Optimization of shield tunneling parameters under controlled surface settlements

The construction of the tunnel usually leads to settlement which normally
depends on the tunnel geometries, geological condition and tunnel construction parameters. The
tunnel construction rate commonly decrease with more strict value of controlled surface settlement,
resulting in an increase of time and budget of tunnel construction. This study develops an
approach to determine the optimal tunneling parameters using genetic algorithm (GA) with varying
allowable surface settlements incorporated with the artificial neural network (ANN). The
ANN is used to construct prediction models of surface settlement and tunnel construction rate. In
this study, the MRTA Blue Line data are used to train the ANN and used as case study for determining
optimal tunnel construction parameters. The results demonstrates that the approach of
combination of ANN and GA can be an efficient tool for application in tunnel construction.
With the data of MRTA blue line, to obtain the maximum construction rate, the penetration rate
and grouting pressure have to change significantly with the variation of allowable settlement.