**3. Methodology**

In this section, the concepts and specifications of the mathematical model development and the measurement and verification integration model for energy efficiency and the carbon emission signatures approach proposed in this work are presented. The integration of the high volume of data used was in real-time in the determination of energy intensity on the realization of the low-carbon product. The proposed architecture of the framework was based on the Leontief Input–Output mathematical model approach for energy savings. The entire architecture's key components are presented here to understand how it works. The components that constitute the architecture are the physical layer, communication layer, middleware layer, database layer, and the application and managemen<sup>t</sup> layer. The moving average over the past year, i.e., the last 12 values, was determined with specific functions for rolling statistics in Pandas in the determination of the time series techniques for the actual data-processing method. We collected Tier 2 automotive company data for 2015 to 2018 and examined the results of IoT-based energy monitoring devices using an asymmetric energy causal mathematical model for the analysis through decomposing into the following components: prediction, testing, and forecasting, for decision making and policy formulation. We used the vehicle body productions' energy intensity with the expected energy consumption and savings, with the expected energy consumption adjustment from the 2015 to 2016 energy performance data, to compare the results.

#### *3.1. Asymmetric Energy–Carbon Emission Causal Mathematical Model*

The asymmetry energy causal mathematical model is an order of the dependency variables characteristics of the energy variables and vehicle production parts from a heuristic assumption. The definition represents inferences from the statistical data of energy consumed in producing one unit (inputs) as an exogenous variable, which is determined by embodied energy intensity per unit as an output variable.

#### Energy Analysis—Input and Output Theory

The following equilibrium structure predicts the future transport manufacturing energy consumption over carbon and energy intensity, thereby entailing probabilistic independence or dependence of the process variables. The adoption of the asymmetric energy–carbon emission causal mathematical models, which are presented in Figure 1, is the heuristic dependency determines the observed probabilistic correlations among variables, or the outcomes of the lowest influence variables. We then manipulate the variables within the equations to produce asymmetric causal equations as follows:

**Figure 1.** Causal graph embodied energy intensity in an input and output equilibrium structure.

*Xij* = transaction from sector i to sector *j*, *Xj* = total output of sector *j*, *£j* = embodied energy intensity per unit of *Xj* (amount of energy consumed in producing one unit), and *Ej* = energy consumed to restate the demand of the reporting periods under a common set of conditions. *Es* = energy savings in Equation (1), *βpeu* = baseline energy usage for a period of use, *ρpeu* = reported energy usage for a period of use, and *A* = sum of the adjustment as Equation (2)

$$E\_S = \beta\_{peu} - \rho\_{peu} \pm A \tag{1}$$

$$A = 1 - \left(1 - R^2\right)^{n-1} / n - p\_p^{-1} \tag{2}$$

Carbon Emissions converted from Energy

Evaluating the carbon footprint relative to the energy consumption during the production of vehicle bodies using applicable emission factors (7.0555 ∗ <sup>10</sup>−<sup>4</sup>) to estimate the carbon emissions signature (CESTM) for the energy mix for a particular year.

> Carbon emission [kgtCO2)]= *£j* ∗ CESTM [kg tCO2/GJ] (3)

The annualized non-baseload CO2 output emission rate is used to convert reductions in hours into avoided units of carbon dioxide emissions. The average carbon intensity factor of 0.4 kgCO2/GJ is adopted for the equivalent of emission reductions from energy efficiency programs that are assumed to affect non-baseload generation (power plants that are brought online, as necessary to meet demand).

$$\text{Energy}\_{\text{consumption}} = f\left(E\_{\text{c}}, \text{CO}\_{2}, \, V\_{op}, \, E\_{i}\right) \text{CO}\_{2} \tag{4}$$

$$\text{Energy}\_{\text{consumption}} = f(E\_{\text{c}}, V\_{\text{b}}, E\_{\text{i}}, N\_{\text{c}}) \tag{5}$$

*Ec* = electricity consumed, *Vb* = vehicle bodies, *Ei* = energy intensity, *Nc* = normalization coefficient in CUSUM savings (895.56) to produce parts body.

Carbon Emission Savings

$$\text{CO}\_{2\text{ savings}} = f\left(E\_{\text{s}} \mid V\_{op\text{\textdegree}} \mid E\_{\text{i}\text{\textdegree}}\right) \tag{6}$$

Considering the variables required in total internal requirements within the model, the equation was X = AX, that is, the total input is equal to the total output. X = AX + D (I − A) X = D, where I is a 3 by 3 identity matrix X = (I−A) − 1D, all consumption is within the industries. There is no external demand.

Producing *P*1 units of *Es* required *Esp*1 units of CO2 savings; producing *P*2 units of *Vop* required *Vopp*1 units of CO2 savings; producing *P*3 units of *Ei* required *Eip*1 units of CO2 savings.
