| With the rapid growth of air traffic and the development of large aircrafts, the higher requirements have been raised on the performance and safety of airport pavements. For the current airport pavements at home and abroad, some of them within their design service life have had various deformations and cracks. Also, these cases have relatively great differences from their appearances and causes. Therefore, researchers are paying more and more attention to the failure mechanism and failure modes of airport pavements. At present, the USA national laboratory is most famous in the world for its full scale airport pavement tests. The FAA obtained precious testing data and records of flexible and rigid airport pavements, which provides the important and basic references for the analysis of pavement failure mechanism and whole service process. However, the common finite element analysis can not fully reflect experimental phonemona of airport pavements. Besides, the current FEA analysis of airport pavement is not only complex but also relatively ideal, leading to a considerable difference between the testing and calculating results. For these two issues, this thesis conducts the following studies by applying neural networks and cellular automata and by starting from the testing data:Radial basis function ANN model is built to simulate the dynamic deformances of invisible pavement layers of the flexible pavement under repeatedly loading, based on the corresponding testing data from FAA. The ANN model establishes the correspondence between the deformations of the asphalt concrete runway surface and the underground layers. Thus, the dynamic deformence process of the flexible pavement can be given out using the ANN model. In the choice of the ANN models, several schemes have been tried according to the physical jugement on practical response of the flexible pavement. Finally, the so-called point-by-point method is used as the input of the ANN model for the pavement surface, in other words, three neiboring points in the pavement surface are taken as the ANN input and a point in the invisible pavement layer is taken as the ANN output. Hence, the ANN model can calculate out the dynamic deformation curves of the invisible layers under repeatedly loading. The judgement of the ANN model in accuracy also utilizes the testing data. It is shown that the ANN model could reflect the dynamic deformation process of invisible pavement layers and the local bulge phenomenon which could be not simulated by the present FEA methods. In order to verify the validity of the ANN model in simulating the deformations of the invisible layers, an experimental mode of the multi-layer system is designed and tested in the lab. The experiment obtains the deformation curves of the individual layers in the cross section of the experimental model, which verifies the rationality and availability of the ANN model in comparson of the experimental and simulating results.According to the dynamic deformations of the pavement obtained from the ANN model, several parameters are defined to reflect the characteristics of the deformations of the surface layer and underground layers in the flexible pavement. In order to quantitatively reflect the contribution of the layer deformations in the total deformation of the flexible pavement, a parameter of deformation integration is introduced. Using the proposed parameters, it studies the relationship between the real-time average thickness and the initial thickness of each layer, the relationship between the real-time deformation amplitude and the average thickness of each layer, and the relationship between the real-time maximum deformation amplitude and the average thickness of each layer.Applying the finite element program ANSYS, it makes a comparison of the layer deformation calculated by the FEA program with that simulated by the ANN model. According to the structural characteristics of the FAA airport asphalt pavement model, a calculation method, so-called as the initial residual deformation accumulation method, is tried to conduct the numerical simulation of the airport asphalt pavement and the parameter analysis. The results show that the method is feasible and and simple when compared with a fine FEA model.Based on the cracking patterns of the concrete pavements tested at FAA, the two-dimensional cellular automata (CA) is proposed to calculate the cracking pattern of the concrete pavement slab. Different from the conventional method, the numerical mode of the CA is developed to describe the structural characteristic of slab. Then, based on the tested cracking patterns of the full-scale pavement slabs, the cracking patterns of slabs with different sizes from the base sabs are calculated through the criterion for matching zone similarity and the criterion for projecting cracking values. The CA method realized the prediction of the cracking pattern of the concrete pavements directly based on the tested cracking patterns of pavement slabs.The thesis tries to simulate the maximum stress distribution of the rigid pavement under moving load using the finite element method. The maximum stress distribution and their corresponding cracking patterns obtained in the simulation are used to judge the cracking patterns mapped by the CA method. The results show that both results are basically identical, but the CA method is simple and efficient compared to the finite element method. |
PDF Full Text Request