Page 76 - Numerical Analysis and Modelling in Geomechanics

COMPRESSED AIR TUNNELLING 57

Figure 2.20 Failure envelope for the soil tested in the experiments.

identified using a parameter identification technique from back analysis of some
measured data from a tunnel. It was shown that the predicted variation of the
permeability of shotcrete as it cures is comparable to that of concrete and cement
paste reported in the literature.
A genetic algorithm and weighted least squares approach was used in the
optimisation procedure in the identification process. A genetic algorithm has the
advantage of simplicity and high efficiency in locating the global optimum
solution in the optimisation process.
The calculated and measured air losses from a tunnel were compared. The
comparison shows that the established relationship for variation of permeability
of shotcrete and the numerical model produce realistic results. The proposed
relationship was verified against a case study. However, although the tunnel
passed through various soil conditions in this case study, the results of the
numerical prediction show the same trend as the measured data in the field. It is
recommended that this relationship and the procedure are tested against other
case studies but case studies on field measurements in compressed air tunnelling
are rare.