**Definition 1.**

$$
\begin{bmatrix} P\_1 \\ P\_2 \\ P\_3 \end{bmatrix} = P\_1 \begin{bmatrix} E\_{s\_{p1}} \\ E\_{s\_{p2}} \\ E\_{s\_{p3}} \end{bmatrix} + \begin{bmatrix} V\_{op\_{p1}} \\ V\_{op\_{p2}} \\ V\_{op\_{p3}} \end{bmatrix} + \begin{bmatrix} E\_{i\_{p1}} \\ E\_{i\_{p2}} \\ E\_{i\_{p3}} \end{bmatrix} \* \begin{pmatrix} P\_1 \\ P\_2 \\ P\_3 \end{pmatrix} \tag{7}
$$

*Total required units of CO*2 *savings*

$$E\_{s\_{p1}} + V\_{op\_{p1}} + E\_{i\_{p1}} \tag{8}$$

**Definition 2.** *The matrix M is the consumption matrix.*

*Applying the Leontief Input–Output Model implies the following:*

$$
\overline{P} = M\_{\overline{P}} + \overline{d} \tag{9}
$$

*The consumption matrix is made up of consumption vectors. The jth column is the jth consumption vector and contains the necessary input required from each of the sectors for sector j to produce one CO2-saving. Vector P = production, vector d = external demand, and vector MP = internal demand.*

*The case is open if d = 0 and closed if d =* 0 *; since cases of d = 0 are rare, the case of d =* 0 *and I* − *M are as follows:*

$$
\overline{P} = (I - M)^{-1} \overline{d} \tag{10}
$$

$$\overline{P} = \left(I + M + M^2 + M^3 + \dots - \dots\right)^{-1} \overline{d} \tag{11}$$

#### *3.2. Energy and Carbon Emissions Efficiency Regression-Based Approach*

The adoption of the regression-based approach (RBA) for this study requires defining the boundary through the production operations, including energy savings commitment. The international performance measurement and verification protocol (IPMVP) framework is used in the determination of energy intensity on the realization of the low-carbon product. The compared consumption was measured before and after implementation to make suitable adjustments following the measurement and verification integration. The processes were carried out as follows:

**Step 1**. Secondary energy data from the electricity generated were obtained for produce vehicles in automotive plants for a 3-year–period was obtained through the quantitative method.

**Step 2**. Baseline year was established using 2015 and 2018 data for tracking the energy performance to capture the energy savings and energy intensity of vehicle production bodies to determine the energy improvement in energy intensity.

In **step 3**. Relevant variables were determined, including output units for the production. The considered variables of production variability, product variability, feedstock quality, and quantity were dependent on the processes and outputs. Energy performance was normalized using a regression approach for these variables to track the energy consumption and baseline year consumption. Before the normalization, variables were identified by observations using the best technical judgments, in this case, the production level and units. Energy performance indicator (EnPI) tools were used to automate the process of evaluating all possible variables for a given year.

**Step 4**. Gathering of energy consumption data was performed for the baseline year and subsequent year for annual reporting. In this study, energy analysis was carried out using the input and output theory for equating energy consumption as the total of energy sources for vehicle part production, excluding feedstock in million thermal units (MMBtu), as presented in Equation (1). Electricity was valued by the primary energy required for the generation, transmission, and distribution of energy. Monthly energy consumption was collected on production data and relevant variable data for the regression model.

**Step 5**. Using regression analysis to normalize the data, the techniques were used to estimate the dependence of actual energy consumption as a dependent variable (kWh per unit of vehicles) for a given period and the production levels of parts as an independent variable, while controlling other variables simultaneously. The regression linear model in Equation (2) was used based on the strength of estimating energy savings through the measurement and verification of projects when the variations in operation conditions included the input data, as follows:

$$E\_{\mathbb{C}} = m\_1 \mathbf{x}\_1 + m\_2 \mathbf{x}\_2 + m\_3 \mathbf{x}\_3 + \mathbf{b} \tag{12}$$

where *m*1, *m*2, *m*3 = kWh per unit of vehicles, *x*1, *x*2, *x*3 = independent variables, and b = energy use when *x*1, *x*2, *x*3 are 0 (kWh per month).

The linear equation is developed to model the energy consumption for a given period when an independent variable is set to zero using a known set of conditions. Comparing the actual energy consumed to the modeled energy consumption can help to estimate the energy performance improvements of the produced parts. Equation (2) presents the energy savings.

**Step 6.** Requires the determination of energy intensity from the baseline year. Equation (3) presents the total improvements in energy intensity depending on the regression method analysis, as follows:

$$E\_i = \sum \left( c\_{i1} \ast c\_{i1} \right) + \left( c\_{i2} \ast c\_{i2} \right) + \left( c\_{i n} \ast c\_{i n} \right) \left( c\_{i 1} + c\_{i 2} - c\_{i n} \right) \\ \quad \ast \quad \mathbf{100} \% \tag{13}$$

where *Ei* = total improvement in energy intensity, *el*1, = modeled baseline energy use per product, *ec*1 = actual energy consumption per product, and n represents the number of products.

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

A balance between the manufacturing sector's expansion and energy efficiency can be achieved through interventions such as cogeneration plants and energy-efficient measures in enterprises, which will give the businesses the chance to become ready for the industrial policy consequences to be incorporated. The predicted energy consumption with carbon emissions quantifies the case company's future energy needs and serves as a benchmark for the implementation of some important energy consumers' small-scale renewable technology needs. When applying Equation (12), the analysis provides a direct energy and carbon relationship, which implies larger tax burdens.

A detailed analysis of the model projects the future climate policy on the vehicle production energy econometric as an industry response to climate policy over a medium to long-term time scale. Figure 2 shows the energy intensity for the vehicle bodies' production model based on the data, will help the industry to become more competitive in lower energy consumption and carbon intensity for production without a greenhouse gas emission constraint.

**Figure 2.** Energy intensity for the vehicle bodies production model based on the collected data.

This paper applied univariate TS analysis as a regression model based on 1086 observations of data with a 385 minimum sample size of vehicle body production energy intensity, as shown in Figure 3. Figure 4 presents a graphic representation of the energy consumption (MWh) and CO2 emissions (Mt), while Figure 5 is the energy (MWh) and CO2 emission (Mt) profiles.

**Figure 3.** Analysis of the energy performance indicator based on energy intensity of the manufactured product.

**Figure 5.** Graphic representation of the energy (MWh) and CO2 emission (Mt) profiles.

Figure 6 shows that there is a significant positive trend that exhibits less than the 10% critical value margin of error, with a 95% confidence and correlation coefficient as the fraction of the total variation in the regression of the results.

**Figure 6.** Significant positive trend that exhibits less than the 10% critical value margin of error with 95% confidence.

Table 1 presents the test statistics and the results for the vehicle body production, the energy intensity with the expected energy consumption, and the savings. Although the variation in the standard deviation is small, the mean is increasing with time, and this is not a stationary series, as presented in Figure 7. It shows that there is seasonality that exhibits less than the 10% critical value margin of error with 95% confidence. in energy (MWh) and CO2 emission (Mt) profiles that exhibits less than the 10% critical value margin of error, with a 95% confidence and correlation coefficient as the fraction of the total variation in regressing the results.

**Table 1.** Test statistics and results for vehicle body production, carbon intensity with expected energy consumption, and energy savings.


**Figure 7.** Seasonality that exhibits less than the 10% critical value margin of error with 95% confidence.

The test statistic is less than the critical values. It is important to note that the signed values should be compared and not the absolute values. TS has even smaller variations in the mean and the standard deviation in magnitude. The test statistic is smaller than the 1% critical value, which is better than in the previous case. In this case, there are no missing values as all the values from the beginning are given as weights. It will not work with the previous values. The mean and the standard deviations have small variations with time. The result is less than the 10% critical value, thus, the TS is stationary with 95% confidence. We can take second or third-order differences, which might obtain better results in certain applications.

Figure 8 is the representation of the simulation of the test and training root mean square error (RMSE) values of five years of energy and CO2 emission prediction using the Dickey–Fuller test statistic to determine heuristically the RMSE as the normalized distance between the vector of the predicted values and the vector of the observed values. The test RMSE was evaluated on unused test data, which use the energy-carbon dioxide causal regression model that was fitted to the training data as a measure of how well fitted the model is, but not simply how well the training RMSE fits the data that were used to train the model. This study proves the expert's prediction that CO2 emissions from transport manufacturing are the largest cause of climate change that is expanding at the quickest rate. At the same time, the expansion of CO2 intensity may result in more greenhouse gas (GHG) pollution from the increased energy use. The test statistic is smaller than the 1% critical value, which is better than in the previous case.

**Figure 8.** The actual trained RMSE and prediction test RMSE.
