Air losses in compressed air tunnelling: a prediction model

Abstract

Data (measurements) from a tunnel in Munich, Germany, were used to develop and validate a model to predict air loss volumes in the process of compressed air tunnelling. In the implemented case study, compressed air was used as a measure to control the groundwater followed by placing of a shotcrete lining as temporary support. Evolutionary polynomial regression (EPR) was used for modelling. EPR is a data-driven method based on evolutionary computing that aims to search for an explicit and structured polynomial model representing air losses. Comparisons made between the actual measurements and the model predictions showed robustness of the suggested model in learning and predicting the behaviour of the system. A sensitivity analysis was also performed on the developed model; this revealed the reliability of the model by correctly presenting the expected effects from air permeability of the soil despite the non-homogeneous and layered nature of the geo-materials involved in this case study.