influence between indoor air temperature control and CO 2 control. In fact, the indoor air temperature control in VAV systems could result in the change of outdoor air rate and affect the CO 2 concentration levels for multi-zone buildings. Conversely, the CO 2 control might also have certain effects on supply air rate and indoor air temperature. In this study, the multi-control strategies of indoor air temperature and CO 2 were investigated with OpenModelica simulation, by simultaneously analyzing the changing patterns of indoor air temperature and CO 2 concentration, for a multi-zone building with VAV systems. The control performance of three typical multi-control strategies, i.e., PID control of thermal comfort and a fixed outdoor air ratio, PID control of thermal comfort and design outdoor air rate, PID control of thermal comfort and occupancy-based demand-controlled ventilation, were compared via the OpenModelica modeling. With the predicted occupancy profiles, the performance of occupancy-based, multi-control strategy for indoor air temperature and CO 2 were evaluated as well. Finally, the energy usage of three multi-control strategies was also analyzed.

Principles of Indoor Air Temperature Control and CO 2 Control
As defined in the ASHRAE standard 55 [28], indoor air temperature should range within the thermal comfort zone to ensure people's thermal comfort status. If the indoor air temperature is too high, some occupants might feel hot, while if the temperature is too low, some occupants might feel cold. Indoor air temperature is the key factor that affects thermal sensations of occupants. Meanwhile, indoor air temperature is influenced both by internal and external parameters, e.g., supply air conditions, outdoor air temperature, the number of occupants, etc. For an air-conditioned thermal zone, the heat balance of the indoor environment can be expressed as Equation (1): where T z is the indoor air temperature of the conditioned zone, K; V z is the volume of the conditioned zone, m 3 ; ρ is the air mass density, kg/m 3 ; C a is the specific heat of air (at constant pressure), J/(kg·K); V sz is the supply air rate for the conditioned zone, m 3 ; T s is the supply air temperature, K; and Q load is the cooling load of the conditioned space, W. As the cooling load varied with the outdoor conditions (i.e., outdoor air temperature) and the internal loads (i.e., occupants, electric devices), the supply air rate or supply air temperature could Energies 2020, 13, 4425 4 of 20 be adjusted to ensure the heat balance of indoor environment and control indoor air temperature. According to different control variables, thermal comfort control strategies could be divided into variable volume control (VAV) and variable temperature control [29]. The premier control method aims to maintain the heat balance by changing the volume of supply air with constant supply air temperature. The other control method (i.e., variable temperature control) tries to change supply air temperature to ensure the heat balance of the indoor environment. Compared to variable temperature control, VAV control is more widely implemented in practical applications, as it is easy to realize [30].
Beside thermal comfort, there is a growing concern about the need to improve the indoor air quality of buildings and reduce the associated ventilation energy consumption. As for the air-conditioned space, the indoor pollutant mainly concerns carbon dioxide generated by occupants' respiration. The mass balance of CO 2 in the ventilation space can be expressed as below: where V out is the outdoor air rate, m 3 /s; C o is the outdoor air CO 2 concentration that can be regarded as constant, g/m 3 ; m e is the indoor CO 2 generation rate, g/s; C z is the CO 2 concentration of the ventilation zone, g/m 3 ; and V is the volume of the ventilation zone, m 3 . Based on Equation (2), the CO 2 concentration level of the zone is affected by the outdoor air rate and the CO 2 generation rate of the zone. At the same time, the CO 2 generation rate is related to the number of occupants inside the zone. According to the ASHRAE standard 62.1-2019 [22], the outdoor air rate can be determined by the occupancy and required outdoor air rate for each person. Consequently, the proper determination of outdoor air rate is the key to maintain indoor air quality.

Multi-Control of Indoor Air Temperature and CO 2
The multiple control of thermal comfort and indoor air quality can be easily implemented for a single room or single-zone buildings, which could be achieved by controlling the supply air rate and outdoor air rate, respectively, according to the thermostat and CO 2 sensor. However, as for multi-zone buildings, the building shares only one air handling unit (AHU) to provide fresh air and cool air to different thermal zones, to concurrently ensure the thermal comfort and indoor air quality, as shown in Figure 1 [31].
Energies 2020, 13, x FOR PEER REVIEW 4 of 21 variable volume control (VAV) and variable temperature control [29]. The premier control method aims to maintain the heat balance by changing the volume of supply air with constant supply air temperature. The other control method (i.e., variable temperature control) tries to change supply air temperature to ensure the heat balance of the indoor environment. Compared to variable temperature control, VAV control is more widely implemented in practical applications, as it is easy to realize [30]. Beside thermal comfort, there is a growing concern about the need to improve the indoor air quality of buildings and reduce the associated ventilation energy consumption. As for the airconditioned space, the indoor pollutant mainly concerns carbon dioxide generated by occupants' respiration. The mass balance of CO2 in the ventilation space can be expressed as below: where Vout is the outdoor air rate, m 3 /s; Co is the outdoor air CO2 concentration that can be regarded as constant, g/m 3 ; me is the indoor CO2 generation rate, g/s; Cz is the CO2 concentration of the ventilation zone, g/m 3 ; and V is the volume of the ventilation zone, m 3 . Based on Equation (2), the CO2 concentration level of the zone is affected by the outdoor air rate and the CO2 generation rate of the zone. At the same time, the CO2 generation rate is related to the number of occupants inside the zone. According to the ASHRAE standard 62.1-2019 [22], the outdoor air rate can be determined by the occupancy and required outdoor air rate for each person. Consequently, the proper determination of outdoor air rate is the key to maintain indoor air quality.

Multi-Control of Indoor Air Temperature and CO2
The multiple control of thermal comfort and indoor air quality can be easily implemented for a single room or single-zone buildings, which could be achieved by controlling the supply air rate and outdoor air rate, respectively, according to the thermostat and CO2 sensor. However, as for multizone buildings, the building shares only one air handling unit (AHU) to provide fresh air and cool air to different thermal zones, to concurrently ensure the thermal comfort and indoor air quality, as shown in Figure 1 [31]. In general, by installing the VAV boxes at each zone, the indoor air temperature could be effectively maintained by changing the supply air volume. As is known, the indoor air quality is heavily dependent on the outdoor air rate. Considering the AC system scheme, all return air from different thermal zones is mixed in the return duct, as expressed by Equation (3): In general, by installing the VAV boxes at each zone, the indoor air temperature could be effectively maintained by changing the supply air volume. As is known, the indoor air quality is heavily dependent on the outdoor air rate. Considering the AC system scheme, all return air from different thermal zones is mixed in the return duct, as expressed by Equation (3): Energies 2020, 13, 4425 5 of 20 where C r represents the CO 2 concentration at the return duct; V sa,i is the supply air rate for zone i; C z,i is the CO 2 concentration of zone i; V sa,tot is the total supply air rate of the system. The CO 2 -based demand-controlled ventilation is one of the strategies that could lower ventilation energy use by reducing over-ventilation of the multi-zone buildings. With the CO 2 -based DCV strategy, the outdoor air rate is determined by the CO 2 concentration level at the return duct. However, according to Equation (3), the CO 2 concentration level at the return duct cannot accurately reveal the CO 2 concentration of multiple thermal zones. Such a controlling approach cannot effectively ensure the indoor air quality, as it might result in poor indoor air quality in certain zones of multi-zone buildings. Therefore, the outdoor air rate needs to be carefully determined to ensure acceptable indoor air quality of different thermal zones, with fewer energy consumptions.

OpenModelica Modeling
Modelica is an equation-based, object-oriented language, which was developed to model complex engineering systems that were described by coupled systems of differential, algebraic, and discrete equations [32]. By using the Modelica prototyping, OpenModelica is an open-source Modelica-based simulation platform, which is robust to dynamic modeling for complex systems, consisting of multi-domains [33]. Different from conventional modeling and simulation programs, Modelica-based simulation program has the advantages of addressing acausal modeling, which can be determined by users [34]. With OpenModelica platform, a large class of library packages for thermo-fluid systems can be implemented for simulating HVAC equipment, e.g., ThermoFlow Library, AirConditioning Library, and Modelica_Fluid Library [35]. For building energy and control systems, the Building Library, developed by the Lawrence Berkeley National Laboratory, is a free and open-source Modelica Library that consists of thermal and fluid domains of the HVAC systems and the typical control strategies [36,37]. Meanwhile, the library supports building envelope and multi-zone simulations. In our study, the simulation platform OpenModelica with Building Library was used to simulate the performance of HVAC systems.

Methodology and Case Study
Through the statement of control methods of thermal comfort and indoor air quality, it was justified to combine typical control techniques to investigate the changing laws of indoor air temperature and CO 2 concentration. In this study, three different combined control strategies were assessed. The indoor air temperature was maintained by adjusting the supply air rate of the air handling unit. A VAV (variable air volume) box was installed at each zone to control the volume of cool air supplied to the zone, so the supply air rate would be dynamically changed to offset the variable cooling loads. Through PID control of the damper position, the supply air rate varied from the minimum to maximum volume, to maintain the indoor air temperature. The relationship between indoor air temperature and supply air rate could be determined by Equation (1).
As for the indoor air quality, the outdoor air was provided to dilute the indoor contaminants. Inside air-conditioned buildings, CO 2 is the main contaminant generated by occupants and other sources. Outdoor air with a lower CO 2 concentration level is supplied to buildings to ensure indoor air quality. The outdoor air rate could be determined by the ratio to supply airflow rate, as shown in Equation (4): Generally, the minimum outdoor air ratio ranges from 15% to 30%. In this study, the outdoor air fraction was fixed as 25%. Thus, the outdoor air rate could be determined by the supply airflow rate.
3.1.2. Strategy 2: PID Control for Indoor Air Temperature and Fixed Outdoor Air Rate The indoor air temperature was also controlled by a VAV box, by changing the supply air volumes. According to the ASHRAE standard 62.1 [22], the required outdoor airflow rate was determined by accounting for people-related sources and area-related sources, as shown in Equation (5): where R p represents the outdoor airflow rate per person to dilute contaminants generated by people, L/s·per; R a is the outdoor airflow rate per unit area to dilute contaminants emitted by floor area, L/s·m 2 ; P z is the designed zone population; A z is the floor area of the zone, m 2 . The values of R p and R a are related to the building types. The uncorrected outdoor airflow rate for the multi-zone buildings are shown in Equation (6): where V ou is the uncorrected outdoor airflow rate of the system, m 3 /s. The outdoor air ratio of the system was determined by Equation (7): 3.1.3. Strategy 3: PID Control for Indoor Air Temperature and Demand-Controlled Outdoor Air Rate The indoor air temperature of the system was also controlled by changing the supply air volume of the VAV boxes. As for the indoor air quality, if the occupancy profile of the different thermal zones could be predicted, the required outdoor air rate of the zone could be determined, based on Equation (5). Within Strategy 2, the outdoor air rate was determined by the design number of occupants, since the accurate occupancy profile could not be obtained. If the occupancy profile of thermal zones could be well predicted, the outdoor air rate would dynamically change with the predicted occupancy. For Strategy 3, the ventilation scheme was demand-controlled ventilation with the Wi-Fi probe enabled occupancy, to avoid possible over-ventilated or under-ventilated cases. The required outdoor airflow rate of the zone could be determined by Equation (8): where P pre,z τ represents the occupants predicted by the Wi-Fi probe at time τ.
As the occupancy profile is known, by comparing the predicted number of occupants of each zone, the zone requiring the largest outdoor air rate would be selected as the critical zone, as expressed by Equation (9): where V oz,i τ is the required outdoor air rate of zone i at time τ.
Different from Strategy 2, P pre,z is the predicted number of occupants rather than the design number of occupants of the zone, and thus the required outdoor air rate was no more constant, but dynamically varied with time. According to the multi-zone ventilation method [22], the outdoor air ratio with predicted occupancy profile could be expressed as:

Case Study
To investigate the performance of control methods on thermal comfort and indoor air quality, three scenarios using different control strategies were studied via the OpenModelica platform. As the real occupancy profiles were difficult to collect, it made the demand-controlled strategy hard to comply with the actual demand of supply air volume and ventilation rate. As mentioned in [38], arobust occupancy prediction method with the Markov time-window inference approach (DMTWI) was proposed. In Strategy 3, the Wi-Fi-probe-enabled occupancy profile was used to determine the outdoor air ratio for ventilation. The occupancy profiles for the office rooms were predicted by the Wi-Fi-probe-based DMTWI, and the predicted occupancy was over 80%, compared to the actual occupancy on-site experiment. The details of the occupancy prediction method are given in previous studies [27,38]. This study mainly focuses on the performance of the occupancy-driven multi-control strategy.

•
Scenario 1: PID control of indoor air temperature and fixed outdoor air ratio to dilute CO 2 contaminant. • Scenario 2: PID controls of indoor air temperature and design outdoor air rate to dilute CO 2 contaminant. • Scenario 3: PID controls of indoor air temperature and demanded-control ventilation scheme to dilute CO 2 contaminant with Wi-Fi probe enabled occupancy.
The performance of three scenarios was compared via the OpenModelica platform, and the OpenModelica models are shown in Figure 2. The OpenModelica modeling predicted indoor air temperature, CO 2 concentration, and supply air rate, under three scenarios. Based on the simulation results, thermal comfort, indoor air quality, and power consumption were compared among the three scenarios. The building model had six similar thermal zones of 10 m × 10 m × 3 m. The supply air temperature was assumed to be 18 • C. The outdoor air CO 2 concentration was assumed to be 350 ppm, and the CO 2 generation rate of a person kept constant at 4.9 × 10 −6 m 3 /s. Based on the ASHRAE standard 62.1 for the office building, the outdoor air required rate per person was 3.0 × 10 −3 m 3 /s, and the area per unit was 0.28 × 10 −4 m 3 /s. The set-points of indoor air temperature and CO2 concentrations were 24 • C and 1000 ppm, respectively. The simulation period started from 8:30 am to 18:00. The initial CO 2 concentration level and indoor air temperature of all zones were set as 1000 ppm and 27 • C, respectively.
To simplify the control loop, this study made some assumptions, such as: The indoor air temperature and CO 2 concentration were uniformly distributed inside the room; 2.
We assumed there were no natural ventilation of all rooms with good air tightness; 3.
The heat transfer through the building envelope were ignored, and heat gains for all zones were assumed to come from occupants, lights, and computers; Under the first two scenarios, two kinds of occupancy profiles were assessed, i.e., the designed occupancy recommended by the ASHRAE standards and the real occupancy profile. As for Scenario 3, the Wi-Fi-probe-enabled occupancy and the real occupancy were compared. Thus, six cases were investigated within OpenModelica to evaluate the control effects via the indoor air temperature, CO 2 concentration levels, and the power energy consumptions. The indoor air temperature and CO 2 concentration level of the rooms were initially set as 27 • C and 1000 ppm, respectively. The air temperature and CO 2 concentration of fresh air were 30 • C and 350 ppm, respectively. To simplify the control loop, this study made some assumptions, such as: 1. The indoor air temperature and CO2 concentration were uniformly distributed inside the room;

Scenario 1: PID Control of Thermal Comfort and Fixed Outdoor Air Ratio
Under Scenario 1, the control methods for thermal comfort and indoor air quality were PID control and fixed outdoor air fraction, respectively. The outdoor air ratio was fixed as 25% to the supply air volume. The performance of the control method was simulated with OpenModelica. The CO 2 concentration level, indoor air temperature, and supply air rate of six zones were investigated through OpenModelica modeling.
The CO 2 concentration levels of the six zones are shown in Figure 3. With the fixed fresh air ratio, the changing laws of CO 2 concentration for six zones were similar and complied with exponential changing. At the beginning of the operation time, the CO 2 concentration increased from 940 ppm to 1280 ppm. After 12:00, the CO 2 concentration tended to be stable at 1280 ppm. Such a control method of outdoor air rate could not efficiently dilute the CO 2 produced by occupants, resulting in poor indoor air quality. occupancy recommended by the ASHRAE standards and the real occupancy profile. As for Scenario 3, the Wi-Fi-probe-enabled occupancy and the real occupancy were compared. Thus, six cases were investigated within OpenModelica to evaluate the control effects via the indoor air temperature, CO2 concentration levels, and the power energy consumptions. The indoor air temperature and CO2 concentration level of the rooms were initially set as 27 °C and 1000 ppm, respectively. The air temperature and CO2 concentration of fresh air were 30 °C and 350 ppm, respectively.

Scenario 1: PID Control of Thermal Comfort and Fixed Outdoor Air Ratio
Under Scenario 1, the control methods for thermal comfort and indoor air quality were PID control and fixed outdoor air fraction, respectively. The outdoor air ratio was fixed as 25% to the supply air volume. The performance of the control method was simulated with OpenModelica. The CO2 concentration level, indoor air temperature, and supply air rate of six zones were investigated through OpenModelica modeling.
The CO2 concentration levels of the six zones are shown in Figure 3. With the fixed fresh air ratio, the changing laws of CO2 concentration for six zones were similar and complied with exponential changing. At the beginning of the operation time, the CO2 concentration increased from 940 ppm to 1280 ppm. After 12:00, the CO2 concentration tended to be stable at 1280 ppm. Such a control method of outdoor air rate could not efficiently dilute the CO2 produced by occupants, resulting in poor indoor air quality. Besides the indoor air quality, the indoor air temperature under PID control was simulated, as shown in Figure 4. During the operation period, the values of indoor air temperature had small variations around the set-point of indoor air temperature (i.e., Tset = 24 °C). The largest discrepancy ranged from −0.5 °C to 0.1 °C. The indoor air temperature rose from 23.5 °C to 24 °C during the first four operation period with few occupants inside the building. After 10:00 am, the indoor air Besides the indoor air quality, the indoor air temperature under PID control was simulated, as shown in Figure 4. During the operation period, the values of indoor air temperature had small variations around the set-point of indoor air temperature (i.e., T set = 24 • C). The largest discrepancy ranged from −0.5 • C to 0.1 • C. The indoor air temperature rose from 23.5 • C to 24 • C during the first four operation period with few occupants inside the building. After 10:00 am, the indoor air temperature was well-maintained around 24 • C. Therefore, the PID control of indoor air temperature could effectively ensure an acceptable thermal environment for occupants by controlling supply air volumes.
Energies 2020, 13, x FOR PEER REVIEW 10 of 21 temperature was well-maintained around 24 °C. Therefore, the PID control of indoor air temperature could effectively ensure an acceptable thermal environment for occupants by controlling supply air volumes. Under Scenario 1, the changing patterns of supply air rate of the six zones were analyzed. As shown in Figure 5, the supply air rate dynamically varied with time during the operation period. Due to the different number of occupants in each zone, the supply air rate was subsequently different. The supply air volumes of Zone 1 and Zone 4 were higher than the other four zones. By assessing the changing curves in Figure 5, it could be summarized that the supply air rate showed an increase at the beginning, as the occupants started entering the building. Then, the supply air volumes decreased as fewer people stayed in the office at lunchtime. In the afternoon, the overall supply air volumes kept stable. After 17:00, the supply air rate decreased, since the occupants gradually left the office. Therefore, the changing patterns of supply air rate complied with the occupancy profile, to maintain indoor air temperature. Under Scenario 1, the changing patterns of supply air rate of the six zones were analyzed. As shown in Figure 5, the supply air rate dynamically varied with time during the operation period. Due to the different number of occupants in each zone, the supply air rate was subsequently different. The supply air volumes of Zone 1 and Zone 4 were higher than the other four zones. By assessing the changing curves in Figure 5, it could be summarized that the supply air rate showed an increase at the beginning, as the occupants started entering the building. Then, the supply air volumes decreased as fewer people stayed in the office at lunchtime. In the afternoon, the overall supply air volumes kept stable. After 17:00, the supply air rate decreased, since the occupants gradually left the office. Therefore, the changing patterns of supply air rate complied with the occupancy profile, to maintain indoor air temperature.
The supply air volume and outdoor air rate of the system are calculated, as shown in Figure 6. The changing patterns of the supply air rate and the outdoor air rate were similar, due to the constant outdoor air ratio of 0.25. The maximum supply air rate was 1.0 m 3 /s at 15:00. The supply air volume and outdoor air rate of the system are calculated, as shown in Figure 6. The changing patterns of the supply air rate and the outdoor air rate were similar, due to the constant outdoor air ratio of 0.25. The maximum supply air rate was 1.0 m 3 /s at 15:00.

Scenario 2: PID Control of Thermal Comfort and Design Outdoor Air Rate
Under Scenario 2, the PID control was used to maintain thermal comfort, while the multi-zone ventilation method was applied to dilute CO2 generated by occupants. Generally, the real occupancy profile of the thermal zones is unknown. Strategy 2 provides the method to calculate the outdoor airflow rate, expressed as Equation (5), and the required outdoor airflow rate was determined by the design number of the occupants of the zone. Therefore, the outdoor airflow was maintained constant during the operation period.
Through OpenModelica modeling, the CO2 concentration of the six zones under the ventilation scheme of Strategy 2 was predicted, as shown in Figure 7. The figure presents the CO2 concentration The supply air volume and outdoor air rate of the system are calculated, as shown in Figure 6. The changing patterns of the supply air rate and the outdoor air rate were similar, due to the constant outdoor air ratio of 0.25. The maximum supply air rate was 1.0 m 3 /s at 15:00.

Scenario 2: PID Control of Thermal Comfort and Design Outdoor Air Rate
Under Scenario 2, the PID control was used to maintain thermal comfort, while the multi-zone ventilation method was applied to dilute CO2 generated by occupants. Generally, the real occupancy profile of the thermal zones is unknown. Strategy 2 provides the method to calculate the outdoor airflow rate, expressed as Equation (5), and the required outdoor airflow rate was determined by the design number of the occupants of the zone. Therefore, the outdoor airflow was maintained constant during the operation period.
Through OpenModelica modeling, the CO2 concentration of the six zones under the ventilation scheme of Strategy 2 was predicted, as shown in Figure 7. The figure presents the CO2 concentration

Scenario 2: PID Control of Thermal Comfort and Design Outdoor Air Rate
Under Scenario 2, the PID control was used to maintain thermal comfort, while the multi-zone ventilation method was applied to dilute CO 2 generated by occupants. Generally, the real occupancy profile of the thermal zones is unknown. Strategy 2 provides the method to calculate the outdoor airflow rate, expressed as Equation (5), and the required outdoor airflow rate was determined by the design number of the occupants of the zone. Therefore, the outdoor airflow was maintained constant during the operation period.
Through OpenModelica modeling, the CO 2 concentration of the six zones under the ventilation scheme of Strategy 2 was predicted, as shown in Figure 7. The figure presents the CO 2 concentration level of six thermal zones, which first increased with the number of occupants from 8:30 to 11:00. During this period, as there were fewer occupants than the design occupancy, the CO 2 concentration Energies 2020, 13, 4425 12 of 20 of the six zones rose from 350 ppm to 1000 ppm. When the real number of occupants was larger than the design number, the CO 2 concentration level was over the set-point (i.e., 1000 ppm). Taking Zone 4 and 5 as an example, the CO 2 concentration levels were higher than the set-point from 14:30 to 18:00. At the beginning of the working period, the CO 2 concentration of the six zones was much lower than 1000 ppm, indicating over ventilation. While under ventilation occurred as the real number of occupants increased beyond the design number. The ventilation scheme with fixed outdoor air rate could not properly ensure CO 2 concentration during the operation period, due to the ventilation scheme with the potential to cause over-ventilated or under-ventilated cases.
Energies 2020, 13, x FOR PEER REVIEW 12 of 21 level of six thermal zones, which first increased with the number of occupants from 8:30 to 11:00. During this period, as there were fewer occupants than the design occupancy, the CO2 concentration of the six zones rose from 350 ppm to 1000 ppm. When the real number of occupants was larger than the design number, the CO2 concentration level was over the set-point (i.e., 1000 ppm). Taking Zone 4 and 5 as an example, the CO2 concentration levels were higher than the set-point from 14:30 to 18:00. At the beginning of the working period, the CO2 concentration of the six zones was much lower than 1000 ppm, indicating over ventilation. While under ventilation occurred as the real number of occupants increased beyond the design number. The ventilation scheme with fixed outdoor air rate could not properly ensure CO2 concentration during the operation period, due to the ventilation scheme with the potential to cause over-ventilated or under-ventilated cases. The indoor air temperature was also analyzed via OpenModelica modeling. As illustrated in Figure 8, the indoor air temperature had a small variation around the set-point of indoor air temperature (i.e., Tset = 24 °C). The largest discrepancy between indoor air temperature and the setpoint varied from −0.05 °C to 0.1 °C. Thus, the indoor air temperature could be effectively controlled to maintain the thermal comfort of occupants by changing the supply air rate. The indoor air temperature was also analyzed via OpenModelica modeling. As illustrated in Figure 8, the indoor air temperature had a small variation around the set-point of indoor air temperature (i.e., T set = 24 • C). The largest discrepancy between indoor air temperature and the set-point varied from −0.05 • C to 0.1 • C. Thus, the indoor air temperature could be effectively controlled to maintain the thermal comfort of occupants by changing the supply air rate.
Meanwhile, the supply air rate of six zones was evaluated, as shown in Figure 9. The figure presents the changing patterns of the supply air rate of the six zones. Due to the dynamic occupancy profile, the supply air rate of each zone varied with time, during the working period. By analyzing the changing patterns of the supply air rate of the six zones, it could be concluded that from 8:30 to 11:00, the supply air rate gradually increased from the minimum value (the required ventilation rate) as the occupancy of the zone changed from the unloaded state to the partially loaded state. During the afternoon, the supply air volume reached the maximum value with fully loaded occupants.
The total supply air rate and outdoor air rate of the air conditioning system were analyzed, as shown in Figure 10. Under Scenario 2, the total outdoor air rate of the system was constant, as the outdoor air rate of each zone was determined under the design number of occupants. While the total supply air rate of the system varied with time, and the maximum supply air rate was 1.3 m 3 /s, occurring at 15:00. At the beginning of the operation period, with a fixed outdoor air rate, the phenomenon of over ventilation appeared, as the real number of occupants was much smaller than the design value. Figure 7 also reflects over ventilation, as the CO 2 concentration levels of the six thermal zones ranged from 350 ppm to 500 ppm between 8:30 to 9:30. Therefore, the fixed outdoor air rate would cause over-ventilation at the beginning of the operation period, which might cause unnecessary wastage of power energy. Meanwhile, the supply air rate of six zones was evaluated, as shown in Figure 9. The figure presents the changing patterns of the supply air rate of the six zones. Due to the dynamic occupancy profile, the supply air rate of each zone varied with time, during the working period. By analyzing the changing patterns of the supply air rate of the six zones, it could be concluded that from 8:30 to 11:00, the supply air rate gradually increased from the minimum value (the required ventilation rate) as the occupancy of the zone changed from the unloaded state to the partially loaded state. During the afternoon, the supply air volume reached the maximum value with fully loaded occupants.
The total supply air rate and outdoor air rate of the air conditioning system were analyzed, as shown in Figure 10. Under Scenario 2, the total outdoor air rate of the system was constant, as the outdoor air rate of each zone was determined under the design number of occupants. While the total supply air rate of the system varied with time, and the maximum supply air rate was 1.3 m 3 /s, occurring at 15:00. At the beginning of the operation period, with a fixed outdoor air rate, the phenomenon of over ventilation appeared, as the real number of occupants was much smaller than the design value. Figure 7 also reflects over ventilation, as the CO2 concentration levels of the six thermal zones ranged from 350 ppm to 500 ppm between 8:30 to 9:30. Therefore, the fixed outdoor air rate would cause over-ventilation at the beginning of the operation period, which might cause unnecessary wastage of power energy.

Scenario 3: PID Control of Thermal Comfort and Demand-Based Ventilation with Wi-Fi-Enabled Occupancy
Different from the fixed outdoor air ratio and design outdoor airflow rate, the outdoor airflow rate was determined by the occupancy predicted by the Wi-Fi probe. According to Strategy 3, the thermal zone with the largest number of occupants was selected as the critical zone, and the outdoor airflow rate was determined by the required fresh air of the critical zone. Strategy 3 was applied to the air conditioning system, severing six thermal zones on the platform of OpenModelica. With the Wi-Fi enabled occupancy, the outdoor airflow rate was first determined by Strategy 3. The outdoor

Scenario 3: PID Control of Thermal Comfort and Demand-Based Ventilation with Wi-Fi-Enabled Occupancy
Different from the fixed outdoor air ratio and design outdoor airflow rate, the outdoor airflow rate was determined by the occupancy predicted by the Wi-Fi probe. According to gy 3,Strate the thermal zone with the largest number of occupants was selected as the critical zone, and the outdoor airflow rate was determined by the required fresh air of the critical zone. Strategy 3 was applied to the air conditioning system, severing six thermal zones on the platform of OpenModelica. With the Wi-Fi enabled occupancy, the outdoor airflow rate was first determined by Strategy 3. The outdoor airflow rate was input to the OpenModelica model, and the CO 2 concentration levels of the six zones were simulated. The CO 2 concentration of the six thermal zones was predicted, as shown in Figure 11. The figure shows that the CO 2 concentration of the six thermal zones was not over 1000 ppm, which indicates that the determined outdoor air volume could effectively ensure the indoor air quality of six zones. Except Zone 2, the CO 2 concentration level of the other five zones generally varied from 850 ppm to 1000 ppm. During 13:00 t0 14:00, Zone 2 had the lowest CO 2 concentration, which was caused by the large discrepancy of occupants between Zone 2 and the critical zone.
Besides the CO 2 concentration of the six zones, indoor air temperature was analyzed as well, as shown in Figure 12. Under Scenario 3, the indoor air temperature could be maintained around the set-point, by effectively controlling the supply air rate. The changing patterns of air temperature for the six zones were similar. From 8:30 to 9:00, there were large variations of air temperature, as the indoor air temperature varied from 23.6 • C to 24.1 • C. After 9:00, the indoor air temperature had small variations around the set-point (i.e., T set = 24 • C), and the discrepancy between indoor air temperature and the set-point ranged from −0.1 • C to 0.1 • C.
The supply air volume under Scenario 3 was simulated, as presented in Figure 13. Compared to the changing patterns of Scenario 1 and 2, the supply air rates of the six zones showed smaller variations during the working period. As the outdoor air volume was determined according to the predicted occupancy profile, the supply air volume of each zone also varied with the occupancy. Meanwhile, the curves of supply air rate for the six zones were similar. At the beginning of the working period, the supply air rate fluctuated. Then the supply air rate for each zone fluctuated around the average value. For example, after 12:00, the supply air rates of Zone 1 and Zone 2 fluctuated around 0.25 m 3 /s and 0.20 m 3 /s, respectively. The overall supply air rates of Zone 4, 5, and 6 varied around 0.15 m 3 /s.
The supply airflow rate and outdoor airflow rate of the air conditioning systems were simulated as well, as shown in Figure 14. The changing curves of the supply air rate and the outdoor air rate were similar, as they both varied with the occupancy profile. Generally, the outdoor air ratio under Scenario 3 was 0.55.
Energies 2020, 13, x FOR PEER REVIEW 15 of 21 airflow rate was input to the OpenModelica model, and the CO2 concentration levels of the six zones were simulated. The CO2 concentration of the six thermal zones was predicted, as shown in Figure  11. The figure shows that the CO2 concentration of the six thermal zones was not over 1000 ppm, which indicates that the determined outdoor air volume could effectively ensure the indoor air quality of six zones. Except Zone 2, the CO2 concentration level of the other five zones generally varied from 850 ppm to 1000 ppm. During 13:00 t0 14:00, Zone 2 had the lowest CO2 concentration, which was caused by the large discrepancy of occupants between Zone 2 and the critical zone. Besides the CO2 concentration of the six zones, indoor air temperature was analyzed as well, as shown in Figure 12. Under Scenario 3, the indoor air temperature could be maintained around the set-point, by effectively controlling the supply air rate. The changing patterns of air temperature for the six zones were similar. From 8:30 to 9:00, there were large variations of air temperature, as the indoor air temperature varied from 23.6 °C to 24.1 °C. After 9:00, the indoor air temperature had small variations around the set-point (i.e., Tset = 24 °C), and the discrepancy between indoor air temperature and the set-point ranged from −0.1 °C to 0.1°C. The supply air volume under Scenario 3 was simulated, as presented in Figure 13. Compared to the changing patterns of Scenario 1 and 2, the supply air rates of the six zones showed smaller variations during the working period. As the outdoor air volume was determined according to the predicted occupancy profile, the supply air volume of each zone also varied with the occupancy. Meanwhile, the curves of supply air rate for the six zones were similar. At the beginning of the working period, the supply air rate fluctuated. Then the supply air rate for each zone fluctuated The supply airflow rate and outdoor airflow rate of the air conditioning systems were simulated as well, as shown in Figure 14. The changing curves of the supply air rate and the outdoor air rate were similar, as they both varied with the occupancy profile. Generally, the outdoor air ratio under Scenario 3 was 0.55.

The Energy Use of the Three Scenarios
Energy use under the three scenarios was analyzed, except for the control performance of indoor air temperature and the CO2 concentration. Energy consumption concerned the power energy of the The supply airflow rate and outdoor airflow rate of the air conditioning systems were simulated as well, as shown in Figure 14. The changing curves of the supply air rate and the outdoor air rate were similar, as they both varied with the occupancy profile. Generally, the outdoor air ratio under Scenario 3 was 0.55.

The Energy Use of the Three Scenarios
Energy use under the three scenarios was analyzed, except for the control performance of indoor air temperature and the CO2 concentration. Energy consumption concerned the power energy of the

The Energy Use of the Three Scenarios
Energy use under the three scenarios was analyzed, except for the control performance of indoor air temperature and the CO 2 concentration. Energy consumption concerned the power energy of the supply air fan and the outdoor air fan. The rated power of the supply air fan (FCU) was 1100 W, with a rated airflow rate of 5000 m 3 /h. The rated power and airflow rate of the fresh air fan (FAU) was 370 W Energies 2020, 13, 4425 17 of 20 and 2700 m 3 /h, respectively. The power energy of the air fan was proportional to the third power of the air volume ratio: where P fan is the power energy of the fan, W; P fan,rated is the rated power energy of the fan, W; G airflow is the airflow rate of the fan, m 3 /h; and G airflow, rated is the rated airflow rate of the fan, m 3 /h. Based on Equation (11), the total power energy of the AC system under the three scenarios was calculated, as presented in Figure 15.
Energies 2020, 13, x FOR PEER REVIEW 18 of 21 supply air fan and the outdoor air fan. The rated power of the supply air fan (FCU) was 1100 W, with a rated airflow rate of 5000 m 3 /h. The rated power and airflow rate of the fresh air fan (FAU) was 370 W and 2700 m 3 /h, respectively. The power energy of the air fan was proportional to the third power of the air volume ratio: 3 airflow fan fan,rated airflow,rated ( ) G P P G   (11) where Pfan is the power energy of the fan, W; Pfan,rated is the rated power energy of the fan, W; Gairflow is the airflow rate of the fan, m 3 /h; and Gairflow, rated is the rated airflow rate of the fan, m 3 /h. Based on Equation (11), the total power energy of the AC system under the three scenarios was calculated, as presented in Figure 15. Through the comparison of the power energy among three scenarios, it was found that the power energy of fans under Scenario 1 was the lowest with the fixed outdoor air fraction. While Scenario 2 consumed the highest power energy with the designed outdoor air rate. The maximum value was 1031 W at 15:00, when the occupancy was fully loaded. For Scenario 3, the outdoor air was determined according to the predicted occupancy, and thus, the power energy was overall lower than Scenario 2. Since the control method of indoor air temperature was the same for three scenarios, the ventilation scheme decided the discrepancies of power energy. Considering the control performance and power energy of the three control strategies, Strategy 3 had the best performance to maintain thermal comfort and indoor air quality, with acceptable energy usage.
Above all, the comprehensive performance of three control methods for thermal comfort and indoor air quality was discussed. Compared to Strategy 3, Strategy 1 and Strategy 2 were easier implemented in actual applications. The difference between the three strategies lay in the control technique of indoor air quality. By comparing CO2 concentration levels of the six zones under three scenarios (i.e., Figures 3, 7, and 11), it could be deduced that Scenario 1 had the worst control performance of indoor air quality, as the CO2 concentration patterns possess exponential change, and the maximum CO2 concentration level reached 1280 ppm. Moreover, the main drawback of Strategy 1 was that outdoor air rate varied with the supply air rate, and low supply air rate induced low outdoor air rate, which caused a poor IAQ for the multi-zone buildings. Under Scenario 2, the CO2 Through the comparison of the power energy among three scenarios, it was found that the power energy of fans under Scenario 1 was the lowest with the fixed outdoor air fraction. While Scenario 2 consumed the highest power energy with the designed outdoor air rate. The maximum value was 1031 W at 15:00, when the occupancy was fully loaded. For Scenario 3, the outdoor air was determined according to the predicted occupancy, and thus, the power energy was overall lower than Scenario 2. Since the control method of indoor air temperature was the same for three scenarios, the ventilation scheme decided the discrepancies of power energy. Considering the control performance and power energy of the three control strategies, Strategy 3 had the best performance to maintain thermal comfort and indoor air quality, with acceptable energy usage.
Above all, the comprehensive performance of three control methods for thermal comfort and indoor air quality was discussed. Compared to Strategy 3, Strategy 1 and Strategy 2 were easier implemented in actual applications. The difference between the three strategies lay in the control technique of indoor air quality. By comparing CO 2 concentration levels of the six zones under three scenarios (i.e., Figures 3, 7 and 11), it could be deduced that Scenario 1 had the worst control performance of indoor air quality, as the CO 2 concentration patterns possess exponential change, and the maximum CO 2 concentration level reached 1280 ppm. Moreover, the main drawback of Strategy 1 was that outdoor air rate varied with the supply air rate, and low supply air rate induced low outdoor air rate, which caused a poor IAQ for the multi-zone buildings. Under Scenario 2, the CO 2 concentration of certain zones was too low, while the other zones were over the set-point of CO 2 concentration (i.e., C set = 1000 ppm). The drawback of Strategy 2 was that over-ventilated or under-ventilated cases occurred for the multi-zone buildings. With the predicted occupancy, the ventilation scheme was demand-based ventilation, as the outdoor air rate was determined by the predicted occupancy. Under Scenario 3, the indoor air quality of multiple zones could be well maintained with CO 2 concentration levels of six zones under the set-point ( Figure 11). As for the control performance of the thermal comfort, indoor air temperature of six zones could be well maintained around the set-point (i.e., T set = 24 • C), as the supply air rate was controlled to ensure the heat balance of the thermal zones (shown in Figures 4, 8 and 12). Besides the performance of thermal comfort and indoor air quality, the energy performance of the three scenarios was compared. The power energy of Scenario 1 was the lowest, while for Scenario 2 the power energy was the highest with the designed outdoor air rate. Compared to Scenario 2, Scenario 3 consumed less power energy, with demand-based ventilation strategy, and the discrepancy between the two scenarios ranged from 14.3% to 54.9%.

Conclusions
In this study, the combined control methods of thermal comfort and indoor air quality were investigated to evaluate the control performance and energy use. Three multi-control strategies were applied to a building with six office rooms corresponding to three scenarios. The indoor air temperature and CO 2 concentration level of six zones were simulated via OpenModelica platform. Meanwhile, the power energy consumptions of the air conditioning system were compared among three scenarios. Concerning the control performance of thermal comfort, the indoor air temperature of six zones could be well maintained under three scenarios with variable air volume. The temperature differences between indoor air temperature and the set-point ranged from −0.5 • C to 0.1 • C, which signified that the VAV could efficiently ensure the thermal comfort of occupants. Through the analysis of CO 2 concentration levels, the indoor air quality of the six zones under Scenario 1 was the poorest, as the CO 2 concentration was over the set-point during the operation period. The ventilation scheme with fixed outdoor air fraction could not provide enough fresh air to dilute the CO 2 generated by occupants. Under Scenario 2, the outdoor air rate was determined as the design outdoor air rate, according to the ASHRAE standard. When the number of occupants was more than the design number, the CO 2 concentration level was over 1000 ppm. The CO 2 concentration of the six zones ranged from 350 ppm to 1000 ppm between 8:30 to 12:00, with fewer occupants inside the building. Different from Scenario 2, the outdoor air rate dynamically varied with the number of occupants. Based on the predicted occupancy with the Wi-Fi probe, the ventilation scheme was demand-based outdoor air rate, under Scenario 3. The CO 2 concentration of the six zones could be maintained around 1000 ppm. Additionally, the total power energy of the three scenarios was analyzed. Although the power energy of Scenario 1 was the lowest, the control method could not provide acceptable indoor air quality for occupants. While the control method of Scenario 2 consumed the highest power energy, this control method might induce over ventilation, due to the use of a design outdoor air rate. Scenario 3 required less power energy than Scenario 2 to maintain acceptable indoor air quality for the six zones, as the ventilation method was demand-based ventilation. Therefore, Strategy 3 (i.e., PID control of thermal comfort and demand-based ventilation) possessed the best performance on controlling thermal comfort and indoor air quality.

Funding:
The work described in this paper was sponsored by the Natural Science Foundation of Jiangsu Province (#BK20190362), and was also supported by National Key R&D Program of China (#2019YFD1100903). Any opinions, findings, conclusions, or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of the NSFC. The work was also supported by Research Grant Council of Hong Kong Special Administrative Region, China (RGC #11209518).