Graduation Year


Document Type




Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Civil and Environmental Engineering

Major Professor

Rajan Sen, Ph.D.

Co-Major Professor

Gray Mullins, Ph.D.

Committee Member

Autar Kaw, Ph.D.

Committee Member

Kandethody Ramachandran, Ph.D.

Committee Member

James Giancaspro, Ph.D.


Accelerated Aging, Aging Behavior, Bond Strength, Mixed-Mode, Pull-Off Test


A laboratory study was conducted to investigate possible quantitative and qualitative methods for determining the bond life of structural epoxies externally bonded to concrete substrates. Three concrete mixtures, two surface preparation methods, and three structural epoxies were used creating eighteen combinations of bonded systems. Direct tension (pull-off) specimens were prepared from which a subset was submerged in heated water per a prescribed accelerated aging protocol for up to 200 days. A novel test fixture was designed to facilitate controlled-rate testing of the pull-off specimens. A total of 324 specimens were intermittently removed from submersion exposure and tested without replacement. An additional 12 concrete prisms were used to determine the direct tensile strength of the various substrates. Additional characterization testing was conducted for use in determining the significant factors contributing to changes in bond strength. These tests included concrete compressive strength and concrete surface profilometry. Bulk epoxy and concrete moisture uptake data was also generated for future use. Plots of the raw data imply bond strength decreases at a relatively rapid rate during early aging then slows with longer exposure periods. Analysis of variance testing was conducted as part of a multiple regression analysis which found that square root of concrete strength, square root of aging time, mean peak height, and matrix tensile strength were significant in explaining variations in bond strength. A numerical model was produced based on raw data from the significant factors. The raw data did exhibit instances of high variability (>10%). The fitted model was shown with a coefficient of determination of 63% but implied rapid initial decreasing strength which agreed with apparent trends in the plotted raw data. Bond life was calculated for each study combination as the point in aging time where bond strength decreases beneath the substrate tensile strength. Methods for transforming data points in order to lower variability were investigated but were shown to be inaccurate and impractical. To provide a direct comparison of bond strength values, isolation of bond strength in mixed-mode specimens was attempted but was determined to not be possible without impractical, complex methods. This conclusion implies that chemical bond and mechanical interlock may not readily be decoupled without similar intensive methods. A secondary objective was to determine bond life through a failure mode analysis approach. An alternative set of pull-off failure modes was first proposed to facilitate further discussion. Inconsistencies in mixed-mode specimen surfaces were noted by the author in previous studies that warranted the secondary investigation. New bi-directional fracture path behavior is conceptually posed which is preliminarily confirmed. This finding implies that the A6 failure mode may be used as an early indicator of impending termination of bond life. Collectively, conclusions from this study indicate key factors that affect bond strength, provide insight into bond degradation behavior under accelerated hygrothermal conditions, present a means for conducting controlled rate bond testing, and offer a quick and robust method for determining the onset of bond life in epoxy-concrete systems without the need for complex analyses or additional characterization testing.