Abstract:
The temperature response of concrete box girder bridges to cyclic solar heating has been a subject of growing concern during recent years. Because of this, research in the field has been performed both theoretically, in the form of methods for the prediction of transient response and experimentally in the form of model studies. This has led to subsequent modifications in the nature of the temperature distribution assumed for the design of concrete box girder bridges. In this thesis the development of ideas initiated from the above mentioned studies is examined. Initially the implementation of quadrilateral finite elements into a two-dimensional heat flow program has been effected and the accuracy of the results so obtained, compared with those for the constant flow triangular element. The transverse thermal stresses induced in a box girder structure as a result of cyclic temperature variations are also examined. This has resulted in the development of a computer program for the computation of two-dimensional thermal stresses. The two-dimensional heat flow theory is extended to the third dimension by the use of a tetrahedron heat flow finite element. A saving in computation time is also achieved by the linearisation of the convective and radiative heat loss laws. The computer program thus developed has then been used to predict the thermal response of both a model diaphragm and a full scale bridge structure. Experimental work in the form of a quarter scale model study of a typical diaphragm segment is presented. The construction, instrumentation and thermal testing of the model segment is also discussed. A comparison of results so obtained with those predicted theoretically has shown good agreement. The instrumentation of a full scale bridge structure for the measurement of temperature is also performed. In order to validate both the two-dimensional and three-dimensional heat flow theories, two regions, a midspan region and a diaphragm region were instrumented. Once again good agreement between theoretical and experimental responses was noted although it was found that for windy days the boundary heat losses were not adequately represented. It is concluded that the vertical temperature gradients at the diaphragm can be more severe than those experienced at the midspan section and can induce reasonably large transverse stresses in the diaphragm. Of the three weather condition parameters recorded at the bridge site viz. solar radiation intensity, shade air temperature and wind velocity, it appears that the wind speed is significant in controlling temperature response. Days with similar solar radiation can result in remarkably different temperature distributions depending on the wind speed at the site.