Identifying locally induced loads in hard rock tunnel linings by distributed optical fibre sensors

Licentiate thesis, 2024

In an infrastructure system, tunnels enable fast means of transportation in
impassable terrain or city areas. However, inspections and maintenance can be
expensive due to long down time and, in hard rock tunnels, the complexity of
the structure may lead to over designing.
This thesis presents an experimental series aiming to reproduce loading conditions
acting on shotcrete tunnel linings in hard rock. The specimens consisted of
two concrete layers, a top fibre reinforced concrete layer simulating a shotcrete
lining, and a 300 mm thick substrate layer simulating rock. The top FRC layer
response was measured using strain sensing distributed optical fibre sensors
(DOFS) installed in two layers. A full factorial design with four factors and
two levels was applied, where the varied factors were load conditions, load size,
lining thickness and tensile bond strength between lining and substrate.
From the experimental results, it is concluded that using DOFS to monitor
loads in tunnel linings are suitable as the global behaviour is captured at
low load magnitudes and that performance indicators, such as cracks, can be
used to identify the load type.
The work presented is part of a research project aiming to develop a machine
learning model, trained on synthetic data from finite element analyses
calibrated with the acquired experimental results, to identify loads and predict
the structural integrity of a tunnel lining. The experimental results will ultimately
be used to verify the machine learning model and, thus, implications
for future finite element and machine learning models are discussed based on
the experimental results.