Date of Award

2020

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Civil and Environmental Engineering

First Advisor

Eric M. Hernandez

Second Advisor

Chris Danforth

Abstract

Fatigue damage is the continuous degradation of a material, primarily due to the formation of microcracks and resulting from the repeated application of stress cycles. Traditionally a fatigue analysis was performed during the structural design stage of a machine or structure; however, more recently there has been increased interest in the monitoring and prognosis of fatigue damage in existing and operating structures. In monitoring, the structure already exists and its mechanical properties can be estimated by processing sensor measurements and non-destructive testing. The traditional approach to fatigue monitoring is to carry out a visual inspection, find macroscopic cracks and then predict their growth. This was often carried out by finding changes in dynamic properties of the system, i.e. changes in modal frequencies, mode shapes, and modal damping. Yet in many cases, by the time the cracks grow to a point where they are detectable, the load bearing capacity of the structure has been greatly reduced. Therefore, a preferable approach is to track fatigue damage on the whole structure prior to the appearance of macroscopic cracks. This would allow for higher levels of reliability, larger lead times and reduced risk. Although no exact figures are available, it is estimated that upwards of 50% of mechanical failures in metallic structures can be attributed to fatigue. Structural health monitoring has been extensively studied for structural systems but hasn't been applied to biomechanical systems where biomechanical failure is consistent with the process of mechanical fatigue.

The objective of this dissertation is to show that state estimation algorithms, i.e. the Kalman filter, can be successfully formulated to estimate fatigue damage in near-real time for structural and biomechanical systems. The Kalman filter combines dynamic response measurements at minimal spatial locations with a structural model to estimate the response of the dynamical system at all model degrees-of-freedom. The estimates of the dynamic response of the instrumented structural systems are subsequently used for fatigue damage diagnosis and prognosis in combination with an empirical S-N curve. By quantifying the uncertainty in both the state estimate and S-N curve, the fatigue damage index becomes bounded based on a user-defined allowable probability of failure.

The main contributions of this dissertation are summarized as follows: i) Development of a fatigue monitoring framework for structural and biomechanical systems; ii) Experimental validation of service life fatigue monitoring in near-real time for statically determinant structures; iii) Uncertainty quantification and propagation of system response and fatigue damage estimates using Kalman filters.

Language

en

Number of Pages

180 p.

Share

COinS