Research On Rapid Determination Method Of CBR Indices For Expansive Soil Filler

Physical treatment technique can bring remarkable economic, ecological andenvironmental benefits during the construction of expansive soil subgrade ,and thekey and precondition of this technique is that whether or not filler is qualified.Nanning Basin is the well-knowen region with widespread distribution of expansivesoil in china. About 1 million cubic meters expansive soil need to be used during theconstruction of outer ring expressway. Because of the local climatic conditions ofplenty of rainfall, the time for construction is very short. In order to use expansivesoil filling subgrade in great force, the technical problem that evaluation of filler canbe fast and accurate obtained must be solved firstly. Therefore, this paper carried outmodified CBR expansive soil filler rapid measurement research.Based on the modified CBR test results and analysis and research of predictiontheory and method at home and abroad, determine the test plan and carry out test work.Test on soil properties, strength and deformation of road performance were performedfor three representative kinds of expansive soil. According to the demand ofprediction, the immersion time of expansive soil samples were determined. (4h, 8h,12h, 16h, 20h, 24h, 28h, 32h, 36h, 40h, 44h, 48h, 56h, 64h, 72h, 80h, 88h, 96h ).Modified CBR testing of 108 samples ( 54 group ) have been completed, whichprovides necessary data information for prediction study.Using exponential, hyperbolic, power function, square roots and logarithm ofRegression parameter model to carry out curve fit about change law of CBR indexesand CBR swelling capacity with immersion time, we found that the exponential modelis most precise. Therefore, it can be concluded that the exponential prediction modelshould be choosed for predicting the modified CBR indexes based on the the testingresults of soaking for 1 day.In order to get the best prediction effect, this paper introduced organic grayneural network model ( OGN ). Using OGN carriing out the prediction analysis of theseries of actual measured data of three kinds soil samples, we found that the CBRindexes for soaking 4 days can be predicted well according to the testing results ( 6data ) of soaking for 1 day and the error is very smaller than the prediction results ofGM (1,1) model. In addition, the prediction of modified CBR indexes of filler can berealized by using OGN model directly or establishing realization between CBR indexes and CBR swelling capacity indirectly. Finally, the prediction method ofmodified CBR for expansive soil filler which use the testing results of soaking for 1day was presented based on OGN model.The rapid measurement method for modified CBR indexes was established topromote the generalization of new physical treatment technology of expansive soilembankment.