Experimental Research On BFRC Mechanical Properties And BP Neural Network Prediction Of Constitutive Model

Fiber reinforced concrete has been one of the hot issues in the field of composite materials research because of its enhanced and toughened mechanical properties and its ability to better solve the durability problems of engineering components under extreme environments such as freeze-thaw,corrosion and salt invasion.It is difficult for traditional concrete structures to avoid cracks during construction or service,which will lead to corrosion of internal reinforcement,deterioration of concrete performance and shortened service life.Adding a certain amount of basalt fiber into concrete can enhance the mechanical properties of toughened traditional concrete structures,control the propagation of microcracks,and improve the durability of concrete structures.This paper takes short-cut basalt fiber concrete(hereinafter referred to as BFRC)as the research object,carries out BFRC compressive strength test,splitting tensile test and triaxial compression test,analyzes the effects of confining pressure,fiber length and volume fraction on the reinforcement and toughening of concrete,and gives the corresponding optimization results.Based on the experimental data,this paper adopts the BP neural network algorithm and realizes the training and prediction of BFRC constitutive model by programming in Matlab.The main research contents of this paper are as follows:1.The existing design methods of fiber concrete mix ratio are analyzed and summarized,and the advantages and disadvantages are compared based on the actual situation.The fiber mixing method is selected to calculate the content of basalt fiber.Basalt fibers with diameter of 15 micron and fiber length of 6 mm,12 mm and 18 mm were selected.Short-cut basalt fibre concrete with fiber volume fraction of 0%,0.2%,0.4% and 0.6% was obtained by adding fibers.Collapse test is used to measure the working performance of basalt fiber concrete.The results show that the higher the volume fraction of the fibers,the smaller the collapse.Adding fibers can delay the settlement of aggregate,improve the cohesion and water resistance of concrete,and show good working performance.2.Based on the existing implementation plans,industry standards and relevant theoretical foundations of compressive strength,splitting tensile and triaxial compression tests of fiber concrete,the compressive strength and splitting tensile test plans for MTS816 material test platform are formulated.The results show that the compressive strength and splitting tensile strength of BFRC are the highest when the volume fraction of BFRC is 0.2%,except for the fiber length of 6mm,but the compressive strength decreases with the continuous increase of the volume fraction of the fiber.The splitting tensile strength of concrete mixed with fibers is always higher than that of miserable fibers,and it can be found from the test data that the splitting tensile strength is stronger than that of compressive strength.3.To further explore the mechanical properties of BFRC under different stress states,compression tests under different confining pressures were carried out for BFRC.The test results show that the failure mode of BFRC specimens is related to the confining pressure and is independent of the length and volume fraction of basalt fibers.Without confining pressure,the specimens are fractured and cracked through;with confining pressure,the specimens are diagonally sheared.The main crack angle increases with the increase of confining pressure.The length and volume fraction of the fibers do not change the failure mode of the specimens,but the width of the main cracks and the number of fine cracks can be reduced and the integrity of the concrete can be improved.With the same fiber length and volume fraction,the peak stress and peak strain of basalt fiber concrete increase in varying degrees with the increase of confining pressure.Continuing to increase the confining pressure,the effect of fiber reinforced concrete gradually decreases.4.BP neural network is used for sample training and model prediction of BFRC constitutive model.The network parameters and topological structure are set according to the test data.70% of the data is taken as training set and 30% as test set to verify the trained network.The result shows that the error between the predicted value and the test value is small and has good consistency.Besides,it also shows that this method can better predict the constitutive relation of BFRC based on sufficient sample data,solve the difficulty of manual modeling,and establish the implicit relationship between stress and strain of BFRC.