Off-campus South Dakota State University users: To download campus access theses, please use the following link to log into our proxy server with your South Dakota State University ID and password.

Non-South Dakota State University users: Please talk to your librarian about requesting this thesis through interlibrary loan.

Document Type

Thesis - University Access Only

Award Date


Degree Name

Master of Science (MS)

Department / School

Civil and Environmental Engineering

First Advisor

Junwon Seo


This paper presents the performance reliability of reinforced concrete beams strengthened with fiber reinforced polymer (FRP) sheets in the form of structural fragility curves. Emphasis is placed on the development of effective strains that can represent FRPdebonding failure. To investigate the effect of parameters related to FRP-debonding failure on the effective strains in a statistical manner, a second-order polynomial function, which is response surface metamodel (RSM), is created based on a large set of experimental database consisting of 230 FRP-debonding failure test beams collected from past literature. The effect of significant parameters related to FRP-debonding failure, including FRP thickness (tf), steel reinforcement ratio (ρ), concrete strength (f’c), beam height (h), beam width (w), span length (L), and shear span (a), on the variation of effective strains has been examined by the RSM. The probability of effective strains in the beams that exceeds a certain debonding limit state has been calculated using RSM coupled with Monte-Carlo simulation (MCS) with 10,000 trials. The exceeding probabilities of FRP-debonding failure also known as fragility curves for the beams are created for each of the parameters. To evaluate performance of the RSM in generating the FRP-debonding fragility curves, these curves are compared to other sets of fragility curves generated from a conventional standard approach. In addition to the use of RSM, three other metamodels—Multivariate Adaptive Regression Splines (MARS), Kriging (KG), and Artificial Neural Network (ANN)—are developed for this study. To evaluate the metamodels’ efficiency, robustness and accuracy in generating the fragility curves, these curves are then compared to those obtained from the experimental database, standard American Concrete Institute (ACI) debonding formulas, and empirical equation proposed in the American Society of Civil Engineering (ASCE) Journal of Bridge Engineering. The results show that the KG, MARS, and ANN yield more accurate fragility curves than the RSM and ACI and ASCE equation. The capability of metamodels in producing consistent outcomes by multiple processes that is known as robustness is measured. It is found that the ASCE equation is selected as the most robust model among the used models and KG, MARS, ACI, and ANN have been ranked in the second, third, forth, and fifth place in developing robust models, respectively. It should be noted that RSM producing the smallest robust result is ranked in the last place. As for the efficiency evaluation, the MARS acts as the most efficient model among all the metamodels. Hence, the MARS yields relatively the most accurate, robust, and efficient FRP-debonding fragility curves compared to other models. It is anticipated that the MARS can be used as the most reliable metamodel to create the FRP debonding-fragility curves, accounting for uncertainty in values of significant parameters in a efficient manner.

Library of Congress Subject Headings

Concrete beams -- Testing Reinforced concrete construction -- Testing


Includes bibliographical references (pages 52-55)



Number of Pages



South Dakota State University



Rights Statement

In Copyright