Study On Gas Migration And Diffusion Law And Concentration Dynamic Prediction In Construction Tunnel

With the implementation of the strategy of Western Development and First Build Roads in China,railway in mountainous areas has developed rapidly.Due to the influence of complex geographical conditions and rich mineral resources in Southwest,gas accidents often occur in railway gas tunnel engineering.Gas concentration is an important indicator to measure the damage degree of tunnel gas.In China,most of gas accidents are related to gas concentration exceeding the limit.Therefore,it is of great theoretical value and practical significance to study the gas migration and diffusion process during tunnel construction,analyze the influencing factors of gas concentration and explore the sequence signature of gas concentration data for controlling gas concentration and preventing gas disaster.This paper takes the gas emission in the construction process of Yuelongmen tunnel of Chenglan Railway as the background,to optimize the aeration control scheme and improve gas monitoring methods;through theoretical analysis,numerical simulation and ELM model prediction,systematically studying the law of gas migration and diffusion,gas space-time evolution characteristics and dynamic prediction of gas concentration during tunnel construction.The main conclusions are as follows:（1）.Through establishing the theoretical model of tunnel gas migration and diffusion and the three-dimensional finite volume numerical model,studying the gas migration and diffusion law and spatial and temporal distribution characteristics under two typical ventilation modes（forced ventilation and gallery ventilation type）of gas tunnel.The results show that according to the flow pattern of forced ventilation air flow,the flow field structure along the tunnel can be divided into vortex section,development section and stable section.Among them,the developing and stable sections have little influence on gas diffusion,the gas concentration distributed uniformly.The vortex section is an important area affecting the gas distribution.The vortex at the arch shoulder and arch foot on the left side of the tunnel face（no duct side）has the effect of accumulation and retention of gas.Adopted gallery ventilation technology,ejection effect occurs when the air flow on both sides of the main tunnel converges in the cross cave,which will entrap the low-speed airflow in the main tunnel,accelerate the gas diffusion rate and improve the ventilation efficiency of the tunnel.When the gas flows into the cross cave from the main tunnel,the concentration decreases by about 40.85%.However,due to the change of air flow pattern and tunnel sidewall during confluence,there are two ventilation"dead zones"and three"vortex zones"near the transverse tunnel,in which the"dead zone"and"vortex zone"of the transverse tunnel have the effect of accumulation and retention on gas diffusion.（2）To optimize the ventilation control scheme,single factor control method was adopted to study the influencing factors of tunnel gas concentration and different air duct layout position,the distance between the air outlet and the palm surface,flow field characteristics under tunnel construction method and gas diffusion law were selected for analysis.The results show that the gas diffusion rate is affected by the air flow velocity and eddy current intensity,and the eddy current intensity is the main factor.When the air duct is arranged in a vault,the wind speed on the palm surface is evenly distributed and the eddy intensity is small.Therefore,when the gas concentration is lower than the allowable concentration during the whole construction stage,air duct layout has the best ventilation effect,followed by arch shoulder layout,arch foot layout has the worst ventilation effect.The gas diffusion effect is inversely proportional to the distance between the air outlet and the palm surface:when the ventilation distance is 1 A^1/2,the gas concentration in the tunnel was lower than the allowable concentration.With the increase of ventilation distance,the gas concentration is reduced to the allowable concentration and the required ventilation time is increased;when the ventilation distance is 4 A^1/2,the jet air generates invalid jet and it takes 5.25h for the gas concentration to be reduced to the allowable concentration.When using the full cross-section method,bench cut method and lower pilot tunnel advanced method,the time required for the gas concentration to be reduced to the allowable concentration is 0.51h,3.15h and 5.32h.Among them,the step method and the lower pilot tunnel advance method have ineffective jet in the construction of jet air.（3）In order to further improve the prevention and control ability of tunnel gas disaster,a prediction model of tunnel gas concentration was constructed by extreme learning machine（ELM）according to the sequence characteristics of gas concentration data.In addition,the dynamic prediction of gas concentration was verified by real time gas monitoring data and the model is validated with on-site test data.The result shows that the ELM prediction model has good learning ability in processing gas concentration data.There is no under-fitting or over-fitting phenomenon and the prediction error of gas concentration is kept within -0.035%～0.032%.The assessment index of test set sample accuracy R²、RMSE、MAE and ME are 0.9678、0.0089%、0.0063%、0.0359%in sequence,which meets the requirements of tunnel construction.It can dynamically predict the gas concentration in the short term and assist the tunnel monitoring system to prevent gas disasters.