Estimating the safety factor of monsoon-affected urban drainage systems using fault tree analysis

By Florida Visser, supervised by Martine Rutten

Many cities in South-East Asia increasingly face flooding problems that frequently inundate
large parts of the city. The main cause is twofold: these cities are often affected by
monsoons, and rapid urbanization quickly alters the runoff generation. Drainage systems can
often not keep up with the changes and start to fail. There are many different failure
mechanisms that cause failure in different parts the system. These mechanisms are often
known to local experts and maintenance workers and are treated one by one. The mutual
relationships of the causes of failure are easily overlooked in this way. For this research, the
capitol of Vietnam, Hanoi, is taken as a case study. This city has grown from 1.7 to 7 million
inhabitants in the last 2 decades. During the monsoon season, heavy rains occur and can
inundate certain locations multiple times per year.

This research aims to provide a more formal approach to the conditionality of the failure
mechanisms: in what way do failure mechanisms relate to each other? This is done by means
of fault tree analysis (FTA). This method connects the different mechanisms using the failure
probability of each specific failure mechanism and connects them using Boolean logic. The
methodology is carried out in two main steps. First, the fault tree is drawn using the input of
local experts. Interviews and a focus group are held with inhabitants of Hanoi that come
from different disciplines, such that their experience is represented in a formal way. Second,
a normative neighborhood of Hanoi is modelled using PCSWMM software. This software
combines the tools of the well-known drainage modelling software SWMM and e.g. QGIS.
Additionally, the software enables the user to execute the scripting tool. This allows me to
perform Monte Carlo simulations, such that the relevant uncertainties can be estimated. In
this way, the sensitivity to certain failure mechanisms is explored, and the failure probability
of these mechanisms can be predicted using characteristic values and ranges.

The formal approach of FTA provides a clear and quick overview of the vulnerable parts of
the system, and their relative contributions to the inundation problems. The result of this
study can be used for local drainage authorities, such that they can make more targeted
improvements.