ECONOMIC, ENVIRONMENTAL AND PRODUCTIVE PRACTICES INFLUENCE ON THE INDUSTRIAL ENERGETIC EFFICIENCY

The industrial energetic efficiency (EE) is recognized as one of the main factors for the reduction of gas emissions that cause the greenhouse effect and for the improvement of the industrial competiveness. Within this context, many papers of the international literature have proposed different indicators of industrial, economic and environmental behavior, so as to promote the EE inside the industries. However, such proposals do not generally check the result of the joint work for more than one indicator in the organizations, making more global analysis more difficult related to EE. This paper aims to check which environmental, economic and industrial practices indicators influence the EE of the industries. The data have been collected from the framework developed by Trianni et al. (2014), that analysed the main energetic efficiency measures for the technologies: motors, lighting, compressed air and HVAC systems (heating, ventilation and air conditioning). A logistic regression model has been adjusted for understand the relationship the economic, environmental and productive practices behavior on the 1 PhD student in Environmental Sciences. São Paulo State University (Unesp). fabio.neves@unesp.br 2 PhD in Management (UFRJ). Sorocaba Engineering School. henrique.ewbank@gmail.com 3 PhD in Mathematical (UNB). São Paulo State University (Unesp). jose.roveda@unesp.br 4 PhD in Industrial Engineering (Politecnico di Milano/Italy) . University of Technology Sydney . andrea.trianni@uts.edu.au 5 PhD in Eletrical Engineering. São Paulo State University (Unesp). fernando.marafao@unesp.br 6 PhD in Mathematical (UNB). São Paulo State University (Unesp). sandra.regina@unesp.br 514 R. gest. sust. ambient., Florianópolis, v. 9, n. esp , p. 513-531, fev. 2020 energetic efficiency. Results suggest that a healthy workplace enables investments in equipment and machinery, allowing the EE inside the industries.

Still, technical actions can be provided in interventions in cross-cutting technologies 7 , because these technologies comprise most of the industrial energy consumption. Among the examples of cross-cutting technologies include electric motor driven systems, which account for about 70% of the worldwide electricity consumption in industrial industries (IEA, 2011;TRIANNI et al., 2014).The industrial lighting is the most disseminated cross-cutting technology and corresponds to about 5% of the worldwide electricity consumption (IEA, 2006).
Compressed air may reach about 10% of the industrial electricity consumption, while the systems HVAC (heating, ventilation, and air conditioning) have an interval of 10 to 20% of the final energy consumption in some industrial contexts TRIANNI et al., 2014).
Energetic efficiency policies are very relevant in saving energy consumption.
Although, the rate of implementation of those policies are very low, not exceeding 50% of the recommended actions (ANDERSON & NEWELL, 2004, BUNSE et al., 2011CAGNO & TRIANNI, 2012). This lack of implementation is due to many barriers, such as those related to economic and information aspects (SORREl et al., 2004;CAGNO et al., 2010;TRIANNI et al., 2013a;TRIANNI et al., 2013b;. These critical factors show that such barriers are not sufficiently transmitted to the industrial decision makers through a more in-depth view of the usefulness of the implementation of measures of energetic efficiency. This shows the impact on the production system, its problems related to the effective implementation, as well as the interactions with other parts where the industrial productivity may present to the decision makers when the main perspectives and characterization are not evident. . Therefore, those crytical factors develop in two main approaches. The first is related to general energy policies, especially regarding costs, and do not explain the real specificity of energetic efficiency measures to be promoted, such as investments affecting energetic and operational issues, consequently impacting the production performance. The second, the energy policy makers in the industries are estimulated to leverage an enhanced understanding to support the decision makers in a clear and efficient way. (SHIPLEY & ELLIOT, 2006;. 7 Technologies related to own manufacture or various industrial production systems Pye & Mackane (2000) recognize that when quantifying the benefits and even the barriers for the implementation of the energy management improvement in the industries help understand the financial opportunities of the investments on energetic efficiency measures. Taking into consideration that the energy economy is a prime factor for the industrial decision making, and therefore it can be seen as productivity increase, environmental conformity reduced costs, production reduced costs, reduction of scrap costs, product quality improvement, better capacity use, higher reliability and higher safety for the employers. These factors are part of the total benefits of a project of energetic efficiency (PYE & MACKANE, 2000). To make the energetic efficiency more convincing beyond the pollution prevention, it is also necessary to understand the inter-relations of the measurement of costs and benefits so that the financial ramifications are understood and can be communicated to the employees of different hierarchy levels, as the better the energy management is the

MATERIAL AND METHODS
The research method used in this paper is the statistic modeling based on a logistic regression model. This approach was chosen in order to explain the amount of saved energy in the industries, according to economic, environmental and industrial production phenomena on the quantity of energy consumed in an industry.

Research framework
The implementation of the research occurred in the following steps: the collection was carried out through the article of Trianni et al. (2014) taking into consideration the economic, environmental and industrial production indicators as independent variables and the amount of saved energy as a dependent variable.
These indicators can be found on Table 1. The amount of saved energy is composed by the proposals of energy consumption improvement proposals of the technologies proposed by the center of industrial evaluation (motors, lighting systems, HVAC and compressed air). It is noteworthy that Trianni et al. (2014) made the survey of these indicators through a broad and systematic literature review.
In the other, with the surveyed indicators and it was made the adjustment of the logistic regression model relating the amount of saved energy to the economic, environmental and industrial production indicators. After getting the results the final analysis has been carried out and the final version of the paper was developed.

Modeling
The influence of the economic, environmental and industrial production factors has been analyzed on the amount of industrial saved energy. For this purpose, a logistic regression model has been adjusted.
According to Draper & Smith (1998), a logistic regression model is a statistic technique in which the probability of a dichotomy result (as adoption and nonadoption) is related to a set of explanatory variables that are hypothesized to influence the result, as represented by the equation 1: where the subscript i denotes the ith observation in the sample, P the probability of the result, β0 is the intersection term, β1, β2,….βk are the coefficients associated to each explanatory variable X1, X2, …., Xk.  To evaluate the significance of the logistic regression, model the F test has been adjusted, and the statistic of the estimated coefficients by the standard error will be equal or different from zero.
In the construction of the logistic regression models it is necessary to select the independent variables that will be part of the model. In general, the problem is to select correctly a set of independent variables that include the variables considered important by the researcher (Mann, 2006, Hair-Junior et al., 2010. In addition, the indicators that had a significant effect on the level of significance of 5% (Draper & Smith, 1998) have been selected to compose the final model. After getting the model that best adjusts to the data, it is necessary to fulfill the premises associated to a linear regression model, so as to consider the developed model valid (DRAPER & SMITH, 1998).

Analysis of the logistic regression model adjusted
At first three criteria have been used to validate the models:

1.
The Anova test accepting significant models with a p<0.05

2.
The lowest value of Akaike information criteria (AIC) found for the different studied models

3.
The largest coefficient of estimation of Pseudo-R 2 of Macfadden According to Macfadden (1977) the ρ 2 (Pseudo-R 2 ) tends to have low values, with the inter value of 0.2 to 0.4 considered excellent, so the ρ 2 can be interpreted as R 2 , but not indicating large values.
All the statistical analysis have been carried out using the computational environment R, version 3.5.3 (R Code Team, 2019).

RESULTS AND DISCUSSION
A descriptive analysis of the indicators studied in the Table 2 shows how their behaviors are in energy management in industries. Costs and waste emissions that differ most from the average for a binary data set (0.5). Thus, it is found that costs are as important as approaching 1 (98.8%), as waste emissions are less important than approaching zero (5.7%).     (1997) stated that the first financial expenses in the implementation of improvement measures of the energetic efficiency can be even more important than the return rates of the amount of saved energy. Therefore, although there are significant costs in the implementation, its financial return rate becomes substantial to reach the amount of saved energy (Table 2).
By analyzing the industrial production indicators, we can see the importance of a workplace with impact on the satisfaction of the employees.
Adequate working conditions and clarifications of the functionality of the implementation of an energy management program of the cross-cutting technologies increase the potential of the energetic efficiency (Raziq & Maulabakhsgh, 2015). The framework demonstrates that the motor and lighting systems are the main technologies that have support to develop an adequate workplace. The large quantity of these technologies in manufacturing industries and in different functional units compel the employees to deal with some environments in which the quality become primordial so that the productive routine and the energy management are not stressing and spoil all the productive system. (Lu, 2016;Schulze et al, 2016;Boyd, 2017).
In relation to the operation and maintenance of the motors, HVAC, compressed air and lighting systems leads to lower expenses and can reach significant ASE. When compared to the costs of integral change of the crosscutting technologies to OM there is a larger acceptance by the managers. In addition, the employees may accept the operation and maintenance more easily to get to the energetic efficiency of some technologies as the HVAC and lighting systems, once they are related to the comfort of the workplace Cosgrove et al., 2017).
These results about the influence of the production on ASE differ from the pointed ones by Alhourani & Saxena (2009). This can be explained by the factor that the authors analyzed primary data, containing qualitative and quantitative indicators from USA industries. These authors stated that the return period, working hours and kinds of recommendation are the more influential indicators in the decrease of the quantity of energy consumed, indicating that the companies invest in recommendations that have a lower return time.
The achieved results from Table 3 indicate that, in order to occur the decrease of the amount of saved energy, first it must occur a structural change of the functionality of the industrial system. This change can be achieved by implementing management systems that achieve energy efficiency in industry, mainly in energy-intensive manufacturing industries. Among the most widespread energy efficiency improvement management systems is ISO 50001, because they follow the same implementation system of ISO 9001: the PDCA (Plan, Do, Check and Action).  Table 4 presents the main industrial sectors that issued the ISO 50001 standard. It is noteworthy that energy-intensive manufacturing was the main sector to issue ISO 50001 in 2017, which corroborates with the findings of a necessary implementation of management systems to achieve energy efficiency (ISO, 2019).

Conclusions
The results showed the characteristics of the amount of saved energy in industries, through a logistic regression model that checks the impact of economic, environmental, and industrial production indicators.
During the adjustment phase of the indicators in the logistic regression model some indicators have been excluded by the test p-value, such as implementation costs, emission reduction, corporate involvement, productivity, kinds of recommendation technologies and check-up frequency.
For the adjusted logistic regression model the indicators working environment and operation and maintenance influenced the amount of saved energy. The adjustment of these two indicators showed that the structural change of the functionality of the industry plays a ruling role in the decrease of the energetic consumption, guided by the managers and employees of the companies.
The practical implications of this paper are to cooperate with the academy through new concepts and guidelines about the amount of saved energy in a general aspect. For the industrial sector the indicators can be guidelines for the efficiency of the adoption of measures for the reduction of the amount of saved energy. For the technical committees this paper becomes relevant to supply information that allows them to improve the strategies of market as well as contribute to the orientation of the consultants of the corporations. Finally, for the governmental sector it can be as a parameter to analyze the indicators that did not influence the amount of saved energy.
Future works may check the influence of environmental, economic and industrial production indicators for each cross-cutting technology separately.
Also, future research may use mixed multiple logistic regression models providing qualitative evaluations of the indicators associated to the influence on the quantity of saved energy.