**Simulation Analysis of a Ventilation System in a Smart Broiler Chamber Based on Computational Fluid Dynamics**

**Shikai Zhang 1, Anlan Ding 1, Xiuguo Zou 1,2,3,4,\*, Bo Feng 2,3,4, Xinfa Qiu 2,3,4, Siyu Wang 5, Shixiu Zhang 1,6, Yan Qian 1,6, Heyang Yao <sup>1</sup> and Yuning Wei <sup>1</sup>**


Received: 3 May 2019; Accepted: 5 June 2019; Published: 6 June 2019

**Abstract:** In this paper, a CFD (computational fluid dynamics) numerical calculation was employed to examine whether the ventilation system of the self-designed smart broiler house meets the requirements of cooling and ventilation for the welfare in poultry breeding. The broiler chamber is powered by two negative pressure fans. The fans are designed with different frequencies for the ventilation system according to the specific air temperature in the broiler chamber. The simulation of ventilation in the empty chamber involved five working conditions in this research. The simulation of ventilation in the broiler chamber and the simulation of the age of air were carried out under three working conditions. According to the measured dimensions of the broiler chamber, a three-dimensional model of the broiler chamber was constructed, and then the model was simplified and meshed in ICEM CFD (integrated computer engineering and manufacturing code for computational fluid dynamics). Two models, i.e., the empty chamber mesh model and the chamber mesh model with block model, were imported in the Fluent software for calculation. In the experiment, 15 measurement points were selected to obtain the simulated and measured values of wind velocity. For the acquired data on wind velocity, the root mean square error (RMSE) was 19.1% and the maximum absolute error was 0.27 m/s, which verified the accuracy of the CFD model in simulating the ventilation system of the broiler chamber. The boundary conditions were further applied to the broiler chamber model to simulate the wind velocity and the age of air. The simulation results show that, when the temperature was between 32 and 34 ◦C, the average wind velocity on the plane of the corresponding broiler chamber (Y = 0.2 m) was higher than 0.8 m/s, which meets the requirement of comfortable breeding. At the lowest frequency of the fan, the oldest age of air was less than 150 s, which meets the basic requirement for broiler chamber design. An optimization idea is proposed for the age of air analysis under three working conditions to improve the structure of this smart broiler chamber.

**Keywords:** smart broiler chamber; ventilation system; wind velocity; age of air; computational fluid dynamics; simulation analysis

#### **1. Introduction**

In summer, a closed broiler house will produce many harmful gases, including ammonia, hydrogen sulfide, carbon monoxide, carbon dioxide, dust, etc. The ventilation system design of a broiler house plays an important role in improving the internal environment of the broiler house (such as temperature and air quality). In a closed broiler house designed with mechanical ventilation, air flow is influenced by the power and quantity of fans; the size, shape, and location of air inlets and outlets; the distribution of temperature; the difference between the indoor and outdoor temperature; and the flow of outside air [1]. In view of the ventilation of a broiler house applying fine breeding, this paper proposes a practical ventilation scheme. The ventilation system of the broiler house should be investigated to examine its design rationality [2].

CFD (computational fluid dynamics) simulation technology is a very effective method to predict and evaluate the performance of ventilation system design [3]. At present, CFD is widely applied to the ventilation simulation of broiler houses. Researchers have extensively employed CFD technology in simulating the environment of livestock houses using two-dimensional models or three-dimensional models under the conditions of natural ventilation or mechanical ventilation [4–6]. Kic et al. [7] used Fluent software to simulate the ventilation in broiler houses in summer and winter. They concluded that the accuracy of three-dimensional simulation was higher than that of two-dimensional simulation.

Some researchers used CFD to compare the strengths and weaknesses of different ventilation schemes for livestock houses and tried to select the optimal solution and further optimize it. Bjerg et al. [8] researched the influence of different modes of air intake on the airflow in a pig house using CFD technology. Norton et al. [9] investigated the improvement of eaves openings of livestock houses by CFD technology. Mostafa et al. [10] simulated and compared four different ventilation schemes of the broiler houses designed with duct ventilation in winter, and concluded that the four schemes could improve the uniformity of indoor wind velocity to 60–70%. Yao et al. [11] used CFD technology to simulate and analyze an airflow problem where a large amount of airflow diffused above the goose house and the ventilation effect on the ground was blocked, and then they proposed an optimized scheme. Li et al. [12] used CFD to simulate and analyze the influence of different opening angles, installation heights, and wind velocity of rectangular air inlets on the airflow distribution in a closed flat broiler house. Coradil [13] used CFD technology to simulate the temperature field in a broiler house heated by a stove.

In order to improve the accuracy of CFD in evaluating the ventilation conditions of a broilers house, some researchers have proposed appropriate settings for different boundary conditions or turbulence models. Blanes-Vidal et al. [14] studied the influence of different boundary conditions on the simulation accuracy in a broiler house, and compared the simulation results with experimental data. Cheng et al. [15] obtained the resistance coefficients under different conditions by comparing the different shapes and sizes of broiler house models, and then simplified the occupied area of caged hens as a porous medium for wind tunnel experiments. They concluded that the result by RNG (the renormalization group) k-ε is more accurate for the full geometry model.Sun et al. [16] applied PIV (particle image velocimetry) technology to the research of a broiler house. Seo et al. [17] simulated the ventilation conditions of the broiler houses with natural ventilation in winter by CFD technology, and improved the original ventilation scheme based on the simulation results. The numerical simulation results shown by the previous researchers have indicated that the relative error of the RNG k-ε turbulence model is the smallest among the turbulence models selected. Therefore, it is effective and feasible to employ CFD simulation technology to optimize the ventilation conditions of existing livestock houses [18].

With the advantages of low cost, high efficiency, and good repeatability, CFD simulation has been widely used in evaluating the ventilation environment of broiler houses. However, researchers have mostly focused on large-scale farms, and there is limited research on the simulation of the ventilation environment of fine breeding broiler house. In this paper, based on the on-site measurement of the structure of a smart broiler house, SolidWorks was used to establish a three-dimensional model of the broiler house. Then, ICEM CFD (the integrated computer engineering and manufacturing code for computational fluid dynamics) was used to mesh the model of the empty chamber and the model of the house with broilers. After that, CFD was introduced into Fluent software to numerically simulate the distribution and velocity of the airflow. The velocity contours of the empty chamber were compared with those of the chamber with broilers. In addition, the ageof airwas also simulated to check the ventilation efficiency of the broiler house. In summary, the design rationality of the ventilation system of the smart broiler house was analyzed and verified in a comprehensive manner.

#### **2. Materials and Methods**

#### *2.1. Research Objects and Measurement Methods*

#### 2.1.1. Layout of Experimental Site

Compared with large broilers farms, the small-scale broiler house in this experiment used a variety of sensors to detect the environmental parameters, which is easier to manage and more intelligent than large farms.The broiler house for the experiment is located in the Jinniuhu Subdistrict, Luhe District, Nanjing City, Jiangsu Province. The experiment started on 20 July 2018 and lasted for 21 days. The broiler house consists of two symmetrical chambers. The four walls around the broiler chamber are made of 55 mm-thick colored steel–polystyrene sandwich plates. Each chamber is 1.9 m in width, 2.9 m in length, with a total area of 5.51m2. The roof is slope-shaped with a height of 1.88 m on the west and 1.77 m on the east, which is convenient for draining rainwater. The ventilation system in the chamber consists of an air inlet, an air outlet, and an internal circulation. The internal circulation refers to the part where the openings at both ends are connected through pipes. Fan A and Fan B were arranged at the air outlet and the internal circulation air inlet, respectively. A Pulin Leshi 400 axial flow negative pressure fan was selected, with a rated air volume of 9000 m3/h and theoretical applicable area of 20–35 m2. Frequency conversion controllers were used to change the frequencies at different temperatures to create an optimal ventilation environment for broilers. Since the internal structures of the two broiler chambers were completely consistent, this paper only analyzed the ventilation structure of one chamber. Figure 1 shows the internal environment of the broiler chamber. Figure 2 shows the structure and dimensions of the chamber. Figure 3 presents the three-dimensional chamber model established by SolidWorks.

**Figure 1.** Internal picture of the experimental broiler chamber.

**Figure 2.** The structure and dimensions of the experimental broiler chamber.

**Figure 3.** The three-dimensional chamber model was established by SolidWorks with (1) chamber door, (2) air inlet, (3) internal circulation front air bellow, (4) internal circulation pipeline, (5) Feeder A, (6) water tank, (7) drinking water pipe, (8) Feeder B, (9) internal circulation rear air bellow, (10) Fan B, (11) Fan A, (12) camera, and (13)air conditioner.

#### 2.1.2. Selection of Measurement Points

According to the average height of broilers and a certain active area, the plane of 20cm above the ground was selected. The selected plane was divided into three lines, with five points in each line. Thus, a total of 15 measurement points was set, as shown in Figure 4. The wind velocity was measured using a Testo 405i handheld hot-wire anemometer with a measurement range of 0~30 m/s and a measurement error of ±(0.1m/s+5%). Since the broilers might interfere with the measurement work in the broiler chamber, the measured data were drawn from the empty broiler chamber after the breeding. The velocity of the fan inlet was acquired by taking an average of thenine points at the air inlet.The measurement points are shown in Figure 4b. Table 1 displays the measured wind velocity data of the nine points at the air inlet under three working conditions. The measured wind velocity data were transmitted to a mobile phone through Bluetooth, and they could be read and stored by the mobile phone.

**Figure 4.** Point distribution: (**a**) The plane distribution position of the detection points; (**b**) The inlet distribution position of the detection points.


**Table 1.** The measured wind velocity values of the nine measuring points at the air inlet.

#### 2.1.3. Ventilation System

Fan A and Fan B are installed at two different air vents and work independently. They perform frequency modulation and temperature control according to the actual indoor temperature. The specific scheme is shown in Table 2. Since the experiment was conducted in summer in Jiangsu, the temperature at night is generally higher than 24 ◦C, so the 10 Hz working condition of Fan A is not met and, therefore, the simulation experiment did not study this. Because the broiler chamber is set outside the chamber, the sun shines directly on it, and sometimes the temperature is higher than 34 ◦C. To ensure the continuity of the experiment, the air conditioner was utilized to avoid high temperatures at which the broiler would have a severe thermal emergency response. At this time, the air conditioner is only used to reduce the indoor temperature, and this paper does not study this working condition. The simulation analysis in this paper is mainly based on five working conditions, as shown in Table 2.

**Table 2.** The experimental plan of fan frequency based on five working conditions. RPM = revolutions per minute.


#### *2.2. CFD Model*

#### 2.2.1. CFD Control Equations

CFD is a kind of numerical simulation under the control of basic flow equations. In the numerical calculation, the air is considered as a continuous, steady, and incompressible Newtonian fluid. The continuity equation is also a mass conservation equation, and any flow movement must satisfy the Law of Conservation of Mass [19].

(1) Mass conservation equation

$$\frac{\partial(\rho \mathbf{u})}{\partial \mathbf{x}} + \frac{\partial(\rho \mathbf{v})}{\partial \mathbf{y}} + \frac{\partial(\rho \mathbf{w})}{\partial \mathbf{z}} = \mathbf{0} \tag{1}$$

where, ρ is the fluid density, and *u*, *v,* and *w* are the components of the velocity vector in *x*, *y,* and *z* directions, respectively.

(2) Momentum conservation equation

$$\frac{\partial(\rho u\_i)}{\partial t} + \frac{\partial(\rho u\_i u\_j)}{\partial \mathbf{x}\_j} = -\frac{\partial p}{\partial \mathbf{x}\_i} + \frac{\partial \mathbf{r}\_{ij}}{\partial \mathbf{x}\_i} + \rho \mathbf{g}\_i + F\_i \tag{2}$$

where, *p* is the static pressure on the fluid microelement body, *t* is time, and *gi* and *Fi* represent the gravity volume force and other external volume forces acting on the microelement body in the *i* direction, respectively. τ*ij* is the viscous stress tensor acting on the surface of the microelement body due to the molecular viscous effect.

#### 2.2.2. Mesh Division of the Broiler Chamber

In order to improve the calculation speed and meshing quality, a three-dimensional model was established in ICEM CFD15.0 according to the coordinates of each point. At first, the 1:1 broiler chamber model was simplified, and the steel frame structure at the top was removed. The internal structure was simplified without affecting the CFD simulation results. Non-structural meshes were selected to mesh the empty broiler chamber. The maximum side length of the mesh was set to 0.05 m. Since the main area of activity of broilers is located at the bottom of the broiler chamber, the ventilation condition on the ground surface is more worthy of our attention. The meshes were increased in density from a 0.6 m-high plane to the bottom of the empty chamber with a total cell number of 7,494,859, and the mesh inspection quality was greater than or equal to 0.34, which meets the calculation requirements. Mesh division of the broiler chamber in ICME CFD is shown in Figure 5.

**Figure 5.** Mesh division of the broiler chamber in ICME CFD (integrated computer engineering and manufacturing code for computational fluid dynamics).

#### 2.2.3. Broiler Model

#### Simplification of Broiler

Cheng et al. [15] simplified a single broiler to three different degrees, including a body only model, a broiler body model, and an ellipsoid model. After simulation and comparison, they finally chose the body only model as the research model. Chen [20] used a block model for CFD research on the broiler chamber. For the model of this project, it was necessary to determine a suitable model. In this paper, 35 seven-week-old yellow feather broilers were studied in the broiler chamber, and the chamber space was small, so geometric modeling could be carried out on a single broiler to improve the simulation accuracy. At the same time, in order to improve the calculation efficiency and accuracy, the broiler was simplified to a certain extent according to the original volume, as shown in Figure 6, (a) shows the body only model, and (b) shows the block model.

**Figure 6.** Schematic diagram of broiler model: (**a**) body only model, (**b**)block model.

Efficiency Calculation and Selection

As shown in Figure 7, 35 measuring points were randomly selected from the position 15 cm away from the ground for comparison of model simulation results, and the corresponding velocity values of the 35 measuring points in the two models were drawn into the comparison chart of efficiency verification of wind velocity, as shown in Figure 8.

**Figure 7.** Distribution of two models for broilers: (**a**) body only model, (**b**) block model.

**Figure 8.** Comparison of efficiency verification wind velocity between the body only model and the block model.

From Figure 8, it can be seen that the simulation results of the body only model and the block model have good consistency, and the trend is consistent. The average absolute error between the body only model and the block model was 0.12, and the error between the two sets of simulation values was small, indicating that the impact on the wind field in the broiler chamber was approximate. Table 3 shows the efficiency comparison of different models.


**Table 3.** The efficiency comparison of different models.

Under the distribution of two different broiler models, the mesh division time and Fluent numerical calculation time were counted, the mesh division number and convergence calculation time length were compared, and an appropriate broiler model was selected for the research in this paper. The computer processors were Intel ®Xeon E7-4830 @2.13 GHz and Intel ®Xeon E5-2660 V4 @2.0 GHz, respectively, with 64.0 GB of installed memory. The mesh division time of the block model and the body only model increased by 0.24 and 4.46 times, and the mesh number increased by 0.11 and 4.35 times, respectively. The mesh quality of the empty chamber was better than the mesh quality of the chamber with the block model, and both of them were better than the mesh quality of the chamber with the body only model. The calculation time of the block model and the body only model using Fluent increased by 1.48 and 5.5 times compared with the empty broiler chamber model. In terms of computational efficiency, the block model of broilers had greater advantages. Therefore, the block model was selected as the prototype to be studied in the next step.

#### *2.3. Solver Parameter Settings*

After the mesh model was completed, Fluent 15.0 software was used to perform the simulations of the empty broiler chamber, the broiler chamber with the block model, and the age of air in the broiler chamber with broilers under different working conditions. The simultaneous study of PIV and numerical simulation for a broiler house was conducted by Sun. Comparing the results of the two, it was found that the standard mean square error between the simulated values by the RNG k-ε turbulence model and the measured values was less than 0.25 and that the simulated values were more accurate [14]. Cheng et al. [17] explored the effects of flow resistance by considering the geometry of the broiler models (full geometry, ellipsoid, and maternal model in which the head, neck, and legs were ignored) and spatial distribution. In the wind tunnel experiment, five full geometric models were used to represent the broilers, and the model verification was performed using numerical simulation. Different turbulence models were evaluated, and the RNG k-ε model showed better performance than the other models. Specific parameters are given in Table 4.


**Table 4.** Specific parameters of solver that were used in Fluent.

#### *2.4. Broiler Chambers*

In the simulation of the wind field in the broiler chamber, the location distribution of broilers had different effects on the wind field. From the video, we randomly selected an arrangement photo of the broiler positions, shown in Figure 9a. As the basis for the distribution of broilers, a simulated experimental arrangement of the broiler chamber with broilers was carried out, as shown in Figure 9b. This paper mainly observed the wind velocity and the wind velocity flow field arrangement in the random distribution area of broilers, compared the velocity of the wind field in the empty broiler chamber, and analyzed the influence of broilers on the wind field in the broiler chamber, so as to make better improvements and optimizations for future research.

**Figure 9.** Experimental arrangement of the chamber with broilers was simulated by the block model. (**a**) one video frame of broiler distribution, (**b**) modeling of broiler distribution in ICEM CFD.

#### *2.5. RMS Error*

RMSE (Root mean square error) is very sensitive to extremely large or small errors in the two sets of data, so it can reflect the approximation of the two sets as well as the extent to which the simulated values deviate from the measured values.RMSE can be calculated by Equation (3).

$$\text{RMSE} = \sqrt{\frac{\sum\_{i=1}^{n} \left(X\_{obs,i} - X\_{mid,i}\right)^2}{n}} \tag{3}$$

where, *n* = 15 is the number of measurement points, *i* = 1,2, ... ... *n*.*Xobs*,*<sup>i</sup>* is the measured value of point *i*, and *Xmodel*,*<sup>i</sup>* is the simulated value.

#### *2.6. Control Equation of Age of Air*

The age of air refers to the time taken for fresh air to travel from the entrance to each mesh cell, which can reflect the ventilation situation in the chamber and the residence time of airflow, thus reflecting the replacement velocity of indoor fresh air. Therefore, the age of air was used as an index to evaluate the air quality in the broiler chamber in this paper [21]. The control equation of steady-state age of air is given as Equation (4).

$$
\nabla(\mu \tau) - \nabla(\Gamma \cdot \nabla \tau) = 1 \tag{4}
$$

where, μ is the velocity of air (m/s), τ is the age of air (s), Γ is the diffusion coefficient, and ∇ = (∂/∂*x*, ∂/∂*y*, ∂/∂*z*).

#### **3. Results and Discussion**

#### *3.1. Wind Velocity Simulation of Empty Broiler Chamber under Di*ff*erent Working Conditions*

#### 3.1.1. Numerical Comparison of Monitoring Points

Figure 10 shows the measured and simulated values of the 15 measurement points in the empty broiler chamber under different working conditions. The simulated values of the 15 measurement points were obtained by the momentum conservation equation in Fluent. Figure 10 shows the results in sequence, from working condition 1 to working condition 5. The maximum absolute error values and RMSE values are shown in Table 5 after the data were exported and compared.


**Table 5.** Error values under different working conditions.

**Figure 10.** Numerical comparison of empty broiler chamber under (**a**)work condition 5; (**b**)work condition 4; (**c**)work condition 3; (**d**)work condition 2; and (**e**)work condition 1.

At the height of Y = 0.2 m, under different working conditions, the difference between the simulated values and the measured values is not large, and the variation rules are consistent. The RMSE in this research was 0.19 m/s, which is close to the result of 0.16 m/s by Yao [16], and the maximum absolute error was 0.27 m/s, which shows that the CFD model and the adopted boundary conditions are suitable for this smart broiler chamber model. Therefore, the boundary conditions corresponding to each working condition were obtained, and then they were applied to a broiler chamber for the simulation experiment.

#### 3.1.2. Analysis of Velocity Contours

Air velocity contours of the empty chamber were arranged in sequence from working condition 1 to working condition 5. As shown in Figure 11, the research of air velocity on the Y-axis plane of the empty chamber was at the height of 0.2 m in order to compare the velocity relationship among different working conditions, the colormap of the contour was unified, and the color in the velocity contour corresponded to the velocity in the colormap. Therefore, the velocity range under the five working conditions was 0~3.5 m/s, and the change rule of air velocity was studied in sequence.

**Figure 11.** Simulated velocity contour of the empty chamber with the colormap on the left.

In Figure 11, according to the contour of the five working conditions, the law of the plane wind velocity field in the broiler chamber at the height of Y = 0.2 m was relatively consistent. The wind velocities at the inlet and outlet were large, and there was obvious convection at the two opposite outlets. The velocity at the inlet was large, and fresh air was fed into the chamber. The five contours all had obvious high-velocity wind areas at about x = 0.6 m from the entrance, which was due to the influence of the water pipe of the water tank, causing the airflow to rise or fall along the pipe wall. Working condition 5 did not open the internal circulation system, and Fan A frequency was low, which can allow a small amount of ventilation. There was no wind at the two corners of the internal

circulation side. The wind field uniformity of the broiler chamber was not high, and the suitable living area for broilers was limited. Fan A and Fan B worked together when the internal circulation system was started, and fresh air flowing to the middle of the broiler chamber circulated to the two fan ports. The air flowing through Fan A was directly exhausted from the broiler chamber, while the air flowing through Fan B was blown out from the other side through the internal circulation pipeline. By comparing working conditions 1 and 2, when the frequency of Fan B increased, the weak wind area in the plane could be effectively reduced.

Under the working condition 1, the plane with Z = 0.2 m in the broiler chamber shows the contour of the wind velocity field on the internal circulation vertical plane. The wind was drawn into the ventilation duct from Fan B, and the airflow direction at the wall near the outlet side formed a certain angle with the wall, which made the wind vector line at the wall denser and improved the wind velocity at the top of the inlet wall. Therefore, a larger vortex was formed in the broiler chamber along the X direction, so that the upper airflow could be driven to the lower part to exit the broiler chamber through airflow circulation. Under these five working conditions, the higher the internal circulation frequency, the more uniform the wind velocity field as a whole, indicating that the design of internal circulation has a great effect on the wind-velocity uniform distribution of the broiler chamber.

#### *3.2. Simulation of the Broiler Chamber with Broilers under Three Common Working Conditions*

Since the simulated and measured values of different working conditions under the empty broiler chamber were close to each other, the determined boundary conditions under the empty broiler chamber were applied to the model of the broiler chamber with broilers. The temperature of the broiler chamber in which broilers are put was relatively higher than that of the empty chamber, so the working conditions 1 and 2 were commonly used during the day in summer and the working condition 4 was used at night. The following is a study on the wind velocity field of the broiler chamber with broilers under three common working conditions. The wind velocity contours are shown in Figure 12.

Y = 0.13 m

**Figure 12.** Simulated velocity contours of the broiler chamber with broilers.

The apparent temperature of broilers was generally analyzed by measuring the average skin temperature of side face, ears, comb, and lower leg, so the height of the velocity contour was selected as Y = 0.13 m (the average height of the middle part of the head in the broiler chamber). In the study from Yahav, the most suitable wind velocity was 2.0 m/s under a high-temperature environment (35 ± 1.0 ◦C). At that wind velocity, the body temperature was the lowest, and overhigh (3.0 m/s) or overlow (0.8m/s) wind velocities will raise the body temperature [21]. Zhang et al. [22] conducted experiments on 42-day-old broilers at 26 ◦C, 29 ◦C, and 32 ◦C with 0, 0.5 m/s, 1m/s, 1.5m/s, and 2.0m/s wind velocities. Under the condition of relatively high temperatures, considering the ventilation benefits, the optimum wind velocity was 1.2 m/s. According to Figure 12, the contours of working condition 2 and working condition 1 were more uniform in the wind velocity field due to the shielding and guiding effects of broilers on airflow compared with the contours of the same working condition of the empty chamber. The wind velocity at the vent can reach 2 m/s, which is favorable for the rapid emission of high-temperature gas. Therefore, when the weather temperature reaches above 30 ◦C, it is more suitable for broilers to gather near the vent. Most of the wind velocity values in the broiler chamber were above 0.875 m/s, which is close to the most suitable wind velocity, and there was almost no windless area, thus ensuring the normal living environment of broilers and avoiding a heat stress reaction. Working condition 5 corresponds to a temperature range of 24~26 ◦C. This condition mainly occurs at night when the air temperature is suitable for broilers and the fan is not necessary to increase the wind speed to cool the broilers, but needs to achieve continuous ventilation of the house.

#### *3.3. Simulation of the Age of Air*

The good ventilation environment in the broiler chamber not only cools the surface of the broiler and makes it feel comfortable, but also needs to provide sufficient fresh air to the broiler chamber. Therefore, the age of air under the broiler chamber was simulated. The age of air with Y = 0.2 m (broiler living plane), Z = 0.2 m (vertical section of internal circulation), and Z = 1.65 m (vertical section of air inlet and outlet) were simulated under three working conditions in turn. The contours of the age of air are shown in Figure 13.

Mean age of air

**Figure 13.** The contours of the age of air in the chamber with broilers.

In Figure 13, according to the contours, the ages of air were as follows: Working Condition 1 < Working Condition 2 < Working Condition 5, which proves that the increase of fan frequency greatly improves the ventilation efficiency of the broiler chamber. It can be seen from the contours of the plane at Y = 0.13 m that the convection area of the inlet was contrasted with the inner side. The ages of air at the inlet and outlet were younger, and fresh air changed faster. The age of air inside the broiler chamber was older than that at the air duct. Working conditions 1 and 2 need to use internal circulation. The age of air in the middle of the broiler chamber was younger than that in working condition 5. The internal circulation promotes the air circulation in the plane. The fresh air with the Z = 0.2 m section first reached the outlet and then reached the internal circulation outlet. The wall corner of the internal circulation inlet continued to move in the two directions of the internal circulation pipeline and wall climbing. The age of air was old, and there were vortexes and weak wind velocity, which is not conducive to the replacement of fresh air. The wall near the internal circulation inlet had an older age of air. At the bottom of section Z = 1.65 m, the air inlet and outlet realized convection ventilation, which was the most efficient area for ventilation in the broiler chamber. Through velocity analysis, the wind velocity was relatively high, and the vortex was generated upward through wall–floor collision, which made a great contribution to the renewal of the upper air.

Yu et al. [23] analyzed the age of air of the cabin kitchen with a length of 5.6 m and a width of 3 m. The air in the cabin was highly disturbed, and the air freshness was higher. Our results showed that the age was within 60 s, the oldest age of air was 255 s, which appeared in the corner, and the age of air in the air inlet and outlet of the broiler chamber was also within 60 s. The oldest age of air in the broiler living area was about 100 s under the working condition 1 and 150~180 s under the working condition 4, which was within a reasonable range. This showed that this ventilation system was conducive to timely updating of the fresh air inside the broiler chamber and to improving the living environment of the broilers.

#### **4. Conclusions**

The conclusions of this research are summarized as follows:

(1) Through actual measurement, the three-dimensional model of the broiler chamber was established and simulated in Fluent software. The comparison between the simulation results and the actual measurement results showed that the RMSE of the wind velocity at each measurement point was up to 19.1%, and the maximum absolute error was 0.27 m/s, which shows the effectiveness of the CFD model.

(2) The wind velocity and the age of air in the broiler chamber were simulated and analyzed. The design of setting different frequencies for the fan met the requirement of setting appropriate temperatures in different periods. In the high-temperature area (above 30 ◦C), the wind velocity at the air inlet and outlet was close to 2 m/s, which is beneficial to the cooling of broilers. The internal wind velocity could be regulated above 0.8 m/s, which meets the ventilation requirement in summer. The

age of air of the broiler chamber could be renewed within about 100s at the bottom of the chamber in a relatively high-temperature environment. When starting the working condition 5 at night, ventilation could be carried out in all parts of the chamber within 150~180 s. Compared with the residential ventilation using fans, the age of air of the broiler chamber in this research was much younger, so the design of the ventilation system in the broiler chamber is superior to the common residential standard.

(3) The height of the air inlet and the center of the fan in the broiler chamber were set to be the same as those in the empty chamber. When the broiler chamber is mechanically ventilated, a large amount of airflow circulates in the lower part, effectively improving the ventilation and cooling effect. As a result, more airflow passes through the surface of the broiler chamber. As the density of polluted gas is small, the polluted gas mostly gathers at the top, and needs to be circulated to the bottom through the vortex and then blown out from the air outlet, thus reducing the efficiency of pollutant emission. Based on this design, an exhaust fan can be applied to the upper part of the broiler chamber to improve the efficiency of pollutant emission.

**Author Contributions:** X.Z. and X.Q. conceived and designed the experiments; X.Z., S.Z. (Shikai Zhang), A.D., B.F., S.Z. (Shixiu Zhang), and Y.Q. performed the experiments and analyzed the data; S.W., H.Y., and Y.W. helped perform the data analysis; S.Z. (Shikai Zhang), A.D., and X.Z. wrote the paper.

**Funding:** This research was funded by the Fundamental Research Funds for the Central Universities of China (KYTZ201661), China Postdoctoral Science Foundation (2015M571782), Jiangsu Agricultural Machinery Foundation (GXZ14002), and University Student Entrepreneurship Practice Program of Jiangsu Province (No. 201810307010P).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **An IoT Integrated Tool to Enhance User Awareness on Energy Consumption in Residential Buildings**

#### **Marco Dell'Isola 1, Giorgio Ficco 1, Laura Canale 1,\*, Boris Igor Palella <sup>2</sup> and Giovanni Puglisi <sup>3</sup>**


Received: 23 October 2019; Accepted: 23 November 2019; Published: 26 November 2019

**Abstract:** Unaware behaviors of occupants can affect energy consumption even more than incorrect installations and building envelope inefficiencies, with significant overconsumptions widely documented. Real time data and an effective and frequent billing of actual consumptions are required to reach an adequate awareness of energy consumption. From this point of view, the European Directive 2012/27/EU already imposed the use of metering and sub-metering systems, setting the minimum criteria for billing and related information based on real energy consumption data. To assess the ability of buildings to exploit new information and communication technologies (ICT) and sensitize both landlords and tenants to related savings, the new European Directive 2018/844/EU promotes the use of a smart readiness indicator. At the same time, basic information about indoor thermal comfort should be also gathered, aimed at avoiding that an excessive saving tendency can determine the onset of issues related to excessively low internal temperatures. In this paper, the authors address the problem of gathering, processing, and transmitting energy consumption and basic indoor air temperature data in the framework of an Internet of Things (IoT) integrated tool aimed at increasing residential user awareness through the use of consumption and benchmark indexes. Two case-studies in which thermal and electrical energy monitoring systems have been tested are presented and discussed. Finally, the suitability of the communication of energy consumption in terms of temporal, spatial, and typological aggregation has been evaluated.

**Keywords:** user awareness; energy consumption; individual metering; feedback strategies; N-ZEB; IoT

#### **1. Introduction**

Encouraging energy savings in residential buildings has been a topic of scientific interest since the 1970s, when the energy crisis made people aware of the possible exhaustion of fossil fuels. Almost 50 years later, it is clear that all intervention addressed to improve the energy efficiency of buildings should be combined with actions aimed at increasing the awareness and participation of end users, also through more frequent and detailed information on energy consumption [1]. In the absence of frequent information, two buildings with similar thermo-physical characteristics and energy performances, even designed consistently with N-ZEB (Net-Zero Energy Buildings) criteria, can consume one twice the other depending upon the occupants' behavior [2]. In recent years, smart devices and information and communication technologies (ICT) allowed the possibility to set up integrated systems to support decisions at the building, district, and city stages, but they are not common owing to the complexity of the problem, the reduced interoperability among the different systems, and high costs [3]. In addition, the effectiveness of user's feedback actions addressed at the energy savings is a still debated topic in the scientific literature.

On the basis of the results of 38 different studies addressing the effectiveness of interventions aimed at encouraging families to reduce their energy consumption [4], two macro-categories are identified, depending on the kind of information provided to families: (i) antecedent strategies; (ii) consequent strategies. Antecedent strategies include media campaigns, workshops, educational conferences, and energy audits for targeted and personalized information [5,6]. It is proven that antecedent strategies raise user awareness, but do not necessarily lead to behavioral changes or sure energy savings. This category includes any type of user feedback for example, real-time feedback, information presented via in-home displays, mobile apps, or online services [7–9]. Feedback actions can be direct, when learned directly from the instrument display (meter, sub-meter, and so on) or indirect, when information on data consumption is preliminary processed before reaching the user. The inconsistencies in behaviors related to the use of energy in families are the result of the following reasons [10]: (i) temporal coherence of decisions, (ii) difficulty in processing consumption data and in assuming simple decisions, and (iii) effects of presentation.

The available literature identifies three key problems related to the feedback: (i) the poor evidence of effectiveness, (ii) the need for involving users, and (iii) the potential occurrence of unwanted consequences. The main finding is that actual in-home displays could not be effective in orientating users' behaviors. Thus, it is necessary to develop and test novel feedback devices accounting for the degree of user involvement [11]. In a recent experimental campaign [12], many interviewed users reported difficulties in the interpretation of the units (kW, kWh) and poor feedback (e.g., lack in the corresponding economic value). This research also highlighted the usefulness of presenting disaggregated data for each device (sub-metering), at least for the most energy-consuming devices (stove, oven, dishwasher, washing machine, dryer, and so on) and of benchmarks with historical consumption.

In this scenario, low-income families, such as those living in public housing, are a particular category of users to be approached in a specific way. A recent experimental study in seven European Union (EU) countries highlighted several problems both in the implementation of smart-metering solutions and in the use of personalized feedback for low-income families in the Mediterranean region [13]. The experimental results proved that the use of smart-meters associated with in-home displays is not so effective. On the other hand, the monitoring of individual electrical devices, the allocation of consumption inside the dwelling, and suggestions for energy retrofits are appreciated. The joint implementation of these measures and the personalization of user feedback resulted in electricity consumption savings varying in the range from 22% to 27% [14]. Unfortunately, the adoption of energy saving strategies in social housing could lead to a potential worsening of comfort conditions [15]. For example, the reduction of the average winter indoor air temperature could result in condensation phenomena and mold. In the same research paper, the authors also point out that providing end-users with information about their energy consumption is more effective when people live in relatively energy-efficient dwellings, but is less useful for users living in public housing.

Rating the performances of energy devices and benchmarking energy consumption is a highly debated topic, as they could represent effective tools to help users in managing their energy costs. In an interesting guidance on energy consumption benchmarks on residential customers of the Australian Energy Regulator [16], indications about displaying the most appropriate electricity consumption benchmarks are given. In the same document, it is suggested that benchmarks tailored on household size, climatic zone, type of heating system, seasonal factors, and so on are more effective and better explain the variability of users' energy consumption. Benchmarking methodologies can be classified in four different categories [17,18]: (i) regression model-based, requiring extensive data sets of buildings having similar characteristics; (ii) points-based rating system, comparing the measured energy consumption to best practice standards; (iii) simulation model-based, in which energy consumption is modeled in the building energy simulation environment; and (iv) hierarchal and end-use metrics, in which set of performance metrics is developed for the underlying system performance data. Kavousian

et al. [19] presented a method to rank residential buildings based on their appliance energy efficiency, as the definition of a scale indicating the position within a distribution is considered to be more definitive than a comparison to a simple value. A literature review on up to date energy benchmarking methods and their performance levels is also provided in the work of [20], where twelve methods for benchmarking building energy consumption, including six black box methods, two gray box methods, and four white box methods, are analyzed, highlighting that many methods, although simple, can still achieve satisfactory performance. In the work of [18], research projects, tools, and programs focused on energy benchmarking methods, energy rating procedures, and classification schemes for the building sector are also reviewed and discussed.

Moreover, these technologies may also result in a series of issues related to the access to confidential information on users' activities and habits, privacy, confidentiality, and availability of data (also considering the greater quantity and vulnerability of data). This is for authorized parties (e.g., utility companies, metering companies); unauthorized parties (e.g., competitors, thieves, real estate owners); and, finally, for final users who are often unable to access and use their own data. In addition, the remote billing brings up problems related to data security and integrity (e.g., the risk of deletion/modification of information). Different privacy preservation techniques may be based on information theory, multiple source energy engineering, and cryptographic network protocols.

It is thus clear that, to achieve energy saving through more frequent and detailed information to the user via individual metering devices and smart technologies, a strategic feedback design is needed.

In this framework, the paper is aimed at addressing the problem of measuring, processing, and transmitting energy consumption data to final users through a suitably designed feedback strategy, without the intention of investigating deeper the social/behavioral causes underlying the eventual energy consumption variation. The development of tailored benchmark indicators applicable to metering and sub-metering in residential buildings is designed and proposed with the aim of making the information about energy consumption to the user more understandable. Three case-studies for thermal and electrical energy are presented and discussed. Finally, the communication strategy about energy consumption in terms of temporal, spatial, and typological aggregation is evaluated.

#### **2. Architecture and Data Transmission**

The integrated Internet of Things (IoT) tool developed by the authors is based on three levels. The first level is represented by metering and sub-metering systems for gathering energy consumption data of electrical, thermal, and natural gas devices (nodes) of the relative plants. The second level is the data concentration by wireless personal area networks (ZigBee protocol) and remote transmission data with the home router connected to the Internet. Smart meters may also directly communicate with the cloud. The third level is the web-based data management providing parallel solutions for data entry, storage, analysis, and processing. In particular, in this latter level, data for user feedback are processed by creating reports (e.g., indirect feedback), as well as real-time displaying via dashboard (e.g., direct feedback). Therefore, the IoT tool combines and stores information and data, as follows:


Figure 1 shows a simplified sketch of the developed IoT integrated tool.

**Figure 1.** Internet of Things (IoT) integrated tool scheme.

In the following, for the sake of completeness, the metering and sub-metering systems generally employed in measuring energy consumptions within residential houses and buildings are further detailed.

#### *Level 1: Smart Metering and Sub-Metering Systems*

As well known, residential buildings are supplied by several energy carriers and their technical plants provide the energy necessary for the fundamental activities performed. For simplicity reasons, for classification of smart metering and sub-metering devices, the authors will refer to the "typical" (i.e., more widespread) configuration of technical plants supplying residential buildings in Italy [21,22]:


Following the above-described classification, metering and sub-metering of the heating system allow the control and monitoring of energy consumption of heating and domestic hot water [23–28]. At the metering level, the smart direct thermal energy meter directly measures the consumptions of the boiler for heating and hot water production services at the building level. On the sub-metering level, the energy consumed by each apartment could be allocated through (smaller) thermal energy meters in the case of ring configuration of the plant. Otherwise, consumptions of single radiators, fan coils, or generic heating elements, when a vertical raising main configuration of the plant is present, are estimated through the so-called indirect accounting devices, such as two-sensors electronic heat cost allocators or insertion time counters. Additionally, the sub-metering of domestic hot water can be performed through a direct thermal energy meter or, alternatively, through a water meter suitable for hot water measurements.

Referring to the electrical system, different measuring devices are available at the metering level: smart electric energy meters, generally made up of a static type sensor with an associated processing system, are employed for fiscal purposes (i.e., metering and billing), but smart non-fiscal devices for monitoring current flows via the electric system phase are also available on the market. Electricity sub-metering multifunction devices also allow to monitor and control energy consumption of single appliances (e.g., refrigerator, dishwasher, oven, hairdryer) and they are often associated with the so-called smart plugs.

The natural gas system generally supplies natural gas for cooking purposes as well as to the main autonomous boiler for heating and, in some cases, the production of domestic hot water. Gas smart meters can be employed to measure the total gas volumes supplied to the house/building at the metering level. In this case, the meter is associated with temperature and pressure sensors, whose signals are processed by an electronic calculation module. On the market, "hybrid" smart gas meters equipped with electronic correction/transmission modules and static ultrasonic or thermal mass are also available. For sub-metering functions (e.g., cooking), small domestic gas meter can be used (e.g., class G2.5), but optimal operational conditions should be adequately considered, as the measured flow rates are often very low. Table 1 shows the technical specifications of the metering and sub-metering systems used by the authors in the case studies.


**Table 1.** Technical characteristics of metering and sub-metering (case studies): accuracy class as per measuring instrument directive (MID).

\* fiscal; \*\* non-fiscal.

In this research project, basic information about indoor air temperature measurements in one or more zones of the apartments was also given. The aim was to allow the end-user to know in which way his energy saving behaviors were affecting, positively or negatively, the indoor air temperature, which was chosen in the present experimental campaign as a simple parameter that the end-user could directly relate to his perceived thermal comfort. In this way, the user could assess if eventual changes in his behavior toward energy savings (i.e., the closure of one or more radiators, no or excessive ventilation [29]) could be related to low indoor temperatures (e.g., below 18 ◦C), which represents an undesirable consequence to be avoided.

#### **3. Methods**

#### *3.1. The Buildings Case-Studies*

For the experimentation of feedback strategies on the consumption of thermal energy for heating, an experimental campaign is currently underway in two social housing buildings belonging to the Italian Territorial Agency for Social Housing (ATER), served by a centralized natural gas system and in a detached house all located in the district of Frosinone (Central Italy). The social housing buildings (building #1 and building #2), both built in the 1970s, have very low energy performance and would require relevant energy retrofit intervention, both to improve the insulation of the building envelope and to increase the efficiency of the heating plant. End-users are mostly low income and elderly, mainly living in single- or two-family units, with limited ability to interact with automation systems, and can be considered a representative sample of the typical Italian social housing user. Buildings #1 and #2 are almost identical in terms of constructive characteristics, inhabitants, floor areas, plan scheme, and

heating system; the only exception is that one of the buildings consists of eight dwellings while the other has nine.

In each building, a thermal energy meter for the direct measurement of the thermal energy produced by the boiler (metering level) and two different indirect heat metering systems were installed (submetering level): (i) insertion time counters compensated with fluid temperature and thermostatic electronic valves controlled by programmable thermostat (building #1); (ii) two-sensor electronic heat cost allocators, mechanical thermostatic valves, and a programmable thermostat (building #2).

With regards to electrical energy consumption, an experimental campaign is currently underway in a detached house (building #3) located in the district of Frosinone (Central Italy) built in the first decade of the 2000s and inhabited by a family of four people. The house is a two-floor detached building, divided into two apartments, of which only one is actually inhabited by the family, but both are served by the main electrical energy meter with a maximum power installed of 4.5 kW. A current clamp meter was installed on the main power line of the sole inhabited apartment (metering level) in the fuse box of the house, whereas on the sub-metering level, two different devices were installed: (i) current clamp meter on the main light's powerline in the fuse box of the house; (ii) smart plugs on the more energy consuming electrical appliances.

In Figures 2 and 3, only two of the investigated buildings are depicted, as buildings #1 and #2 are almost identical.

**Figure 2.** Case-study building #1.

**Figure 3.** Case study building #3 with location of devices (red dots: smart-plugs, green dots: current clump meters).

Specific survey questionnaires were administrated to the inhabitants of the buildings, to assess user's attitude to adopt energy saving strategies and to interact with monitoring and control systems. In the survey, the user could assign a vote to seven statements, attributing a grade from 1 (not at all in agreement) to 10 (fully agreed), based on their level of concordance with the statement. For each

question, there was also the option "I don't know". Each user was also given the opportunity to express general considerations in a special note field. The questions given to families are listed below:


A number of 18 family members (i.e., one for each of the families taking part to the experimental campaign) took part to the survey, namely, eight from building #1, nine from building #2, and one for building #3. In Figure 4, information about the composition of the investigated sample of families is given, for a calculated average number of family members of 2.6 people per dwelling (close to the national average of 2.4 [21,22]). Questionnaires were provided at the end of the first heating season following the installation of the metering and temperature control devices. Authors required answers from the householders (i.e., the person or the people managing the house).

**Figure 4.** End-users living in investigated buildings, sample composition.

#### *3.2. Designing Information Strategies*

Effective strategies for improving end-users' awareness in "smart homes" are influenced by numerous aspects, such as the following [30]: (i) the quality of the perceived interaction (e.g., speed, brevity/easiness); (ii) information efficiency (e.g., accuracy and completeness); (iii) usability (e.g., ease of use, intuitiveness, user satisfaction); (iv) the aesthetics; (v) the usefulness (e.g., offered functions); and (vi) acceptability (e.g., low cost, number of potential users). The feedback of monitored data should reach end-users over time and the most adequate way to allow the full understanding of the phenomenon, before it is irreversible or no longer visible, linking it to specific retrofit actions [31]. To identify the most effective feedback, the authors analyzed the features shown in Table 2.

In the technical practice, the simplicity and/or cost of information system is sometimes favored, while in others, the completeness and/or the effectiveness of the information is favored. The different types of feedback can have very different costs and customer satisfaction levels, but a crucial issue should be the awareness and immediacy of information to lead users at performing higher energy savings. A unanimous judgment of end-users is the greater appreciation of a detailed, frequent, and actual feedback. Therefore, authors decided to differentiate between direct and indirect feedback: (i) by using frequent, synthetic, and immediate information in the case of direct feedback; (ii) by providing detailed and disaggregated information for each consumption area (i.e., bedrooms, living, bathroom, kitchen), for each energy carrier (i.e., thermal energy, electrical and natural gas), and for the device/system in the case of indirect feedback.


**Table 2.** Types of feedback.

Other aspects positively evaluated in the technical literature are the diagnosis of faults and malfunctions, the comparison with historical consumption, simplicity, and effectiveness in understanding user information. Therefore, for indirect feedback and for each consumption area, the authors presented the following: (i) historical consumption benchmark, (ii) benchmark with average consumption of other users (building average), and (iii) theoretical expected consumption obtained on the basis of the specific characteristics of the user (e.g., characteristics of energy systems, type of user) and of climate data. To enhance the communication effectiveness, pie charts (for allocation) and bar charts (for comparisons with previous periods and with other users) were prepared.

#### *Benchmarking Indices and Energy Saving Tips*

Suitable tailored benchmarking indices were built by the authors as the ratio between the measured and the expected energy consumption of the room/appliance, as per Equation (1).

$$\text{Benchmarking index} = \frac{\text{measured consumption} - \text{expected consumption}}{\text{measured consumption}} \cdot 100\tag{1}$$

In the following, the benchmarks designed for the case studies (i.e., heating and electricity services) are presented and discussed separately, owing to the corresponding peculiarities.

In the heating service case studies (buildings #1 and #2), the performance benchmarks were calculated as the ratio between the energy consumption measured at actual conditions of use (i.e., operational rating) and the estimated primary energy consumption adjusted to the actual conditions of use and climate (i.e., tailored rating), for each room, apartment, and the whole building [32].

A different approach was adopted for the electricity case study (building #3). In fact, it is well known that energy consumption of an appliance strongly depends on its use, which, in turn, relies on the number of family components, characteristics of the house (e.g., floor area and outdoor spaces), and on the end-user (e.g., income, work, age, presence of children and/or elderly people). These characteristics are then employed to build reference baseline values within certain descriptive distributions and adopted for benchmarking purposes. On the other hand, the benchmarking methodology proposed by the authors in this paper is based on a simplified hybrid approach. The expected electricity consumption

is built with a "bottom-up" approach considering each electrical appliance installed in the house. Basic information is collected from the technical booklet of the appliances (i.e., energy consumption for cycle/energy labels/power) and then used to calculate the expected consumption through statistical data on the use of household appliances. In fact, for the purposes of the present research, it was necessary to use a simple representative indicator for the expected energy consumption of each electrical appliance. In this way, the methodology could be simply implemented in an IoT application for energy monitoring. To this aim, it was decided to consider the number of family components as the most representative parameter for the estimation of the expected energy consumption of electrical devices. This hypothesis, although introducing a certain level of simplification (e.g., for the calculation of consumption for lighting), is considered to be reliable for the calculation of the expected consumption of large electrical appliances, as the number of family members directly impacts their frequency of use. In this context, in order to determine the expected energy consumption of each electrical appliance, the authors made a preliminary analysis of statistical data about electrical energy use from the Italian National Institute of Statistics (ISTAT) and Italian National Agency for New Technologies, Energy, and Sustainable Economic Development (ENEA). In particular, regarding the Italian energy consumption, ISTAT provides statistical data about the following: (i) number of cycles per week for given electrical appliances for different numbers of family components; (ii) lights turn-on period; and (iii) expected expenditure for different numbers of family components.

In particular, Figure 5 shows the data about dishwasher and washing machine usage per week and the trend of the expenditure for electrical energy of different family sizes.

**Figure 5.** Data usage for main appliances [21,22].

Suitable usage coefficients were calculated by normalizing all the statistical data about energy use/expenditure with respect to reference number of family components (i.e., 2.4 for Italy [21,22]). This means that, for example, the expenditure coefficient for a family of four components was determined by dividing the expected statistical expenditure (estimated to be about 1908 €) for the expenditure expected for a family with 2.4 family components (i.e., about 1706 €). Usage coefficients were employed as a base to calculate, for each electrical appliance, the expected time of use (expressed in hours) in the reference period (year). The expected energy consumption is then built either per Equation (2) or per Equation (3), depending on the type of appliance and on the data available in the technical booklet.

$$\text{expected consumption} = \text{num}\_{\text{c, w}} \cdot \text{52} \cdot \text{cons}\_{\text{cycle}} \tag{2}$$

$$\text{expected consumption} = \text{LIC} \cdot \text{num}\_{h,w} \cdot \text{52-power}\_{\prime} \tag{3}$$

where *numc*, *w* is the number of weekly cycles of the given electrical appliance (calculated as a function of the usage coefficient [21,22]), *conscycle* is the electrical energy consumption per cycle of the appliance declared by the manufacturer and retrievable in the technical booklet (expressed in kWh/cycle), *UC* is the usage coefficient of the appliance, *numh*, *<sup>w</sup>* is the expected weekly hours of use of the appliance, and *power* is the declared power of the appliance expressed in kW. Note that the use of Equation (2) was applied to all of the electrical appliances whose energy consumption per cycle was given in the technical booklet (e.g., dishwasher, washing machine, dryer), while Equation (3) was employed for all the other appliances. Number cycles/functioning hours per week were determined as the weighted average of the statistical data given in the work of [21,22] or, in the case in which data were not available, by making suitable assumptions.

In this context, the baseline values (i.e., expected consumption) employed to calculate the benchmarking indicators will be updated as soon as a change is made in electrical devices (i.e., replacement of an appliance with a new one, replacement of light bulbs), in the insulation of the building, or if the number of family components changes at some point.

For the present analysis, the tips presented in the indirect feedback were selected from a set of suggestions expressly provided by the ENEA as part of the dissemination actions about energy efficiency in residential buildings [33]. The set of selected tips is general advice targeting the correct management of each heating, cooling, lighting, and electrical systems, and was given when the measured consumption of each room/appliance was found to be higher with respect to the expected one (i.e., calculated energy consumption).

Finally, two new direct and indirect feedback strategies were adopted by the authors based on the following aspects: (i) the results of the survey; (ii) the personal interaction between the authors and the participants of the experimental campaign, which occurred during the informative meetings; (iii) the characteristics of the installed energy systems and measuring devices; and (iv) the analysis of existing scientific literature regarding the best practices in feedback about energy consumption.

#### **4. Results and Discussions**

Through the administration of specific designed surveys in the ATER buildings, the authors gathered useful information about users' attitudes to adopt energy saving strategies and to interact with monitoring and control systems. The response rate to the questionnaires provided was 100%. Figure 6 shows the list of the questions together with the overall analysis of the answers obtained. With regard to the installation of monitoring and control systems, the users, although they declare themselves satisfied (100%) and quite familiar with such systems (64%), were wary of the potential effectiveness in terms of savings (71%). As for indoor temperature perception, most users feel that they do not perceive too high (71%) or too low (78%) indoor temperatures.

The characteristics of the feedback designed within this work are summarized in Table 3.

#### *4.1. Direct Feedback*

The dashboard built for direct feedback (daily frequency) is made up of two sections. The first one for sub-metering shows the energy consumption (kWh and %) of each room/appliance, using a bar graph. In the second section (metering), through a multi-scale display, the user can simultaneously access the energy data consumed by the apartment (in kWh and in €) and the corresponding CO2 emitted (in kg). In this way, the user receives, in real time, information about his own energy consumption, the related costs, and the environmental impact, as well as on their distribution among different environments, leading, at the same time, to adopting efficient behaviors at the energy, economic, and environmental levels. A daily frequency of this feedback was chosen by the authors owing to metering and sub-metering devices' characteristics and to the related costs of data transmission (e.g., battery consumption). Figure 7 shows the dashboard developed by the authors to display the daily energy consumption of a typical user both for heating and for electrical energy consumption.

**Figure 6.** Results of the analysis of surveys.


**Table 3.** Technical specification of direct and indirect feedback.

#### *4.2. Indirect Feedback*

As above reported, through indirect feedback (monthly frequency), users also receive information on their consumption in terms of performance indices, personalized suggestions, comparison with historic consumption, and so on. In this way, the authors also tried to capture the users' attention by using emoticons, colors, and "user friendly" information.

**Figure 7.** Direct feedback: (**a**) heating, (**b**) electricity.

To all users involved in the experimentation, the authors provided indirect feedback sheets, collecting impressions and suggestions. In particular, from the meetings carried out, the following emerged:


Figures 8 and 9 show the form designed by the authors for indirect feedback of heating divided into six sections:


**Figure 8.** Indirect feedback for heating service.

Figure 9 shows the informative sheet for indirect feedback of electrical energy consumption.

In the electrical energy case study (Figure 9), no comparison with other user was possible, as the family lives in a detached house. A ring chart shows the share of energy usage for each monitored appliance in the reference period (month).

Tables 4 and 5 show the usage coefficients calculated as described in the methodology section for the available data [22] and the calculated expected energy consumption of each of the electrical appliances installed in the building #3 case study, respectively.

**Figure 9.** Indirect feedback for electrical energy.




**Table 5.** Expected energy consumption of main electrical appliances of the analyzed case-study (ISTAT, 2014).

<sup>a</sup> calculated based on the works of [21,22]; <sup>b</sup> hypothesized based on user's declaration; <sup>c</sup> the expected consumption of the refrigerator is the declared annual energy consumption from the energy label.

It is underlined that the expected consumption in the present research has a high degree of tailoring, as it is assumed that the user has preliminary specified the technical characteristics (i.e., energy consumption per cycle and/or energy label) of each appliance installed. It is thus possible to state that the benchmark indices presented in this research are built with a tailored bottom-up approach based on the actual characteristics of the house analyzed and not with a top-down approach (i.e., a statistical approach), thus replicating the commonly applied methodology to determine the "tailored rating" for heating energy consumption applications.

#### **5. Conclusions**

In this work, the authors presented the first results of a study aimed at designing a proper feedback strategy for IoT energy consumption monitoring systems, so as to increase user awareness on energy consumption. In this framework, two case studies were presented and discussed related to heating and electrical energy consumption monitoring in residential buildings and specific tailored benchmarks indices were developed to present energy consumption data at the sub-metering level. To this aim, frequent informative meetings were carried out with a total of 18 end-users in three different buildings.

The direct interaction between the authors and the end users over the campaign, together with the provision of simple questionnaires, showed that the huge amount of measured data and the complexity of the monitored systems make analysis and feedback particularly complex for non-skilled users. Compatibly with the scientific literature, most of the users were not able to understand the measuring units expressed in kWh (or similar), while they showed particular interest in the consumption data expressed in euro and in percentage.

The design and the content of the designed feedback strategy was greatly appreciated by the final users, whose attention on the desired information was captured through simple information (e.g., different colors and emoticons). Owing to the characteristics of the sample of users (mostly elderly people with limited ability to manage technical information), these last favored the simplicity and immediacy of information, which has been given in the following form:


All the users showed particular appreciation for the benchmarking indicators developed at the sub-metering level (generally not available neither in the energy bills nor in the energy monitoring systems available on the market), as this information allowed them to precisely locate the main energy consuming rooms/appliances of the house and to take specific actions in order to reduce their consumption.

Personalized suggestions were also given to specific participants whose consumption was found to be much lower than that expected, in order to make them aware that incorrect energy-saving practices, such as the closure of most radiators, can dangerously penalize the indoor air temperature, thus also affecting the perceived thermal comfort, which was not the purpose of the present campaign.

The adopted IoT technologies demonstrated high potential in terms of energy savings, as effective and frequent feedback contributes significantly to motivating and supporting change in occupant behavior. The analysis of the data showed some incorrect behavior that users were not aware of, such as excessive ventilation of some rooms (e.g., entrance, bathrooms, and kitchens), incorrect management of the thermostatic valves, and incorrect management of some domestic appliances.

It is highlighted that the benchmarking methodology developed by the authors for the purpose of the present campaign could be very useful for researchers and software developers looking for targeted strategies for energy benchmarking. Referring to benchmarking for electricity consumption, in fact, the authors proposed a simplified, detailed approach to estimate the energy consumption of a given appliance, which could be simply implemented in an IoT application for energy monitoring. In fact, this methodology requires the knowledge of the family composition and basic information about the appliance, retrievable by the technical booklet, by specific databases, or directly by the smart device.

Further progress of the present research is represented by the development, currently ongoing, of specific diagnostic routines for natural gas, electricity and heating systems that, based on the load profiles measured by the system, will allow the detection of possible faults or incorrect management, and will automatically recommend the most suitable personalized suggestion.

**Author Contributions:** Conceptualization, M.D. and G.F.; methodology, M.D., G.F., and L.C.; formal analysis, L.C. and B.I.P.; investigation, M.D., G.F., and L.C.; resources, M.D. and G.P.; data curation, L.C.; writing—original draft preparation, M.D., G.F., L.C., and B.I.P.; writing—review and editing, B.I.P. and G.P.; supervision, M.D., G.F., and G.P.; project administration, M.D. and G.P.; funding acquisition, M.D. and G.P.

**Funding:** This work was developed under the projects "Ricerca di Sistema Elettrico PAR 2016" funded by ENEA (grant number I12F16000180001) and "PRIN Riqualificazione del parco edilizio esistente in ottica NZEB" funded by MIUR (grant number 2015S7E247\_002). The authors wish to thank ATER of Frosinone, the Territorial Agency for Social Housing, for the technical support during the on field experimental campaign.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Hybrid Ventilation System and Soft-Sensors for Maintaining Indoor Air Quality and Thermal Comfort in Buildings**

**Nivetha Vadamalraj 1, Kishor Zingre 2,\*, Subathra Seshadhri 1, Pandarasamy Arjunan <sup>3</sup> and Seshadhri Srinivasan <sup>3</sup>**


Received: 9 December 2019; Accepted: 31 December 2019; Published: 16 January 2020

**Abstract:** Maintaining both indoor air quality (IAQ) and thermal comfort in buildings along with optimized energy consumption is a challenging problem. This investigation presents a novel design for hybrid ventilation system enabled by predictive control and soft-sensors to achieve both IAQ and thermal comfort by combining predictive control with demand controlled ventilation (DCV). First, we show that the problem of maintaining IAQ, thermal comfort and optimal energy is a multi-objective optimization problem with competing objectives, and a predictive control approach is required to smartly control the system. This leads to many implementation challenges which are addressed by designing a hybrid ventilation scheme supported by predictive control and soft-sensors. The main idea of the hybrid ventilation system is to achieve thermal comfort by varying the ON/OFF times of the air conditioners to maintain the temperature within user-defined bands using a predictive control and IAQ is maintained using Healthbox 3.0, a DCV device. Furthermore, this study also designs soft-sensors by combining the Internet of Things (IoT)-based sensors with deep-learning tools. The hardware realization of the control and IoT prototype is also discussed. The proposed novel hybrid ventilation system and the soft-sensors are demonstrated in a real research laboratory, i.e., Center for Research in Automatic Control Engineering (C-RACE) located at Kalasalingam University, India. Our results show the perceived benefits of hybrid ventilation, predictive control, and soft-sensors.

**Keywords:** indoor air quality (IAQ); hybrid ventilation; demand controlled ventilation (DCV); internet of things (IoT); soft-sensor; convolution neural networks

#### **1. Introduction**

The building sector in India currently contributes to ~37% of the total energy consumption of the nation and predicted to further increase by 8% annually due to recently proposed construction of 40 billion m2 by 2050 which are driven by rapidly growing population and urbanization [1]. Statistics suggest that space cooling alone contributes to about 40–45% of the total building energy consumption in India. Consequently, energy optimization for space cooling maintaining thermal comfort has attracted significant attention recently (see [2,3] and references therein). While the importance of energy consumption is often exacerbated, the indoor air quality (IAQ) is not usually discussed [4]. However, IAQ

is an important parameter that determines the productivity and performance of occupants in the building. The IAQ refers to the air quality within and around buildings and structures, especially it relates to the health and comfort of building occupants (https://www.epa.gov/indoor-air-quality-iaq/introductionindoor-air-quality). The American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) defines IAQ as Air in which there are no known contaminants at harmful concentrations as determined by cognizant authorities and with which a substantial majority (80% or more) of the people exposed do not express dissatisfaction (http://cms.ashrae.biz/iaqguide/pdf/IAQGuide.pdf?bcsi\_scan\_ C17DAEAF2505A29E=0&bcsi\_scan\_filename=IAQGuide.pdf).

It is widely perceived that understanding the common pollutants and controlling them could reduce health concerns, which can either be short- or long-term. Common short-term effects include dizziness, headaches, and fatigue, whereas long-term effects could be respiratory diseases, heart diseases, and even cancer. There are many surveys which have shown that both IAQ and thermal comfort are not well maintained by current ventilation systems (see, for example, in [5–8]). This is mainly because IAQ, thermal comfort, and energy consumption are competing objectives. As an increase in IAQ means more fresh air induction which will increase the energy consumption and can lower down thermal comfort as well. Therefore, maintaining IAQ, thermal comfort and minimizing energy is a major challenge in buildings, but important for achieving energy goals as well as occupant performance in buildings.

There are two procedures commonly used for maintaining IAQ: (i) Ventilation Rate Procedure and (ii) Indoor Air Quality Procedure [9]. The later aims to strike a good balance between energy savings and IAQ. Furthermore, it provides direct reduction of indoor contaminants. Similarly, in practice, there are three approaches to improve IAQ: (i) source control, (ii) improved ventilation, and (iii) air-cleaners. In source control, we eliminate individual sources of pollution or reduce their emissions and is considered an effective method for maintaining IAQ. Ventilation improvements means forcing fresh air into the buildings by opening windows and doors when weather permits, but this may lead to energy loos. Air cleaners aim to remove the particle from indoor air and uses filtering blocks to remove pollutants. Among these methods, forcing fresh air is by far the cheapest method for maintaining IAQ. However, optimal fresh air should be infused to reduce energy consumption.

Over the years, several control approaches for maintaining both IAQ and thermal comfort have been studied but with certain limitations. Among those, few studies implemented model predictive control (MPC) to investigate the combined (i) thermal comfort and CO2 optimization by regulating fresh air [10], (ii) IAQ (particle concentration) and energy optimization [11], (iii) multi-objective optimization of IAQ (particulate matter) and energy consumption in subway ventilation system [12], (iv) optimization energy and IAQ with focus on CO2 [13], and (v) energy optimization and air-quality through fresh air induction [3]. More recently, the role of demand controlled ventilation(DCV) on IAQ and energy savings was studied [14]. Another study implemented MPC for energy optimization and monitoring carbon dioxide in commercial HVAC systems with an Internet of Things (IoT) based control [15]. A combined control for IAQ, energy and thermal comfort was proposed in [16] for variable air volume (VAV) controlled systems. Similarly, thermal comfort and IAQ in chilean schools with surveys was studied in [17]. The control of IAQ in direct expansion air-conditioning system was studied in [18]. However, most of these investigations have concentrated only on temperature, humidity, and carbon dioxide levels while discussing IAQ.

As for sensors, a WiFi-enabled IAQ monitoring and control system for buildings was proposed in [19]. A soft-sensor for measuring carbon dioxide content in the building by fusing carbon dioxide and PIR sensor was proposed in [20]. A soft-sensor for estimating cooling load using long short-term memory (LSTM) was proposed in [21]. A review of sensors for measuring Indoor Air Quality (IAQ) was presented in [22]. There are many studies on sensor design for IAQ or studying sensor installation/monitoring issues reviewing which are out of the scope of the paper. More recently, the use of soft-sensors for IAQ modeling was studied in [23] wherein sensor readings was used to train deep subspace network to model IAQ parameters. A soft-sensor for detecting urban IAQ was proposed

in [24] using Bayesian networks. These results illustrated that by combining compact sensors with databased techniques capabilities of sensor could be overarched and they can be made more smarter. Albeit, such significant advantages, the role of soft-sensor has not been studied extensively.

A review of the literature reveals three gaps in control and monitoring in buildings. First, maintaining IAQ, thermal comfort and optimizing energy simultaneously is a challenge that has not been fully addressed in the literature. In particular, the ones concerning IAQ parameters such as Volatile Organic Compounds, Carbon-dioxide and others have not been combined with energy optimization based control and thermal comfort. Second, although low-cost sensors are available their reliability and performance is limited. To overcome this, soft-sensors which combine compact sensors with data analytics methods are gaining popularity. The advantage of soft-sensors is that they can also predict the variables and can also model variables which are not measured, e.g., cooling load. In addition, they provide smartness to the control. Third, the use of soft-sensors within control has not been reported. Our main objective is to address these research gaps in the literature by proposing a soft-sensor based ventilation method that could optimize energy consumption while maintaining IAQ and thermal comfort. The main contributions can be summarized as below.


The paper is organized into five sections. Section 2 presents the problem formulation and challenges. The hybrid ventilation system, soft-sensor design, and control techniques are presented in Section 3. Deployment results of the different components are presented in Section 4. Conclusions and future directions of the investigation are discussed in Section 5.

#### **2. Problem Formulation**

The objective is to design a novel hybrid ventilation system to maintain IAQ, thermal comfort in terms of set-point temperature and minimize energy consumption at the Center for Research in Automatic Control Engineering (C-RACE) at the International Research Center (IRC), Kalasalingam University, India. The lab houses faculty offices, startups in the building automation laboratory, research labs, researchers and visitors as shown in Figure 1. The current system consists of fans and variable refrigerant volume (VRV)-based air-conditioning system (package units) providing space cooling controlled by a thermostat which turns ON or OFF depending upon the user defined temperature limits. The lab has air quality issues due to the use of chemicals for making printed circuit boards in the startups, presence of human beings, particulate matter suspended due to tests conducted in electric vehicles laboratory and other contamination. Furthermore, some parts of the lab does not have air-conditioners rather fans for circulating air. There are exhaust fans for pushing the indoor air outside, but fresh air induction is currently only by opening doors.

The VRV systems are located in the electric vehicle lab (EV lab), building automation lab, research center on artificial pancreas, and software development lab. There is also the IAQ lab from where the hybrid ventilation system is to be implemented for maintaining IAQ, thermal comfort and energy optimization. Additionally, we plan not to make modifications to existing systems for example, the VRV system control or installing fresh air dampers in the space to avoid costly retrofits. In our analysis, the IAQ is modeled as

$$J\_{IAQ} = f(PM, CO\_2, TVOC, H, T) \tag{1}$$

where *PM*, *CO*2, *TVOC*, *H*, and *T* denote the particulate matter, carbon dioxide(measured in ppm), total volatile organic compounds (measured in ppb), humidity, and temperature, respectively. These

factors are time-varying and their dynamic is complex to model. The control variable that can influence the IAQ is the fresh air induction rate into the rooms *uFA*, the fraction of total mass flow rate of fresh air that can be supplied to the room *mFA*, and is given by

$$
\mu\_{FA} = (1 - d) \times \mathfrak{m}\_{FA} \tag{2}
$$

where (1 − *d*)*mFA* denotes the fresh air fraction that is supplied to the room. Similarly, to control the temperature and thermal comfort, we aim to change the control from being a simple thermostat to have additional degrees of freedom. This is required as there is no coarse control on the space possible with thermostat. To this extent, we could modify the time for which the air-conditioner is turned ON or OFF to modulate the average cooling power being supplied and this provides finer control over the temperature with pulse-width modulation sort of control. To this extent, we modify the control variable from being set-temperature to ON/OFF time. Assuming that the period for which the energy optimization is *Tp* which depends on the time constant of the room, then the control variable *uAC* models the time for which the AC is turned ON and is given by

$$
\mu\_{\rm AC} = \frac{t\_{\rm ON}}{T\_p} \tag{3}
$$

where *tON* is the time-period for which the AC stays ON. However, a turn-off AC could cause the air circulation to go down, thereby making occupants to feel thermal discomfort. This could be avoided by turning ON fans during such times. This procedure of turning on AC and fans, we term it as toggling. The energy optimization problem then could be implemented as a toggling action between the AC and fan. The duty ratio for the fan is

$$
\mu\_{\rm Fan} = \frac{T\_p - t\_{\rm ON}}{T\_p} \tag{4}
$$

We used eight fans in our work. Their heat generation was not considered in our study as they were very minimal compared to the room size and cooling load.

The energy consumption in AC and fan is given by

$$J\_{Power} = \mu\_{AC} \times P\_{AC} + \mu\_{Fan} \times P\_{Fan} \tag{5}$$

where *PAC* and *PFan* are the power ratings of the fan and air-conditioner, respectively. The thermal comfort of the occupant is modeled using the temperature bounds on the room which is given by

$$T\_{\min} \le T(k) \le T\_{\max} \tag{6}$$

where *T*(*k*) is the temperature at time-instant *k* and *Tmin*, *Tmax* denote, respectively, the minimum and maximum temperature supplied by the occupant based on their thermal comfort. Determining that the temperature is within the user defined comfort band requires model of the zone corresponding to the ON/OFF times of the air-conditioner and fan. However, fan does not vary the room temperature, but only the skin temperature of the occupant. Therefore, the zone temperature depends on the current room temperature *T*(*k*), ambient temperature *w*(*k*) and stray heating due to occupancy, lighting loads and others modeled as a disturbance term *v*(*k*). Following [2], we model the room temperature dynamics to be

$$T(k+1) = aT(k) - b\mu\_{AC}(k) + cw(k) + v(k)\tag{7}$$

where *a*, *b*, *c* denote the thermal time constant of the room, the parameter that models the influence of air-conditioner cooling energy, and effect of weather on the room temperature, respectively. Similarly, there exists a minimum and maximum time for which the air-conditioner can be turned ON/OFF. This is modeled as

$$
\mu\_{A\mathcal{C}\_{\min}} \le \mu\_{A\mathcal{C}}(k) \le \mu\_{A\mathcal{C}\_{\max}} \tag{8}
$$

where *uACmin* and *uACmax* denote the minimum and maximum duration for which AC can be turned ON/OFF for the considered duration. Based on the model, we select *Tp* to be 15 min as a change in control input gets reflected after 8–12 min of air-conditioner operation. The problem of maintaining IAQ, reducing energy consumption and maintaining user thermal comfort can be modeled as a multi-objective model predictive controller given by

$$J = \min\_{u\_{AC}, u\_{FA}} \sum\_{k=k+1}^{k+N\_P} f\_{IAQ}(k) + f\_{Pown}(k)\tag{9}$$

$$\text{s.t.}$$

Building thermal dynamics: *T*(*k* + 1) = *aT*(*k*) − *buAC*(*k*) + *cw*(*k*) + *v*(*k*), Thermal comfort constraints: *Tmin* ≤ *T*(*k*) ≤ *Tmax*, Control input limits: *uACmin* ≤ *uAC*(*k*) ≤ *uACmax* , Fresh air limits: *uFAmin* ≤ *uAC*(*k*) ≤ *uFAmax* , TVOC limits: 0 ≤ *TVOC*(*k*) ≤ *-VOC*, *CO*<sup>2</sup> limits: *min CO*<sup>2</sup> ≤ *CO*2(*k*) ≤ *max CO*<sup>2</sup> , PM limits: 0 ≤ *PM*(*k*) ≤ *-PM*.

where *-TVOC*,*-CO*<sup>2</sup> , and *-PM* are the bounds on the IAQ and could be computed as per the ASHRAE standard. The problem in (9) has the following challenges.


In what follows, we address the challenges above and modify the existing VRV system into an effective ventilation scheme.

**Figure 1.** C-RACE laboratory layout located at Kalasalingam University, India (D–Door, DD–Double Door).

#### **3. Novel Hybrid Ventilation System and Soft-Sensor**

To implement the MPC with multiple objectives, we first decompose the problem into two parts: (i) Energy optimizer and (ii) IAQ optimizer. Both the optimizers are coupled through the fresh air induction, which influences the thermal dynamics of the zone thereby the energy consumption. Our idea of energy optimizer is to design a device that switches between air-conditioner and fan to maintain thermal comfort but reduce energy consumption using a predictive controller. To this extent, we first require sensor measurements on temperature, humidity and occupancy (to detect cooling loads). Then, the next step is to design the energy optimizer based on the fresh air flow which will be discussed later.

#### *3.1. Sensor Module: Energy Optimizer*

The sensor module for the energy optimizer is shown in Figure 2. It consists of a temperature and humidity sensor which we use a DHT11 sensor which can measure both temperature and humidity. To measure occupancy, we use carbon dioxide sensor and passive infra-red (PIR) sensor. The carbon dioxide sensor has delays to measure the occupancy and PIR sensor have errors in counting. To overcome this, we build a soft-sensor on top of the sensor to validate the CO2 levels (see Section 3.3).

Sensor module: PIR, CO , Humidity and Temperature (DHT11) conditioner

Relay-board to switch between Fan and AirToggling unit and the sensors are interfaced to the Unit

**Figure 2.** The sensor and controller units used for energy savings.

The PIR motion sensor, carbon dioxide sensor, temperature, and humidity sensor are interfaced to RPi. The RPi is connected to the Ethernet and it acts as a gateway to transmit the sensed information to the server. We use the Message Queuing Telemetry Transport protocol at the application layer to transmit the sensed information. The server uses SQL routines to store the information in the database.

#### *3.2. Energy and IAQ Optimizer*

The second step is to implement the control that saves energy, maintains thermal comfort and guarantees IAQ. However, as mentioned earlier these are competing objectives and also there are challenges (C1)–(C5) that needs to be addressed. To address challenges (C1) and (C2), we first decouple the problem of maintaining IAQ and energy optimization plus thermal comfort by designing a hybrid ventilation system. Next, we solve the problems independently but exploit the power of IoT to couple them. In what follows, we describe the different elements used in our implementation.

The hybrid ventilation system consists of two parts: energy optimizer and IAQ optimizer. To decouple the multiple objectives but coordinate the actions, we select the fresh air infusion flow rate as the variable coupling the energy and IAQ optimizers. One can observe that the problem of fresh air induction is quite challenging as the function to approximate IAQ is time-varying and

depends on too many parameters. To overcome this, instead of designing a fresh air damper, we use a demand controlled ventilation system for maintaining the IAQ. In this paper, we select the RENSONs' Healthbox3.0 based on analysis performed with various scheme. The Healthbox3.0 is a DCV that controls fresh air induction based on TVOC. (measured in ppb), CO2 (measured in ppm), humidity, and temperature. The installation of the RENSON Healthbox 3.0 in the C-RACE lab is shown in Figure 3. Consequently, the DCV will introduce fresh air depending on the IAQ in the particular room. Still particulate matter is not considered and we install PM2.5 sensor additionally to measure this variable. The Healthbox 3.0 is installed in the IAQ lab of C-RACE and it samples air from the different labs to change the fresh air induction. The system has inbuilt sensors to measure the IAQ variables which could be transmitted to database through simple SQL scripts by writing the variables into an excel or CSV files.

**Figure 3.** RENSON Healthbox 3.0 DCV Installation.

The next step is to connect the DCV to the IoT and other interfaces. However, the sensors need to be calibrated. To this extent, we first install the mobileApp provided by RENSON and then the calibration is performed. The flow information about the fresh air is then passed on to the Gateway (RPi) which uses this information to optimize the energy and also for the soft-sensor to compute carbon dioxide. The hybrid ventilation scheme is shown in Figure 4.

As the IAQ objective is now decoupled with the DCV, the energy optimizer solves the following optimization problem to implement the MPC.

$$J = \min\_{u\_{AC}} \sum\_{k=k+1}^{k+N\_p} P\_{AC} u\_{AC} \tag{10}$$

s. t.

Building thermal dynamics: *T*(*k* + 1) = *aT*(*k*) − *buAC*(*k*) + *cw*(*k*) + *v*(*k*) Thermal comfort constraints: *Tmin* ≤ *T*(*k*) ≤ *Tmax* Control input limits: *uACmin* ≤ *uAC*(*k*) ≤ *uACmax* ∀*k* ∈ {*k* + 1, . . . , *k* + *Np*}

The problem in Equation (10) is a linear programming problem and we used Gnu Linear Programming Kit (GLPK) to implement the controller in RPi. The implemented controller obtained weather forecasts using webservices in Python, and the cooling load estimates were obtained based on the CO2 predictions from soft-sensor. To address challenge (C4), i.e., absence of coupling between energy and IAQ optimization, we use the flow information and this coupling is solved inherently. Moreover, the hybrid ventilation system also implemented the IoT architecture for realizing the controller addressing the challenge (C5). The only problem to be tackled now is the development of soft-sensor for IAQ which is presented in the rest of the section.

**Figure 4.** Schematic of the novel hybrid ventilation system.

#### *3.3. Soft-Sensor for IAQ and Cooling Loads*

As stated earlier, IAQ and cooling load due to fresh air induction due to DCV is a challenging problem due to complex dynamics of these variables. Finding dynamical equations for modeling it is a challenging task. Therefore, it makes perfect sense to develop models using databased methods to estimate the IAQ and cooling load. In this context, time-series data of IAQ variables is available from the hybrid ventilation system designed by us. Therefore, data required for building soft-sensors required for our application is available to us as shown in Figure 5. These are IAQ sensor readings (temperature, RH, CO2 and TVOC) collected from the Helathbox 3.0 for 10 days with a sampling interval of 15 min. To study the dependence of the parameters, we studied the correlation among the parameters (see, Table 1). One can see that there is no strong correlation between the parameters. Therefore, the univariate forecasting model is used to simplify our analysis. Similarly, the cooling load should be estimated from coupling variable fresh air flow in individual zones. However, cooling loads are also affected by occupancy, stray heating loads such as lighting, computers etc. Therefore, building soft-sensors for cooling loads using data-based methods requires rethinking. In this paper, we use the zone thermal dynamics to estimate the cooling load as a function of fresh air flow and heat generated in the zone as will be illustrated in this section.

**Table 1.** Correlation matrix of indoor air quality (IAQ) parameters.


**Figure 5.** IAQ measurements for 10 days with 15 min interval.

#### 3.3.1. Soft-Sensors for Cooling Load Measurement

The cooling load is also due to fresh air induction and the effect of cooling load is measured from the prediction equation using the zone thermal dynamic model as

$$Q(k) = T(k+1) - aT(k) + b\mu\_{A\mathbb{C}}(k) + cw(k) \tag{11}$$

The above equation gives an estimate of the cooling load as well as the effect of fresh air induction. The fresh air induction and the raise in temperature due to weather, current room temperature, and control input, i.e., the time for which the air-conditioner is kept ON/OFF is used to train the cooling load estimator. We use a simple radial basis function neural networks (RBFNN) to estimate the cooling load.

#### 3.3.2. Soft-Sensors for IAQ

As our idea is to design a soft-sensor for IAQ with univariate analysis, we consider the use of deep neural networks as they tend to outperform other models used in computer vision. In recent years, deep neural networks have been shown to outperform previous state-of-the-art machine learning approaches in various application domains [25]. Convolutional Neural Networks (CNN) is a special type of deep neural networks which is mainly used to handle 2D input data such as images. A CNN architecture is built by stacking three main types of layers: *convolutional*, *pooling*, and *fully-connected*. The convolutional layer maps the input data to a feature map by performing the convolution operation (dot product between the input and a filter) by sliding over the input data. The output of convolution operation is passed through an activation function that controls neuron activation. Rectified Linear Unit (ReLU) is a commonly used activation function to introduce non-linearity into the neuron's output. The pooling layers are often included between the successive convolutional layers. They are useful in reducing the dimensionality of feature maps to avoid overfitting. Max pooling is commonly kept as it is the most dominant spatial relationship. Next, the pooling layer output is flattened and given as input to the fully connected dense layer that computes the final model output.

Our CNN model architecture for short-term forecasting is shown in Figure 6. It takes the past IAQ sensor readings as input and performs a one-step prediction. In our experiments, the past 24 sensor readings (6 h long) are used as input. Next, we added two 1-dimensional convolutional layers (kernel size 2) with ReLU activation function that learns the trend and seasonality in the sensor data. This is followed by a 1-dimensional max pooling layer (pool size 2) that reduces the feature map dimension into half. Consequently, the pooling layer output is fed to the fully connected layer that yields the final prediction. We implemented our CNN architecture in Python 3 using the keras library (https://keras.io/). An early stopping method was also used to optimize the training process.

**Figure 6.** The convolutional neural network (CNN) model architecture for forecasting IAQ sensor data.

#### *3.4. Calibration of IAQ Sensors Using MobileApp*

To calibrate the sensors, a mobileApp provided by Renson was used. It is an Android application for Renson Healthbox 3.0 that allows the user to access and calibrate the device. Calibration is done by establishing connection between the device and mobile. The calibration tool performs auto-configuration and calibration is done for the actuators and sensors using the mobileApp automatically. Once installed, the Renson Healthbox 3.0 could connect to cloud and other platforms to port the data. There is also a portal provided by the Healthbox 3.0 for porting the data. The calibration is for control valves that open the vanes for supplying fresh air and takes up to 3–6 min. This will adjust the nominal air-flow rate to individual zones as well. The calibration procedure with the mobileApp is shown in Figure 7.


**Figure 7.** Calibration with MobileApp for Renson Healthbox 3.0.

#### **4. Results**

#### *4.1. Soft-Sensors for IAQ*

The soft sensor effectiveness was validated using the data collected from the hybrid ventilation system in Figure 4. We used 80% of our dataset to training our CNN models and the remaining 20% for testing. The descriptive statistics of the collected data is shown in Table 2. We summarize the performance of our short term forecasting model for each IAQ sensor using Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) in Table 3 and visualization in Figure 8. The forecasting error of TVOC is high compared to other sensors due to no seasonality in the raw data. This requires further investigation, possibly extending our CNN model with Long Short-Term Memory [26]. The forecasting error of all other sensors is less than 5%.


**Table 2.** Descriptive statistics of the collected IAQ sensor data.

**Table 3.** Comparison of short-term forecasting results using mean absolute percentage error (MAPE) and mean absolute error (MAE).


**Figure 8.** Visualization of IAQ sensor forecasting using CNN model.

As an example, consider the scenario wherein the objective is to maintain both IAQ and minimize power consumption. Suppose, if the IAQ is bad, then the Healthbox 3.0 which is a DCV introduces more fresh air which increases the cooling load. However, the DCV controls the IAQ parameters efficiently. On the other hand, the increase in cooling load is forecasted as increase in cooling loads and fed to the toggler which is a model based controller. It fuses this information into the model and predicts future temperature evolution in a particular zone and then tries to optimize the energy supplied to the individual zones based on temperature set-points provided by the user by adjusting the duty cycle of the toggler. This way both IAQ and energy is optimized by the hybrid ventilation scheme.

#### *4.2. Indoor Air Quality*

The performance of the DCV with the proposed energy optimizer is evaluated. For this study, we first collect information on the ventilated space over different periods. Then, we compare the results with and without hybrid ventilation system. It was seen that without the hybrid ventilation system the TVOC, carbon dioxide, temperature, and humidity levels were not maintained, as there was no control on IAQ and space cooling.

The variations in humidity with the hybrid ventilation scheme and the energy optimizer is shown in Figure 9. One can see that the humidity is maintained within acceptable values of 60 even during the nights when the ventilation system is shut out. This is due to the pre-conditioning created by the fresh air introduced by the proposed hybrid ventilation system. Contrary to this before installing the Healthbox 3.0, the humidity cannot be maintained and the behavior is almost random.

**Figure 9.** Humidity variations with the hybrid ventilation system.

The variation in CO2 levels with the hybrid ventilation system installed is shown in Figure 10. The CO2 levels are well maintained with the proposed energy and IAQ optimizer. A clear pattern of the variation is shown and the maximum value reaches 850 ppm though there are approximately 5–6 occupants in the room. This results shows the capability of the proposed system to maintain CO2 levels. While the Helathbox maintains the IAQ, the energy optimizer also coordinates its action by using the fresh-air flow as the variable.

**Figure 10.** The carbon dioxide levels with Hybrid Ventilation system.

The variations in TVOC with the proposed hybrid ventilation system is shown in Figure 11. One can see that the TVOC is maintained within a ppb of 1000 during normal days and exceeds to 1400 only when the system is turned OFF. This results shows the ability of the system to maintain IAQ by limiting the TVOC.

**Figure 11.** The Total Volatile Organic Compound levels with Hybrid Ventilation system.

#### *4.3. IAQ and Thermal Comfort*

As seen in the results, one can see that the IAQ variables are maintained within acceptable values throughout the period, and the temperature is also maintained within user-defined comfort values by the system. The occasional spurts represent the night times when the system is kept off, but the temperature is maintained well within the user defined comfort margin of 20–25 degree ◦Celsius.

#### *4.4. Energy Optimizer*

The performance of the energy optimizer for a 12 h period with a sampling time of 15 min is shown in Figure 12. Here, the control input is the duty cycle, i.e., *uAC* = *tON Tp* . A value of *uAC* = 0.5 means that the air-conditioner should be operated for 7.5 min within 15 min. However, this gives rise to frequent switching and transient energy gets wasted. Therefore, to overcome this, we use an averaging approach, i.e., the control inputs computed at two time-instants are implemented for a time-frame of 30 min. For example, *uAC*(*k*) = 0.5 and *uAC*(*k* − 1) = 0.3, where *k* − 1 was not implemented to avoid transient energy loss, then a total of 12.5 min is turned ON during the *k*th time-instant. Such an averaging may sometimes result in the zone temperature to raise above the comfort band of 28 ◦C. However, considering that the time constant of the zone is very high in the order of 12–13 min, this is negligible period for which the thermal comfort could be breached by a small margin, but the savings in transient energy is quite high. Therefore, we implement the wait and apply strategy to reduce the transient energy consumption.

**Figure 12.** Variations in zone temperature, control input, heating due to cooling load, and ambient temperature (*x*-axis: 1 unit = 15 min).

The energy consumption was recorded with and without the proposed control strategy. We found that an average saving of 19.2–21.4% was observed with the proposed control strategy over conventional thermostat control while maintaining the IAQ. These savings are for 12 h period in a working day. The savings are mainly due to switching between fan and air conditioner. This result illustrates the energy savings potential of the proposed hybrid ventilation and control scheme.

Our results showed that the proposed hybrid ventilation scheme achieved good IAQ and energy savings without breaching the comfort band proposed by the occupant. We also showed that an energy savings of up to 19–21% can be achieved, and the IAQ parameters were maintained within bounds specified. We also demonstrated the role of soft-sensors and implementation aspect of the proposed hybrid ventilation scheme, i.e., demand controlled ventilation, sensors for measuring IAQ parameters and energy relevant data, communication architecture and other aspects were also presented. The results demonstrated the value of soft-sensor to measure parameters like cooling loads, TVOC, carbon dioxide concentration and others which were not possible before. An additional benefit of soft-sensor is the ability to generate short-term predictions which could be used to plan ventilation schedules.

#### **5. Conclusions**

This paper presented a novel design of a hybrid ventilation system for optimizing energy consumption in variable air volume (VRV) systems while maintaining indoor air quality (IAQ) and the thermal comfort of the building occupants. We showed that the problem is multi-objective, multi-time step optimization problem as the objective to optimize both IAQ and energy requires a model predictive controller (MPC). As such implementing MPC schemes on low-cost hardware is a challenge and the absence of sensed information/predictions on IAQ and cooling loads posed additional challenges. To overcome this, we proposed IoT based sensing that was extended with deep learning tools to design soft-sensors that provided measurements on parameters, which could not be measured with physical sensors. Then, the multiple objectives was decoupled into IAQ and energy savings. To achieve IAQ we proposed a novel hybrid ventilation system which used RENSONs' Healthbox 3.0, which is an off-the-shelf demand controlled ventilation. On the other hand, we designed an MPC in Rasperry Pi using the Gnu Linear Programming Kit (GLPK) which provided optimal solution to the multi-time step optimization problem to realize the energy optimizer. We provided results which showed the performance of soft-sensors, IAQ and energy optimization capabilities of the hybrid ventilation scheme while meeting user defined comfort bands. Extending the hybrid ventilation scheme to other IAQ measures and including thermal comfort with predictive mean vote or other standards is the future course of this investigation.

**Author Contributions:** Conceptualization, K.Z. and S.S. (Seshadhri Srinivasan); methodology, S.S. (Seshadhri Srinivasan) and P.A.; software, N.V. and S.S. (Subathra Seshadhri); validation, S.S. (Subathra Seshadhri), N.V. and K.Z.; writing—original draft preparation, S.S. (Seshadhri Srinivasan); writing—review and editing, P.A., K.Z. and S.S. (Subathra Seshadhri); supervision, S.S. (Seshadhri Srinivasan); project administration, S.S. (Seshadhri Srinivasan) and S.S. (Subathra Seshadhri); funding acquisition, S.S. (Seshadhri Srinivasan) and K.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** This project is supported by the project Resilient and Optimal-micro-energy grid (ROME): Funded by Department of Science and Technology and Research Council of Norway, Norway through Resilient and Optimal Micro-Energy grid INT/NOR/RCN/ICT/P-05/2018.

**Acknowledgments:** The authors thank RENSON Belgium for providing support with Healthbox 3.0. In particular, we would like to thank Makarand Kendre from RENSON India for his support in the initiative. The authors also thank Northumbria University, UK for providing support to do the open access publication. The authors also thank the Center for Research in Automatic Control Engineering, International Research Center, Kalasalingam University for its support to conduct the research.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**

The following abbreviations are used in this manuscript.


#### **References**


c 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
