STORM WATER DRAINAGE MODEL COUPLED WITH FLOOD DEPTH MAPPING: A NEW APPROACH TOWARDS SOLUTION OF URBAN DRAINAGE PROBLEMS Mir Mostafa Kamal* and M Fazle Rabbi** * River Hydraulics Division, Surface Water Modelling Centre, Dhaka-1206, Bangladesh ** Water Resources Division, Surface Water Modelling Centre, Dhaka-1206, Bangladesh ABSTRACT Storm water drainage and its management in many large cities in the world are becoming increasingly problematic. Unplanned or poorly planned urban development along with man made intervention in the natural drainage systems are compounding the fact.
With the advent of very fast computing facilities, sophisticated mathematical modelling tools have come forward to solve urban drainage problems. A study was carried out for the Central part of Dhaka, the capital of Bangladesh, regarding the severe waterlogging problem recurring every time due to even a moderate rainstorm. An urban drainage modelling software, MOUSE of the Danish Hydraulic Institute (DHI) of Denmark was used in conjunction with Geographic Information System (GIS).
Model based waterlogging maps (flooding depth and extent map) produced for September and October 1996 rain showed close matches to the real situation. Graphical two-dimensional animation of waterlogging and its improvement under several alleviation schemes were of practical help in visualising the real situation. The study results suggest that a meaningful solution for urban drainage problems can be obtained with better confidence of optimisation using the coupled use of an urban drainage model and flood depth mapping.
KEYWORDS Storm water; Dhaka City; Mathematical Model; Sustainable Management INTRODUCTION Background Storm water drainage and its management in many large cities in the world are becoming increasingly problematic. Unplanned or poorly planned urban development along with man made intervention in the natural drainage systems are compounding the fact. With the advent of very fast computing facilities through 4th generation computer, sophisticated mathematical modelling tools have come forward to solve the urban drainage problems.
However, almost all relevant modelling software are designed for planning and designing of sewerage system for a design storm. It is hardly possible to foresee the impact of instant improved drainage facilities along with future urban development on the urban flooding and waterlogging problems. Dhaka is the capital of Bangladesh and nerve centre of its socioeconomic and political activities of the country. Its present population is about 9. 6 million and has a population growth of nearly 4% per annum. Dhaka is going to be one of the ‘Mega Cities’ in the world by the year 2000.
Therefore, urban development has been inevitable to cope up with the increasing thrust of the huge population. However, the urban development did not take place as systematically as it should be. Unplanned or poorly planned urbanisation comprising abolishment or constriction of natural stormwater carrying channels, filling up of low-lying areas and natural ditches – which were used to act as reservoirs, etc. – are the major causes for the stormwater drainage congestion in and around the Dhaka City. Moreover, stormwater drainage facilities did not improve significantly to accommodate the ‘too apid growth’ of the City. As a result, the City has been experiencing worsening waterlogging problem for the last decade. The problem of stormwater drainage was highlighted in the months of September and October 1996 when the City dwellers faced the worst ever waterlogging in Dhaka City. Human activities were almost suspended for few days due to the unusual waterlogging problem. The Central part of Dhaka City was the most suffered area. Case Study A study was carried out for the Central part of the Dhaka City regarding the severe waterlogging problem (SWMC, 1997).
An urban drainage modelling software was used in conjunction with Geographic Information System (GIS) Model based waterlogging maps (flooding depth and extent map) produced for September and October 1996 rain showed close matches to the real situation. Several alleviation schemes were tested to see improvements in waterlogging in each of the scheme. Graphical two-dimensional animation of waterlogging and its improvement under different alleviation schemes were of practical help in visualising the real situation.
Objectives The principal objectives of the study comprised an understanding of many elements of a complex drainage system of the Dhaka City, and development of a dynamic mathematical modelling system to investigate causes and countermeasures for the recurrent waterlogging problem in the City. ABOUT THE DHAKA CITY DRAINAGE SYSTEM Profile of the Study Area The Dhaka City is bounded by four rivers: Balu on the east; Tongi Khal on the north; Turag on the west; and Turag-Buriganga on the south. The drainage of the City areas is mostly depended on the water levels of the peripheral rivers.
The major drainage channels (locally known as Khal) in the City are Dholai khal, Gerani khal, Segunbagicha khal and Begunbari khal, which collects catchment runoff and drains to the peripheral rivers. Figure 1 shows the rivers and major drainage channels in and around the Dhaka City. The drainage network of the Dhaka City is very complex. For this study, a small but important area was selected for the study. The study area includes Motijheel commercial area, the nerve centre for Bangladesh’s economy, and the Government Secretariat.
The Segunbagicha khal, the main drainage channel of the area carries catchment runoff to the Balu river. The khal after originating from a park travels a distance of 3. 4 km to cross a road through a sluice gate and then meets with another drainage khal before draining finally into the Balu river. The khal covers a drainage area of 4. 54 km2 upto the sluice gate. The Segunbagicha khal was originally a natural channel. A rectangular-shaped concrete conduit has replaced a portion of it (first 2. 1 km). Figure 2 shows the study area.
Description of the Drainage System Components of the Drainage System. Various components of the Dhaka City drainage system comprised of primary, secondary and tertiary (or road-side) drains – open or closed; flood protection schemes including sluice gates and pumping station. A typical layout plan of secondary drainage system is shown below: Manhole Underground Sewer Pipe Catchpit Walkway Connection between Catchpit and Manhole Road Divider Surface Drain Road Width Figure 1: Rivers and Major Drainage Channels in and around Dhaka City Features of the Presenting Problem.
Flood inundation or waterlogging in the Dhaka City can be classified into two types. One results from high water levels of the peripheral rivers, the other is caused by runoff from storm rainfall in the City area. During the monsoon from May to October, the drainage of the Dhaka City is mostly dependent on the water levels of its peripheral river systems. During this period, river water levels generally remain higher than the internal drainage level. A wide range of rainfall intensities that prevail during the monsoon period severely restricts an effective surface drainage within the City area.
The situation worsens when System consists of underground pipes, circular & rectangular, and open channels. Model area 454 ha Circular Rectangular Sluice Gate Open Figure 2: Study Area monsoon runoff generated from short duration and high intensity rainfall combines with high water level in the river systems. Flooding due to rivers normally takes place in the low-lying fringe areas once in every five to ten years. Waterlogging caused by local rainfall occurs in the built-up areas of the City several times a year.
This is mainly the result of inadequate existing drainage facilities and their improper operation and maintenance. The Dhaka drainage system serves less than 25% of the urban area. The remaining area drains through overland flow into adjacent watercourses or into low areas, also causing flooding. Water then passes away from the low areas either by infiltration or evaporation. Dhaka Water Supply and Sewerage Authority (DWASA) are the responsible agency to develop and manage the drainage systems for the Dhaka City. DWASA has identified the following deficiencies in the system (Mott MacDonald, 1993): • • • • • unplanned urbanisation, expansion of the urban areas, increases in built-up areas and metal roads; filling of low-lying areas to construct buildings, with no or little provision for drainage; blockage of the main drainage systems of the urban area (khals) by unauthorised constructions; insufficient storm sewers constructed in the extensions to the urban area; lack of maintenance to the system; lack of coordination among the different organisations engaged in the development works; • solid waste disposal in the storm sewer. METHODOLOGY OF MATHEMATICAL MODELLING
Modelling Software Storm Water Drainage Modelling Using MOUSE. The drainage system of Dhaka City consists of a combination of closed conduits and open channels. MOUSE, a package for modelling of urban drainage and sewer system developed by Danish Hydraulic Institute (DHI), has been selected for the study. It is an integrated modelling system that can simulate both surface runoff from rainfall and unsteady pipe and open channel flows (DHI, 1995). In MOUSE, surface runoff hydrographs are computed in the Hydrological Module for selected rainfalls for all subcatchments of a system.
Then in the Hydraulic Module the runoff result file containing runoff hydrographs from each subcatchment is used as input for the pipe flow simulation. Each subcatchment runoff hydrograph is attached to node(s) located within the subcatchment of the drainage system. Catchment parameters needed for each subcatchment comprise total subcatchment area, degree of impervious area, time of concentration, tme-area curve and name of node in the pipe model to which the surface runoff is directed. The description of the hydraulics of the system, i. e. lows, water levels, velocities are computed based on the fully dynamic pipe flow model, solving the full St Venant’s differential equations. The Hydraulic Module consists of components like flow over the streets, storage of water on the surface and flow through the pipe system The nodes in the model include both networks of the underground pipe and hydraulic description of the street system. The description of the street system can be divided in three parts: • • • The main streets where overland flow is assumed to occur, the hydraulic description is based on a fully ynamic approach; The local street inside the subcatchment, which were assumed to provide surface storage only without overland flow; The connections between street system and pipe system, taking into account the flow restrictions defined by the catchpits. GIS Modelling. Inundation maps are the easiest and most effective media for visualising flooding. MOUSE produces output in the form of graphics and tables. To produce inundation maps under different scenarios, MOUSE results need to be analysed spatially using GIS. ARC/INFO, one of the world’s most established and well-known GIS package, has been used in the study.
Dynamic Waterlogging Maps. Two separate networks, one for underground pipes and the other for street networks, have been modelled. Mainly water levels on the streets cause flooding on the streets and the adjoining areas. The model simulates water levels at discrete locations along the streets. The simulated water level time series from MOUSE was exported to ARC/INFO to generate 3D water surfaces. In the absence of detailed topographic information, an indicative Digital Elevation Model (DEM) was developed for the model area and compared with the model-based water surfaces to produce waterlogging maps (flooding depth and extent map).
City Street Flooding Mechanism Urban street flooding may cause from numerous reasons. Rainfall starts as overland flow on the street before entering underground pipe system through catchpits. The duration of flooding on the street depends on its elevation and the intake capacity of the catchpits and underground pipes. If the intake capacity of the catchpits are limited, large volumes of runoff will be transported along the surface during high rainstorms even the underground sewer network may have sufficient capacity.
In contrast, insufficient drainage capacity of the sewer pipes may also cause surface flooding even if the catchpit intake capacity is adequate. Both of these phenomena are of interest to the study of the Dhaka City drainage system and accordingly, the adopted methodology is described in the following flow chart. Hydrological Model Rainfall Catchment Area Impervious Area Runoff Hydrograph Street Model Road/Street Road/Street Storage & Routing Water Level Discharge Inundation Map Pipe Flow Model Catchpit Manhole Catchpit Manhole Storm Sewer Pipe Sewer Flow Routing
Water Level Discharge DATA USED FOR MODELLING Rainfall Data Hourly Rainfall data for the 1996 monsoon (till October) have been used for model simulations. An IntensityDuration-Fequency (I-D-F) curve for Dhaka City developed by the JICA (JICA, 1987) was employed to generate synthetic rainstorms using the Alternating Block Method (Chow et al, 1988). The rainstorms having a statistical return period of 2 and 5 years with a total duration of two hours have been used. Hydrometric Data Currently water levels and discharges for sewer network system are not monitored by DWASA.
Land Elevation and GIS Data At present no land elevation data is available for the built-up areas of the Dhaka City. A survey was conducted to measure the elevations of top of the manholes. Usually, tops of the manholes are flushed with the road level. The roads and drainage networks were digitised from maps (1:20,000 scale). JICA in 1987 prepared a 1m contour for the built-up areas of the Dhaka City. Lacking detailed land level information, these were combined with the surveyed levels of the manhole tops and an indicative Digital Elevation Model (DEM) was developed.
MODEL SETUP AND CALIBRATION Set up of the Model The level of discretization of an urban drainage system depends on pipe networks (primary, secondary and tertiary) and nodes (manholes, structures etc. ) which are to be described in the model. The selection usually depends on the objective of the study, geometrical criteria for the network of the system and hydrological criteria. Considering availability of data, time and resources along with the objectives of the study, the modelling area was schematised as shown below. A four-vent sluice gate is located at a road crossing, which is also described in the model.
Although no Ramna m Pipe Length = 424Park Pipe Size = 1350 mm to 1750 mm Catchment area = 55 ha Shantinagar Pipe Length = 1014 m Pipe Size = 600 mm to 1200 mm Catchment area = 60 ha Shantinagar-DIT road Pipe Length = 2246 m Pipe Size = 750 mm to 1700 mm Catchment area = 54 ha Inner Circular Road Pipe Length = 1498 m Pipe Size = 600 mm to 1500 mm Catchment area = 41 ha Outer Circular Road Pipe Length = 1473 m Pipe Size =750 mm to 1700 mm Catchment area = 42 ha Box Culvert Sluice Gate Highcourt area Pipe Length = 461 m Pipe Size = 600 mm to 1350 mm Catchment area = 13 ha
Secretariate Pipe Length = 854 m Pipe Size = 600 mm to 1500 mm Catchment area = 48 ha Stadium Area Pipe Length = 468 m Pipe Size = 600 mm to 900 mm Catchment area = 26 ha Matijheel Area Pipe Length = 1019 m Pipe Size = 600 mm to 1220 mm Catchment area = 42 ha Banga Bhaban-Gopibag Pipe Length = 954 m Pipe Size = 750 mm to 1500 mm Catchment area = 73 ha permanent pumping station exists, provision has been kept to pump out water. The pumps were operating during the waterlogging problem of September and October 1996. A pump of 5 m3/s capacity was included in the model setup.
Calibration of the Model Model calibration in general involves judicious selection of parameters to reproduce the response of a physical system within the range of desired accuracy. In practice it is often done by trial-and-error adjustments of parameters. As there is no measured water levels and discharges available for the drainage system being studied, no calibration could be possible. However, outputs of MOUSE were analysed spatially using GIS to produce waterlogging maps for the model area under different hydrological conditions. The maps were then compared with the physical conditions existed during September and October 1996.
Reasonable agreement could be found between the simulated waterlogging maps and real situations. A sample waterlogging map for October 6, 1996 is shown here (Shaded area indicates waterlogging). In order to assess the uncertainty in the estimation of the model parameters, a sensitivity analysis was carried out. Surface runoff model was run for three different time of concentrations (TOC): 10, 20, and 40 min. The TOC of 20 min gives velocities in the range of 0. 2-0. 3 m/s under 5-year 2-hour rainstorm event and selected for all subcatchments.
MODEL SIMULATIONS FOR ALLEVIATION MEASURES Simulation under Present Drainage Condition Two important deficiencies were observed in the drainage system. The underground pipes were clogged by 6070 percent and replacement of Segunbagicha khal by concrete channel was incomplete at 5 locations. In an effort to find a solution to the existing problems, two potential alleviation scenarios were tested using the model. One was to use pumps temporarily during critical periods in problematic areas and the other was to implement necessary excavation for certain portion of the Segunbagicha khal.
Model results illustrate that the impact of pumping is mostly local. Use of pumps at Shantinagar crossing would reduce the duration of flooding from 24 hours to 7 hours at or near Shantinagar under the October 1996 hydrological condition. However, depth of flooding would hardly be affected. Excavation of the unfinished locations of the box culvert would reduce the duration of flooding at Shantinagar area from 24 hours to 18 hours and maximum depth of flooding from 75 cm to 65 cm under the October 1996 hydrological condition.
Comparing the results, it is evident that without proper maintenance of the drainage pipes no major improvement of the present waterlogging problem can be expected. Simulation under Design Condition The model was also applied to see the effectiveness of the full implementation of an ongoing drainage improvement project (JICA, 1990). The project includes rehabilitation of main, secondary and tertiary khals and pipe drain, construction of new sewer lines and construction of pump stations at the outlet of some major drainage khals.
Model results indicated that the severity of the existing problem would be reduced to a greater extent under design condition. However, some problems for parts of the model area would not be resolved. The Shantinagar crossing, the most suffered area, might require further attention to improve the drainage congestion. This might be due to the fact that the drainage capacity of the existing sewer lines from the Shantinagar to box culvert via Bijoynagar road were not adequate. Absence of any major lakes or retention storage in the study area could be other reasons.
Model results showed that provision of pumping facility at or near Shantinagar area could help reduce the duration of flooding in the area. CONCLUSIONS AND RECOMMENDATIONS Finding a solution to the present waterlogging problem and future management of the drainage systems of the Dhaka City, the study was just a beginning. The results are encouraging and indicated that the new approach of modelling used in the study would be effective for sustainable management for the drainage system. Lacking detailed key information, in most of the cases experience and judgement of relevant engineers were applied to set up the model.
Proper calibration and verification of the model could not be possible due to the non-availability of relevant data. In some cases, calibration parameter such as impervious area of a catchment had to be estimated. Hence, the model results are primarily qualitative and need to be considered as indicative only. Application of mathematical modelling at a detailed level would require a planned, organised and systematic approach. The present status of data availability is inadequate for such a detailed level of modelling. Establishment of a digital database and establishing a data monitoring networks for short- and long-term is recommended.
It is estimated that a number of 6 to 10 water level gauging stations in the concrete channel and 15 to 20 flow sensors at secondary underground pipes should be adequate for short-term monitoring. Few more rain gauges should also be installed in the model area to record rainfall at 5 minutes interval. The results will help establishing a methodology for the entire Dhaka City. REFERENCES Chow V. T, Maidment D. R. and Mays L. W. (1988). Applied Hydrology, International Edition, McGraw-Hill Book Company, New York. DHI (1995). Technical Reference Manual of MOUSE, Danish Hydraulic Institute, Hoersholm, DK-2970, Denmark.
JICA (1987). Study on Storm Water Drainage System Improvement Project in Dhaka City. Japan International Cooperation Agency, Dhaka Water Supply and Sewerage Authority, Dhaka, Bangladesh. JICA (1990). Updating Study on Storm Water Drainage System Improvement Project in Dhaka City, Japan International Cooperation Agency, Dhaka Water Supply and Sewerage Authority, Dhaka, Bangladesh. Mot MacDonald (1993). Tertiary Drainage in the RAJUK Area, Dhaka Metropolitan Development Programme, Dhaka, Bangladesh. SWMC (1997). Storm Water Drainage Modelling for Dhaka City: A Pilot Study. Surface Water Modelling Centre, Dhaka-1206, Bangladesh.