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

Venkat Bhethanabotla, Ph.D.

Committee Member

Gray Mullins, Ph.D.

Committee Member

Kandethody Ramachandran, Ph.D.

Committee Member

Alberto A. Sagüés, Ph.D.


ANOVA, Composite, Model, Oxygen, Permeation


Research has proven that wrapping structural elements with fiber-reinforced polymers (FRPs) slows down chloride-induced corrosion rate but does not stop it. Although FRP has been used for corrosion repair for over 40 years, predictive corrosion rate models for FRP-concrete systems are limited in literature. This dissertation presents an oxygen diffusion-based predictive framework to estimate corrosion rate in FRP-concrete systems.

The dissertation has three main components. First, a core framework that is referred to as Khawaja-Sen-Bhethanabotla (KSB) model. In the KSB model, statistical methods were used to extend experimental data on oxygen permeation coefficients. These data were for one and two layer configuration in a controlled environment. The development initially investigated the significance of potential factors on oxygen permeation coefficients. The core framework ultimately predicted the oxygen permeation coefficient for FRP-concrete system. The prediction was as a function of the identified significant factors: FRP type (Carbon or Glass); number of layers; and water to cement (w/c) ratio. The effects of marine environment and workmanship were incorporated by calibrating the initial predictions against measured metal loss data. These data stemmed from specimens repaired using one to four FRP layers. The specimens were 1/3 pile models that experienced marine environment conditions for over three years.

The KSB model may be used to predict uniform corrosion rate in FRP-concrete. This prediction uses the concept of equivalent oxygen permeation coefficient and equivalent thickness for FRP-concrete. These equivalent parameters are related to the permeation coefficients and thicknesses for the FRP and concrete individually. The KSB model development used data from uncracked concrete and FRP-uncracked concrete specimens. Therefore, the KSB model may be used directly to predict corrosion rate for FRP-uncracked concrete. Although the specimens were uncracked, the KSB model may still be applied with FRP-cracked concrete. In this cracked case, however, a simplified conservative assumption has to be implemented. Namely, cracked concrete provides no resistance to oxygen permeation.

Given the KSB model uses the equivalent FRP-concrete parameters, two investigations are needed to mitigate the said simplified cracks assumption: (1) investigation for cracks effect on steel corrosion rate in cracked concrete elements only (without bonded FRP) and, (2) investigation for cracks effect on steel corrosion rate in FRP-cracked concrete members. The second component of this dissertation aimed to provide the first required investigation. This second component quantified cracks impact on corrosion propagation by comparing corrosion damage in cracked concrete vs. uncracked concrete. The analysis used 1/3 pile models metal loss data and an empirical model to quantify cracks influence. The evaluation was followed by back-calculating the oxygen permeation coefficients of cracked concrete. These were then compared with related data in literature.

The second component results showed cracks caused significant damage in 0.33 w/c specimens. They also indicated that smaller cracks area can still be a sign and/or trigger significant corrosion. The back-calculated cracked concrete permeation coefficient was one order of magnitude higher than the uncracked counterpart (0.33 w/c). Uncracked and cracked concrete permeation coefficients are comparable in 0.5 w/c. Note that these coefficients are part of the data required towards refining the simplified cracks assumption in the KSB model. However, they may currently be used to predict corrosion rate in cracked concrete (without bonded FRP). This prediction is limited by assuming uniform corrosion similar to the KSB model.

Ultimately, the third component of this dissertation aimed to address the uniform corrosion assumption in the KSB model and the second component. This refinement was performed by introducing a capacity reduction factor. It reflected local corrosion and stress concentration effects. The development relied on destructive tension tests data. These data were for the corroded reinforcement retrieved from the 1/3 pile models. Capacity reduction factors were determined for both FRP wrapped and unwrapped concrete specimens. The local corrosion and stress concentration effects were more pronounced in the unwrapped elements. Finally, the application of this dissertation and the links between its components are illustrated by numerical examples.