H. Huang and D. Zhang / Quantitative Geotechnical Risk Management for Tunneling Projects in China 6
the key principle of the method and its
application into the tunnel case are presented
briefly below.
3.1. Fuzzy Fault Tree Analysis (FTA) —Hazard
Identification
It has been widely recognized that the damage
of the tunnel is not likely to be caused only by
a single hazard. There might be a chain effect
between hazards. The fault tree (FT) is always
built to systematically understand the growth
path of a catastrophic event. A typical fault
tree for the damage of the cutter of the earth
pressure balance (EPB) shield machine in
tunneling is shown in Figure 3 (Yan, et al.,
2009). The top event can be triggered by a
combination of the sub-event serially or
parallelly. In this case, the cutter damage can
be triggered by three major sub-events, i.c.,
poor ground condition, irrational construction
and shield factor. In addition, the shield factor
could be further triggered by three "sub-sub-
events", Note that the cutter damage at the top
of the tree also can be a sub-event for a more
serious event, such as cutter failure or failure
of the EPB machine.
S==e
—
Figure 3. An example of fault tree analysis (FTA) for the
ccutter damage of EPB shield machine
When the events that cannot be further
divided, i.e, basic events, are available, the
probability of the occurrence of the top event
can be calculated from Eq. 3 below,
P.=1-J][l- Po) @)
P(M) is the probability of the occurrence of a
minimal cut sets of the events that could
directly trigger the occurrence of the top event.
The independency between minimal cut sets is
assumed in the calculation, However, the
probability for the basic event is usually
difficult to be quantified, Hence, the fuzzy set
theory is adopted to cope with it, A triangular
possibility distribution of the probability of the
occurrence of the basic event is used and
plotted in Figure 4 (Bian and Huang, 2006).
Then the fuzzy probability of the top event can
be expressed as triangular fuzzy numbers and
the parameters. It reflects the robustness of the
calculated probability of the top event.
Figure 4, Possibility distribution of basic event probability
¢ main basic events affecting the
occurrence probability of the top event can be
determined and some effective measures are
verified by sensitivity analysis to reduce
occurrence probability of the basic events and
the top event. The sensitivity of basic event
can be evaluated by the index V; as below:
(4)
where jig is the occurrence probability of the
top event, y,, is the average occurrence
probability of the basic event x,
3.2. Conditional Markov Process (CMP) for
Soil Distribution Probability
The uncertainty in the tunneling can be largely
attributed to the uncertainty of the spatially
varied soils along the tunnel longitudinal
direction, The limited site investigation in
terms of the borehole numbers is the major
source that creates the soil uncertainties. It is
customary to linearly characterize soil layering
between boreholes. However, the tunnel
failures are usually caused by the
underestimation of the complex distribution of
the layered soils. In view of this limitation, the
conditional Markov process (CMP) can be
adopted to fully utilize the existing borehole