Air losses through the tunnel face in compressed air tunneling : a case study

In this study field measurement data from Feldmoching tunnel in Munich were used to develop a model with the capability of predicting the air losses in compressed air tunnelling technique. In this tunnelling project, compressed air was used to control the groundwater and Shotcrete was used as temporary support. The final permanent lining was added after completion of the temporary support and was done in free air condition. An evolutionary approach for modelling the air losses from the face of the tunnel was adopted in this study. Evolutionary polynomial regression (EPR) is a data-driven method based on evolutionary computing aimed to search for polynomial structures representing the target system. In this technique, a combination of the ge-netic algorithm (GA) and the least square method is used to search for the most feasible structures and the appropriate parame-ters of the selected model structures. Data from different tunnel lengths were used in the training phase. The trained model was then used to predict the air losses during the excavation of the points of the tunnel data from which were not used in the train-ing stage of the modelling process. The results of the EPR model predictions were compared with the experimental data. It was shown that the suggested technique can learn and generalize the relationship between appropriate soil and tunnel parameters to predict air losses for different geological and geometric conditions that were not previously seen in the training stage by the EPR, with a good degree of consistency with actual field measurements