GPR similarity criteria and model experimental study on artificial freezing engineering

摘要

Nowadays, the artificial freezing method is an important technique to construct in deep alluvium and fracture rock layers. It was needed to develop a highly efficient, highly precise and non-destructive testing method for freezing wall monitoring. Based on the partial difference function of radar waves, the similarity transformation method was used to deduce the relationship between model experiments and prototype detection. Similarity criterias of GPR EM fields was proposed, and the similarity preconditions of GPR EM field were suggested. The preconditions given in this paper compare the travel time and amplitude of GPR signals between model experiments and prototype detections. Higher frequency antennas were commonly used in GPR model experiments. For instance, in the Artificial Freezing Engineering (AFE) GPR model, according to the feature of its EM parameters at frequencies form 10MHz to 1000MHz, such as similar magnetic permeability, similar dielectric constant and slightly lower electric conductivity, the similarity criteria could not be satisfied when the prototype media were used in model experiments; and the amplitude of electromagnetic (EM) fields could not be strictly similar to that of prototypes. It was deduced that in AFE the travel time and characters of anomalies in GPR profiles are similar to that of prototypes, while the amplitude of GPR signals would be much stronger than that of prototypes. By the model experiment and field exploration, the characteristics of typical anomalies in AFE, such as freezing pipe and defects, were obtained and reported in this paper. It is found that they played an important role in the construction of the connecting passage of Shanghai Yangtze River Tunnel. The research findings have an important theoretical value for AFE and the permafrost region engineering. The Similarity Criteria and analysis method can also be used in other areas where the prototype media are used in GPR model experiments.