1. Introduction
Large amounts of carbon dioxide are emitted to the atmosphere by transportation systems, energy production, industry, and buildings. Owing to the urgent need to reduce the ecological footprint of these activities, alternative solutions to the use of fossil fuels should be taken.
The solution involves two decisive factors, namely energy efficiency and the substitution of the use of polluting fossil energy sources by clean renewable energy sources such as the sun, wind, and biomass. Energy efficiency should be a concern of policy-makers and other players in the energy market, as well as of the technicians who design, accompany, implement, and maintain electrical and mechanical installations.
In buildings, Heating, Ventilation and Air Conditioning (HVAC) is the main energy-consuming equipment [
1]. Thus, energy efficiency can be improved by reducing the energy consumption of already installed equipment, making its use more rational and achieving the thermal comfort for the occupants of the air-conditioned spaces. HVAC energy demand is directly related to the indoor temperature setpoint, air infiltration, incident solar radiation, window type, window-wall ratio, internal loads, building type and climate [
2]. An adequate evaluation of the thermal performance of the building envelope is needed [
3,
4], because it will influence the thermal load of the air-conditioned compartments and, therefore, the HVAC performance. Indeed, the building envelope data (type of elements and their dimensions) is an important step in a building thermal behavior numerical simulation [
5]. In this type of simulation, it is important to define the interior and external walls, floors, ceilings, transparent surfaces and other interior details characteristics, and the thermal conductivity, the thermal capacity, and the specific mass of the building envelope elements [
5,
6].
In Portugal, after reaching its highest peak in 2005, final energy consumption has steadily decreased in the last decade [
7]. In 2015, the electrical energy consumption in the State buildings sector in Portugal represented 5.3% of the total National electrical energy consumption [
8]. Typically, the energy consumption of HVAC systems corresponds to approximately 50% of the total energy consumption of a public service building, such as the University campus buildings employed in this work [
7]. The increasing use of HVAC systems in this type of buildings also makes it increasingly important to use efficient control systems in order to ensure a compromise between the lowest possible energy consumption and the best possible level of thermal comfort. Therefore, designing HVAC control systems that have these goals has become an important area [
9].
ISO 7730 defines thermal comfort as “the condition of the mind that expresses satisfaction with the thermal environment” [
10]. In fact, the sensation of satisfaction is very subjective, and it can be described as a mental condition or state of mind, constructed by the individual from various extrinsic and intrinsic factors. The diversity of opinions and sensations is due to the subjectivity inherent in the reactions of each human being, who responds differently to similar stimuli, based on previous cultural experiences and habits, namely the habitual climate in which he/she lives, his/her physical constitution, his/her conditions of physical and psychological health at the moment, and the clothing used, among others.
Thermal comfort indexes have been developed over the years by several researchers. The most well-known is called PMV (Predicted Mean Vote). This index was developed by Fanger [
11], and it was first included in ASHRAE standard 55 in 1981, also present in the 2013 revised version [
12], and later on in ISO 7730 in 1994, also present in the 2005 revised version [
10]. The PMV index is based on the thermal balance of the human body with the environment that surrounds it, being therefore a so-called rational approach to the concept of thermal comfort. According to Fanger and ISO 7730 standard, the predicted mean vote is the expected average of the values voted by a large group of people exposed to the same environmental space, with similar activity and clothing [
10,
11]. ASHRAE created a scale to be used in surveys, in which occupants of the spaces choose an integer value (vote) between −3 and +3. Zero means that the person feels comfortable, in thermal equilibrium with the environment, negative values correspond to cold thermal discomfort sensations, and positive values correspond to hot thermal discomfort sensations, according to the seven-point thermal sensation scale: −3, cold; −2, cool; −1, slightly cool; 0, neutral; +1, slightly warm; +2, warm; +3, hot [
10]. The equations used to calculate the PMV index, according to a set of parameters, are presented in ISO 7730. There are six parameters, four of them environmental parameters (air temperature, mean radiant temperature, air humidity and air velocity) and two personal parameters (metabolic rate and clothing insulation) [
10]. The PMV index was adopted as a standard of thermal comfort by ASHRAE, which elaborated comfort requirements for the occupied spaces, creating three comfort categories, category A (−0.2 ≤ PMV ≤ +0.2), category B (−0.5 ≤ PMV ≤ +0.5) and category C(−0.7 ≤ PMV ≤ +0.7) [
10,
12].
The use of the PMV index in the study of human thermal comfort revealed some shortcomings in its practical application. The thermal balance model overestimates the occupants’ responses under high temperature conditions, and underestimates them in low temperatures [
13]. Consequently, different thermal comfort models, named adaptive models, were developed. These adaptive models are based on field studies, while the thermal balance model was based on laboratory studies [
14]. The concept of adaptive thermal comfort is based on the principle that when changes occur in environmental conditions that generate thermal discomfort, people react psychologically, physiologically or through behavioral changes in order to restore their comfort (please see [
13,
15,
16,
17]). In Yao et al. [
18], the authors developed a theoretical model of thermal comfort, which they called the Adaptive Predicted Mean Vote (aPMV) model. This model is based simultaneously on non-psychological and psychological factors. They proposed an adaptive coefficient that reflects the psychological, physiological, and behavioral human adaptations to the individual perception of indoor thermal comfort evolution. The aPMV model and its development and theoretical foundation are presented in detail in Yao et al. [
18].
Following this work, other approaches were developed where this model was applied in practical cases, or in which new adaptive coefficients were proposed. For example, in Singh et al. [
19], an adaptive thermal comfort model that considers both the season and the climate zone where the building is built is developed. In Conceição et al. [
20], an adaptive thermal comfort model is applied to evaluate the thermal comfort of classrooms in a school, in winter and summer conditions. This model was experimentally developed using simultaneous indoor and outdoor environmental parameter monitoring and a questionnaire survey during typical summer, mid-season and winter days of a year. The indoor environmental variables (air temperature, air relative humidity and mean radiant temperature) and the outdoor air temperature were measured by sensors. The questionnaire survey was used to obtain the thermal sensation responses of the occupants accordingly to climate variations and their adaptation to it. The difference between the thermal responses of the occupants and those obtained by the Fanger model can calculate the adaptive coefficient. The results obtained show a relation between the adaptive coefficient and the outdoor temperature, thus a dependence of the type of climate where this coefficient was experimentally obtained. Comparing the results obtained with the PMV index with the results obtained with the aPMV index, it is shown the aPMV extends the comfort range 9% on average. In Conceição et al. [
14] another adaptive thermal comfort model is developed, where its adaptive coefficient also depends on outdoor temperature. The results obtained allow to conclude that the developed model ensures that the classrooms analyzed have good thermal comfort levels. Other researchers have also developed, analyzed, and applied this new adaptive thermal comfort concept. For more details, please see papers [
21,
22,
23].
Current HVAC control systems can be divided into two types: air temperature regulator (ATR) and thermal comfort regulator (TCR). Most TCR controllers use as a performance criterion the PMV comfort index, or the effective temperature. When it is desired to simultaneously control energy consumption and maintain thermal comfort and indoor air quality within the values considered adequate by the standards, the operation of the HVAC system control becomes complex [
24]. In the optimization of HVAC control systems, different strategies have been used. In Afroz et al. [
25] a rather extensive review of the modeling techniques is presented, proposed in the last 10–15 years, used to improve the HVAC systems control strategies. In this study, the weaknesses, strengths, applications, and performances of these modeling techniques are discussed. During the last decade, HVAC control systems modeling techniques have become more complex, taking into account, simultaneously occupants’ thermal comfort, indoor air quality, regulatory requirements, and energy savings. This kind of research is not usually found in the literature, because developing an effective and accurate model to represent reality becomes a huge challenge for researchers [
25]. Another perspective in the literature is the absence of rigorous simulation tools to perform complex topology buildings energy calculations, making it possible to design and compare several HVAC control strategies [
26]. There is also a lack of information in literature on the simulation of the thermal behavior of complex buildings with a large number of air-conditioned spaces, as in the public building presented in this study. Homod et al. [
6] proposed a HVAC control system based in a Takagi-Sugeno PMV fuzzy model which is tuned by Gauss-Newton nonlinear regression algorithm. Their main conclusions are the possibility of directly control the thermal sensation of the conditioned space, and indirectly control associated variables like air temperature and air relative humidity. Ferreira et al. [
27] proposed an artificial neural network to estimate the PMV index more efficiently, used as a restriction in a predictive control scheme. The obtained models show good input variables estimation accuracy and ensure good levels of comfort in conditioned spaces with energy consumption savings. This approach was refined in [
28], minimizing not the electricity but its price, employing schedules, and incorporating self-powered wireless sensors [
29] and an intelligent weather station [
30] in the hardware. In Ku et al. [
31], a study with a wireless sensor network using PMV as a variable to be controlled is presented. The used control model is based on an adaptive neurofuzzy inference system and a particle swarm algorithm. The experimental results obtained in this study shows the effectiveness of the control methods used, while keeping the PMV value within the expected range, and energy savings of more than 30%. Xu et al. [
32] developed a computationally efficient PMV model approximation and a “model-based periodic event-triggered mechanism (ETM)” in order to optimize the handling of the uncertainties of the building operation. In Conceição et al. [
24], a numerical simulation of the performance of an HVAC system control based on the PMV index, applied in a building with complex topology is developed. In that study, the thermal comfort of the building’s occupants and the HVAC energy consumption are evaluated. The application of the proposed control technique guarantees, simultaneously, good thermal comfort performance and energy savings.
In recent years, research that intends to improve, simultaneously, the energy efficiency of HVAC systems, occupants’ thermal comfort and indoor air quality has been developed. In 2009, Guo and Zhou [
33] found that the use of better system operation and control algorithms will allow HVAC systems to be operated in an energy saving mode, as well as providing good thermal comfort and air quality levels to occupants, according to standardization requirements. Moon and Jung [
34] developed a control algorithm that consists of two artificial neural network based predictive and adaptive models, that shows an improvement in building thermal comfort and energy efficiency, in the cooling season. Homod [
35] developed a novel control algorithm based on the Takagi-Sugeno-Kang Fuzzy inference strategy. Its performance was tested, and the obtained results showed an improvement in indoor thermal comfort, as well a significant energy consumption reduction. Schirrer et al. [
36] presents a nonlinear modular model predictive optimal control that simultaneously controls the heating and cooling of the spaces in a low-energy office building. The results showed that thermal comfort for the user and building energy efficiency are achieved with excellent control performance and robustness.
The aim of this work is to develop a new type of HVAC control system based on aPMV index and evaluate its performance numerically, by comparing it with the more usual HVAC control system based on PMV index. In this work, the comparison with the most widely used indoor air temperature control is not done, because thermal comfort also depends on other variables, rather than air temperature. The purpose of this new control system is to obtain simultaneously good human thermal comfort levels and energy savings, in a real public building. The inputs of this system are the indoor air temperature, indoor air relative humidity, mean radiant temperature, indoor air velocity, and outdoor air temperature, among other variables.
In this work, HVAC systems commonly applied in University buildings with control based on PMV and aPMV indexes are used. A numerical simulation is performed by software that simulates the building’s thermal behavior with complex topology, in transient thermal conditions, for summer and winter conditions. This is discussed in
Section 2. The set point values for the PMV and aPMV HVAC control are −0.7 in winter conditions and +0.7 in summer conditions. The indoor air temperature, the PMV index and the aPMV values obtained in several compartments, representative of the building use, are presented and discussed in
Section 3, as well as the HVAC systems total energy consumption. Conclusions are drawn in
Section 4.
4. Conclusions
In this work, the performances of HVAC systems, controlled using the PMV and the aPMV indexes, are obtained for a University building, using a software that simulates the building thermal behavior, under transient conditions. In this evaluation, the indoor air temperature, the PMV and aPMV indexes, and the HVAC power consumption in all compartments of the building, are numerically obtained, for summer and winter conditions. The PMV index and aPMV index values for each compartment are obtained from the average values of the environmental variables previously calculated for them. This numerical simulation takes into account the influence of the surrounding buildings on the thermal behavior of the building under study. In this work, the results obtained for four selected compartments are presented, as representative of the study carried out.
According with the obtained results, the performance of the HVAC system with aPMV control has lower total daily energy consumption than the HVAC system with PMV control, either in summer conditions or in winter conditions, for the selected compartments. The aPMV control system achieves the lowest total daily energy consumption of 95.5 kWh vs. 107.3 kWh, in summer conditions, and 83.0 kWh vs. 93.6 kWh in winter conditions. The energy consumption level is higher in winter conditions than in summer conditions for compartments with shading structures, and it is lower in winter conditions than in summer conditions for compartments exposed to direct solar radiation.
During the occupation period, the use of HVAC system with control by the aPMV index guarantees, both in summer conditions and in winter conditions, that the selected compartments are thermally comfortable, within categories C, or even B [
10], by negative aPMV values in winter conditions, and, most of the time, by positive aPMV values in summer conditions. This is more difficult to achieve using PMV control.
Regardless of the control system used (the PMV index or the aPMV index) in the HVAC systems operation, the calculated indoor air temperature values are around 20 °C, in winter conditions, and around 27 °C, in summer conditions, as recommended by the Portuguese standard [
43] as comfortably acceptable.
In conclusion, in Mediterranean climates, HVAC systems with aPMV control perform better than HVAC systems with PMV control, not only improving occupants’ thermal comfort levels, but also reducing energy consumption. However, aPMV control can only be used in this type of climate, because the aPMV model was specifically developed for a Mediterranean environment. For other environmental conditions, HVAC systems with PMV control, developed by Fanger for all HVAC systems, should be used.