Estudo experimental sobre a qualidade da gasolina comercializada no Estado da Paraíba com o uso da norma ABNT 13992:2008 e do planejamento fatorial.

Abstract

The quality of gasoline in Brazil is defined by a set of physical and chemical characteristics, specified in the standards of the National Agency of Petroleum, Natural Gas and Biofuels (ANP). The aims of this study is to construct an evaluation of the quality of gasoline sold in the state of Paraiba (PB). The universe researched consisted of collecting gasoline stations in marketing fuel in the city of Campina Grande (CGA) and the federal highway BR 230 in PB, between the cities of Cabedelo and Cajazeiras. Forty-eight (48) gasoline stations were researched in CGA and seventy-one (71) on BR 230 PB, in the sample space of three hundred and sixty-nine (369). To collect and analyze the samples we used the factorial experimental planning, with the inclusion of specific control variables. The independent variables were four (04): distance (Cabedelo/CGA and CGA/Cajazeiras), locality (center and periphery) of the gasoline station, type of gasoline sold (unleaded and leaded), and the season in which the supply occurred (summer and winter). To run the sample test, in order to state whether it is in compliance (C) or non-compliance (NC), we used the procedures described by the Brazilian Standard (NBR) of ABNT 13992:2008. The quality of gasoline was investigated for three properties: volumetric percentage of ethanol mixed with gasoline, the values of the density of gasoline, and its visual appearance, color and presence of solids. All tests were performed at the Energy and Thermal Machines Laboratory (ETML), Federal University of Campina Grande (UFCG) between 2010 and 2012. The results show that for CGA, all variables have no significant influence on the quality of gasoline. With regard to the quality of gasoline in highway BR 230 PB, we found noncompliance (NC) on the results of the specific mass (ME). The result shows that the independent variable distance was statistically significant at the level of 95% with regression coefficient (R2) 88%, resulting, in a mathematical model expressed. Thus, it is possible to conclude that the likelihood of non-compliance for specific mass is increases the greater the distance from the position in relation to the zero mark in Cabedelo. The results obtained in this study were compared with the database in Brazil’s and Paraíba’s ANP and proved compatible with it.