You are on page 1of 7

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008

Integrated urban water systems modelling with a simplified


surrogate modular approach
Z. Vojinovic1*, and S.D. Seyoum2
Department of Hydroinformatics and Knowledge Management, UNESCO-IHE, Institute for Water Education,
Westvest 7, 2611 AX Delft, The Netherlands
*Corresponding author, e-mail: z.vojinovic@unesco-ihe.org

ABSTRACT
In a complex urban environment, modelling tools are needed to describe the complex waterrelated interactions, and to allow management strategies to be developed. Typically, two
types of models are used: simplified (or strategic) and detailed ones. Simplified models are
normally used for strategic planning purposes, whereas, detailed models are needed to
describe the systems performance according to the specific local needs and objectives. This
paper addresses the issue of setting up conceptual modelling platform to replicate the larger
part of urban water cycle in order to evaluate different strategies and scenarios for integrated
urban water management at the planning level. It explores the use of integrated simplified
models in a procedure where the model parameters of these models are derived either from
the measurements or from a detailed physically-based model results. The paper presents an
ongoing research work undertaken within the work package 1.2 of the EU funded SWITCH
project (further information about this project can be found at: www.switchurbanwater.eu).
Apart from the discussion of concepts of simplified integrated modelling, this paper also
presents some of the preliminary results obtained form a case study where an existing
simplified modeling system (CITY DRAIN) is used as a starting point to develop a more
comprehensive tool.

KEYWORDS
Urban water systems; sewer systems; receiving waters; integrated modeling;

INTRODUCTION
Integrated urban water management has emerged as an important concept for several reasons.
First, there is the growing need to manage the urban water cycle on a global basis. Second, a
range of alternative technologies to process different aspects of the urban water cycle are
becoming available. Third, advances in urban hydroinformatics enable different phases of
the entire cycle to be modeled and to use such models to optimise each phase locally and in
the global context. In particular, the advances in urban hydroinformatics have made
significant impacts on the development of new strategies for urban water management. The
use of computer models pervades all aspects of water management, supporting wealth
creation through products and services, contributing to many improvements in the quality of
life. As a result, there is a growing increase in demands for better use, productivity, flexibility,
robustness and quality of urban water modelling systems. In the urban water sector, no
serious investment decisions are being made without the use of computer models to evaluate
various scenarios. In this context, computer modelling of the urban water cycle is aimed at
understanding and predicting the behaviour and performance of the component and integrated
systems so that the effective solutions to structural and operational problems can be derived
and evaluated within a decision-making framework.
The paper presents some of the concepts for setting up the simplified integrated modelling
framework for integrated urban water systems modelling in order to evaluate different
Vojinovic and Seyoum

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008


strategies and scenarios for integrated urban water management at the planning level. A brief
review of sewer component of a simplified modeling approach embedded within the CITY
DRAIN model is given followed by a case study where the main objective was to evaluate its
capabilities to be used as a platform for simplified integrated modeling purposes.

SIMPLIFIED INTEGRATED MODELLING


The need for integrated analysis has been discussed in literature (see for example, Rauch et al.
1998; 2002) and it has also been formulated within the EU Water Framework Directive
60/2000 with a water-quality orientated view on the whole system which calls for new ways
of assessing its performance.
Models can be employed to meet many different objectives during the planning, design and
operation phase of urban water systems, and thus different types of models are needed to
correspond to different needs. Mathematical models in general can range from simple
equations to complex software codes including many equations and conditions over the time
and spatial domain.
Generally speaking, mathematical models used in urban water application are aimed at
facilitating various aspects such as planning, operations and design. Certainly, depending on
the purpose and objectives of the work different modeling approaches can be applied.
Typically, these approaches differ with respect to the level of complexity and sophistication
of equations used to describe the governing processes. The main difference in applying
different modelling approaches is the amount of data required, the information that can be
obtained from the model, the sophistication of the analysis performed and the simulation
period.
Although the basic principles are known, the development of integrated models is still a
challenging task. The main bottleneck is the complexity of the total system that prevents a simple
linkage of the existing physically based models of the individual subsystems to an entity (Rauch et
al, 2001). Another problem encountered when creating an integrated model is the fact that the
existing models are quite complex and require sophisticated integration algorithms to solve them.
This results in long calculation times, making these models impractical to use, especially within
optimisation problems where a lot of simulations need to be performed. Model integration is
also a challenging task due to the reasons related to data formats, compatibility of scales,
ability to modify source codes, etc. To overcome such difficulties, there are attempts by
commercial software companies to develop the links between different detailed physicallybased models (Moore et al, 2004) to enable development of integrated modeling platforms.
However, efforts in instantiating such models and their linking and running, which usually
takes a substantially long period of time, is still impractical for many applications especially
for those applications concerning strategic planning purposes where not only integration
between different components of urban water cycle is needed but also simulation of different
scenarios and optimization is necessary too.
What is therefore needed is the ability to undertake a holistic analysis of the urban water
cycle by setting up relatively simple models with reasonable accuracy. Due to the limitations,
which make detailed physically-based models based on the conservation of mass and
momentum, inefficient and impractical for integrated modelling for strategic planning
purposes, there is a need to replace such models by fast surrogate (an approximate substitute)
models and to link them up within the same platform. Implicit in this is a requirement both to
2

Integrated urban water systems modelling with a simplified surrogate modular approach

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008

Recipient

Groundwater

understand and to be able to model not only the individual urban water processes but also
their interactions in a relatively simplified modeling framework (Figure 1).

WWTP
Reservoir

Sewer
System

Drinking water
distribution

Water
treatment

Treatment

Overflow

Figure 1. Schematisation of the urban water cycle and interactions between different
components.
Several researchers have attempted to respond to the challenge of developing simplified
integrate modeling. For example, the wastewater side of urban water cycle, which involves
interaction between the sewer system, treatment plant and recipients has been described by
integrated urban wastewater modelling tools such as KOSIM-WEST (Solvi et al, 2005),
SIMBA (Freni et al, 2003), Cosmos (Calabro, 2001) and CITY DRAIN (Achleitner et. al.
2007). Such models use simplified conceptual hydrologic flow routing methods (for example,
in the case of CITY DRAIN Muskingum flow routing model was used) which might be
found more applicable for the routing of free surface flows and less applicable to the routing
of surcharged pipe network flows. Zoppou (2001) has emphasized the importance of
estimating urban flows accurately for integrated modelling for the reasons that pollutant
concentrations and loads cannot be estimated without having estimated the flows and also
procedures to mitigate quantity and quality are often complementary.
It is apparent from literature review of modelling tools for integrated urban wastewater
systems that, most of the modeling tools employed conceptual models to estimate the urban
flows due to the advantage they possess from a computational intensity point of view.
However, due to their conceptual nature, these models cannot provide accurate results
compared to the physically based models especially in conditions of surcharge flows, surface
flooding and significant backwater effect. Since the level of accuracy of the urban stormwater
modelling is fundamental to the overall accuracy of integrated urban water systems modelling,
for the development of simplified integrated urban water system model the emphasis should
be given to the ability of simulating urban drainage component as accurately as possible.
The conceptual model represents the hydrological processes that are seemingly important in
the system using a simplified, conceptual representation. To use these models for estimating
urban water flows, it is necessary to estimate the model parameters relevant to these systems.
The model parameters for gauged drainage systems are generally estimated by undertaking
either trial-and-error or an automatic calibration process. However, this is not possible for
ungauged systems due to the absence of rainfall and runoff data. If a form of equations which
can correlate model parameters to the drainage system characteristics can be derived, such
equations could be then used to estimate model parameters for ungauged drainage systems.
An attempt to develop such equations would be to derive a large number of model parameters
that correspond to a large number of gauged catchments via calibration process and then to
Vojinovic and Seyoum

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008


derive a mapping function which can provide the parameter values for a given combination of
catchment characteristics. In some other cases where limited data (or measurements) exist an
approach would be to develop a physically-based model, calibrated it and generate the time
series data that could be used for derivation of parameters of a simplified model, Figure 2.
Similar methodology was applied by Meirlaen et al (2001) and it can be seen as a process of
transferring the knowledge contained within the physically based model into the conceptual
model by means of such virtual data.
The requirement for a strategic model is that it needs to be less complex and computationally
less expensive when compared to the physically-based models but still be able to produce
reasonably accurate outputs (or replicate the outputs of the deterministic models) for strategic
planning purposes. Setting up a simplified integrated urban water modelling framework can
be accomplished by the flowing steps: developing a conceptual model for each component of
urban water cycle, calibration based on observed data (or on virtual data generated by the
physically-based model) and integration between the simplified models of each component.

CONCEPTUAL SEWER MODEL


Currently, a research has been being undertaken to develop a simplified modelling
framework with associated tools for strategic planning purpose under the EU funded
SWITCH project. In this research, the use of an open source model such as CITY DRAIN has
undertaken and some efforts have been made in testing the capability of such model to
replicate the behaviour of a physically based model. For this purpose the sewer component of
CITY DRAIN model is used to model urban sewer systems and results are compared to the
results of physically based model (in this case, the MOUSE modeling system, which is a
product developed by DHI Water & Environment was used). The intention here is to evaluate
strengths and weaknesses of a simplified modellling approach (such as the one embedded
within the equations of CITY DRAIN model) and to come up with plans for future
improvements in order to make such tool more useful for strategic planning purposes.

Reality

Existing
Catchment and
Sewer system

Existing
Wastewater
Treatment Plant

Existing River
(receiving water)

Complex WWTP
Models

Physically Based
River Models

Simplified
Conceptual
WWTP Model

Simplified
Conceptual
Model

Data Collection

Physically Based
Models

Physically Based
Catchment and
sewer model

Data Generation

Simplified
Conceptual
Models

Conceptual
Catchment and
sewer model

Integrated Wastewater System


Figure 1. From Reality to Simplified Conceptual Models (Adopted from Meirlaen et al,
2001).
4

Integrated urban water systems modelling with a simplified surrogate modular approach

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008

CITY DRAIN is open source software developed for the purpose of integrated modelling of
urban drainage systems, Achleitner (2006). The model was developed within the
Matlab/Simulink environment, enabling a block wise modelling of different parts of an
urban drainage system. The open source structure of the software allows for potential
addition of blocks and/or modification of existing blocks (and underlying models) according
to specific needs (Achleitner et al, 2007).
The computation in CITY DRAIN is based on a fixed discrete time step approach where each
subsystem uses the same time increment, usually predetermined by the temporal resolution of
the rainfall data used. Models implemented for hydraulics and mass transport are formulated
for a discrete time step t. A simple loss model is used to account for losses. Volume of
water that can be retained due to initial loss is represented by a basin. The volume of rain
exceeding the basins volume is considered to be the effective precipitation, contributing to
the catchment surface flow. Permanent losses such as evapotranspiration can be either
considered acting all the time or during dry weather only. The volume per time step to be
evaporated is limited by the initial loss specified. The Muskingum method is used for flow
routing in the catchment, sewer and river blocks. The system (catchment, channels and
conduits) is considered as a whole. The model has separate blocks for handling the combined
and separate sewer systems. The model also has blocks for combined sewer overflow
structures and pumping stations. The hydraulic calculations for these structures are based on
the continuity equation and on maximum flow capacities. From this modeling approach, it
follows that pressurized flow, surface flooding and backwater effects cannot be modelled by
CITY DRAIN. It is our intention to improve the existing version of CITY DRAIN model to
account for such effects in a simplified way.

CASE STUDY
The sewer component of CITY DRAIN model was used to model urban catchments with
different characteristics. The model is manually calibrated and verified using data generated
by the MOUSE model. The results from a small urban catchment in Japan which covers an
area of about 35.43 hectares with population of 9004 inhabitants and with pipes diameter
ranging from 0.6m to 1.65m is shown here. This catchment has been subject to free surface
flow and surface flooding conditions. The calibration and verification for event based
simulation produced good fit to the data for free flow conditions; however, as expected the
model failed to produce meaningful results for surface flooding condition (Figure 3 and Table
1). For surface flooding condition, a retention basin was introduced within the CITY DRAIN
model in order to retain the flow, which exceeds the pipe full capacity of the outfall pipe.
Such alteration had improved the calibration and verification results. Table 1 shows the
values of coefficient of determination R2 and the Nash-Sutcliffe efficiency E.
Table 1. Coefficient of determination R2 and the Nash-Sutcliffe efficiency E for CITY
DRAIN model results.

Calibration
Verification

Free flow condition


Efficiency
R2
0.99
0.99
0.88
0.87

Vojinovic and Seyoum

Surface flow condition


R2
Efficiency
0.95
0.95
0.90
0.89

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008

Event 2

2
Mouse

1.8

CD

1.6

MOUSE
CD

Discharge [m 3/s]

1.6
D is c h a rg e [ m 3 /s ]

Verification
2

1.2
0.8

1.4
1.2
1
0.8
0.6

0.4

0.4
0.2

0
19, 2:38

19, 3:50

19, 5:02

19, 6:14

19, 7:26

19, 8:38

19, 9:50

0
14,12:00 14,13:12 14,14:24 14,15:36 14,16:48 14,18:00 14,19:12 14,20:24 14,21:36 14,22:48 15,00:00
Time [day, hr:minute]

Time [day, hour:minute]

Calibration

Verification

Event 2
14

Mouse

Discharge [m3/s]

12

CD

10
8
6
4
2
0
19, 4:48

19, 6:00

19, 7:12

19, 8:24

19, 9:36

Time [day, hour:minute]

Calibration when there is surface flooding (with no retention basin provided)

Calibration (with retention basin included)


Verification
Figure 3. Calibration and verification results for free flow and surface flooding conditions.

DISCUSSION
CITY DRAIN model provides several advantages and disadvantage compared to the
physically-based models. The significant advantages are that it is easy to build and set up, its
data requirements are relatively less significant and its computational time is significantly
lower. The modular approach enables to represent a catchment in a semi-distributed way.
The parameters of the sewer component include the initial loss (mm), the permanent loss
(mm/day), the number of time steps and the Muskingum parameters K(s) and X. In the
calibration process overlapping effect of the parameters K and the number of steps has been
observed. As the number of subcatchments, combined overflow structures and pumping
6

Integrated urban water systems modelling with a simplified surrogate modular approach

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008


stations increased, number of calibration parameters increase by many fold which makes the
process of manual calibration difficult.
CITY DRAIN model performed reasonably well for free flow conditions (with and without
surcharge) and in catchments with moderate to high slope. For surface flooding and surcharge
conditions and in flat catchments the model is not expected to produce reliable results owing
to its conceptual nature. The retention block introduced appeared to be a relatively effective
measure in order to replicate surcharge flow conditions.

CONCLUSION
For the development of a simplified integrated modelling platform an existing tool CITY
DRAIN has been used as a starting point. For improvements of this tool, the emphasis has
been given to the development of algorithmic procedures which can better describe the
behavior of floods and overflow surcharges. The preliminary results suggest the potential of
such modeling approach to be used as a tool for strategic planning purposes. Issues for further
improvement of this model will include provision of an automated procedure for optimal
parameter estimation based on evolutionary algorithms, uncertainty assessment and provision
of methodology for estimation of model parameters for case of ungauged catchments. Also,
further improvements will be also made towards better and more convenient graphical user
interface.

REFERENCES
Achleitner, S., Mderl, M. and Rauch, W., (2007), CITY DRAIN - An open source
approach for simulation of integrated urban drainage systems, Environmental Modelling
& Software, Volume 22, Issue 8, August 2007, 1184-1195.
Calabro P.S. (2001) Cosmoss: conceptual simplified model for sewer system simulation. A
new model for urban runoff quality. Urban Water, 3, 3342
Freni G, Milina J, Maglionico M, DiFederico V (2003) State for the art in Urban Drainage
Modelling, CARE-S (EU project CARE-S Computer Aided REhabilitation of Sewer
networks: Project Number EVK1-CT-2001-00167), report D7, 2003
Meirlaen J., Huyghebaert B., Sforzi, F. Benedetti L. and Vanrolleghem P., (2001), Fast,
simultaneous simulation of the integrated urban wastewater system using mechanistic
surrogate models, Water Science and Technology, Vol 43 No 7 pp 301309
Moore R, Tindall and Fortune D, 2004, Update on the HamonIT project: The OpenMI
standard for model linking, Proceedings of the 6th International conference on
hydroinformatics, World scientific Publishing Company.
Rauch W., Bertrand-Krajewski J. L., Krebs P., Mark O., Schilling W., Schtze M. and
Vanrolleghem P. A., (2002), Deterministic modelling of integrated urban drainage
systems. Water Science and Technology, 45 (3),81-94.
Rauch, W., Aalderink, H., Krebs, P., Schilling, W., & Vanrolleghem, P., (1998),
Requirements for integrated wastewater models driven by receiving water objectives,
Water Science Technology, 38(11), 97104.
Solvi A.-M., Benedetti L., Gill S., Schosseler P., Weidenhaupt A. and Vanrolleghem
P.A.(2005), Integrated urban catchment modelling for a sewer-treatment-river system,
10th International Conference on Urban Drainage, Copenhagen/Denmark, 21-26 August
2005
Zoppou, C. (2001). Review of urban storm water models. Environmental modelling and
Software, 16(3), 195-231

Vojinovic and Seyoum