Reliability Based Light-weight Design Of TBM Cutterhead

Tunnel Boring Machine (TBM) is often used in extreme environments. As one of the core parts of TBM, cutterhead directly affects the construction efficiency, stability and lifecycle of a TBM, thus the reliability during construction must be ensured. Research on structure reliability is totally different from traditional design method which is generally based on safety-factor. Reliability based structure design is to measure the scale of reliability from the point of malfunction probability. This paper combines FEM, statistics and construction data all together to analyze the strength and stiffness reliability of a TBM cutterhead. To achieve this, a cutterhead is first designed by some complicated theory and then used to change its plate thickness to get a proper reliable and light-weight TBM cutterhead. The main contents are as follows:Research on cutterhead reliability. As to strength and stiffness reliability of a TBM cutterhead, implicit performance function makes it impossible to get reliability by mathematical methods. Monte Carlo Simulation (MCS) is the basic method to get structure reliability; however, MCS needs unacceptable time cost to perform FEA calculation repeatedly in ANSYS. So it's naturally to get a substitution function for the whole FEA calculation. By attempts on several learning machines, Support Vector Machine for Regression (SVR) is used to train the substitution function. Once the function is obtained, we can use SVR-MCS to calculate cutterhead reliability.Research on TBM cutterhead design method. TBM cutterhead design involves multi-subject knowledge and it's quite complicated. This paper is based on the former research achievements of our whole team. Random cutter load is studied and some assumptions are put forward to simplify the big load distribution problem. Based on all these, an general parametrical APDL program is presented.Research on probability design method. To select important plate thickness as optimization variables, probability design method is introduced to make sensitivity test of all the plate thickness and those affects the strength and stiffness most are selected as optimization variables. After that, these variables together with random input loads are grouped as input variables, strength and stiffness are output, sample data between input-output variables is collected by FEA calculation in ANSYS.Light-weight optimization by Particle Swarm Optimization (PSO). Use the strength and stiffness reliability of cutterhead as constraints, its volume (weight) as goal, PSO is the optimization tool. Finally, select the best set of variables to check the actual reliability under certain distribution of cutter loads in ANSYS. This paper presents a practical method of Design-APDL-SVR-MCS-PSO. By a test optimization over Yintao T9 TBM cutterhead, the result shows that the method has very high simulation accuracy. While both strength and stiffness reliability stays higher than 0.99, the cutterhead weigh is optimized by 2.43%, which is about 2.02t. It shows that method of this paper can be generally used to other cutterhead designs.