On-board battery condition diagnostics based on mathematical modeling of an engine starting system

On-line estimation of the battery potential and ability of successfully starting a given engine under specific environmental conditions is a major problem in predicting maintenance strategy of any cars fleet. Battery aging and low operation temperatures are important parameters that may lead to starting failure when this is needed. Several indirect methods have been proposed to evaluate the battery condition. State-of-charge (SOC), state-of-health (SOH), cold cranking amperage (CCA), and internal resistance, are commonly used worldwide to characterize a battery condition. In many cases these indirect methods don't provide reasonable conclusions and practically poor correlation has been found between the predictions and the real battery performance. In the present paper, we propose a new quasi direct approach that is based on mathematical modeling of dynamic behavior of a starting system. This approach is based on updated models of the starting system elements, while special attention was given to the battery dynamic behavior, friction forces between piston and cylinder, and the overrun clutch. The main criterion for prediction of successful engine starting is a steady state average crankshaft angular velocity, obtained from the model. The battery dynamic parameters are obtained from deep battery loading with a special device. The device loads the battery with a high current for a very short period of time, not causing any damage to it. Tests based on our concept, were compared with experimental results performed with Ford Transit Diesel FT-4 engine. Predictions were found to be above 95% of confidence in a wide range of experimental conditions.