Testing and Analysis of Temperature Field in a Three-Cell Concrete Box Girder Model

To study the temperature field in a three-cell concrete box girder, an on-site testing model was constructed. One hundred and fifty-one temperature sensors were embedded in the model, and temperature monitoring was carried out for a duration of 597 days. The experimental results show that the temperature field distribution of the three-cell concrete box girder is uneven. The side web show periodic variations in temperature differences along the thickness direction. The middle web is mainly influenced by the internal air temperature of the box cells, showing small temperature differences in the thickness direction. The horizontal temperature distribution of the top plate exhibits a wavy shape. Furthermore, this study proposes a new numerical simulation method for the temperature field of bridges. Traditional finite-element simulation methods typically use solar radiation heat exchange and atmospheric convection heat exchange to simulate thermal boundary conditions. However, due to the uncertainty of cloud cover and wind speed, these can lead to large errors in numerical simulation results. This study proposes a refined numerical thermal simulation method using measured temperatures as boundary conditions and compares it with traditional methods. The comparative analysis results of different simulation methods on the spatial aspect are described as follows: when simulating the temperature field of the box girder in August and January, the use of a numerical model with measured temperature boundaries significantly reduces the testing errors at various measurement points inside the box girder compared to using traditional thermal boundary numerical models. The comparative analysis results of two methods on the time aspect are described as follows: selecting typical measurement points and verifying the temperature time history for these points on both August 23, 2022 and January 30, 2022, the model with measured temperature boundaries can reduce the error statistics for the selected measurement points.