This thesis addresses the topic of ongoing (excessive) deformations observed in balanced cantilever prestressed concrete bridges all over the world. Many authors attribute this behavior to the time-dependent phenomena of creep and shrinkage. Balanced cantilever bridges are classified as creep-sensitive structures, and for that reason, a detailed analysis of the long-term structural behavior, such as deformations and prestress losses, is recommended. However, during the design of these bridges, commonly used code-based models generally tend to underestimate the long-term creep and shrinkage effects. Additionally, various (simplifying) assumptions are made when modeling these bridges, making their actual creep and shrinkage behavior unclear.
This work aims to investigate whether the long-term effects of creep and shrinkage are indeed a plausible explanation for the excessive and ongoing deflections detected in a specific balanced cantilever bridge in the Netherlands: the Rooyensteinse Brug. This bridge, inaugurated in 1977 in Zoelen, currently exhibits a deflection at midspan of 0.43 m, more than two times what was anticipated in bridge design. To investigate this behavior, a detailed two-and-a-half-dimensional finite element model was developed, incorporating a time-dependent phased analysis accounting for the construction phases. Creep and shrinkage effects were incorporated into the concrete material model through creep compliance and shrinkage strain curves.
A sensitivity study is conducted to analyze the impact of: (i) different code-based models for creep and shrinkage, (ii) accounting for the large creep and shrinkage model uncertainties, (iii) different maturities on the creep compliance curve, and (iv) considering cross-sectional variability in the drying characteristics. The main findings showed that commonly used code-based models, including the current standard Eurocode 2, significantly underestimate the long-term (multi-decade) deflections observed in the Rooyensteinse Brug by 30%. Notably, only the RILEM B4 model (B4), considered the most theoretically grounded creep and shrinkage model, was able to capture the long-term deflection trend reasonably well. Further, acknowledging the inherent uncertainties in B4 significantly widened the range of potential deflections and prestress losses. Using the bounds of the 90% credible interval, the midspan deflections after 60 years range between 0.514 m and 2.91 m, compared to a mean of 0.895 m. The prestress losses range between 18% and 30% in the shear-critical zone, against a mean of 23%. Additionally, accounting for the cross-sectional variability in drying characteristics led to an improvement in both the short-term (first 10 years of service life) and long-term (multi-decade) deflection prediction when comparing the results to in-situ measurements, with differences of 14% and 6.1%, respectively. Based on these predictions, it is expected that the current deflection trend of the Rooyensteinse Brug will continue to decrease linearly (log-scale) for the next two decades.
This study demonstrates that a detailed finite element model incorporating time-dependent phased analysis, in combination with the RILEM B4 model and accounting for cross-sectional variability, can explain the observed behavior in the Rooyensteinse Brug. The results support the hypothesis that the long-term effects of creep and shrinkage are the cause of the ongoing trend of excessive deformations in this balanced cantilever prestressed concrete bridge.