Rock creep under long-term stress may exhibit three stages:decay creep,steady-state creep,and accelerated creep.The Cvisc model in FLAC 3D software can effectively simulate the decay and steady-state creep stages of rocks but shows significant limitations in describing the accelerated creep stage.To address this,the viscosity coeffi-cient of the Cvisc model was modified to create a non-steady version capable of capturing the characteristics of ac-celerated creep.A corresponding three-dimensional creep equation was derived.To obtain the creep parameters of the non-steady Cvisc model for surrounding rock,a BP-PSO algorithm-based neural network inversion model was established.This algorithm was validated using monitored vault settlement data from field tunnels.A comparative analysis of the vault settlement prediction performance of the non-steady Cvisc model and the original Cvisc model was conducted.The results demonstrate that the non-steady Cvisc model significantly outperforms the original model in fitting the accelerated creep stage,enabling a more accurate description of the entire creep process,including the accelerated stage.
关键词
岩石力学/非定常Cvisc模型/三维蠕变方程/时效损伤/参数反演
Key words
Rock mechanics/Non-steady Cvisc model/Three-dimensional creep equation/Time-dependent damage/Parameter inversion
维普
万方数据