A methodology for mechanical diagnostics and prognostics to assess durability of ground vehicle suspension systems

Muhammad Haroon, Purdue University

Abstract

Mechanical vehicle systems operate in highly variable loading environments. Anticipated and unanticipated degradations occur in components and subsystems due to the variability in static and dynamic loading. Degradation can lead to excessive warranty costs and overly aggressive maintenance schedules. Data interrogation methodologies are needed for identifying loads and faults in suspensions, tires, and other vehicle components and predicting the rate at which faults grow to help design more durable systems. The goal of the research presented in this dissertation is to develop a data interrogation methodology for analyzing the dynamic response of vehicle suspension systems in order to detect, locate, and quantify damage and then predict the growth of damage in component(s) and/or system(s) using experimental and operating data. It is shown that empirical relations exist to predict the growth of damage based on the current status of the damage and internal loads, which can be used to develop damage growth models for suspension system diagnostics and prognostics. It has been recognized in this research that damage causes changes in the internal loading of the components of mechanical systems, and it is important to track such changes for objective diagnosis and prognosis of these systems. Consequently, restoring forces are used to characterize the internal forces in system components, in terms of frequency and amplitude, and to track the changes in internal forces for diagnosis and prognosis. A major constraint in mechanical vehicle system data interrogation is that the input to the tire-patch during operating tests is difficult to measure or estimate. In addition, vehicle systems are highly nonlinear at the component and sub-system level and mechanical damage frequently introduces further nonlinearity into mechanical systems. Consequently, nonlinear methods that utilize passive operating response data in an event-driven manner, based on the frequency bandwidth of the response, and exploit the changes in nonlinear behavior to identify damage are employed. Experimental data from suspension sub-system and full-vehicle tests along with simulation data from a full-vehicle MSC.ADAMS model have been used to show that damage causes a redistribution of internal component loads, which affects how damage grows over time. Restoring forces have been used to quantify the changes in the internal loads with damage, and this information has been combined with damage information to generate empirical damage growth prediction models.

Degree

Ph.D.

Advisors

Adams, Purdue University.

Subject Area

Mechanical engineering

COinS