RISK ANALYSIS OF FLOODING OF URBAN DRAINAGE SYSTEMS

Linmei Nie and Wolfgang Schilling, Solfrid Engene Royset and Jan Hovden

Department of Hydraulic and Environmental Engineering,

Norwegian University of Science and Technology, 7491 Trondheim, Norway

(linmei.nie@bygg.ntnu.no and wolfgang.schilling@bygg.ntnu.no)

Abstract: Floods are considered as to be among the most severe natural disasters because of their unpredictability, their overwhelming power and the difficulty to assess the related risks. Consequently, flooding damage is inevitable. This paper focuses on the identification of potential flood hazards in urban drainage systems. All together, 68 individual hazards are identified. Then the flooding causes and modes are analysed by two main methods: fault tree and event tree. The frequencies of natural events and operational obstacles are estimated based on measured climate data and operational records. Consequences of flooding are classified based on the people at risk, the property damages and environmental impacts.  As a result, the flooding risks are described in combination with frequency analysis and consequence analysis, which can be used as a basis for rational flooding mitigation strategies. Trondheim municipality in Norway is used as a case study. The study indicates that it is valuable to analyse the risks of flooding in such an integrated way in order to control and reduce the risks as low as reasonably practical (ALARP). The study also reveals that the joint risk of different hazards is significant and definitely deteriorates the flooding situation. Needless to say, that the value of such analysis is a function of the availability, comprehensiveness and quality of the applied data.

Keywords: risk analysis, integrated risks, fault tree, root cause, contributing factors, event tree, joint risk

1    INTRODUCTION

Risk analysis of flooding has always been one of the core topics of research and one of the main tasks of water management because of the extreme impacts of floods on human beings, their properties and their activities. When floods strike cities, consequences are particularly severe due to the density of population and the value of property. Historically, emphasis is given to extreme floods in natural watercourses. Hydrologic models have been developed and consequences have been evaluated in order to formulate mitigation strategies and take effective protective action. This focus has undoubtedly increased awareness and has played a significant role in formulating rational floodplain management schemes.

However, extreme disasters occur with a low frequency. Another type of flooding is due to comparatively small runoff events in areas where a floodplain cannot be delineated, i.e. in urban drainage systems. In cities, literally every street is a floodplain, and it is practically impossible to impose their restricted use due to a certain level of flood risk. These floods are caused by insufficient capacity of sewers, lack of maintenance, operational faults or planning flaws. Their extent is smaller and their appearance less spectacular. But they happen more frequently and, therefore, result in vast economic damage. According to the annual report of Trondheim municipality, the average annual sewer damage caused by draining water in the last ten years (1990-1999) were 0.77* million krone. In some cases, the combination of small incidents can trigger a big event (T.M.1988-99). There is also a tendency that flooding damages per individual case increase, because many basements have been improved and more valuable property are stored there, some of them even being changed to living space rather than used only as storage.

Nevertheless, there seems to be no models available that can be used to study the joint risk of natural events and operational obstacles in urban drainage systems. One reason might be the obvious difficulty to calibrate such a modelling because of incomplete operational records and the limited quality of the registered data. In short, there is a shortage in risk management methodology and models regarding to the assessment of integrated flooding risks in urban drainage system.

In recent years the frequency of urban flooding seems to increase. In Nordic counties with their cool climate these incidents often happen in winter or early spring and can be caused by:

(1) Long periods of moderate rainfall on frozen surface because of the contribution of runoff from higher impervious iced areas.

(2) Long periods of gradual snowmelt saturating the soil and with subsequent heavy precipitation, a large part of the precipitation will contribute to surface runoff.

(3) Rapid snow melting combined with rainfall.

All these phenomena have frequently occurred in Trondheim in recent years (Milina J. and Selseth I. 1999).

Flood research requires analysis of causes, assessment of effects and proposal of solutions. It appears that a coalition of multiple research disciplines is necessary to tackle to challenge. In addition it is interesting to study the progress carried out in fields with similar problems such as industrial accident investigation procedures and requirements for risk analysis (Kjellen 2000, NSF 1991), where an urban drainage system can be defined as a tightly coupled production system and its risk of flooding be analysed as an “industrial accident”.

The basic concepts applied in this paper are lined out in next part.

2    BASIC CONCEPTS OF RISK ANALYSIS

2.1   Definition of risk

Risk designates the possibility that events might occur, which is undesired for humans or the environment. It is expressed with the probability and the consequences of the undesired events (NSF, 1991). However, for different systems and different purposes, this definition is introduced and applied in different ways. Here in this paper, we define risk as a function of probability of events and their countable damage and other consequences, which are uncountable in monitory unit. It can be expressed as:

Risk = f (probability * damage) + other consequences

Although this definition is an arithmetic expression it cannot be simply understood as a quantitative value, because the consequences can be economic damage and/or other impacts on human beings, their properties and activities as well as the environment. Some of them are too complex to be evaluated merely as a numerical value, while others might be potential and long-term effects.

Moreover, it should also be noted that the risk has a time dimension built into it because the magnitude of a hazard and the capability of its resisting body will vary with time.

2.2   Risk analysis precedures

Risk analysis is a systematic approach for describing and/or calculating risk. The procedures applied in this paper involve:

l       Hazards Identification;

l       Causal analysis;

l       Consequence analysis;

l       Description of risk.

2.3   Basic terminology

There is an established terminology widely used in literatures. In this paper, we define some key terms with the following interpretations:

l       Hazard means a potential threat. It might happen, or it might be controlled or reduced by maintenance or protection measures.

l       Incident indicates a small operational obstacle without significant damage.

l       Event expresses either big operational obstacle or a natural disaster with severe consequences.

l       Accident can be between of an incident and an event with small or medium damage.

3   RISK ANALYSIS

3.1   System definition

Flooding is defined as temporary inundation of all or part of a floodplain along a well-defined channel, or temporary localised inundation occurring when surface water runoff moves via surface flow, swales, channels, and sewers toward well-defined channels (Walesh, 1989). With this definition one can argue that there can be no flooding if there is no surface runoff. However, even during dry weather condition large damage can occur due to clogging of pipes and backwater into basements. In wet weather case, sewer obstacles definitely deteriorate any flooding situation, or might cause flooding that, without the obstacle, would not have happened. With these consequences in mind, we have to consider sewer obstacles as flooding hazards and thus have to expand the definition of flooding given above.

Urban drainage systems are complex socio-technical systems, which relate to climate, topography, land use, human activities and environment. The way they perform depends on natural conditions (e.g. rainfall), their technical concept (e.g. design flow), and the way they are managed and operated at different levels. Therefore, the total flooding hazard in an urban drainage system comprises natural risks, technical problems and human errors. (Fig.1.)

3.2    Identification of hazards

Related to above specified three main domains, technical hazards are identified following a longitudinal process ranging from preliminary investigation and fieldwork, design, construction, operational and maintenance stage. Human errors, ranging from strategy error to mis-operation, can occur at any stage and at any level. Total 68 hazards are covered in the checklist (Nie et al., 2000). Examples are given in Table 1.

     Table 1  Examples of urban flooding hazards

* Type A and B represent the physical (natural and technical) hazards and human factor respectively.

3.3    Causal analysis

Following the principle of fault tree, above hazards identified are ranked into several categories based on the contribution of a hazard to flooding event, which starts with a flooding event and finds the prerequisite conditions downwards to root causes, see table 2.

Table 2 Simplified fault tree in tabular format

* The same hazards codes are introduced in table 2 as in project reports.

The root causes, i.e. the most basic causes for flooding, are mostly human factors, while the others are considered as contributing factors.

Apparently, flooding can be triggered by one or more hazards. In order to find the flooding events, a physical flooding event tree is constructed and depicted in (Fig.2) based on following criteria:

l       Including accidental events;

l       In the dry weather condition, the damage is predominated by sewer incidents;

l       In moderate wet weather condition, the joint risks of natural events and sewer incidents are the control situation;

l       In extreme weather condition, omit the impacts of sewer incidents;

l       Catastrophic flooding might be aggregated by all possible risks.

In order to assess the probability of flooding, we assume that all the risk events are independent, and then calculate the corresponding frequency. The given frequencies of precipitation and temperature are calculated based on the data in Svarttjørnbekken gauge station nearby Trondheim. The frequency of snow melting is estimated based on the simulation of HBV model for the same period data. The frequency of major (river) flooding event is equal to its design frequency. However, the frequency of pipe clogging, manhole incidents and total sewer incidents, is estimated based on its operational records.

In term of the independent assumption of events, the probability of flooding for each branch is the multiplication of joint events (“and” events), and total probability of flooding is equal to the sum of probability of all branches (“or” events). This event tree portrays the possible flooding modes in study area. One has to keep in mind, however, that the individual frequencies are uncertain due to various reasons (e.g. non-representative gauge station, incomplete incident records, etc.).

3.4    Consequence analysis and risk description

There are three principal categories consequences regarding to flooding: the potential loss of human life; the potential property damage and corresponding economic losses; and the environmental damages. The dimensions of consequences are evaluated by five groups: insignificant, minor, Moderate, major and catastrophic.

The risks of flooding events are estimated by combining frequency and consequence analysis, see table 3. The descriptive indicators, low, medium and high, are applied.

             Table 3   Estimation of flooding risks

*L-low, M-medium, H-high.

4    DISCUSSION AND FURTHER RESEARCH

Most flooding events in urban drainage systems are obviously triggered by more than one reason. A catastrophic storm can cause flooding immediately, and a sewer obstacle simultaneously will certainly degrade the situation. A poorly planned and maintained drainage system is indeed the main cause of frequent urban flooding. In that case, it may also lead to frequent and severe consequences although rainfall is moderate (Mark, 1997). In addition, unfavorable combination of rainfall and snow melting as well as a high receiving water level can deteriorate a critical situation. Therefore, further research should focus on the analysis of such integrated risks.

A flood can potentially happen at any time. It is a cognitive process to be aware of the flooding risk in a comprehensive way. This process consists of research and analysis of the risk, and operation and management towards an active and defensive direction in order to control the hazard before it releases. Flood risk in river planes is largely depending on natural conditions and long-term planning. This study focuses on flooding in urban drainage systems where operational factors play an important role, too. Here it is often the joint risk of natural and operational conditions that determine the risk of urban flooding.

In our study, we propose a principle approach and demonstrate it for an example, i.e. apply a flood event tree including both qualitative and quantitative analysis. However, the event tree is case-dependent, the frequency estimation is fully depending on the quality and comprehensiveness of the underlying data, and the numerical result are time variable due to the continuous changes in urban drainage systems and their management.

Finally, and not covered in this paper: any risk analysis should always be followed by a plan that focuses on risk reduction and protection measures.

Acknowledgement

The authors are grateful for the interest and support from Department of Water and Waste Water in the Municipality of Trondheim. We especially thank Vidar Kristiansen, Olav Nilsen, Odd Atle Tveit and Svein Husby for their patience and support.

References

[1]    Asbjornslett, Bjorn E. and Rausand, Marvin (1997), Assess the vulnerability of your production system, NTNU Report.

[2]    Blaikie Piers, Cannon Terry etc. (1994), At Risk, Natural Hazards, People’s Vulnerability and Disasters, Routledge. Taylor and Francis Group.

[3]    Kjellen, Urban (2000), Prevention of Accidents through Experience Feedback, Taylor and Francis.

[4]    Mark, O., “Dhaka Flood Studies” (1997).

[5]    Milina, Jadranka and Selseth, Ingrid (1999), Analysis of flooding condition and calculating of recurrence interval for flooding in 1999 in Trondheim, STF 22 A99310.

[6]    Nie, Linmei., Schilling W., Hovden J. and Royset S. (2000), Project Report of Risk Analysis of Urban Flooding, NTNU.

[7]    NSF (1991), Norwegian Standard on Requirements for Risk Analysis, Norwegian Standardization Association.

[8]    TM, Annual reports. Trondheim Municipality, Water and Wastewater Department (1988-1999).

[9]    Wales Stuart G. (1989), Urban Surface Water Management, A Wiley-Interscience Publication.