Keywords

11.1 Introduction

The human factor is the fundamental piece to understanding energy consumption, as people consume energy. Engineering systems often do not consider people as variables in time and less the differences that may exist between a group of people [1]. In this document, we integrate the people directly into a temperature controller that reduces energy consumption while using as the main indicator the thermal sensation of users.

Thermal comfort is an essential concept in building design according to ASHRAE Standard 55 [2], thermal comfort occurs when the person feels contented with the surrounding environmental temperature. People’s efficiency and productivity are associated with thermal comfort; at the same time, the thermal comfort level is estimated through the metabolic rate [3]. Having a good understanding of these relationships is important to improve the general conditions of a room and the quality of life of the occupants, and in some cases can save energy.

The predicted mean vote (PMV) methodology has the objective of predicting the thermal sensation in buildings [4]. The prediction is based on equations developed in 1960 which are described in the paper [5] The literature on PMV applied in fuzzy logic is extensive. Table 11.1 shows the studies in detail, classifying them according to the setpoint type (user temperature input, automatic setpoint, or not specified ‘NS’) and if taking in consideration the energy consumption on the control loop to do energy optimization, not in consideration or not specified.

Table 11.1 Literature review
Full size table

Table 11.1 shows that most of the articles (15 out of 20) consider energy optimisation. Table 11.1 shows how fuzzy logic is almost always implemented to optimise energy consumption. In most of the identified research studies (12 out of 20 papers) users are not allowed to choose a setpoint-temperature. Rather, the researchers used fuzzy logic to automatically choose a temperature setpoint. Dovjak and Shukuya [11] apply fuzzy logic in thermal comfort control, such as temperature, and combine a level of thermal sensation, similar to how the PMV equation works. The PMV equation takes physical values and gives a thermal sensation value. However, in this study, they do not consider the thermal sensation and compared with the PMV model it.

According to [26] ASHRAE’s [26] in the 1970s when the first steps were taken to develop a thermal comfort model, the concept of PMV was developed. The main concept in the PMV model is the body’s thermal neutrality, The scale, 0 represents thermal neutrality, 1, 2 and 3 represent different levels of sensation of heat and − 1, − 2 and − 3 levels of sensation of cold.

Thermal comfort null (equal to 0), only when the heat generated minus heat transfer to the environment (L = Qgenerate − Qtransfer) is equal to 0. A body thermal load of 0 represents a heat transfer level to keep the temperature of the body stable, and a heat transfer at a comfortable value. Six factors directly affect thermal neutrality: Metabolic Rate (W/m2), Clothing insulation (dimensionless), Air temperature (°C), Radiant temperature (°C), Airspeed (m/s), Humidity (dimensionless). The first two factors, metabolic rate and clothing insulation depend on the users and correspond, respectively, to the activity they carry out and to the clothing they wear, the other 4 factors depend on the environment.

In the PMV model, it is necessary to calculate the predicted percentage of dissatisfaction (PPD). The equation that models PPD is an inverted Gaussian and is a function of the PMV, the equations are described in detail in the literature [5]. When PMV is 0 the PPD is equal to 5% (minimum); the PMV model keeps the PPD lower than 10%, which corresponds to 0.5 and − 0.5 PMV. In the literature, they suggest using the PPD as KPI for the building’s thermostats [2].

11.2 Methodology

The chosen methodology combines the PMV model with fuzzy logic to set the temperature setpoint, with two objectives: to reduce energy consumption and improve thermal comfort. The control logic was tested in a simulated virtual environment and then tested in a real environment.

11.2.1 Simulation Components

Simulink provides access to control components in fuzzy logic, which is the type of control we have chosen for controller derived. The innovative approach is the input, a fuzzy variable ranging from cold to hot, which the controller can interpret. In the other studies, carried out in the literature review, the temperature setpoint is a numerical variable, while here we allow the input as a fuzzy-variable. The second innovation is to allow more than a single input of more than one user and give the system the tools to prioritise medium comfort.

Figure 11.1 shows the model designed in Simulink with the main components: (A) the user thermostat interface, (B) the thermostat (fuzzy temperature controller), (C) the boiler and (D) the house heat transfer system. The novel proposition of this project is the user interface, allowing the users to express their thermal sensation. The thermostat uses fuzzy logic and fuzzy-inputs to set a temperature setpoint.

Fig. 11.1
A schematic diagram represents the simulation interface between the highlighted main components A, B, C, and D. Some of the other labeled parts are customer input P M V, daily temperature variation, environment, and indoor average.

Simulation interface is done in Simulink

Full size image

11.2.2 Room

The model of the room in Fig. 11.2 is made with equivalent thermal components, thermal resistances, thermal inertias for the walls and ceiling and the thermal inertia of the air inside the room. To simulate the heat transfer the model uses material normally used in construction in the UK [27]. In this model, the only source of heat is a water heater that transmits heat to the room. The model takes the external temperature as input data, which transforms from digital to an analogue signal and then determines the internal temperature. The room temperature differential equation cannot be solved explicitly. Therefore, MATLAB uses numerical methods to provide a solution at each time step.

Fig. 11.2
A schematic diagram of the simulation room design that includes a temperature sensor, air roof convection, half roof air conduction, roof thermal mass, half atmosphere conduction, roof atmosphere convection, roof atmosphere leakage, wall atmosphere leakage, half wall air conduction, wall thermal mass, half wall atmosphere conduction, wall atmosphere convection, window atmosphere leakage, half window air conduction, window thermal mass, half window atmosphere conduction, window atmosphere convection and temperature source atmosphere.

Room design model, all the thermal components are part of the heat transfer circuit

Full size image

11.2.3 Boiler

The boiler system was built with the tools supplied by Simulink and with the parameters of the ASHRAE [27], such as the water temperature, the calorific value of the fuel, and the type of boiler, and the efficiency of combustion and the humidity of the inlet air. Figure 11.3 shows the boiler components: (a) the fuel pump, (b) the air compressor, and (c) the combustion chamber.

Fig. 11.3
A schematic diagram of the simulation boiler system that highlights three main components, namely, fuel pump, air compressor, and combustion chamber.

Boiler system model: a the fuel pump, b the air compressor, and c the combustion chamber

Full size image

11.2.4 Test Real Environment

A fundamental part of this project was to compare the PMV with the relative thermal comfort level and actual thermal sensation expressed by a user. For this, it would be advisable to implement a survey that allows users to provide relevant information for the study, such as the level of experienced thermal comfort, clothing factor, and metabolic activity level. The survey follows the standard structure [2].

The control system was tested in an office where users could express their thermal preferences on the ASHRAE scale. Figure 11.4 shows a diagram of how we proceeded in the experiment, in this case, we worked with a human operator who controlled the HVAC system. The test was done at Teesside University, lasted 2 weeks and 5 people participated.

Fig. 11.4
A schematic experiment diagram includes a group of people, a thermometer, a mailbox, and buttons. User thermal sensation and room conditions point to thermal comfort fuzzy logic and further to signal, human operator, and on and off buttons.

Schematic diagram from the data collection/experiment

Full size image

During the first week of testing, the participants expressed their thermal sensation in the survey while the temperature of the office was not modified. During the second week the developed controller was tested, and the responses of the users were used to modify the temperature. In this case, a human operator entered the resulting setpoints from the developed controller and manually controlled the existing thermostat based on what the fuzzy temperature controller indicated.

11.3 Results and Discussion

The two results of this study are: (1) The results of the MATLAB simulation show that fuel consumption was reduced by 22% when implementing a controller with 5 multiple users and 31% with a single user (Fig. 11.5); (2) the results of the test in the offices show that the comfort of the participants was improved between 14 and 17% (Fig. 11.6).

Fig. 11.5
3 line graphs of gas versus time for normal gas consumption, 4-person gas consumption, and 1-person gas consumption. Graphs a to c plot an upward trend line each. Three horizontal, linear, and parallel lines pass through all the graphs in different colors between 60 and 120 kilowatt-hours.

Gas consumption comparison, a normal gas consumption (106 kwh), b the new controller 4 users gas consumption (83 kwh) and c new controller one person gas consumption (73 kwh)

Full size image
Fig. 11.6
A grouped bar graph of thermal sensation versus hot, warm, slightly warm, neutral, slightly cool, cool, and cold. Two bars denote weeks 1 and 2. Two rectangular boxes highlight the center bar plot as a and the center 3 bars as b. The peak value is neutral for weeks 1 and 2.

Thermal sensation survey answer comparison, thermal sensation neutral (a), and a slight degree of discomfort (b)

Full size image

The improvement in thermal sensation was 14% if we compare only the ‘neutral’ responses but if we include the ‘Slightly warm’ or ‘Slightly cool’ responses, the improvement is 17%. ‘Slightly warm’ or ‘Slightly cool’ levels are points at which a person can adapt to that temperature because their level of discomfort is not very great, for example wearing or taking off a sweater. In the PMV model, the PPD is kept below 10%, that is, the PMV is between − 0.5 and 0.5, for this reason taking box a would be more restrictive and on the other hand taking box b is more flexible than the PMV model.

11.4 Conclusion

The information shown in this document serves as a framework for the development of a fuzzy controller based on thermal comfort theory. The methodologies that allow the application of the theory of thermal comfort in practice were presented, as well as the tools that MATLAB offers to simulate laboratories.

On the subject of energy efficiency, there are many things that we still do not know. Integrating the human factor in the modelling of energy efficiency can yield good results. For example, in HVAC systems, the goal should be to keep people comfortable, rather than maintaining a fixed temperature. This way we can avoid unnecessary energy consumption and improve the thermal sensation of building users. The next step will be to test the controller in a larger and more complex environment to obtain conclusive results.