Prediction of crack widths in NZS 3101 and its significance in the seismic design of connections

Abstract

Cracking of reinforced concrete structures due to bending or tension has usually great significance on structural behaviour. Design codes limit crack widths for durability and aesthetic purposes. The research of cracking in reinforced concrete with the aim of finding appropriate models for Serviceability Limit State (SLS) design of crack widths is hot topic for more than 100 years. Numerous theoretical and empirical models have been proposed, however, both approaches have strong limitations. Most of the theoretical approaches fail to address the complexity of bond between reinforcing bars and concrete and its importance on structural cracking, while the accuracy of empirical models always depend on and are limited to the database that was used for the calibration. The empirical crack width model in NZS 3101 is largely based on Frosch (1999).

Present paper provides details of a comparative statistical analysis of the NZS 3101 crack width model and other code models, using an international crack width database. It is demonstrated that none of the studied models is robust. All models overestimate the small actual crack widths, and underestimate the large actual crack widths. This observation has great significance in seismic design, when the expected crack widths are large and can be well beyond the limits set in SLS design. One special application is the seismic design of post-installed anchors, which does not directly address the actual crack widths despite crack width being a driving parameter for post-installed anchor seismic qualification. Present paper provides literature review and discussion of these topics.