Published Online: 8 September 2009
Page Count: 6
Graduate Research Assistant, Virginia Tech, Blacksburg, VA
Weyers, R. E.
Charles E. Via, Jr. Professor, Virginia Tech, Blacksburg, VA
Sprinkel, M. M.
Associate Director of Materials Research, Virginia Transportation Research Council, Charlottesville, VA
(Received 6 November 2007; accepted 7 August 2009)
Deterioration of steel reinforced concrete bridge decks by chloride induced corrosion in North America is of major concern. Chloride from applications of sodium, calcium, and magnesium chloride during winter maintenance operations initiates corrosion, which subsequently spalls the reinforcing steel cover concrete. Initial service life prediction models were diffusion based deterministic/empirical models, which significantly overestimate service lives. Probabilistic solutions provided more realistic estimates. However, chloride corrosion service life models for reinforced concrete elements have not been sufficiently validated against actual field performance. The objective of this study was to validate a probabilistic model using field survey data from ten bridge decks built in Virginia. Field surveys included damage assessments as the sum of spalls, delaminations, and patches. Damage surveys were conducted initially and 3 years later. Chloride content and clear concrete cover depth surveys were conducted to establish cover depth, surface chloride concentration, and chloride diffusion constant global variations. Validation comparisons were performed considering the ten decks as representative of a construction era and as individual decks. Results demonstrated that the probabilistic model accurately predicts global performance but not individual deck performance. Damage predictions of individual decks were greater than and less than the field assessed condition. Differences ranged from an overestimate of 64 years to an underestimate of 11 years. Underestimates averaged 14 years and overestimates averaged 43 years. The large overestimates of 44, 62, and 64 years occurred for estimated service lives of 80–100 years. Insufficient number of model input parameters of surface chloride and particularly the diffusion constant is the plausible reason for the lack of prediction accuracies for the individual decks.
Paper ID: JAI101519