Updating deterioration models of reinforced concrete structures in carbonation environment using in-situ inspection data

Abstract

Physical models for carbonation-induced reinforcement corrosion and concrete cracking have been proposed mainly based on accelerated carbonation tests. However, it has been criticised for not being representable of the natural corrosion process, and may not be applicable to real structures, so it is meaningful to calibrate and modify these models with in-situ inspection data of reinforced concrete (RC) structures in actual carbonation environment. The available in-situ data with the current inspection technologies (non-destructive) include concrete cover thickness, concrete strength, carbonation depth, reinforcement corrosion proportion, cracking proportion, etc. Based on Bayesian updating theorem, the paper proposes an integrated procedure for the updating of carbonation-induced deterioration models with inspection data, and then the durability assessment of existing RC structures in remaining service lives. In the procedure, the carbonation depth model and the probability of reinforcement corrosion are firstly updated with the inspected data, and then the un-carbonated depth model, the cracking probability and the critical corrosion depth model for cracking are updated subsequently. The methodology is illustrated using a case study of a RC industrial building with an age of 33 years. The influences of prior distribution and inspection data amount on the updating effect are discussed. Especially, the uncertainty associated with the number of corroded samples due to the error in corrosion inspection is discussed, and its effect on updated corrosion probability is investigated.

KEYWORDS:

• Bayesian updating
• carbonation environment
• deterioration models
• durability assessment
• inspection data
• reinforced concrete structures

Acknowledgements

The research described in this paper was supported by the China National Nature Science Foundation (grant nos. 51778337 and 51890901). These supports are gratefully acknowledged. However, the views in this paper represent those of the authors, and do not represent the views of the sponsoring organisation.