Experimental Study On Dynamic Resilient Modulus Of High Liquid Limit Silt Under Dry-Wet Cycles

In the actual operation of subgrade,the humidity state of subgrade soil is in constant cyclic change due to the influence of groundwater level change,drying evaporation and precipitation.Especially in the south of China,the subgrade soil is seriously affected by the environmental climate,which is often under dry and humid conditions.The dry-wet cycles have a significant impact on the working performance of the subgrade,leads to differential settlement and other road diseases.Besides,it will also reduce road usage life and increases the cost of road maintenance.Therefore,it is necessary to study the stiffness attenuation characteristics of subgrade soils under repeated dry-wet cycles to ensure its durability and stability.And then promoting scientific and reasonable subgrade design method.Some researchers have studied the attenuation mechanism of the resilient modulus under different dry-wet cycles.However,these test methods are difficult to simulate the dry-wet cycles characteristics of subgrade soil in real cases.Therefore,a new test method for dry-wet test was developed in this study.Then,the dynamic triaxial tests of subgrade soils under different dry-wet cycles and different working conditions were carried out,and also the variation law of resilient modulus with various factors was explored.Finally,based on the existing research,the resilient modulus prediction model of subgrade soil was established,which comprehensively considered the number of dry-wet cycles,humidity parameters,stress parameters and state parameters.It can be seen from the experimental results that the resilient modulus is positively correlated with compaction degree and confining pressure,and negatively correlated with the number of dry-wet cycles,deviatoric stress and moisture content.The dry-wet cycles have an important influence on MR at the first time,and the attenuation range of the resilient modulus is 27%-54%.After 5 dry-wet cycles,the resilience modulus tends to be stable and finally decreases by more than 63%.In addition,the dry-wet cycles damage is more sensitive to the change of moisture content,and the larger the target moisture content is,the larger the reduction range of MR will be,and even the specimen will be damaged.However,the resilient modulus has a low sensitivity to the compaction degree and the change of stress state under dry-wet cycles.Finally,a prediction model of resilient modulus considering the number of dry-wet cycles is established.It has been proved that the model has the advantages of simplicity,clear parameters and high prediction accuracy.A three-layer ANN method was established.The physical properties(liquid limit,plastic limit,plasticity index,0.075mm passing percentage,maximum dry density,optimum moisture content),state variables(degree of compaction,relative moisture content,number of dry-wet cycles)and stress variables(confining pressure,deviatoric stress)of subgrade soils were selected as input variables.The MR was directly used as the output variable.The results show that the developed optimized ANN method has the advantages of fast speed,strong generalization ability and good accuracy in MR estimation.