Robust cylinder health monitoring for internal combustion engines

Mert Geveci, Purdue University


The goal of the research presented in this dissertation is to introduce a class of robust cylinder health monitoring algorithms for internal combustion engine applications. The proposed cylinder health monitoring approach is based on the crank-angle domain measurements of crankshaft rotational speed on a multi-cylinder internal combustion engine. An analytical study of the crank-angle domain dynamics of a flexible drivetrain model is utilized to show that for a significant class of powertrain configurations, the existence of a non-zero first harmonic component of crankshaft rotational speed is indicative of an irregular torque distribution among the cylinders. The same analytical approach is also utilized to show that the Nth harmonic component of crankshaft rotational speed of an internal combustion engine with N equally spaced cylinders can be used to estimate the total inertia of the drivetrain. This result is used to design a class of cylinder health monitoring algorithms that are capable of detecting and measuring the amount of imbalance present in one of the cylinders of a multi-cylinder engine even when the total inertia of the drivetrain is not known. Experimental results from three distinct powertrain configurations, as well as a detailed numerical study of systems with random parametric variations, are provided to demonstrate the effectiveness and robustness of the method to parametric variations in the governing dynamics. The proposed approach is capable of detecting and measuring imbalance in the presence of significant unmodeled variations of the powertrain dynamics, as well as a vibration absorber failure.




Bernhard, Purdue University.

Subject Area

Mechanical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server