Date of Award


Document Type

Campus Access Dissertation


Civil and Environmental Engineering


Civil Engineering

First Advisor

Juan M. Caicedo


The state-of-the-art in modal identification for civil structures is limited to the estimation mode shapes with a low spatial resolution. Modal coordinates are identified only at the location of sensors, which are fixed at particular locations on the structure. Increasing the number of sensors in the structure is an alternative to increase the spatial resolution. Unfortunately, this increases the cost of the instrumentation and the data to be transmitted and processed. Another common alternative is to use numerical algorithms to expand the modal coordinates to non-measured degrees of freedom. However, mode shape expansion techniques could introduce errors in the indentified modes. The original contribution of the research presented on this document is the formulation, evaluation and validation of an innovative methodology for Modal Identification using Mobile Sensors (MIMS). This methodology estimates mode shapes with high spatial resolution. The scope of this research is in the formulation of the methodology and the validation in a laboratory environment. The methodology was formulated for sinusoidal, impulse and ambient vibration. Emphasis was given on ambient vibration since this is the preferred approach for testing civil infrastructure. The results of this research conclude that MIMS has the potential to be used on modal identification.

A Fast Mode Identification (FMI) technique was derived from the fundamentals of MIMS. FMI uses stationary sensors and only needs a fraction (13% or less) of the computational power required with current state-of-the-art methods. Results indicate that FMI is more precise than current techniques for the identification of mode shapes.

The robustness of the methodologies to key factors such as noise in the sensors and the accuracy of input parameters (i.e. natural frequencies and damping ratios) is studied. The understanding of the behavior or the methodology to the variation of these factors is an important step for a successful validation and implementation. The results of this study show that the methodologies are robust to noise in the sensors, and errors on input parameters in the case of ambient vibration.

The methodologies are experimentally verified at the University of South Carolina Structures Lab using a uniform simply supported steel beam. Results of the proposed methodologies were compared with those of current accepted techniques. MIMS successfully estimated spatially dense mode shapes using only two sensors. FMI was successful at identifying the experimental mode shapes with limited computational requirements.