**Design and Optimization of an Integrated Turbo-Generator and Thermoelectric Generator for Vehicle Exhaust Electrical Energy Recovery**

#### **Prasert Nonthakarn 1,\*, Mongkol Ekpanyapong 1, Udomkiat Nontakaew <sup>2</sup> and Erik Bohez <sup>1</sup>**


Received: 27 May 2019; Accepted: 9 August 2019; Published: 15 August 2019

**Abstract:** The performance of turbo-generators significantly depends on the design of the power turbine. In addition, the thermoelectric generator can convert waste heat into another source of energy. This research aims to design and optimize an integrated turbo-generator and thermoelectric generator for diesel engines. The goal is to generate electricity from the vehicle exhaust gas. Electrical energy is derived from generators using the flow, pressure, and temperature of exhaust gases from combustion engines and heat-waste. In the case of turbo-generators and thermoelectric generators, the system automatically adjusts the power provided by an inverter. Typically, vehicle exhausts are discarded to the environment. Hence, the proposed conversion to electrical energy will reduce the alternator charging system. This work focuses on design optimization of a turbo-generator and thermoelectric generator for 2500 cc. diesel engines, due to their widespread usage. The concept, however, can also be applied to gasoline engines. Moreover, this model is designed for a hybrid vehicle. Charging during running will save time at the charging station. The optimization by variable van angles of 40◦, 50◦, 62◦, 70◦, and 80◦ shows that the best output power is 62◦, which is identical to that calculated. The maximum power outputted from the designed prototype was 1262 watts when operating with an exhaust mass flow rate of 0.1024 kg/s at 3400 rpm (high performance of the engine). This research aims to reduce fuel consumption and reduce pollution from the exhaust, especially for hybrid vehicles.

**Keywords:** thermoelectric generation; turbo-generator; exhaust heat recovery

#### **1. Introduction**

Currently world energy consumption is continuously increasing, and this is associated with a high rate of environmental damage due to the use of fossil fuels. This has been a concern in light of a shortage of energy resources and environmental protection. In the past few years, a pressing need has arisen to develop green energy systems and breakthrough technologies to solve these problems. Since transportation is one of the main sectors responsible for the utilization of energy, transmissions, and new vehicle components are being developed in order to achieve maximum energy savings, and as a consequence less fuel consumption.

The prime source for powering vehicles is the internal combustion engine, or ICE. The principle approach for releasing energy is the conversion of fuel in the combustion chamber into usable energy. Unfortunately, the ICE has poor energy conversion efficiency. In the case of gasoline, diesel, and hybrid electric vehicles, the percentage of utilized fuel during running is around 25%, with the remaining 75% lost in the form of exhaust gas enthalpy. This lost energy is crucial, and is largely due to exhaust gas from waste heat (40%), the cooling system (including heat radiation, 30%), and friction (5%) [1]. Consequently, the amount of energy lost from the exhaust and cooling systems is twice the amount of

energy than is actually utilized for mechanical energy. It can be estimated that 20% of the lost energy can be converted to only 10% of the total electricity [2]. In this case, the exhaust gas or cooling system can be used to generate energy, and thus fuel efficiency is increased.

In order to increase fuel efficiency, various energy saving technologies have been developed. Many researchers have focused on the utilization of the exhaust gas of vehicles for power generation. Engine exhaust gas energy recovery is one of the more promising technologies, through simultaneous energy saving and emission reduction. Engine exhaust gas has a certain pressure and heat due to its high pressure and high temperature, which can be used to generate electricity for power storage, thereby reducing fuel consumption and increasing energy recovery efficiency. Moreover, the high-grade mechanical energy of the remaining pressure can be recycled through the exhaust gas directly expanding, which is known as the direct recovery method, and through low-grade thermodynamic energy used in heat transmission, known as the indirect recovery method. Each method has different measurements and solutions for energy recovery [3].

This study focused on designing and optimizing turbo-generator coupling with a thermoelectric generator for 2500 cc. diesel engines. The work is composed of two main categories: Electrical energy recovery, and mechanic energy recovery units. The design proposes to convert exhaust gas energy to electrical power by mechanical exhaust energy, which is then converted to electricity through a turbine, and further proposes to convert waste heat directly to electric energy. When using diesel engines without turbos, there is no obstacle that will reduce the pressure and flow rate of the exhaust stream from the combustion unit. Electrical energy recovery is a way to utilize recovered energy to supply electrical power to the automobile's interior equipment, or to bring this energy to charge the battery for further use. Mechanical energy recovery, on the other hand, is a way to recover energy from the exhaust gas, and is directly attached to the inverter in order to make the power available to the vehicle. Power can be generated by the different turbine geometry components; for example, the rotor, nozzle and volute. Electrical energy is collected during driving time.

The structure of this paper is as follows. First, the turbo-generator model and thermoelectric generator are presented. Next, the simulation results are illustrated. Then, the experimental results, which validate the idea, are discussed. Finally, we conclude the paper and make recommendations for future work.

#### **2. Exhaust Gas Heat Recovery Technology**

The energy emitted by the engine exhaust gas is pressure and heat. Exhaust gas recovery transforms the high pressure and heat back to reusable energy in the form of electrical power. The high speed of hot waste gas produces a great amount of mechanical energy that can be utilized for power generation; this is known as the direct method. At present, waste heat recovery technology from the engine by the direct method is available, and is known as mechanical turbo-compounding and electric turbo-compounding. Mechanical turbo-compounding using a conventional turbocharger depends on the turbine to generate energy from the exhaust gas flow expelled from an internal combustion engine. Alternatively, thermal energy derived from the exhaust gas is indirectly converted to additional electricity using the Rankine cycle (organic or steam), and is directly recovered through heat transmission of the thermoelectric generator.

Energy recovery using the direct technology method is easier to assemble than the indirect method. Electrical turbo-compounding can be easily assembled, and can be directly connected to the turbocharge axle or directly attached to the secondary turbocharge shaft [4–6]. Melanie Michon [7] reported the results of study on the comparison of an automotive turbo-generator exhaust gas energy recovery system using low voltage and high voltage machines. Aman M.I. Mamat [8] developed a highly efficient turbine suitable for a low pressure turbine, or LPT, using turbo-compounding in small and large segment passenger cars that use gasoline engines with a capacity of 1.0 L. The turbine was designed for two purposes, to recover the energy of exhaust gas at low pressure with ratios of 1.05–1.3, and to drive a small generator with fixed operation conditions at 50,000 rpm. The resulting

power output is up to 1.0 kW. Commercial Computational Fluid Dynamics (CFD) software (ANSYS Unveils Release 16.0) was used in simulation to determine the low-pressure turbine performance. Hence, the performance was verified with comprehensive turbine testing derived from the Imperial College turbine test rig (cold-flow test facility). The caterpillar concept has been considered in electrical turbo-compounding research [9–11]. These researchers have shown that using a high-performance turbocharger reduces fuel consumption by 5% in basic engine cycles (brake specific fuel consumption). Besides that, it can reduce fuel consumption overall with by up to approximately 9%–10%. Electrical turbo-compounding is designed as part of the turbo-compound and is directly attached to the exhaust pipe, to obtain exhaust gas heat flowing out with a generator running at a very high rotational speed. The turbine design in this case will produce a large amount of power when compared with using only a turbine for the compressor driver. The energy in the exhaust residual during usual use is taken to drive the high-speed generators that are included in the same unit, which will transform the remaining energy to electric power. Energy recovery technology can achieve high power from exhaust gas, and work as a power plant. The maximum energy recovery from the exhaust is approximately 1.8 kW at the initial stage. Bowman Power Group Ltd. presents turbo-generators for power generation (power plants) [12]. Bowman designed and applied a turbo-generator for installation in a power plant generator, which brings the electrical power produced to the electric grid parallel to the original alternator. The turbo-generator engine is mounted with a 250 kW to 1 MW engine that is equipped with a turbocharger, and mounted after the turbocharger. The installation can produce around 10%–20% more power. Controlled Power Technologies Ltd. (Basildon, UK). Federal-Mogul Controlled Power Ltd. (Michigan, USA).

Unit 4, Westmayne Industrial Park, Bramston Way, Laindon designed and built a turbo-generator integrated gas energy recovery system (TIGERS) that has the same functionality as the turbo-generator [13]. TIGERS are water-cooled generators coupled to an exhaust driven turbine. The small turbo-generator is installed at the exhaust pipe. This pipe is another pipe created for flow control to the generator. The turbo-generator uses the valve to increase or decrease the amount of exhaust passing through this series of turbo-generator. The opening of the more or less valve is controlled by the control system relative to the velocity of the exhaust gas engine. The exhaust gas flows through the turbine, causing the power transmission to be powered by the generator. Power output depends on the engine load. The system produces up to 600 watts of electricity power.

Other technologies, such as the Rankine cycle (organic or steam indirect), produce power indirectly. Beside these, thermoelectric generator technologies obtain energy directly through heat transmission. The Rankine cycle uses the principle of changing the status of liquids. At any time, the liquid may be pressurized and heated, and it transforms into pushed vapor due to expansion. The waste heat recovery (WHR) evaporator transforms the energy in the exhaust by installing it in a position after the manifold exhaust in the production phase, by which the working fluid expands in the turbine. The efficiency of energy recovery is based on the efficiency of the heat exchanger. Furthermore, using the organic Rankine cycle to recover energy requires a large space to install the system, and it is difficult to assemble the parts [14,15].

A thermoelectric generator is another way of bringing the energy from the exhaust gas back to a usable state, and is convenient and easy to install in an engine. The Peltier–Seebeck effect describes how a thermoelectric generator transforms the heat in the exhaust gas into electric power. The current arises from the difference of temperature between one side of the thermoelectric generator and the muffler. It is a device with no moving parts (a solid-state device) that directly transforms the heat to electricity. The advantage of a thermoelectric module without moving parts is that the power-generating device is simple, robust, reliable, modular, and maintenance free. Moreover, its production of electricity is environmentally friendly.

X. Liu [16] proposed simulation and experimental results of positioning of the thermoelectric generator installation. In this work, the thermoelectric generator is placed between the catalytic converter and muffler, where there is no system parts to block the flow of exhaust. As a result, it does not cause the exhaust pressure to reverse inside the exhaust pipe, and there will be a large area for heat transfer, which improves the performance of the heat transmission. Tae Young Kim [17] studied the efficiency of the heat released with recycled exhaust gas. The experiment used 40 thermoelectric generators equipped with a large diesel engine, and operated with the engine at a rotation cycle of 1000, 1500, and 2000 rpm. The thermoelectric generator was installed on the top and bottom of the exhaust gas channel series. The experimental results illustrated that the energy released directly varies based on the engine load and speed, and the maximum energy value at 2000 rpm is 119 watts. Xiuxiu Sun [18] compared two characteristics of thermoelectric generators: Series installation and parallel installation. The results of the experiments show that parallel installation provides a better performance than series installation. Hua Tian [19] presented the effect of the energy that has been released, and the efficiency of thermoelectric generators. The results conclude that energy and efficiency can be increased by increasing the heat source temperature, and reducing the heat in the desired cooling areas. The performance is based on the ratio of two types of materials—Skutterudite and bismuth telluride—to make thermoelectric generators reliable and easy to use. Ilker Temizer [20] developed a prototype for recycling heat. In this work, the thermoelectric generator is mounted with a diesel engine. Experiments were carried out looking at different engine speeds and different engine loads. The speed of the engine rotation was set at five different levels, and the engine load was set at two different levels. The experimental results show the maximum output was 156.7 watts at the maximum speed and maximum engine load. Changxin Liu [21] studied prototype generation by installation of a thermoelectric generator with a car exhaust muffler. The experimental results found that, when operating the engine with a heat pipe at a temperature of 473 K, the thermoelectric generator can produce an energy output of to 250 watts, with a heat efficiency of 5.35%.

Based on the hybrid electric charge for a hybrid car, it was found that the average electric charge would be 1–5 h per 100 km driving distance. For charging, the car will be parked at a power station, which wastes time. The average speed of a used car driving on the highway is 100 km/h. At that speed, the engine speed is 3500 rpm.

Contemporary's cars have a higher level of power consumption than ever before, because of the equipment used to drive the engine and because of electrical demands such as the air conditioning system (using an electric compressor), electric power steering, electronic throttle control, electric windows, and power seats. In addition, the equipment included for convenience when driving such as phone chargers, refrigerators, heaters, and car entertainment equipment also require electricity.

In this research, the recovery of energy in the exhaust is different from previous works. It can summarize with the major differences as shown in Table 1. The main difference is this research will design the turbo-generator and thermoelectric for 2500 cc. diesel engines of light-truck.


**Table 1.** Summary of parameters for turbo-generators and the thermoelectric generator model.

TG (turbo-generator), TEG (thermoelectric-generator), and n/a (not available).

#### **3. Turbo-Generators and Thermoelectric Generator Modeling**

The turbo-generators and thermoelectric generator system developed by the designing and installing of turbo-generators after exhaust manifold and installing the thermoelectric generator in the next position. Therefore, this section outlines the process of design for turbo-generators, followed by turbo-generator development.

#### *3.1. Exhaust Energy*

External heat balance is based on measurements of mechanical energy, and energy taken outside by the cooling liquid and exhaust gases. The external balance equation takes the following form [2]:

$$Q = Q\_{\varepsilon} + Q\_{cl} + Q\_{w} + Q\_{n} + Q\_{r} \tag{1}$$

where *Q* is the total heat inputted to the engine (J/h), *Qe* is heat that can be exploited, *Qch* is cooling loss, *Qw* is the loss of exhaust, *Qn* is an incomplete loss, and *Qr* is the residual of the balance/remaining. The loss of exhaust is the largest amount of energy derived from heat and pressure, which can be calculated from the following equation:

$$Q\_w = \dot{m} \times c\_p \times \Delta T\_\prime \tag{2}$$

where . *m* is the exhaust gas mass flow rate, *cp* is the specific heat of exhaust gas, and Δ*T* is the exhaust gas temperature difference.

To understand an overview of all the energy generated by the internal combustion engine using the functional principle of the Otto cycle, and to be able to see and understand the energy mix in the exhaust gas released along with the exhaust muffler, Figure 1 shows the relationship of pressure and volume in the cylinder box of the combustion engine within four strokes. The bottom part of the figure shows the piston and the cylinder that move up and down during operation of a four-stroke engine. The piston continuously moves from the bottom death center to the top death center. The movement causes various strokes, including the intake, compression, combustion, and exhaust strokes, each of which work together with the displacement of the piston and closing/opening of valve. Briefly, the energy in the exhaust gas in the figure is the area of the exhaust energy (area F-G-H-I) that is released from the discharge of the exhaust stroke in the spitting. The piston moves up from the bottom death center to the top death center at the same time that the exhaust valve opens. This makes the head of the piston press the exhaust gas. Then, the exhaust gas with heat and high pressure flows out of the cylinder through the manifold exhaust to the exhaust pipe and to the pipe end, respectively.

**Figure 1.** Exhaust energy

#### *3.2. Flow Chart for Calculating the Exhaust Mass Flow of an Internal Combustion Engine*

This flow chart (Figure 2) shows the process for calculating the mass flow rate of the exhaust gas, which will be used to design the rotor blades in the next stage.

The operating exhaust conditions are setup for a 2500 cubic centimeter diesel engine with the speed of the engine 3400 rpm (normal high use). Table 2 shows the flow characteristics of the exhaust gas that changed when the engine speed changes. To summarize, the increased engine rotation speed affects the mass flow rate and flow velocity, where the speed of the engine varies from 1000 to 5000 rpm. The exhaust mass flow rate calculated as the following:

$$
\dot{m}\_{\text{exhaust}} = \dot{m}\_{\text{air}} + \dot{m}\_{fuel} = 0.1024 \text{ kg/s.}
$$

**Table 2.** Mass and velocity flow of exhaust gas on difference engine rotation speed.


#### *3.3. Specific Design*

In the first step, the dimension design of the parts of the rotor used the details following the instructions as shown in Table 3 and Figure 3.


**Table 3.** Design parameters of the rotor.

The equation below describes the relationships of the theoretical work for turbines:

$$L\_0 = \frac{k\_1}{k\_1 - 1} \times R\_{sp} \times T\_s \times \left[1 - \left(\frac{P\_s'}{P\_s}\right)^{\frac{k\_1 - 1}{k\_1}}\right] \tag{3}$$

$$\frac{T\_s'}{T\_s} = \left(\frac{P\_s'}{P\_s}\right)^{\frac{K\_1 - 1}{K\_1}},\tag{4}$$

$$L\_o = \frac{k\_1}{k\_1 - 1} \times R\_{sp} \times (T\_s' - T\_s),\tag{5}$$

where *L*<sup>0</sup> is the theoretical work by energy per mass of exhaust gas (*kJ*/*kg*), *k*<sup>1</sup> is the value of adiabatic exhaust gases, *Rsp* is the constant value of the exhaust gases relative to the gas constant to the molecular weight of the gas, *Ps* is the pressure of exhaust gases behind the turbine, *P <sup>s</sup>* is the pressure of exhaust gases in front of the turbine, *Ts* is the temperature of exhaust gases flowing into the turbine, and *T <sup>s</sup>* is the temperature of exhaust gases flowing out from the turbine.

**Figure 3.** Velocity triangles for the rotor and stator.

The following equation explains the size of the work occurring in the turbine, which is related to the decrease of the temperature after the exhaust gas flows through the turbine.

$$N\_l = G\_5 \times l\_l = G\_5 \times l\_0 \times \eta\_{l,t} \tag{6}$$

where *Nt* is the theoretical power (*W*), *Gs* is the intensity of exhaust gas flow (*kg*/*s*), and η*<sup>t</sup>* is the efficiency of the turbine in isentropic.

The Euler equation shows the energy transfer in the rotor, which can be represented as a product of the torque by the angular velocity:

$$\dot{W} = \overline{\omega} T \mathcal{Q} = \dot{m} (\mu\_1 \mathbf{c}\_{01} - \mu\_2 \mathbf{c}\_{02}) = \dot{m} . \mathbf{c}\_p (T\_{10} - T\_{20}) , \tag{7}$$

where . *<sup>W</sup>* is the energy transfer, *<sup>Q</sup>* is the heat transfer per unit mass flow, <sup>ω</sup> is the angular velocity, . *m* is the mass rate of flow, (*u*1*c*<sup>01</sup> − *u*2*c*02) is the rate of change of angular momentum, *cp* is the specific heat at a constant pressure, and (*T*<sup>10</sup> − *T*20) 10-T20 is the temperature difference of the inlet and outlet.

#### *3.4. Formula for One Dimensional Calculation*

Once the energy transfer has been determined, the mass flow of exhaust gas required to achieve the specified power can be calculated from:

$$\mathcal{W} = \frac{Power}{\dot{m}\_{\text{exhust}}}.\tag{8}$$

*Energies* **2019**, *12*, 3134

The blade speed at the inlet can be calculated from the velocity triangles, in which the relative work output is:

$$
\mu\_2 = \sqrt[4]{w}.\tag{9}
$$

The absolute velocity at the inlet:

$$v\_2 = \frac{u\_2}{\sin \alpha\_2}.\tag{10}$$

where α<sup>2</sup> is absolute gas angle at radius.

The radius at the inlet of the rotor has the value of:

$$r\_2 = \frac{60u\_2}{2\pi \times rpm}.\tag{11}$$

The relative velocity at the inlet:

$$
\Delta w\_2 = \sqrt{v\_2^2 - u\_2^2}.\tag{12}
$$

The rotor shroud radius and hub radius are:

$$r\_{3\circ} = 0.75r\_2.\tag{13}$$

The rotor hub radius:

$$r\_{\mathfrak{A}\mathfrak{h}} = kr\_{\mathfrak{A}\mathfrak{e}}.\tag{14}$$

where *k* is the hub-tip ratio at the inlet impeller.

The hub blade speed:

$$
\mu\_{3\text{l}} = \frac{2\pi r\_{3\text{l}} rpm}{60}.\tag{15}
$$

The shroud blade speed:

$$
\mu\_{\mathfrak{H}} = \frac{2\pi r\_{\mathfrak{H}} rpm}{60}.\tag{16}
$$

The mean of the exit radius is equal to its square value using the Balje diagram:

$$r\_3 = \sqrt{0.5(r\_{3\kappa}^2 + r\_{3l}^2)}.\tag{17}$$

Figure 4 shows the Balje diagram used to calculate the value of the exit radius. The value of Ω*<sup>s</sup>* efficiency is in range 0.2–0.8 and the recommended value of *<sup>r</sup>*<sup>3</sup> *<sup>r</sup>*<sup>2</sup> is in range 0.53–0.66.

Consequently, the blade speed at the mean exit radius is:

$$
\mu\_3 = \frac{2\pi r\_3 rpm}{60}.\tag{18}
$$

The blade width at the outlet:

$$b\_3 = r\_{36} - r\_{3h}.\tag{19}$$

Using mass balance, the inlet blade height can be determined as:

$$b\_2 = \frac{\dot{m}\_{\text{exhaus}}}{2\pi r\_2 v\_2 \cos \alpha\_2}.\tag{20}$$

**Figure 4.** Balje diagram for a radial turbine [Reproduced with permission from [22], Seppo, A. Principles of Turbomachinery; John Wiley & Sons: New York, NY, USA, 2011.].

Using the velocity diagram for calculating the flow angle of relative velocity at the exit:

$$\beta\_3 = \tan^{-1} \left( \frac{\mu\_3}{\upsilon\_3} \right). \tag{21}$$

The flow angle of relative velocity at the hub and shroud at the outlet are:

$$
\beta\_{3h} = \tan^{-1} \left( \frac{u\_{3h}}{v\_3} \right), \beta\_{3\epsilon} = \tan^{-1} \left( \frac{u\_{3\epsilon}}{v\_3} \right). \tag{22}
$$

Figure 5 shows an overview of the equations used for calculating the dimensions of the turbine, respectively, where the process for calculating values is continuously correlated using the MATLAB program.

**Figure 5.** The dimension design of turbine.

#### *3.5. The Dimension Design*

The values of the calculated components from the previous sections are summarized and shown in Table 4 in order to create the turbine model as well as applying the values for the turbo-generator drawing.


**Table 4.** The results from the one-dimensional calculation of the radial turbine.

#### *3.6. Drawing the Turbo-Generator*

The turbo-generator parts use calculated values from Table 3 for drawing in Computer Aided Design (CAD) models of the rotor and volute with the size and shape shown in Figure 6. The overview of all parts is shown in Figure 7.

**Figure 6.** Rotor and volute design.

Part of the turbo-generator model:

**Figure 7.** The turbo-generator components.

The size of the new turbine design is shown in Table 5, and compared with the turbine of current vehicles. The differences of dimension occur when the inlet blade angle increases and the outlet blade angle decreases.


**Table 5.** The geometric features of the turbine compared with turbines in current commercial cars.

\* Toyota Hilux model.

#### *3.7. Drawing and Construction of a Thermoelectric Generator*

The output voltage is directly proportional to the temperature change, which is the principle of a thermoelectric generator using the phenomenon characteristics known as the Seebeck effect [16], and displayed as the following equation:

$$V = a\Delta T,\tag{23}$$

where α is the Seebeck coefficient *VK*−<sup>1</sup> and Δ*T* is the temperature difference of two sides of the surface in *K*.

The Reynolds number derived from the value of the heat source from the engine exhaust gas is:

$$R\_{\mathfrak{e}} = \rho v D / \mathfrak{\mu}\_{\mathfrak{r}} \tag{24}$$

where ρ is the density, ν is viscosity, *D* is the equivalent diameter, and μ is the viscosity of the fluid flowing through the tube.

The heat transfer coefficient of the hot side is determined by:

$$h\_{\mathfrak{e}} = N\_{\mathfrak{u}} k\_{\mathfrak{e}} / D\_{\mathfrak{h}}.\tag{25}$$

The Nusselt number is defined as the ratio of convection heat transfer to fluid conduction heat.

$$N\_{\rm u} = \frac{h\_{\rm c} D\_{\rm h}}{k\_{\rm c}} \, ^\ast \tag{26}$$

where *Nu* is Nusselt number, *ke* is thermo conductivity of exhaust, and *Dh* is the hydraulic diameter. The following equation shows convection on the plate:

$$N\_{\rm ll} = 0.664 \Big( \mathrm{Re}^{0.5} \times \mathrm{Pr}^{0.33} \Big), \tag{27}$$

where Re is the Reynolds number and Pr is the Prandtl number.

The heat conversion efficiency of waste heat recovery is calculated using the energy from the thermoelectric generator divided by the heat inserted into the thermoelectric generator.

$$\eta = \frac{P\_{\text{output}}}{\dot{m} \mathcal{C}\_p (T\_{\text{in}} - T\_{\text{out}})} \, ^\prime \tag{28}$$

where <sup>η</sup> is the conversion efficiency, *Poutput* is the thermoelectric generator power output, . *m* is the exhaust gas mass flow rate, *Cp* is the exhaust gas specific heat, *Tin* is the exhaust gas system inlet temperature, and *Tout* is the exhaust gas system outlet temperature.

The equation to calculating the power generated by the thermoelectric generator is:

$$P = N a\_{\rm pu} I \Delta T\_{\rm l\xi\chi} - I^2 N \mathcal{R}\_{\rm p\nu\nu} \tag{29}$$

where *N* is the number of thermoelectric couples employed, α*pn* is the Seebeck coefficient, *I* is, respectively, the electric current, Δ*Tleg* is the thermoelectric leg temperature difference, and *Rpn* is the value of the thermoelectric resistivity couple.

The equation for the system efficiency, η*,* can be calculated by:

$$
\eta = \frac{p\_{\rm TEG}}{P\_{\rm engine}} \times 100\%\_{\rm \prime} \tag{30}
$$

where *pTEG* is the thermoelectric generator maximum output power and *Pengine* is the power of the engine.

The efficiency of a TE module used as a generator can be approximated by the following relationship, where *Z* is a material property, *Tc* is cold temperature, *Th* is hot temperature, and *T* is (*Th* + *Tc*)/2.

$$\eta\_{TE\text{max}} = \frac{W\_{\text{elcc}}}{Q\_{\text{h}}} = \frac{\Delta T}{T\_{\text{h}}} \cdot \frac{\sqrt{1 + ZT - 1}}{\sqrt{1 + ZT + \frac{T\_c}{T\_{\text{h}}}}} \cdot \tag{31}$$

$$Z = \frac{\left(\alpha\_p - \alpha\_n\right)^2}{\left(\left(\lambda\_p \rho\_p\right)^{1/2} + \left(\lambda\_n \rho\_n\right)^{1/2}\right)^2},\tag{32}$$

where α*<sup>p</sup>* is the Seebeck coefficient corresponding to *p*, α*<sup>n</sup>* is the Seebeck coefficient corresponding to *n*, λ*<sup>p</sup>* is thermal conductivity corresponding to *p*, λ*<sup>n</sup>* is thermal conductivity corresponding to *n*,ρ*<sup>p</sup>* is electrical resistivity the corresponding to *p*, and ρ*<sup>n</sup>* is electrical resistivity corresponding to *n*.

The power output is:

$$P = Q\_{\rm li} - Q\_{\rm c} = I^2 R\_{\rm L} \tag{33}$$

where *I* is the electrical current in the generator circuit, *RL* is the electric resistance of semiconductor couple, *Qh* is the heat absorbed from heat source, and *Qc* is the heat absorbed from the cold source.

Power outputs and conversion efficiencies were calculated applying the numerical results, the electric current and absorbed heat shown in the equations below:

$$P = N \Big[ \alpha\_{\text{pu}} (T\_h - T\_c)I - I^2 R \Big],\tag{34}$$

$$R = \frac{L}{A} (\rho\_n + \rho\_p)\_\prime \tag{35}$$

$$
\eta = \frac{P}{Q\_h} \tag{36}
$$

where *P* is the output power, *N* is the number of thermoelectric elements in the module, and α*pn* is the Seebeck coefficient. *L* is the length of the legs and *A* is the cross-sectional area.

The equation for the thermocouple conversion efficiency is:

$$
\eta = P / Q\_{\text{h}}.\tag{37}
$$

where *Qh* is the absorbed and *P* is the output power.

Another important part of the process for designing a thermo-electric generator is finding the heat transfer correlations, which is calculated from the heat of the engine exhaust flowing through the exhaust pipe. The design will focus on the suitability and ease of installation with real engines in a limited space to make the best heat transfer efficiency and to not extremely affect the pressure of the exhaust. Figure 8 illustrates the heat transfer correlations of the thermo-electric generator.

**Figure 8.** Heat transfer correlation of the thermoelectric module.

Energy balance of cooling heat exchanger calculated by:

$$Q\_{\mathcal{L}} = m\_{\mathcal{c}} \mathcal{c}\_p (T\_{\mathcal{c}+1} - T\_{\mathcal{c}}),\tag{38}$$

$$Q\_{\mathcal{L}} = h\_{\mathcal{c}} A\_{\mathcal{c}} (T\_{\mathcal{c}} - (T\_{\mathcal{c}} + T\_{\mathcal{c}+1})/2),\tag{39}$$

where *hc* is the heat transfer coefficient for the coolant heat exchanger, *Ac* is the heat transfer area for the coolant side, *mc* is the mass flow rate, *Tc* is the coolant water temperature, and *Tt* is the temperature of the cold side of the thermoelectric module.

Energy balance of exhaust heat exchanger calculated by:

$$Q\_h = m\_h c\_p (T\_h - T\_{h+1})\_\prime \tag{40}$$

$$Q\_h = h\_h A\_h (T\_h - (T\_h + T\_{h+1})/2 - T\_i) \tag{41}$$

where *hh* is the heat transfer coefficient for the exhaust heat exchanger, *Ah* is the heat transfer area for the exhaust side, *mh* is the mass flow rate, *cp* is the specific heat of the exhaust, *Th* is the temperature of the exhaust, and *Ti* is the temperature of the hot side of the thermoelectric module.

Heat and cooling transfer of the system calculated by:

$$Q\_c = SIT\_c + K(T\_h - T\_c) - 0.5I^2R\_\prime \tag{42}$$

$$Q\_h = SIT\_h + K(T\_h - T\_c) - 0.5I^2R\_\prime \tag{43}$$

where *S* is the Seebeck coefficient, *R* is the internal resistance, *K* is the thermal conductance of the module, and *I* is the total current of the generator.

Calculations for efficiency of a thermo-electric generator:

The exhaust gas mass flow rate was 0.1024 kg/s, the exhaust gas specific heat *cpg* was 1*kJ*/*kg*.*k*, and the supplied heat was: *the exhaust gas* = *mgcpg*Δ*T* = 0.1024 × 1.148 × (110 − 115) = 587.776 W, *electrical power output* = *VI* = 16 × 5 = 80 W

$$efficiency\text{ of thermalectic generator} = \frac{electrical\text{ power output}}{heat\text{ supplied by the exhaust gas}} = 13.61\tag{44}$$

The thermoelectric module in Figure 9 composed of two units with dimensions of 240 mm × 100 mm × 300 mm. It is a combination of three main parts: The water cooling box, heat exchange box, and thermoelectric plate (*TECI* − 12706). The water cooling box has eight parts with dimensions of 70 mm × 40 mm × 300 mm. It was used as a thermo-electric plate to reduce the heat by removing the heat from the water using the radiator. The heat exchange box has dimensions of 100 mm × 100 mm × 300 mm to transfer the heat from hot water to the 80 thermoelectric plates. The design will focus on the suitability and ease of installation with real engines in limited space to make the best heat transfer efficiency and to not extremely affect the pressure of exhaust.

**Figure 9.** Thermoelectric generator module.

#### **4. Simulation Results and Optimization**

#### *4.1. Turbine Performance*

The steps of the CFD simulation started from design and drawing, which was shown in Figure 10 by modeling the adjustment of the vane angle at 40◦, 50◦, 62◦, 80◦, and 85◦ respectively. Then, the step was the model preparation for using in the simulation process. The step started from the CFX-Design Modeler procedure which is an important step of ANSYS simulation. The total number of faces that resulted from this procedure was 358 faces, separated into turbine 282 faces, exhaust gas 28 faces, and vane 48 faces. After that was the surface preparation procedure using the section of the CFX-Meshing. The total value of the model was 914,018 nodes and 3,461,517 elements, divided into turbine 509,734 nodes and 2,545,575 elements, exhaust gas 132,988 nodes and 688,147 elements and vane 271,296 nodes and 227,768 elements.

**Figure 10.** Model preparation. (**A**) CFX-design modeler and (**B**) CFX-meshing.

The next step was the process of simulating the flow condition in the model. The simulation simulated the exhaust flow that flowing into the inlet of the turbine case and simulated the flow out of the front. There are various configurations in this step including the exhaust flow rate, pressure temperature, and flow characteristics details as follows: The CFX-Pre procedure had three domains and three interfaces, the exhaust domain had a static inlet, an inlet mass flow rate 0.1024 kg/s, and outlet pressure at 101.325 kPa. Turbine domain configured a rotation speed between 15,000 and 35,000 rpm and the vane domain was defined as the static flow at 400 ◦C as shown in Figure 11.

**Figure 11.** CFX-pre simulation.

The results of the simulation shown in Figure 12. In the simulation, there were eight different vane angle simulations including not vane, vane angle 40◦, 50◦, 62◦, 80◦, and 85◦, and commercial 1, 2. The simulation was set with a rotation speed from 15,000 to 35,000 rpm.

**Figure 12.** Moment on the turbine.

The torque on the turbo blade was 0.628 N – m. This was used to calculate the power, using this equation:

$$P = T \times 2 \times \pi \times \frac{N}{60} = 0.628 \times 2 \times \pi \times \frac{52000}{60} = 3401 \text{ watt.} \tag{45}$$

Table 6 shows the comparison of torque and power for adjusting the inlet blade at various angles without the vane and commercial turbines.

**Table 6.** Turbine performance.


The results of the simulation for the input parameter exhaust inlet was 0.1024 kg/s (3500 rpm; engine rotation speed), for the pressure outlet was 1.1 bars, and for the temperature was 400 ◦C. The inlet angle set to five variables with a vane angle of 40◦, 50◦, 62◦, 70◦, and 80◦ as shown in Figure 13. The results shown in Table 5 and Figure 14A suggest torque and power increased if the rotation speed increased. The turbine could generate from 1937 watts of power with an engine speed of 4200 rpm, when the vane angle varied from 40◦ (1937 watts) to 62◦ (3401 watts). After the vane angle increased to more than 62◦, a reduction of power occurred.

**Figure 13.** The variable vane angled; 40◦, 50◦, 62◦, 80◦, and 85◦.

**Figure 14.** Generated power results. (**A**) Results showing the vane angle and power and (**B**) results showing the rotation speed and power.

The results from the simulation of the relationship between the generated power and rotation speed are shown in Figure 14B, the power output increased to 3401 watts at 25,000 rpm.

#### *4.2. Pressure and Path Line of Flowing*

The value of the pressure is shown in Figure 15. Pressure reached the maximum at the inlet and was lower at the outlet. The value at the inlet was 1.67 atm, at the volute was 1.46 atm, at the turbine blade was 1.20 atm, and at the outlet was 1 atm.

**Figure 15.** Pressure at all turbine walls.

The value of the pressure was varied as shown in Figure 16. The highest pressure was 1.5 atm and the lowest at 1 atm. Pressure in the turbine blade in Figure 17 shows the characteristic of the exhaust gas flow in the gap between the rotor and volute, expressed as a flowing line from the entrance through the rotor blade and flowed out in front of the rotor.

**Figure 16.** Pressure at the turbine blade.

**Figure 17.** Path line of flowing at the turbine blade.

The sound speed at the turbine blade was 490–530 m/s. This sound speed was an acceptable value for the shock wave design as shown in Figure 18.

**Figure 18.** Sound speed at the turbine blade.

#### **5. Experimental Results**

#### *5.1. Experimental Setup*

The experimental facility shown in Figure 19 used in this study was based on a real internal combustion engine, and experiment testing works without load. The experimental design for the experiment consists of four main parts: The engine building, turbo-generator unit, thermoelectric-generator unit, and parameter output measurement. The engine was based on a 2500 cc. diesel engine (Toyota 2LII model). The engine consisted of a water-cooling system, oil lubricating system, and engine electric system. The turbo-generator consisted of a turbine, volute, vane, and generator. The connection of the turbine to the generator used coupling and the gear reducer rotation speed. The thermoelectric-generator consisted of a water-cooling box, heat exchange box, and thermoelectric plate. Finally, the performance measurement aimed to find the engine rotating speed, turbo-generator rotating speed, temperature and pressure measurement, exhaust gas measurement, and finally the power measurement of the turbo-generator and thermoelectric-generator.

Each experiment performed 10 times with the same condition. The values from results recorded after the engine runs 20 min so that the engine was working in the heat condition. The speed of the engine was adjusted with step 100 rpm started from 1000 to 3600 rpm. The values recorded consisted of the engine rotation speed, inlet and outlet temperature of the turbo-generator, inlet and outlet temperature of the thermoelectric-generator, the power value from the turbo-generator, and the power value from the thermoelectric-generator.

**Figure 19.** The real internal combustion engine test.

#### *5.2. The Measurement Intrument*

The set of instruments for measuring values from the experiments were created specifically. The main components were divided into three parts: The processing and interpret sections, data recording section, and the measuring sensors detailed as follows:

The first section was the processing section using the STM32F407 microcontrollers for receiving the values from the sensor and displaying the values with the LCD monitor.

Since you have not mentioned about the details of this sensors/apparatus, so that in order to understand the reported results the measurement accuracy of the discussed

The second part was the section for receiving values from the microcontroller to store in the SD card. The SD card module recorded the values in two ways: When the value changed and recorded the values every second.

The third part was the sensor that sent various values in both the digital and analog signal to the microcontroller composing of the following values: Engine rotation speed, turbo generator rotation speed, inlet temperature of the turbo generator, outlet temperature of the turbo generator, inlet temperature of thermoelectric, outlet temperature of thermoelectric, inlet pressure of the turbo generator, outlet pressure of the turbo generator, output power turbo generator, and output power of thermoelectric. The details of the sensor used as following: Engine rotation speed and turbo generator rotation speed used the hall effect proximity switch (NJK-5002C), inlet temperature of the

turbo generator, outlet temperature of the turbo generator, inlet temperature of thermoelectric, and outlet temperature of the thermoelectric used resistance temperature detector (PT100 RTD), inlet pressure of the turbo generator, and outlet pressure of the turbo generator used MEMS pressure sensor (XGZP701DB1R), output power turbo generator, and output power of thermoelectric used current detection sensor module (WCS1800).

These instruments are comparable to the standard instruments and adjusted to the precise value to solve the problem of variance of experimental results, as the results of each trial result, the value of the measurement instrument is different, but there will be a tendency or appearance of the same rise or decline. This causes different values due to a number of factors such as how long it takes to start an extended engine to heat up the inside of the engine, and the heat in the exhaust is more as well. The acceleration of the engine at each time of the trial is not fixed. Sometimes the speed of the cycle may be accelerated or sometimes slow acceleration. This affects the exhaust pressure and the heat of the exhaust, and the installation in experiments does not start at the same heat, can affect the various values that come out. Therefore, it uses multiple experiments to find the average value that makes the measured value more reliable. Before the actual experiments, the trial operator was experimented 20 times to make the actual experiment the most accurate. In conclusion, each test was made when the engine was at a normal temperature (before starting the engine). The values recorded starting from the operating temperature of the engine that was 80 degrees Celsius. The engine acceleration gradually accelerated from the light to the peak. In each trial, there was a time equivalent of 1 h. The experiment would repeat 20 times in each experiment and then select 10 times of experimental results with the lowest variance and maximum reliability. The selected variance with uncertainty between ± 0.1%–3.0%.

#### *5.3. Temperature of the Exhaust Turbo-Generator*

The temperature experiment on the turbo-generator was undertaken by running the engine and adjusting the engine idle speed to the maximum speed. The temperature sensor measurement was installed at the inlet and outlet of the turbo-generator as shown in Figure 20. The graph shows the correlation value between the rotation speed engine and exhaust gas temperature. The trial operation repeated 10 times for calculating the average value, as shown in the picture. The experiments consisted of measurements inlet and outlet exhaust gas temperature in the turbo-generator and recorded the values from the starting point of the engine (600 rpm) and accelerated the speed of the engine to the maximum speed (4700 rpm). The X-axis displays the engine rotation speed in rounds per minute (rpm). The Y-axis displays the exhaust gas temperature in degrees Celsius. The red line shows the outlet exhaust gas temperature and blue line shows the inlet exhaust gas temperature.

**Figure 20.** The inlet and outlet exhaust temperature.

The temperature of the exhaust gas depends on the rotation speed of the engine and the engine running time. At the beginning of the rotation, the exhaust gas temperature was lower. It will increase as the speed of the engine increases, as a result of the heat combustion of exhaust gas. The inlet temperature was higher than the outlet temperature since the direct proportion characteristics.

#### *5.4. Pressure of the Exhaust Turbo-Generator*

Pressure measurements of the turbo-generator were taken by running the engine and adjusting the speed of the accelerator from idle speed to the maximum speed. The measurements were taken at the inlet and outlet of the turbo-generator as illustrated in Figure 21. The graph shows the correlation value between the rotation speed engine and exhaust gas pressure. The trial was operated 10 times for calculating the average value as shown in the picture. The experiments consisted of the measurements inlet and outlet exhaust gas pressure in the turbo-generator and recording the values from the starting point of the engine (600 rpm) and increase the speed of the engine to the maximum speed (4700 rpm). The X-axis displays the engine rotation speed in rounds per minute (rpm). The Y-axis displays the exhaust gas pressure in kilopascals. The red line shows the inlet exhaust gas pressure and the blue line shows the outlet exhaust gas pressure.

**Figure 21.** Inlet and outlet exhaust pressure.

The rotation speed of the engine and runtime engine defines the pressure of the exhaust gases. During the initial stage, the rotation speed of the pressure was low. It will increase as the speed increases, as a result of the exhaust gas flow rate. The outlet pressure was lower than the inlet pressure.

#### *5.5. Rotation Speed of the Turbo-Generator Test Based on a Variable Vane Angle*

The rotational test of the rotation speed of the turbo-generator was undertaken by running the engine and adjusting the accelerator speed from idle to maximum speed, along with an adjustment of the vane angle testing set. The results are shown in Figure 18. The graphs show the correlation value between the rotation speed engine and the turbo-generator rotation speed with 10 times the repeat experiments for calculating the average value, as shown in the graph. The experiments measured the engine rotation speed in the pulley position of the engine, measured the turbo-generator rotation speed in the axis position of the turbo-generator and recorded the value from the starting point of the engine at a lighter speed (700 rpm), and then accelerated the engine speed to the maximum speed (3600 rpm). The X-axis shows the engine rotation speed in rpm. The Y-axis shows the turbo-generator rotation speed in the rpm. The red line shows a 52◦ vane angle turbo-generator rotation speed, the green line showing a 62◦ vane angle turbo-generator rotation speed, and the blue line displaying a 72◦ vane angle turbo-generator rotation speed.

Figure 22 shows three different conditions of the engine speed and vane angle. When the vane angle was 62◦, speed increased at a stable rate and had a tendency to continuously increase. When the vane angle was 52◦, the turbo-generator started to rotate at a high engine speed. When the vane angle was 72◦ the turbo-generator speed increased as the engine speed increased, and it tended to decrease at a higher engine speed. Increasing the number of vanes decreased the axial flow velocity and increased the turbulence energy [23].

**Figure 22.** Rotation speed of the turbo-generator test.

#### *5.6. Power of the Turbo-Generator Test Based on Adjusting the Vane Angle*

The power test of the rotation speed of the turbo-generator was undertaken by running the engine and by adjusting the accelerator speed from idle speed to maximum speed. The vane angle was adjusted for each testing set as shown in Figure 19. The graphs display the correlation value between the rotation speed engine and the turbo-generator power by performing 10 iterations of the experiment and calculating the average value. The experiment consisted of measuring the engine rotation speed in the pulley position of the engine and the turbo-generator power in the axis of output power of turbo-generator. Then, the values recorded from the starting point of the engine at 700 rpm and accelerated the speed of the engine to the maximum speed (3600 rpm). The X-axis is a rotation speed engine in rpm. The Y-axis is the turbo-generator power in watts. The red line represents the 62◦ vane angle turbo-generator rotation power, the green line represents the 72◦ vane angle turbo-generator rotation power, and the blue line represents the 52◦ vane angle turbo-generator rotation power.

Figure 23 illustrates the correlation value of the engine rotation cycle with the capacity outputted from the turbo-generator with the angle adjustment of three vane levels. The total output of energy was increased when the cycle speed of the turbo-generator engine was increased with adjusting the angle of the vane at 62◦ and 72◦. The engine could produce energy from the beginning at low speed up to the maximum speed while the turbo-generator adjusted the angle of the vane at 52◦ must be added to the engine rotation cycle to 2400 rpm to start producing power output. As shown in the graph, the turbo-generator with the vane at 62◦ had a capacity of producing the maximum energy starting from 150 watts at the engine rotation around 900 rpm and increased by rotation cycle to 900 watt at 3600 rpm.

**Figure 23.** Power of the turbo-generator test.

Turbine efficiency:

$$
\eta\_t = \frac{T\_{00} - T\_{20}}{T\_{00} - T\_{2s}} = 51.1\%. \tag{46}
$$

Total Shaft Output Power:

The reduction of turbo-generator energy can be divided into three speed of different power: Idle speed (145 watts), running speed (870 watts), and high speed (890 watts).

#### *5.7. Power of the Thermoelectric-Generator Test*

The power output from the thermoelectric generator shown in Figure 24. The graph shows the relationship value between the engine running time and the thermoelectric-generator output power, with 10 repeat experiments for calculating the average value. The experiments consist of measuring the engine running time when starting the engine and measuring the thermoelectric-generator output power in the power cord position. The values were recorded from the starting point of the engine at 700 rpm, and accelerated the speed of the engine to the maximum speed (3600 rpm). The X-axis is the engine running time, showing the value in seconds. The Y-axis is the thermoelectric-generator output power display value in watts. The red line is the inlet exhaust gas temperature, the yellow line is the outlet exhaust gas temperature, the green line is the output voltage, the purple line is the output current, and the blue line is the output power. The line-graph illustrates the exhaust temperature and power when running the engine. Overall, the power output is related to the temperature difference. To begin with, the power output started at 271 watts, before it increased to just under 392 watts at a higher temperature. This fell slightly when the engine was stopped, before dropping to its lowest point of 288 watts at the end of testing.

**Figure 24.** Power of the thermoelectric-generator test.

Figure 25 shows the correlation between the rotation speed of the engine and output power of the whole system. The trial operation was repeated 10 times for calculating the average value. The X-axis represents the engine rotation speed. The Y-axis represents the output power. The red line represents the energy from the turbo-generator. The green line represents the energy from the thermo-electric generator, and the blue line represents the total power from the turbo-generator and thermo-electric generator. The total power saving when combining both the turbo-generator and thermo-electric generator units was 1262 watts.

**Figure 25.** Total output power.

#### **6. Conclusions**

Turbo-generators and thermoelectric generators are ideal for the recovery of waste energy in exhaust gas. The temperature and flow pressure of the combustion engine is transformed into electric power. That power can be converted into electrical energy to support the electrical supply in a vehicle. Normally, exhaust gas is discarded to the environment as waste gas. The main difference between this research and previous works is the type of energy from exhaust gases for recycling. Most of previous works used either pressure or heat whilst this research focus on the discharge of energy from the exhaust with both pressure and heat recovery to the form of electrical energy. In addition, this research designed the prototype for a 2500 cubic centimeters diesel engine, which is widely used. The research aimed to design a turbo-generator and thermoelectric generator that could convert exhaust gas energy to electrical energy. The turbo-generator model could generate up to 870 watts of power at 3400 rpm (top speed of the engine), and the thermoelectric generator could generate up to 392 watts of power. Both systems combined could generate up to 1262 watts of power. The electric power can be used in electric charging; generally, the alternator will produce 12 volts, or 35 amperes (420 watts), to support in-vehicle usage. The experiments were a comparison of the results from the simulation from the design and optimization of the vane angle of 40◦, 50◦, 62◦, 70◦, and 80◦ with the results of the installation and the actual test. The result shows that everything was consistent in the design with the angle vane 62◦. Besides that, in the simulation process, the result shows that the angle vane gave the maximum torque and power, which corresponded to the same direction as the outcome of the actual trial. The integrated turbo-generator and thermoelectric generator could potentially be used as part of the charging system. Energy recovery during driving can be used for battery charging and as an energy storage device. In this model, the turbo-generator and thermoelectric generator had the capacity to generate up to 1262 watts that could be utilized in hybrid vehicles. The power from the recovery of waste energy could support the electrical needs of other parts. At present, the device converts the power into electricity, and for the convenience of the driver, it could be utilized for features such as electric air conditioning, electric power steering, refrigerators, and mobile phone chargers.

**Author Contributions:** P.N. contributed to the conceptualization, methodology, validation, and data collection, and wrote the paper; M.E., U.N. and E.B. made corrections to the paper and gave some useful recommendations.

**Funding:** This research was funded by Asian Institute of Technology, School of Engineering and Technology, Thailand.

**Acknowledgments:** The authors would like to thank the Rajamangala University of Technology Srivijaya for scholarship in doctoral study.

**Conflicts of Interest:** 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* **Classifying the Level of Energy-Environmental E**ffi**ciency Rating of Brazilian Ethanol**

#### **Nilsa Duarte da Silva Lima, Irenilza de Alencar Nääs, João Gilberto Mendes dos Reis \* and Raquel Baracat Tosi Rodrigues da Silva**

Postgraduate Program in Production Engineering, Universidade Paulista - UNIP, Dr. Bacelar Street 1212, 04026002 São Paulo, Brazil; nilsa.lima@stricto.unip.br (N.D.d.S.L.); irenilza.naas@docente.unip.br (I.d.A.N.); raquel.silva@stricto.unip.br (R.B.T.R.d.S.)

**\*** Correspondence: joao.reis@docente.unip.br; Tel.: +55-11-5586-4145

Received: 28 February 2020; Accepted: 14 April 2020; Published: 21 April 2020

**Abstract:** The present study aimed to assess and classify energy-environmental efficiency levels to reduce greenhouse gas emissions in the production, commercialization, and use of biofuels certified by the Brazilian National Biofuel Policy (RenovaBio). The parameters of the level of energy-environmental efficiency were standardized and categorized according to the Energy-Environmental Efficiency Rating (E-EER). The rating scale varied between lower efficiency (D) and high efficiency + (highest efficiency A+). The classification method with the J48 decision tree and naive Bayes algorithms was used to predict the models. The classification of the E-EER scores using a decision tree using the J48 algorithm and Bayesian classifiers using the naive Bayes algorithm produced decision tree models efficient at estimating the efficiency level of Brazilian ethanol producers and importers certified by the RenovaBio. The rules generated by the models can assess the level classes (efficiency scores) according to the scale discretized into high efficiency (Classification A), average efficiency (Classification B), and standard efficiency (Classification C). These results might generate an ethanol energy-environmental efficiency label for the end consumers and resellers of the product, to assist in making a purchase decision concerning its performance. The best classification model was naive Bayes, compared to the J48 decision tree. The classification of the Energy Efficiency Note levels using the naive Bayes algorithm produced a model capable of estimating the efficiency level of Brazilian ethanol to create labels.

**Keywords:** biofuel policy; efficiency rating; ethanol; data mining

#### **1. Introduction**

Developing renewable energy is one of the leading global interests in promoting sustainability and environmental quality, including modern electricity grids worldwide, which have begun to rely more heavily on renewable energy sources [1–3]. Ethanol is a renewable fuel produced by the fermentation of sugarcane extract and molasses. The product has a lower carbon footprint, is biodegradable, and has greater energy-environmental efficiency (renewable energy) compared to oil due to its sustainability in the production chain with better use of natural resources [4–10]. Ethanol is one of the main biofuels consumed in Brazil.

Biofuel partially (or entirely) replaces fossil fuels in engines (flex vehicles) [11]. The addition of 27% ethanol in gasoline (Cgasoline, with the addition of anhydrous ethanol fuel) has been mandatory in Brazil since 2015 [8,12–14]. Such an initiative expanded the country's consumption of biofuels and increased the energy matrix's sustainability. It also supported the goal of reducing 37% of GHG emissions by 2025, compared to 2005 levels [2,12,15,16]. Differently, some European Union countries also include electromobility (electric cars) in their GHG emission reduction forecasts [17].

The production of biofuel from sugarcane is supported by the National Agency of Petroleum, Natural Gas and Biofuels' (ANP) sustainable development programs to achieve goals to reduce greenhouse gas emissions. Therefore, annual national targets were developed by the Brazilian National Biofuel Policy (RenovaBio) [12,15]. These goals are established in units of Decarbonization Credits (CBios) calculated from annual mandatory targets defined individually for each producer and distributor according to their participation in the fossil fuel market [12,16,18–21].

The certification of biofuel production assigns different marks to each biofuel producer and importer, in a value inversely related to the carbon amount of biofuel produced. The note reflects the individual contribution of each producing agent to mitigate a specific amount of GHG concerning its fossil substitute (in tons of CO2 equivalent). The total emission is compared with that of the equivalent fossil fuel (gasoline, for ethanol), resulting in a final score (Energy-Environmental Efficiency Rating), characterizing the mitigation of emissions. This note generates CBios for biofuel producers and importers; with the decarbonization of the Brazilian energy matrix, there is a mechanism for the commercialization of these CBios linked to the carbon intensity of biofuels [2,12,16,18].

The incentive to reduce pollutant emissions in the biofuel chain goes far beyond the use of flex vehicles by consumers [9,11,22], but it is directly linked to decarbonization credits given to biofuel producers and distributors, because although renewable, it depends on how sugarcane is produced [5,8,23–26]. Alkimim and Clarke [27] showed that the carbon debt of deforestation in Brazilian biomes for ethanol production was equivalent to 608 Mg CO2 ha−<sup>1</sup> in the Amazon, 142 Mg CO2 ha−<sup>1</sup> in the Cerrado, and 212 Mg CO2 ha−<sup>1</sup> to the Atlantic Forest with the respective return time of 62, 15, and 22 years. In this context, it is essential to integrate dissemination strategies, clear and comprehensive, on what is the level of energy-environmental efficiency of biofuel producers and distributors [22,28,29], in order to raise awareness and increase the number of certificates at RenovaBio, the membership of which is not mandatory.

Predictive models are essential for the biofuel chain for both the consumer and the distributor as this classification of the level of efficiency can generate labels of their energy-environmental performance, increasing transparency and environmental responsibility concerning a product such as ethanol. The use of data mining to assess the performance of classifiers in the sugarcane sector is quite broad. It ranges from classifiers for mapping sugarcane planting [30] to deep learning techniques for the detection of sugarcane diseases [31] and classification of crop yield characteristics with neural networks used both in the recognition and in the grading of satellite images of sugarcane plantations [32].

The present study aimed to assess and classify energy-environmental efficiency levels to reduce greenhouse gas emissions in the production, commercialization, and use of biofuels certified by the Brazilian National Biofuel Policy (RenovaBio).

#### **2. Material and Methods**

#### *2.1. Characterization of the Study*

The study considered certification data for the production or efficient import of biofuels from the Brazilian National Biofuel Policy (RenovaBio) regulated by the National Agency of Petroleum, Natural Gas and Biofuels [12,33,34]. The Certificate of Efficient Production of Biofuels includes an Energy-Environmental Efficiency Note, resulting from the technical profile informed by a calculation validation spreadsheet (RenovaCalc) [2,12,33,34], and is linked to the volume of biofuel produced and marketed, generating the Decarbonization Credits (CBio) of the RenovaBio to reduce carbon emissions and improve the performance of biofuels. In addition to offering CBio, it is generated by the difference between the CO2 equivalent emission of fossil fuels (baseline) and its biofuel substitute [12,18]. The biofuels evaluated in the study were anhydrous ethanol and hydrous ethanol. Only the database of the certificates of the efficient production or import of biofuels approved in 2019 was evaluated [12].

#### *2.2. Classifying the Biofuel Energetic-Environmental E*ffi*ciency*

The classification levels of the biofuel environmental efficiency were performed based on the Energy-Environmental Efficiency Rating. Such a rating is the result of the certificates of the production or efficient import of biofuels informed through the calculation (RenovaCalc), linked to the volume of biofuel produced and commercialized, generating Decarbonization Credits (CBio) in the RenovaBio's program.

The energy-environmental efficiency level parameters were standardized according to Table 1. The dataset was categorized according to the Energy-Environmental Efficiency Rating (RenovaBio) [12], the rating scale varying between lower efficiency and high efficiency + (highest efficiency), and an example of the label's design of the energy-environmental efficiency for the performance of a certificate is given (Figure 1).


**Table 1.** Parameters of the level of the Energy-Environmental Efficiency Rating.

<sup>1</sup> Based on the RenovaBio; <sup>2</sup> Efficiency Classes.

**Figure 1.** Energy-environmental efficiency label design for performance certificates.

The calculation and indication of the environmental energy efficiency score (E-EER) of the certification for efficient production of biofuel, made available by the ANP, was discretized into categories of levels (classes) (pre-processing of the dataset) for the classification (data mining) and to create one of the labels of the energy-environmental performance (post-processing) of Brazilian ethanol. This labeling system allowed it to be classified into five classes (A+, A, B, C, and D) to provide consumers with a differentiation of the ethanol consumed from different producers, regions, or states.

The RenovaBio Program allows producers and importers to be able to declare the energy-environmental efficiency of their product, which is economically attractive for decarbonization and the competitiveness of biofuels in the oil market, with a complex and solid structure (Figure 2) [2,12,18,33,34]. The label may be shown at fuel pumps to consumers with a validity of one to three years, a validity that is applied to the Certification of Efficient Production of Biofuels when approved by the ANP. It can also endorse the information and increase transparency in the biofuel market at the consumer level, helping to make a purchase decision. The objective of energy-environmental labeling is to encourage Brazilian sugar and alcohol industries to develop innovations and improvements beyond the minimum levels of efficiency. However, it is expected that

more ethanol producers will be able to adhere to the ANP certifications of RenovaBio [2,12,19,33,34], and consequently, the labeling system can be improved with the inclusion of more data on the platform.

**Figure 2.** Plan of the emission and trade of decarbonization credits (CBios) created by the RenovaBioProgram, adapted by Klein et al. [19].

#### *2.3. Classification of Model Prediction*

Data mining applies to this study, through techniques (algorithms), for the classification of the levels (classes) of energy-environmental efficiency in the search of strategic information that allows the extraction of implicit information existing in the databases, contributing to the process of identifying and classifying new patterns [35–37]. The steps of the data mining method were selection, pre-processing, data mining, and post-processing (knowledge filtering, interpretation and explanation, evaluation, and knowledge integration) for knowledge discovery from the classifiers [36,37]. The results obtained could be used in information management, information processing, decision making, and process control.

The data contained in the databases could be used to learn a specific target concept [35–38]. The tasks performed by data mining techniques and machine learning, the classification, build models that can be applied to unclassified data to categorize them into classes, to relate the meta attribute (whose value will be predicted) and a set of forecasting attributes [35–38].

The data were assessed in the ANP database for the registration of certificates of the production or efficient import of biofuels approved and included in the RenovaBio program in 2019 [12,33,34]. We considered only anhydrous ethanol and hydrated ethanol products, generating two products for the same biofuel producer and importer. The data pre-processing was performed in Excel spreadsheets for further processing in the data mining software Weka © (Waikato Environment for Knowledge Analysis) Version 3.8.4 [39–42]. The attributes used to build the predictive model were: "biofuel-type", "state", "eligible-volume (%)", "emission factor", "Energy-Environmental Efficiency Rating", and "LER" (Level of Efficiency Rating). Figure 3 presents the modeling process used to classify the Energy-Environmental Efficiency Rating.

**Figure 3.** Schematic of the modeling process used to classify the Energy-Environmental Efficiency Rating of Brazilian ethanol. E-EER, Energy-Environmental Efficiency Rating; LER, Level of Efficiency Rating.

A classifier is a mapping from unlabeled instances to (discrete) classes. Classifiers have a form (classification tree) plus an interpretation procedure (including how to handle unknown values). Most classifiers also can provide probability estimates (or other likelihood scores), which can be thresholded to yield a discrete class decision, thereby taking into account a cost/benefit or utility function [43,44].

During the pre-processing of the data, the dataset was extracted from the RenovaBio Program (ANP) platform, selecting only the data on the ethanol product and organized in a spreadsheet. The implementation of the supervised filter Resample was applied to maintain the distribution of classes in the subsample and to reach a uniform distribution for comparing the data not submitted to the filter (noResample). The filter Resample produces a random subsample of a dataset using either sampling with replacement or without replacement. The filter was made to maintain the class distribution in the subsample or to bias the class distribution toward a uniform distribution.

The J48 decision tree classification algorithms ("*weka*.*classi fiers*.*trees*.*J*48 − *C*0.25 − *M*2" without the filter and in relation to "*data* − *bio f uel* − *weka*. *filters*.*supervised*.*instance*.*Resample* − *B*0.0 − *S*1 − *Z*100.0" using the resample filter) and the naive Bayes algorithm ("*weka*.*classi fiers*.*bayes*.*NaiveBayes*" without the filter and in relation to "*data* − *bio f uel* − *weka*. *filters*.*supervised*.*instance*.*Resample* − *B*0.0 − *S*1 − *Z*100.0" using the resample filter) were used. The model validation was done using cross-validation (10 folds) applied to the dataset of 27 instances and six attributes ("biofuel-type", "state", "eligible-volume (%)", "emission-factor", "energy-environmental-efficiency-rating", and "LER" (Level of Efficiency Rating)). For each sample, the known class label was compared with the prediction of the learned class model.

In supervised learning, each data input object is preassigned a class label. The main task of supervised algorithms (J48 and naive Bayes) is to learn a model that produces the same labeling for the provided data [43–45]. The decision tree algorithm is a widely used algorithm for classification, which uses attribute values to partition the decision space into smaller subspaces in an iterative manner, and the decision processes can be represented graphically as a tree. Each possible decision is covered and represented as a branch, and a complete decision process is essentially a path or branch from the root node to a leaf [43,44]. Naive Bayes is a classification algorithm widely used in

problems due to its simplicity, effectiveness, and robustness, being a probabilistic approach based on assumptions that resources are independent of each other and that their weights are equally important [46]. They can better represent the complex relationships between input variables found in real problems [46]. Probabilistic inference can be studied as an approach based on the assumption that decision variables follow probable distributions. The essence of a Bayesian classifier is to estimate the probabilities of all alternative models or hypotheses, given data as evidence, and then to find the most probable classification to be assigned to each new input [44,46].

In the step of the post-processing for filtering, interpretation and explanation, evaluation, and knowledge integration generated by the algorithms for knowledge discovery from the classifiers, the metrics of the performance of the algorithms were used. Post-processing procedures usually include various pruning routines, rule quality processing, rule filtering, rule combination, model combination, or even knowledge integration [47].

The last step is to evaluate the prediction, and such an analysis was made based on the performance values obtained through the test of the prediction model [48]. The performance evaluation measures of the prediction models used were the confusion matrix, accuracy, precision, and recall and the correlation coefficient between classes (Matthews Correlation Coefficient (MCC)) for testing with the resample filter. The Kappa statistic measured the learning capacity of the algorithm.

The confusion matrix presented a matrix with results obtained during the test phase of the model, and it was used in models that used classification algorithms. Considering a confusion matrix of a hypothesis, it offered an adequate measure of the classification model, by showing the number of correct classifications versus the predicted rankings for each class, over a set of instances. The number of correct answers, for each class, was located on the main diagonal of the matrix, and the other elements represented errors in the classification.

The precision represents what has been classified correctly. The values obtained in correctly classified instances and incorrectly classified instances are determinant for predicted accuracy, since they display the values of correct classification and incorrect classification obtained by the algorithm (Equation (1)).

$$Precision = \frac{TP}{(TP + FP)}\tag{1}$$

where TP = True Positive; FP = False Positive . The sensitivity (recall) signifies the proportion of wrong classifications or the occurrence of defects. In addition to accuracy and precision, its value varies from 0 to 1, with values closer to 1 being indicators of a good performance prediction model obtained by Equation (2).

$$(Sensitivity(Recall)) = \frac{TP}{(TP + FN)}\tag{2}$$

where TP = True Positive; FN = False Negative.

The Kappa statistic is a metric that compares an observed precision with an expected precision (random chance). It is a measure used to deal with multi-class and unbalanced class problems. The Kappa statistic can be defined as a measure of the degree of agreement between two categorized datasets. The Kappa result varies between 0 and 1. The higher the Kappa value, the stronger the bond [49] (Equation (3)).

$$Kappastatistic = \frac{P\_O - P\_E}{1 - P\_E} \tag{3}$$

where *PO* = proportion of observed agreements; *PE* = proportion of agreements expected by chance.

The Matthews Correlation Coefficient (MCC) is a correlation coefficient between the dependent classes and represents a measure of quality. Unlike accuracy, precision, and sensitivity, its value ranges from −1 to 1, where values closer to −1 are indicators of a poor prediction model, values equal to 0 indicate that the prediction model is entirely random, and values closer to 1 are indicators of a prediction model with good performance (Equation (4)).

$$\text{MCC} = \frac{TP \ast TN - FP \ast FN}{(TP + FP)(TP + FN)(TN + FP)(TN + FN)} \tag{4}$$

where TP = True Positive; TN = True Negative; P = False Positive; FN = False Negative.

#### **3. Results**

#### *3.1. Classification of the Energy-Environmental E*ffi*ciency Level for Biofuels*

The **If-Then** classification rules are presented related to the energy-environmental efficiency level for biofuels.

The decision tree generated by the J48 algorithm presented the following classification rules (Figure 4): **If** the energy-environmental efficiency deficiency (E-EER) was higher than 60.3, **then** the classification was A (high efficiency). **If** the Energy-Environmental Efficiency Rating (E-EER) was less than or equal to 60.3, **then** the rating depended on the state where the ethanol was produced. **If** the state was Goiás (GO), **then** the classification was C (standard efficiency). **If** the state was São Paulo (SP), **then** the classification was B (average efficiency). **If** the state was Mato Grosso do Sul (MS), **then** the rating was B (average efficiency). The results indicated that the model classified with precision above 60.

**Figure 4.** Decision tree for classifying the level of the energy-environmental efficiency of biofuels.

The performance of the classifiers in predicting the classes of the E-EER level using the J48 decision tree algorithm (Table 2) showed 74.07% of instances classified correctly and 25.93% for those classified incorrectly, and the learning capacity of the algorithm was 0.56 for the Kappa statistic. Class A showed the best 100% True Positive (TP) rate, with a 0.07 False Positive (FP) rate, a precision of 0.93, and high recall. However, for Class C, the precision was zero, not classifying any.


**Table 2.** Classifier performance (J48 decision tree) for scale class prediction models of the level of the energy-environmental efficiency of biofuel.

WA = Weighted Average. TP rate = True Positive rate. FP rate = False Positive rate. MCC = Matthews Correlation Coefficient.

The performance of the naive Bayes algorithm showed 81.48 of instances correctly classified and 18.52% for those classified incorrectly, with a learning capacity of 0.70 for the Kappa statistic. It classified Classes A and B with high performance, with an accuracy of 1.00 and 0.73, respectively (Table 3). Comparing the two algorithms, naive Bayes presented better performance indexes concerning the J48 decision tree.


**Table 3.** Naive Bayes classifier performance for scale class prediction models of the level of the energy-environmental efficiency of biofuel.

WA = Weighted Average. TP rate = True Positive rate. FP rate = False Positive rate. MCC = Matthews Correlation Coefficient.

The decision tree generated by the J48 algorithm for the dataset with the application of the resample filter (Figure 5) presented the following classification rules (Figure 4). **If** the Energy-Environmental Efficiency Rating (E-EER) is higher than 60.1, **then** the classification is A (high efficiency). **If** the Energy-Environmental Efficiency Rating (E-EER) is less than or equal to 60.1, **then** the classification depends on the eligible volume (%) of the biomass. **If** the eligible volume is higher than 97.43, **then** the rating is B (average efficiency). **If** the eligible volume is less than or equal to 97.43, **then** the rating is C (standard efficiency) (Figure 5). The results indicated that the use of the resample filter had higher weight for the appropriate attribute volume during the evaluation of the E-EER and that ethanol production depended directly on this volume.

**Figure 5.** Decision tree for the energy-environmental level classification of biofuels using the Resample filter. GO, Goiás; SP, São Paulo; MS, Mato Grosso do Sul.

The performance of the classifiers in the prediction of the classes of the E-EER level using the J48 decision tree algorithm and with the application of the resample filter (Table 4) presented 81.48% of instances correctly classified and 18.52% for those classified incorrectly, and the algorithm learning ability was 0.70 for the Kappa statistic. Classes A and B showed high prediction with values of 1.00 and 0.78, respectively. However, the forecast for the C class was 0.40.


**Table 4.** Classifier performance for scale class prediction models of the level of the environmental efficiency of biofuel with the resample filter.

WA = Weighted Average. TP rate = True Positive rate. FP rate = False Positive rate. MCC = Matthews Correlation Coefficient.

The performance of the naive Bayes algorithm showed 77.78% of instances classified correctly and 22.22% for those classified incorrectly, with a learning capacity of 0.64 for the Kappa statistic (Table 5). It classified Classes A and B with high performance, with an accuracy of 0.77 and 0.90, respectively (Table 5). The J48 decision tree showed better performance indexes when compared to naive Bayes.


**Table 5.** Classifier performance (naive Bayes) for scale class prediction models of the level of the environmental efficiency of biofuel using the resample filter.

WA = Weighted Average. TP rate = True Positive rate. FP rate = False Positive rate. MCC = Matthews Correlation Coefficient.

The confusion matrix for the J48 decision tree algorithm presented results with positive gains for the classes when the resample filter was applied, going from 20 to 22 strikes, specifically for Class C (Table 6). This gain in the resample increased the normal balance or distribution of the dataset, shown in Table 4. The results of the Matthews Correlation Coefficient (MCC) proved this gain with values of 1.00 for A, 0.85 for B, and 0.79 for C.

**Table 6.** Confusion matrix of the classification of the level of energy-environmental efficiency with the J48 decision tree.


The confusion matrix for the naive Bayes algorithm also showed results with positive gains for the classes when the resample filter was applied, going from 20 to 21 hits, specifically for Class C, with two hits (Table 7). However, this gain did not have good accuracy when we observed the results of Matthews Correlation Coefficient (MCC) (Table 5), which presented values of 0.80 for A, 0.64 for B, and 0.41 for C, when compared to the J48 decision tree algorithm in the same analysis condition, which performed better.


**Table 7.** Confusion matrix of the classification of the level of energy-environmental efficiency with naive Bayes.

The results showed that the classification using the naive Bayes algorithm was better than J48, in the approach without the resample filter. It classified the minority class (C) better and presented a higher degree of agreement (Kappa statistic). It also indicated high performance by the Matthews correlation coefficient concerning the J48 decision tree algorithm. However, the use of the resample filter in both algorithms improved the distribution of classes in the confusion matrix (Tables 6 and 7), especially Class C.

#### **4. Discussion**

The decision tree signalized the production state importance. The E-EER value was higher for the most productive states (SP, MS). However, Brazilian ethanol producers' certification was still small concerning the high number of states and the volume of ethanol produced in Brazil. The classification of the energy-environmental efficiency of Goiás state did not show its status as one of the states that produces the most ethanol in Brazil [50]. The number of certified producers was still less than it should be. Such a public policy (RenovaBio) was implemented in 2018, so far with little adhesion by the producers and importers in producing states.

The comparison of the J48 decision tree and naive Bayes algorithms, applied to the dataset without using filters, showed differences in the performance indicators. The naive Bayes algorithm, when compared to the J48 decision tree, presented better performance results, predicted Class C better (50% higher in the TP rate), and had better learning capacity (higher Kappa statistic value), indicating that the prediction model had a good performance with this algorithm. However, the sensitivity of the model decreased. The application of the Resample filter in the pre-processing of data for classification with the J48 decision tree algorithm was the one that showed the best performance compared to naive Bayes, as the Kappa statistic was high and the MMC for all classes was higher, indicating that the prediction model performed well.

The data mining technique is also applied in the area of environmental impacts of sugarcane production, for predicting the energy produced and the environmental impacts. Artificial intelligence methods, artificial neural networks, and adaptive neural fuzzy inference system models are also used to predict the environmental impacts of the life cycle and energy output of sugarcane production on planted farms [51].

Integrated approaches based on complex systems for forecasting the growth of sugarcane based on meteorological parameters using extreme machine learning and neural networks were able to show a more generalized model of forecasting for the growth of sugarcane, bringing benefits to industry and the community [52].

Brazil is a major producer of sugarcane and a major consumer of ethanol, with intensive production in order to meet the demands of biofuel (clean energy) and to reduce the use of fossil fuel (oil). However, several efforts in public policy must come into synergy regarding the growing consumption of biofuel and the sustainable development of the sugarcane chain, mainly regarding its energy-environmental efficiency, use of inputs, and agricultural processes [5]. The ethanol energy-environmental performance labels are applicable in this context, as they would help consumers understand how the chain is evaluated mainly in terms of environmental impact and also collaborating with the transparency of public sector policies.

The most evaluated and used energy efficiency labels are for buildings, appliances, equipment, and lamps. These labeling systems are part of public energy-saving policies; since their implementation, there have been improvements in the evaluation standard, and they have also impacted consumption behavior [53–60]. The implementation of energy efficiency measures can guarantee a sustainable economy. In this context, the energy efficiency labeling program for buildings is generally designed with performance processes and standards, with a rigorous database. Eco-labeling is another system with an approach based on the environmental performance of products that also influences consumer choice [61]. Batista et al. [57] investigated the contribution of labeling to reducing the electricity consumption of buildings and noted that conventional buildings that adopt measures such as painting walls and ceilings white, in addition to using smoked glass, were sufficient to raise the rating to an A level.

The evaluation of buildings generally includes energy classification schemes and shows the difference between the Brazilian scheme and those applied by other countries to improve the labeling method [62]. Other studies have included a review of international building energy efficiency codes and labeling schemes to establish standards for the assessment and classification of buildings in terms of energy performance [63]. Lopes et al. [60] reviewed studies on energy efficiency policies and regulations for buildings, highlighting how the Brazilian program can be improved compared to the American and Portuguese programs; however, this labeling system does not inform about the reduction of GHG emissions.

Another study evaluated two new proposals for an energy efficiency label and a new method for assessing the energy efficiency of public lighting systems. The main difference between the proposals was the number of parameters evaluated. However, the current evaluation system evaluates only one parameter (energy efficiency index), and the study's proposals recommend five parameters: lamps, energy efficiency index, light pollution, renewable energy contribution, and light dimming [64].

All of these studies demonstrated that the labeling system can be implemented and improved for consolidation and active contribution to consumer behavior. However, it could also increase the contribution of the ethanol production sector with greater participation with the goal of reducing greenhouse gases and reducing energy use. The biofuel sector could benefit from the implementation of an energy-environmental labeling system. With a successful approach, it could increase adherence to the RenovaBio program and consequently increase the sector's decarbonization credits.

#### **5. Conclusions**

The Brazilian National Biofuel Policy (RenovaBio) is one of the main strategies to encourage the reduction of pollutants in the renewable energy sector from sugarcane. For this reason, the implementation of simple labeling can impact consumer behavior and increase the transparency of the incentive program to reduce environmental impacts.

After testing two classifiers, the best model evaluated was naive Bayes without the use of the resample filter compared to the J48 decision tree, also with the use of the resample filter. The classification of the Energy Efficiency Note (RenovaBio) levels using a Bayesian classifier, the naive Bayes algorithm, produced a model capable of predicting the efficiency level of Brazilian ethanol producers and importers certified to create labeling. The rules generated by the models were capable of estimating the classes according to the scale discretized into high efficiency (Classification A), average efficiency (Classification B), and standard efficiency (Classification C), with more accurate

forecasts for the observed classes. These results could generate an ethanol energy-environmental efficiency label for the end consumers and resellers of the product.

However, RenovaBio's database of ethanol was small, concerning the records of efficiency scores registered in the program, as adherence to the program is still voluntary, and the implementation is recent, hindering deeper learning in the classification of labels. Besides, a more in-depth analysis could improve the model's forecast in the generation of energy-environmental labels for biofuels.

**Author Contributions:** Conceptualization, N.D.d.S.L. and J.G.M.d.R.; methodology, N.D.d.S.L. and I.d.A.N.; software, N.D.d.S.L. and I.d.A.N.; validation, N.D.d.S.L., I.d.A.N., J.G.M.d.R., and R.B.T.R.d.S.; formal analysis, N.D.d.S.L. and I.d.A.N.; investigation, N.D.d.S.L., J.G.M.d.R., and R.B.T.R.d.S.; resources, I.d.A.N. and R.B.T.R.d.S.; data curation, N.D.d.S.L. and J.G.M.d.R.; writing, original draft preparation, N.D.d.S.L. and I.d.A.N.; writing, review and editing, I.d.A.N., J.G.M.d.R., and R.B.T.R.d.S.; visualization, I.d.A.N., J.G.M.d.R., and R.B.T.R.d.S.; supervision, I.d.A.N. and J.G.M.d.R.; project administration, N.D.d.S.L. and R.B.T.R.d.S.; funding acquisition, I.d.A.N. All authors read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Brazilian funding organizations: Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Capes Grant Number: 88882315049/2019-01, and Conselho Nacional de Desenvolvimento Científico e Tecnológico - CNPQ Grant Number: 159842/2019-0

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

#### **References**


© 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/).

*Article*
