Statistical approach in studying the air/fuel ratio variation based on oxygen sensors' signals

Etim U. Ubong, Kettering University
Boyan N. Dimitrov, Kettering University


This study examines the test results from a spark ignition engine exhaust gas oxygen(EGO) sensors′ output signals obtained from a V6,3.8 liter, BUICK engine fitted with individual oxygen sensor per cylinder exhaust runner. The goal is to find which factors and what combination of factors may influence the feeding of an engine with fuel. A LabView data acquisition equipment is used to record the individual cylinder′s exhaust gas oxygen sensor′s output voltage signal and the main flow(stock) exhaust gas oxygen sensor voltage signal. A statistical treatment of the variability of the mean values of the six cylinders′ voltage signal is compared with the stock EGO sensor′s voltage signal. The factors that provide stable systematic error are established. The variations between the mean signal values on the left and right banks of cylinders are compared. Some significant differences between the mean values in air-fuel mixture of the cylinders with this configuration of the exhaust manifold have been found. Tables and graphs illustrate the statistic results. The major conclusions are: the two banks of this V-engine are not equally supplied by air-fuel mixture; there are cases where the right bank has a leaner mixture than the left, namely at moderate speed, high load, and high-octane gasoline, and there are cases when both banks have the same air-fuel mixture. A brief model for an overall estimation of air-fuel mixture to the two banks is proposed based on the statistical estimation used in this work. This approach can also be used to highlight the deficiency in the intake and exhaust manifolds designs.