Analysis Of Design Factors Impacting Joint Plain Concrete Pavement Faulting And Development Of Forecasting Model Using LTPP Data

Concrete pavement faulting affects the comfortableness of driving and road trafficcapacity. In order to decrease the adverse effects of faulting, and control faulting duringdesign and maintenance management, it is necessary for us to know the impact factors andtotally grasp the development law of faulting. In China, due to the lack of faulting and itsrelated data accumulated over a long time, it is difficult for deeply research into thedevelopment law and forecasting of faulting. The United States AASHTO’s concretepavement design method in MEPDG put forward faulting forecasting model, but it needsmany design parameters to calibrate. The lack of enough faulting data in our pavementmanagement system leads MEPDG concrete pavement faulting model cannot successfullyapply in our country in a short term.There are a lot of faulting data in the United States Long-Term Pavement Performance(LTPP) database, and those data can be used in to build fault model. Although the road trafficconditions, pavement structure types and environment in our country is not the same withUnited States, but the development law of faulting still has inner links and similarity. So it isnecessary to predict the faulting based on the faulting data from the LTPP database, thus wecan master the impact factors and development law of faulting, which is feasibility andpractical.This thesis analyze the pavement design and engineering practices of united states,combine with LTPP database, a faulting database is established to study the development offaulting. Considering the impact factors on faulting and make a sensitivity analysis, a cementpavement forecasting model is established with BP neural network and orthogonal design isconsidered. The paper’s idea, methods and main results are as follows:(1) Analyze the MEPDG model building and its software, discuss the feasibility of itsapplication in short-term. The behavior of concrete pavement fault in America is studiedbased on LTPP data.(2) To study the mechanics mechanism of concrete pavement in the development offaulting, deeply probe into the reasons of the occurrence of faulting, and refine the threedifferent steps of faulting’s development process. The influence of factors to faulting is also studied In addition, the influence of main factors to faulting is also discussed combining withthe quantitative studies of MEPDG model.(3) Put forward extraction method of faulting analytical samples and establish a set ofthorough treatment method to faulting samples’ quality from the prospective of time andspace. Analyze the relationship between correlated variable and faulting of faulting samples.Using the infiltration rate of joint allowing water quantize the joint condition, coefficient ofbasement washing quantize joint form, load transfer ability of the edge, support ability of theshoulder quantize shoulder form and drainage rate of rainfall quantize drainage capacity. Dataquantitative principle is put forward and LTPP data are converted into variables related tofaulting so that faulting sample analyzing database can be reestablished.(4) Sensitivity analysis to the factors is conducted based on the new database, analysesthe qualities of data and experiment designs’ influence on sensitivity analysis. Then studythe importance ranking of different type of factors under different experiment designs. Andfinally put forward some design suggestions to faulting for different climate partition.(5) Combined with sensitivity analysis results, the BP neural network prediction researchto the development of the faulting is carried out. As a result, optimization idea of BP neuralnetwork prediction model is presented and a reasonable BP neural network prediction modelof different climate partition is established. At the same time, combined with orthogonalexperiment design, analysis thought and method to quantizing the predict results of BP neuralnetwork is put forward and the faulting calculation formula of pavement is given out based onthe predict result of BP neural network.Theoretical analysis and practical engineering problems of concrete pavement arecombined in this paper. The result is helpful for learning developing rules of faulting and theprediction model can be further corrected through the accumulated process of faulting data inour country, so it has values of certain theoretical research and engineering application.