nio final report

48
Project Report on Hydrodynamics and pollutant transport off Mumbai coastal region submitted by Prabhas Ranjan Gupta Indian Institute of Technology Guwahati under the guidance of Dr. P. Vethamony Scientist-G Physical Oceanography Division National Institute of Oceanography, Goa India 28 th May to 28 th July 2010

Upload: prabhas-gupta

Post on 06-May-2015

164 views

Category:

Engineering


0 download

TRANSCRIPT

Page 1: NIO final report

Project Report

on

Hydrodynamics and pollutant transport off Mumbai

coastal region

submitted by

Prabhas Ranjan Gupta

Indian Institute of Technology Guwahati

under the guidance of

Dr. P. Vethamony

Scientist-G

Physical Oceanography Division

National Institute of Oceanography, Goa

India

28th May to 28th July 2010

Page 2: NIO final report

1

Acknowledgements

I would first like to thank Dr. P. Vethamony for giving me an opportunity to learn and work under his

expert guidance. I am grateful to him for giving me much of his precious time and patiently

discussing the progress of the project at every stage. This small but important work would have

been impossible but for his continuous support, guidance and encouragement.

I would also like to express my sincere thanks to Dr. S.R. Shetye, Director, NIO, for providing me all

the facilities needed for the successful completion of this project work.

Special thanks are due to Mr. Vinodkumar for helping me learn the specifics of MIKE 21. I would like

to thank Mr. V.M. Aboobacker for carefully revising the draft of this report and giving me his time

and insightful feedbacks during the project. I am also grateful to Mr. V.R. Renjith, Mr. K. Sudheesh,

Ms. Betty John and others in the POD lab for extending their generous help whenever I was in

need. Lastly, I would like to give a special thanks to Mr. Saheed P.P. for giving me his valuable

advice and inputs from day one and for maintaining a cheerful atmosphere in the lab.

Page 3: NIO final report

2

Contents

Page

Acknowledgements 1

List of figures 4

List of tables 5

List of abbreviations 6

List of symbols 8

Chapter 1 Introduction 9

1.1 Coastal waters and marine outfalls 9

1.2 Study area: Mumbai coastal region 10

1.3 Review of water quality modelling studies 12

1.4 Objectives of the present study 13

Chapter 2 Simulation of off-shore hydrodynamics 14

2.1 Hydrodynamical modeling 14

2.1 MIKE 21 flow model 15

2.3 Model setup 16

2.3.1 Model domain and grid size 16

2.3.2 Time step and length of simulation 16

2.3.3 Bathymetry 17

2.3.4 Bed roughness coefficients 18

2.3.5 Momentum dispersion coefficients 19

2.3.6 Wind data and wind friction factor 20

2.3.7 Initial and boundary conditions 22

2.4 Model calibration, validation and simulation 23

2.4.1 Model calibration 23

2.4.2 Model validation 23

Page 4: NIO final report

3

2.4.3 Final simulations 25

Chapter 3 BOD-DO modeling of wastewater outfalls 26

3.1 The importance of dissolved oxygen 26

3.2 Effluent dispersal processes 26

3.3 Processes affecting DO levels 29

3.3.1 Reaeration 29

3.3.2 Photosynthesis and respiration 29

3.3.3 BOD exertion 30

3.3.4 SOD exertion 31

3.4 Model setup 31

3.4.1 ECO Lab 31

3.4.2 Wastewater disposal scenario and outfall parameter values 32

3.4.3 Description of chosen template and process rate estimates 33

3.5 Final simulations 35

Chapter 4 Results and discussion 37

4.1 Southwest monsoon season 37

4.2 Northeast monsoon season 40

Chapter 5 Summary and conclusion 43

References 44

Page 5: NIO final report

4

List of figures

Title Page

Fig. 1

A marine outfall: shown above is the Boston outfall, which discharges into

Massachusetts bay

9

Fig. 2 Bandra and Worli outfalls and Mumbaiโ€Ÿs drainage zones 10

Fig. 3 DO levels along Mumbaiโ€Ÿs west coast 11

Fig. 4 Interpolated bathymetry for the model domain 18

Fig. 5 Wind rose diagram for Worli during 1 Aug to 31 Aug, 2009 (SW monsoon period) 21

Fig. 6 Wind rose diagram for Worli during 22 Oct to 22 Nov, 2009 (NE monsoon period) 21

Fig. 7 Plot of initial surface elevation 22

Fig. 8 Comparison between measured and simulated water levels at Worli 24

Fig. 9 Comparison between measured and simulated u-component of currents off Worli 24

Fig. 10 Comparison between measured and simulated v-component of currents off Worli 24

Fig. 11 BOD exertion curve (schematic) 30

Fig. 12 Variation of (a) BOD and (b) DO at the outfall location during the SW monsoon 37

Fig. 13 BOD distribution after 72 hours of discharge for (a) case 1 (b) case 2 39

Fig. 14 Relative magnitudes of processes 39

Fig. 15 Tidal variations in plume pattern: (a) flood tide (b) ebb tide 40

Fig. 16 Comparison between (a) BOD and (b) DO levels at the outfall during the two

study periods, viz. SW monsoon (Aug) and NE monsoon (Oct-Nov) seasons 41

Page 6: NIO final report

5

List of tables

Title Page

Table 1 Basic and hydrodynamical parameter values used for final simulations 25

Table 2 Outfall parameters 32

Table 3 ECO Lab process constants 35

Table 4 BOD maximum and DO minimum at outfall location during SW monsoon 38

Table 5 BOD maximum and DO minimum at outfall location during NE monsoon 42

Page 7: NIO final report

6

List of abbreviations

AOML Atlantic Oceanographic and Meteorological Laboratory

AD Advection-Dispersion

ADI Alternating Direction Implicit

BOD Biochemical Oxygen Demand

CBOD Carbonaceous Biochemical Oxygen Demand

CFL Courant-Friedrichs-Lewy

CPU Central Processing Unit

DHI Danish Hydraulic Institute

DO Dissolved Oxygen

DS Double Sweep

ECMWF European Center for Medium-range Weather Forecast

FC Fecal Coliform

GODAE Global Ocean Data Assimilation Experiment

HD Hydrodynamic

IRS Indian Remote-Sensing Satellite

LAT Lowest Astronomical Tide

MISST Multi-sensor Improved Sea Surface Temperatures

MLD Million Litres per Day

MSL Mean Sea Level

MWF Mean Wind Field

MT Mud Transport

NBOD Nitrogenous Biochemical Oxygen Demand

NEERI National Environmental Engineering Research Institute, India

NHO Naval Hydrographic Office, Dehradun, India

NS Navier-Stokes

OTPS OSU Tidal Prediction Software

PA Particle tracking

Page 8: NIO final report

7

SOD Sediment Oxygen Demand

ST Sediment Transport

UTM Universal Transverse Mercator

Page 9: NIO final report

8

List of symbols

C Chezy number

Cd Drag coefficient (bed resistance)

CD Drag coefficient (wind friction)

CR Courant number

d Time-varying water depth

E Eddy viscosity coefficient

f Wind friction factor

g Acceleration due to gravity

h Water depth

H Henryโ€Ÿs law constant

M Manning number

NE Northeast

p, q Flow fluxes in x and y directions, respectively

Six, Siy Source impulses in x and y directions, respectively

SW Southwest

t Time, temperature

u, v Current velocities in x and y directions, respectively

x, y Cartesian space coordinates

โˆ†x, โˆ†y Grid spacing in x and y directions, respectively

โˆ†t Time step for numerical simulations

ฮฉ Coriolis parameter

ฯa, ฯair Density of air

ฯw Density of water

ฮท Surface elevation

๐œ Effective shear stress

ฮธ Arrhenius temperature coefficient

Page 10: NIO final report

9

1. Introduction

1.1 Coastal waters and marine outfalls

Coastal waters provide for societyโ€Ÿs critical needs: they have, since mankindโ€Ÿs early days, been an

easily accessible resource of food and are also likely to serve as the major source of drinking water

in the near future as inland freshwater resources become increasingly scarce. Coastal industries

are now withdrawing sea water for their coolant plants as this need cannot be met entirely by

groundwater. Moreover, coastal waters are also among the worldโ€Ÿs most ecologically important

domains. They support a rich biodiversity of marine lifeforms and form an integral component of the

Earthโ€Ÿs overall ecosystem.

At the same time, coastal waters are also seen as an almost infinite sink for anthropogenic wastes

due to their high dilution and dispersion characteristics. They are now being increasingly used for

this purpose, with coastal cities in several developing countries considering the construction of

marine outfalls (Figure 1) to dispose off their municipal sewage and industrial effluents. Developed

countries have already been using this method of waste disposal since several decades (Gupta et

al., 2006).

Figure 1: A marine outfall: shown above is the Boston outfall, which discharges into Massachusetts

bay (from Hunt et al., 2010)

Page 11: NIO final report

10

Marine pollution occurs when wastewater discharges from such outfalls get concentrated in a

limited area instead of being dispersed over large areas and at longer distances away from the

coast. The resulting degradation in marine water quality is a matter of serious concern because of

its potential impacts on human and aquatic ecosystem health (Gupta et al., 2009). This gives rise to

the need for an accurate prediction of effluent dispersal patterns and water quality around existing

and proposed marine outfalls.

1.2 Study area: Mumbai coastal region

Figure 2: Bandra and Worli outfalls and Mumbaiโ€Ÿs drainage zones (from Kumar et al., 2000)

Page 12: NIO final report

11

The first scheme for the marine disposal of wastewater from the entire Bombay municipal area

through outfalls at Bandra and Worli was envisaged in 1970 (NEERI, 1995a). The city of Mumbai,

with a population of over 18 million (Sawant et al., 2007), generates more than 2700 million litres

per day (MLD) of sewage from seven drainage zones (Vyas and Vyas, 2007). Five of these,

discharge their sewage into the Arabian Sea waters off the west coast. Sewage disposal at Bandra

and Worli is being done through two 3.4 km long outfalls with a diffuser system at their end. Primary

pre-treatment, comprising of screening and degritting procedures, is given to the sewage before it is

disposed off into the sea (Gupta et al., 2006).

Figure 3: DO levels along Mumbaiโ€Ÿs west coast (from Kumar et. al., 2000)

Page 13: NIO final report

12

Regions such as Malad and Mahim, which receive about 50% of the wastewater discharge from the

city, have been reported as highly polluted, going by the Dissolved Oxygen (DO) levels of their off-

shore waters (Kumar et al., 2000). Other stretches of the Mumbai coast are affected by moderate

DO depletion (Figure 3). Biochemical Oxygen Demand (BOD) and nutrients have been reported low

in most of the regions. Fecal Coliform (FC) levels, considered to be the best pollution indicator for

coastal waters, have been reported high in 75% of areas with values ranging from 103-104 counts

per 100mL.

1.3 Review of water quality modelling studies

Coastal systems are extremely complicated and include many variables and processes which

exhibit highly non-linear interactions. However, due to recent advances in computing power,

physical and ecological models have emerged as one of the most potent tools for predicting

pollutant dispersal and water quality variations (James, 2002). As a result, several outfall modeling

studies have been conducted in the past for different regions along Mumbaiโ€Ÿs coast. Some of these

are discussed below.

A project feasibility and environmental impact assessment study was undertaken by NEERI in the

1990โ€Ÿs, wherein it was concluded that construction of outfalls at Bandra and Worli would eliminate

the DO depletion problem near the coast and considerably improve the ecological health of the

near-shore waters in the region (NEERI, 1995b). DO and BOD levels near the diffusers were

reported as expected to be above 4mg/L and below 3mg/L respectively during most of the tidal

cycle. Although high BOD levels near the diffuser (>20mg/L) were also predicted, their probability of

occurrence was reported to be very low (less than 0.01). In a later study, Kumar et al. (2000)

predicted Fecal Coliform concentrations for several hypothetical outfall discharge scenarios at

Bandra and Worli. Considerable reductions in waste field sizes were reported for secondary pre-

treatment schemes and outfall lengths greater than 8 km.

A comparison of far-field dilutions was undertaken by Gupta et al. (2006), in which values obtained

through dye experiments were compared with the predictions of empirical and numerical models.

Sasamal et al. (2007) carried out an analysis of sewage and industrial pollution distribution in Thane

Creek, Mumbai using Indian Remote-Sensing Satellite (IRS) data, in which pollutant plumes were

identified and spread areas were approximately estimated. Vyas and Vyas (2007) simulated the

spread of Fecal Coliform (FC) patches around sewage outfalls at Worli, Bandra, Versova and Malad

during the monsoon season, for different lengths of outfalls and various levels of land pre-treatment.

More recently, Vijay et al. (2010) carried out hydrodynamic and water quality simulations for Malad

creek and evaluated various wastewater discharge scenarios, including a proposed 3.4 km long

outfall at Erangal, using the MIKE 21 flow model. The study concluded that diverting effluents from

Page 14: NIO final report

13

wastewater treatment facilities to the proposed ocean outfall would significantly improve the water

quality of Malad creek.

1.4 Objectives of the present study

Keeping in mind the above developments and the available time frame, the following objectives

have been defined for this study:

Simulating hydrodynamics off the Mumbai coastal region.

Simulating Biochemical Oxygen Demand (BOD) and Dissolved Oxygen (DO) concentrations

around an existing outfall at Worli, Mumbai.

Comparing water quality around the outfall for periods with contrasting met-ocean

conditions, namely, the southwest monsoon and the northeast monsoon seasons.

Comparing water quality around the outfall for atleast two discharge cases.

Page 15: NIO final report

14

2. Simulation of off-shore hydrodynamics

2.1 Hydrodynamical modeling

In coastal systems, it is the physical system that is more fundamental and which โ€œsets the stageโ€ for

the chemical and biological systems (James, 2002). Physical conditions, which include currents,

tides, waves, turbulence, light, temperature, salinity, bed materials and suspended particles,

determine the transport and dispersion of all suspended and dissolved material in the sea, including

contaminants and nutrients, and affect all biological activity from the transport of larvae and bacteria

and the growth of phytoplankton to the behaviour of fish. Therefore, a necessary condition for

modeling any estuary or coastal region is that the physics be properly represented. Hydrodynamical

modeling is thus considered to be a prerequisite for ecological investigations (Babu et al., 2005).

The oceans are forced by a variety of mechanisms. In addition to the periodic gravitational forces of

the moon and sun that generate the tides, the ocean surface is subject to wind stress that drives

most ocean currents. Local differences between air and sea temperatures generate heat fluxes,

evaporation, and precipitation, which in turn act as thermodynamical forcings capable of modifying

the wind-driven currents or of producing additional currents (Cushman-Roisin, 2010).

Hydrodynamical modeling of the ocean system involves solving the well-known Navier-Stokes (NS)

equations for fluid flow after incorporating additional terms that include the effects of the Earthโ€Ÿs

rotation. In case of near-shore waters, as the horizontal dimensions of the ocean are much larger

compared to its depth, the flow system is treated as two-dimensional, and stratification and density

variations are neglected (Boussinesq approximation). This assumption is implemented by depth-

integrating the Navier-Stokes equations, which results in the so called โ€žshallow water equationsโ€Ÿ

(Haidvogel and Beckmann, 1998).

The model used for hydrodynamical simulations in this study is the MIKE 21 Flow Model. The form

of the continuity equation and the depth-averaged momentum equations in the x and y directions

used in this model is as follows (DHI, 2009b):

Continuity equation (mass conservation):

๐œ•๐œ‚

๐œ•๐‘ก+

๐œ•๐‘

๐œ•๐‘ฅ+

๐œ•๐‘ž

๐œ•๐‘ฆ=

๐œ•๐‘‘

๐œ•๐‘ก (1)

x-momentum equation:

๐œ•๐‘

๐œ•๐‘ก+

๐œ• ๐‘2

๐‘•

๐œ•๐‘ฅ+

๐œ• ๐‘๐‘ž

๐‘•

๐œ•๐‘ฆ+ ๐‘”๐‘•

๐œ•๐œ‚

๐œ•๐‘ฅ+

๐‘”๐‘ ๐‘2+๐‘ž2

๐ถ2๐‘•2โˆ’

1

๐œŒ๐‘ค ๐œ• ๐‘•๐œ๐‘ฅ๐‘ฅ

๐œ•๐‘ฅ+

๐œ• ๐‘•๐œ๐‘ฅ๐‘ฆ

๐œ•๐‘ฆ โˆ’ ๐›บ๐‘ž โˆ’ ๐‘“๐‘‰๐‘‰๐‘ฅ +

๐‘•

๐œŒ๐‘ค

๐œ•๐‘๐‘Ž

๐œ•๐‘ฅ= ๐‘†๐‘–๐‘ฅ (2)

Page 16: NIO final report

15

y-momentum equation:

๐œ•๐‘ž

๐œ•๐‘ก+

๐œ• ๐‘ž2

๐‘•

๐œ•๐‘ฆ+

๐œ• ๐‘๐‘ž

๐‘•

๐œ•๐‘ฅ+ ๐‘”๐‘•

๐œ•๐œ‚

๐œ•๐‘ฆ+

๐‘”๐‘ž ๐‘2+๐‘ž2

๐ถ2๐‘•2โˆ’

1

๐œŒ๐‘ค ๐œ• ๐‘•๐œ๐‘ฆ๐‘ฆ

๐œ•๐‘ฆ+

๐œ• ๐‘•๐œ๐‘ฅ๐‘ฆ

๐œ•๐‘ฅ โˆ’ ๐›บ๐‘ โˆ’ ๐‘“๐‘‰๐‘‰๐‘ฆ +

๐‘•

๐œŒ๐‘ค

๐œ•๐‘๐‘Ž

๐œ•๐‘ฆ= ๐‘†๐‘–๐‘ฆ (3)

where, p and q are the fluxes in the x and y directions respectively, h is the water depth, d is the

time-varying water depth, t is time, g is acceleration due to gravity, ฮท is surface elevation, ฯw is the

density of water, ฮฉ is the latitude-dependent coriolis parameter, pa is the atmospheric pressure, f is

the wind friction factor, C is Chezyโ€Ÿs bed resistance coefficient, ๐œ๐‘ฅ๐‘ฅ , ๐œ๐‘ฅ๐‘ฆ , ๐œ๐‘ฆ๐‘ฆ are the components of

effective shear stress, and Six and Siy are the source impulses in the x and y directions.

The following sections describe details of hydrodynamical simulations using the MIKE 21 flow

modeling system.

2.2 MIKE 21 flow model

MIKE 21 Flow Model is a modeling system for 2-D free surface flows developed by the Danish

Hydraulic Institute (DHI), Denmark. It is capable of simulating hydraulic and environmental

phenomena in lakes, estuaries, bays and coastal waters. It is usually used for systems, where

stratification can be neglected (DHI, 2009a).

The hydrodynamic module (MIKE 21 HD) is the basic module in the MIKE 21 Flow Model. It

simulates water level variations and flows in response to a variety of forcing functions such as tides

and winds. The HD Module can be applied to a wide range of hydraulic and related phenomena,

which include:

modelling of tidal hydraulics

wind and wave generated currents

storm surges

MIKE 21 HD is a very general hydraulic model and can be easily set up to describe specific

hydraulic phenomena. The module outputs current speed and direction, that has vital importance in

applications such as dispersion of substances, water quality management, oil spill analysis,

sediment transport, etc. Hence this module is the basis for using the other MIKE 21 Flow Model

modules, such as the Advection-Dispersion module (AD), the Sediment transport module (ST), the

Mud transport module (MT), the Particle tracking module (PA) and the Environment module (ECO

Lab).

Page 17: NIO final report

16

The numerical scheme of the HD module makes use of the Alternating Direction Implicit (ADI)

technique to integrate the equations for mass and momentum conservation in the space-time

domain. The equation matrices that result for each direction and each individual grid line are

resolved by a Double Sweep (DS) algorithm (DHI, 2009b).

2.3 Model setup

MIKE 21 HD requires the following input data for hydrodynamic simulations (DHI, 2009a).

Model domain and grid size

Time step and length of simulation

Bathymetry

Bed roughness coefficients

Momentum dispersion coefficients

Wind data and wind friction factor

Initial and Boundary conditions

2.3.1 Model domain and grid size

The MIKE 21 Flow Model is a finite difference model with constant grid spacing in the x and y

directions, i.e. a uniform rectangular or square grid is used to solve the shallow water equations.

Grids must be large enough to avoid perturbations along the boundary and small enough to allow

accurate description of the complex features like reef and channel systems.

In the areas where the flow is very complex due to the complexity in the bathymetry, high resolution

grids need to be used for a better representation of flow. The model domain should be selected in

such a way that the region of interest and the measurement locations for calibration and validation

are away from the boundaries (Vinodkumar, 2010).

The model domain extends from 70ยฐ 50โ€ŸE to 73ยฐ 03โ€ŸE and from 18ยฐ 15โ€ŸN to 19ยฐ 57โ€ŸN. The model

domain was selected in such a way that the region of interest (Mumbai coastal region) and the

measurement location for calibration and validation (Worli) are away from the boundaries. The grid

size was chosen as 500m (in both x and y directions) based on earlier works using model grids of

similar size (Sharif, 2009). The number of grid points in the x and y-directions were taken as 460

and 360, respectively. The model domain thus covers an area of about 41,400 km2, in which the

land area is about 3900 km2.

2.3.2 Time step and length of simulation

Time step refers to the time step used in the numerical solution of the model equations. Although

better results can be obtained reducing the time step, this often leads to large CPU running times.

Thus, a compromise between the numerical accuracy of solution and CPU running time has to be

Page 18: NIO final report

17

worked out. Another consideration is the Courantโ€“Friedrichsโ€“Lewy (CFL) stability criterion, which

defines the Courant number CR as:

๐ถ๐‘… = (๐‘ ร— โˆ†๐‘ก)/โˆ†๐‘ฅ (4)

where, ๐‘ = ๐‘”๐‘• is the wave celerity. Here, h is the water depth at a given grid point, and g is

acceleration due to gravity.

โˆ†t is the time step for the computation, and

โˆ†x is the grid spacing

As the information (about water levels and fluxes) in the computational grid travels at a speed

corresponding to the celerity, the Courant number expresses how many grid points the information

moves in one time step. Normally, this measure should be close to 1. However, a Courant number

of upto 5 is usually allowable. Higher Courant numbers may be allowed for model domains in which

the bathymetry is very smooth (DHI, 2009a).

For the hydrodynamic simulations in this study, a time step of 30 seconds was eventually selected

after several trials. This corresponds to a Courant number of 1.93. These values were considered to

provide a good balance between the numerical stability, accuracy and running time of the

simulation. It should be kept in mind that the in-built criterion check uses the maximum depth in the

bathymetry to calculate the Courant number. Thus, the value of CR at other grid points is always

less than the value displayed.

It was decided to simulate hydrodynamics and study effluent dispersal patterns during periods of

contrasting met-ocean conditions. Two periods, viz. 1st August to 31st August, 2009 and 22nd

October to 22nd November, 2009, were chosen for this purpose. While the former is representative

of the southwest monsoon season, the latter heralds the onset of the northeast monsoon.

2.3.3 Bathymetry

Bathymetry refers to the depth of the sea bed referred to a chosen datum. It is usually provided in

nautical charts as depth values at distinct points. These charts use the Lowest Astronomical Tide

(LAT) level as their datum, which is usually established from a long time series of water level

observations. This datum is employed because such charts are used for navigational purposes, and

knowing the minimum water depth at any point helps mariners to navigate vessels safely (DHI,

2009a). However, for hydrodynamic modeling purposes, referencing the bathymetry to Mean Sea

Level (MSL) is more useful since other input data, such as boundary tidal levels, is referenced to

Mean Sea Level. MSL is defined as the average of hourly water level measurements taken over a

time period of 18.6 years. This accounts for variability due to all possible factors. However, the

definition of MSL makes it a local datum, different for each tidal station.

Page 19: NIO final report

18

Figure 4: Interpolated bathymetry for the model domain

The bathymetry used for model simulations is presented in figure 4. Bathymetry data for the model

domain was sourced from nautical chart no. INT 7334 published by the Naval Hydrographic Office

(NHO), Dehradun. The depth values were digitised using the bathymetry editor in MIKE ZERO and

interpolated onto the model grid points using bilinear interpolation (DHI, 2009d). The Universal

Transverse Mercator (UTM) map projection for UTM Zone 43 was used for this purpose. The

bathymetry thus obtained was relative to LAT level. To reference the bathymetry to MSL, the

average of differences between LAT and MSL for the model domain was added to all grid points.

The maximum depth in the bathymetry is around 105 m and occurs in the southwest corner of the

domain, at a distance of roughly 225 km from the coast.

2.3.4 Bed roughness coefficients

The sea bed sediment is usually mobile and its roughness may impact the flow and therefore act as

a feedback between the sea floor and overlying flows. Sea bed roughness depends upon the shape

and size characteristics of sediment particles. Friction from the sea bed gains importance in shallow

waters where it can act significantly to retard the flow. In MIKE 21, the shear due to bed resistance

is implemented as (Lambkin, 2010):

๐œ๐‘ = ๐œŒ๐ถ๐‘‘๐‘ˆ2 (5)

where, ฯ is the density of water, U is the depth-averaged velocity and Cd is a drag coefficient.

Page 20: NIO final report

19

The drag coefficient Cd is specified as:

๐ถ๐‘‘ = ๐‘”/๐ถ2 (6)

where, g is the acceleration due to gravity and C is the Chezy number.

Bed roughness is considered as a primary calibration variable within the HD module. The bed

roughness can be specified using either the Chezy number or Manning number. However, Manning

numbers, if specified, are converted to Chezy numbers for use in the above equation (DHI, 2009a):

๐ถ = ๐‘€ ร— ๐‘•1

6 (7)

where, h is the water depth.

Thus, the running time of the simulation increases when Manning numbers are specified. The

Manning number as used in MIKE 21 is the reciprocal of the Manning number as understood in

hydraulics. Hence, in MIKE 21, smaller values of M (or C) imply larger bed resistance for the same

flow velocity and water depth, and vice-versa. For this study, the default value of Manning number,

M=32, was applied over the whole model domain.

2.3.5 Momentum dispersion coefficients

The finite spatial and temporal resolutions used for the numerical solution of flow dynamics result in

finer scales of motion being left unaccounted for. Examples include molecular motion and fluid

motion due to turbulence. Also, the depth-integrated Navier-Stokes equations necessarily imply

neglecting vertical velocity gradients in the case of shear flows. Since these non-resolved scales

also contribute to momentum transfer, it becomes necessary to account for them in some way to

correctly model the flow. The effective shear stresses (๐œ๐‘ฅ๐‘ฅ , ๐œ๐‘ฅ๐‘ฆ , ๐œ๐‘ฆ๐‘ฆ ) in the momentum equations

(Eqns. 2 and 3) incorporate the momentum fluxes due to turbulence, vertical integration and sub-

grid scale fluctuations. In MIKE 21 HD, these stresses are calculated using the eddy viscosity

concept (DHI, 2009a).

The eddy viscosity coefficient E can be taken as constant or can be defined using the Smagorinsky

scheme, which expresses E as a time-varying function of local gradients of depth-averaged velocity:

๐ธ = ๐ถ๐‘ 2โˆ†2

๐œ•๐‘ข

๐œ•๐‘ฅ

2

+1

2 ๐œ•๐‘ข

๐œ•๐‘ฆ+

๐œ•๐‘ฃ

๐œ•๐‘ฅ

2

+ ๐œ•๐‘ฃ

๐œ•๐‘ฆ

2

(8)

where, u and v are depth-averaged velocity components in the x and y directions respectively, โˆ† is

the grid spacing and Cs, the Smagorinsky factor, is a constant to be chosen in the interval of 0.25 to

1 (DHI, 2009a). Note that E has dimensions of L2 T-1. To avoid stability problems, the Eddy viscosity

coefficient must satisfy the following criterion:

Page 21: NIO final report

20

๐ธ. โˆ†๐‘ก/โˆ†๐‘ฅ2 โ‰ค 12 (9)

In the present study, the velocity-based Smagorinsky scheme has been employed and a constant

value of Cs = 0.4 has been applied over the domain for all simulations.

2.3.6 Wind data and wind friction factor

Winds blowing over the ocean surface transfer momentum to the waters below, generating wind-

driven currents. Winds can also induce flows indirectly, by causing vertical motions of the ocean

boundary layer, a phenomenon known as Ekman pumping, which leads to vertical displacements in

the thermocline that eventually produce geostrophic currents.

Momentum transfer from the atmosphere to the ocean takes place through wind stress, which is the

horizontal force of the wind on the sea surface. Wind stress T is calculated from:

๐‘‡ = ๐œŒ๐‘Ž๐‘–๐‘Ÿ ๐ถ๐ท๐‘ˆ102 (10)

where, ฯ represents density, U10 is the wind speed at 10 m above the sea surface and CD is the drag

coefficient (Stewart, 2008).

The drag coefficient CD is estimated from measurements and generally increases with increasing

wind speed. However, it has been reported that values of CD stabilize and even decrease as wind

speeds approach hurricane speeds (Powell et al., 2003; Donelan et al., 2004; Moon et al., 2004).

From these studies it can be concluded that CD values upto 2.6 ร— 10โˆ’3 may be used for strong and

moderate ocean winds.

Zonal (u) and meridional (v) wind speeds over the domain for the periods of study were obtained

from the CERSAT database maintained by IFREMER, France (Bentamy and Croizรฉ-Fillon, 2006).

The Blended Mean Wind Fields (MWF Blended) product was used, which is a combination of wind

fields derived from scatterometer data and ECMWF wind analyses (CERSAT, 2010). The spatial

and temporal resolutions of the gridded data are 0.25ยฐ and 6 h, respectively. The data was

subsequently interpolated onto the model grid points to generate input wind fields for the HD

module. The wind rose diagrams at Worli (72ยฐ 43โ€Ÿ 35.1โ€E 19ยฐ 03โ€Ÿ 31.6โ€N), which is the measurement

location used for model calibration, are presented in figures 5 and 6. Winds speeds shown are in

m/s.

From these diagrams, it can easily be seen that the predominant wind direction in the months of

August and October-November is south-westerly and north-easterly respectively, which is expected,

as these periods correspond to the SW and NE monsoon seasons.

Page 22: NIO final report

21

Figure 5: Wind rose diagram for Worli during 1 Aug to 31 Aug, 2009 (SW monsoon period)

Figure 6: Wind rose diagram for Worli during 22 Oct to 22 Nov, 2009 (NE monsoon period)

Page 23: NIO final report

22

2.3.7 Initial and boundary conditions

As the depth-integrated shallow water equations are partial differential equations, they require

specification of initial and boundary conditions for their solution. Boundary conditions specify the

values of dependant variable or its derivatives at the boundaries of the mathematical model domain.

Since MIKE 21 HD solves for surface elevations and the flux densities in x and y directions, either

water levels or fluxes have to be specified at all open boundaries. Land boundaries do not require

such specifications since no flow occurs through them. Further, since water levels near the shore

are mainly influenced by tides, these can be obtained using global or regional tidal models.

Accuracy of boundary conditions is important to ensure a consistent specification of flow properties

at the interface between the interior and the exterior of the model domain.

The boundary conditions at the north and south boundaries were specified as water levels, which

were predicted using the OSU Tidal Prediction Software (OTPS). For the points on the western

boundary, which is located far away from the shore, fluxes were calculated using geostrophic

currents obtained from AOML (Atlantic Oceanographic and Meteorological Laboratory) databases.

Specification of initial conditions requires providing the elevation of the water surface at the zeroth

time step. If differences in initial boundary surface elevations are not too large, an average value

can be computed and applied as a constant over the whole model domain. However, significant

variations in boundary levels were observed in this case and hence the initial surface elevations

were directly generated using the OTPS tidal model.

Figure 7: Plot of initial surface elevation

Page 24: NIO final report

23

Figure 7 shows the water surface elevation at the start of the simulation. Note that the northern

coastal regions in the model domain are experiencing high tide, while the southern ones are

experiencing low tide.

2.4 Model calibration, validation and simulation

2.4.1 Model calibration

Model calibration is defined as the fine tuning of model parameters until model results and field

measurements are within an acceptable tolerance. Calibration is necessary to ensure that the

model correctly reproduces actual flow conditions. Calibration is accomplished by modifying the

boundary conditions, improving the hydrometeorological forcing input and by adjusting the

calibration parameters, viz. bed resistance, wind friction factor and eddy viscosity. The relative

importance of a parameter to the calibration process depends upon the model outputโ€Ÿs sensitivity to

that parameter. Usually, the model output is calibrated using data on water levels and currents, as

these are easier to measure compared to other quantities like fluxes. Simulated values are

compared to measured values in order to get a good agreement between the two.

In this study, model calibration was done using measured water levels and currents off Worli (72ยฐ

43โ€Ÿ 35.1โ€ E 19ยฐ 03โ€Ÿ 31.6โ€ N). Data was available for the period from 22nd October to 22nd November,

2009. After evaluating several combinations of calibration parameters, it was found that a Manning

number of 32 and a wind friction factor equal to 0.0026 gave the best results. The model output was

not very sensitive to changes in eddy viscosity, and a value of 0.4 has been used in the final

simulations.

2.4.2 Model validation

Calibration is followed by a process called validation, wherein model output is compared with

measured parameter values at several locations within the domain. This provides an indication of

the modelโ€Ÿs sensitivity and gives confidence that the results produced fall within the desired range of

accuracy and are applicable over the model domain.

Comparison between measured and simulated water levels off Worli is presented in figure 8. It can

be seen that simulated water levels match very well with measured water levels. Figures 9 and 10

show the comparison between simulated and measured values for zonal (u) and meridional (v)

components of currents off Worli.

The simulated currents yield a good match, in general, with the measured values. The

discrepancies may be because currents measured at a specific depth do not truly represent the

depth-averaged currents which the model simulates. The significant amplitude difference in the

case of meridional current may be due to lack of information on flow directions. The model assumes

Page 25: NIO final report

24

that the flow is perpendicular or nearly perpendicular to the boundary, a condition which might not

hold true for this particular model domain.

Figure 8: Comparison between measured and simulated water levels at Worli

Figure 9: Comparison between measured and simulated u-component of currents off Worli

Figure 10: Comparison between measured and simulated v-component of currents off Worli

Page 26: NIO final report

25

2.4.3 Final simulations

The final HD simulations for the months of August and October-November were carried out using

the values of input parameters given in Table 1.

Table 1. Basic and hydrodynamical parameter values used for final simulations

Basic parameters Selection

Simulation period (SW monsoon) 1-8-2009 (00:00 h) to 31-8-2009 (00:00 h)

Simulation period (NE monsoon) 22-10-2009 (00:00 h) to 22-11-2009 (18:00 h)

Time step and Max. Courant No. 30 s, 1.93

Bathymetry Digitized from NHO Nautical Chart INT 7334

Coriolis forcing option Applied, as domain lies away from the equator

Flooding and Drying depths 0.1m and 0.2m, respectively

Hydrodynamic parameters Selection

Water levels OSU Tidal Prediction Software

Fluxes (geostrophic currents) AOML Databases

Bed Resistance 32 (Manning number)

Eddy Viscosity 0.4

Wind Used ECMWF Blended MWF

Wind friction factor 0.0026

Page 27: NIO final report

26

3 . BOD-DO modeling of wastewater outfalls

3.1 Importance of dissolved oxygen

Dissolved oxygen refers to molecular oxygen present in the dissolved state in water. The

concentration of dissolved oxygen is a critical attribute of aquatic ecosystems, as it determines the

suitability of natural waters for many types of organisms. Dissolved oxygen is vital, because it is

required for respiration and cellular metabolism. Sufficient DO levels in natural water bodies are

thus necessary for sustenance of aquatic and marine communities and for preserving the health of

the ecological system.

The principal inputs affecting the DO levels of aquatic systems are municipal sewage and certain

industrial waste discharges. Dissolved oxygen depletion basically begins with the input of oxygen

demanding wastes into a water body. If oxygen is consumed at a rate that exceeds the

replenishment rate of DO from the atmosphere or from photosynthesis, the DO concentration can

decrease to less than a few mg per litre, leading to anaerobic conditions. Fish mortality, unpleasant

odours and other aesthetic nuisances are the common symptoms of low DO concentrations

(Thomann, 1987). Another important aspect of sewage disposal in natural waters is the

phenomenon of Eutrophication. Municipal sewage contains a high amount of nutrients which can

lead to excess growth of algae, which would in turn further deplete the DO levels in the water body.

Therefore, reducing the severity of oxygen depletion in the vicinity of pollutant discharge points,

such as sewage outfalls, is a major concern.

For modeling the dissolved oxygen in an natural system, the quantity and strength (in terms of

BOD) of discharged effluents has to be considered. In addition, internal sinks of DO, which

continuously remove a part of the dissolved oxygen, must also be included. These comprise of the

oxygen demand due to respiration processes and the benthic oxygen demand which originates from

the sediment bed (the benthic zone). One also has to take into account oxygen replenishment,

which takes place concurrently through the processes of reaeration and photosynthesis. Further, in

case of highly dynamic flow systems like the oceans, advection and dispersion processes play a

major role in reducing the impact of sewage on dissolved oxygen, by ensuring the effective

dispersal of discharged effluents. All these processes are now discussed in greater detail in the

following sections.

3.2 Effluent dispersal processes

Effluent dispersal occurs due to advection and dispersion. Advection refers to the transport of a

dissolved or suspended substance due to the bulk motion of the fluid medium. Dispersion is a

general term which refers to the scattering of fluid particles due to both random processes like

Page 28: NIO final report

27

molecular and turbulent diffusion, and due to the effects of velocity gradients. Advection is the main

phenomenon responsible for spreading of pollutant in rivers and coastal waters. In a given amount

of time, the distances over which mass is carried by dispersive transport are usually not as great as

those covered by advection (Hemond and Fechner-Levy, 1999).

In advection, the chemical is transported at the same velocity as the fluid, by which it is meant that

the center of mass of the chemical moves by advection at the average fluid velocity. Flux density

due to advection (Ja) is equal to the product of a chemicalโ€Ÿs concentration in the fluid (C) and the

velocity (V) of the fluid:

๐ฝ๐‘Ž = ๐ถ๐‘‰ (11)

Flux density is the mass of chemical transported across an imaginary surface of unit area per unit of

time.

Turbulent fluid motions contain constantly changing swirls of fluid, known as eddies, of many

different sizes. Turbulent diffusion arises from the random mixing of the air or water by these

eddies. This type of motion neither augments nor impedes the advective motion of a chemical

(Hemond and Fechner-Levy, 1999). Turbulent diffusion has the net effect of carrying mass in the

direction of decreasing chemical concentration. Fickโ€Ÿs first law is typically used to describe the flux

density of mass transport by turbulent diffusion (Jt):

๐ฝ๐‘ก = โˆ’๐ท(๐‘‘๐ถ/๐‘‘๐‘ฅ) (12)

for one dimension. Here, D is called the turbulent (or eddy) diffusion coefficient. Its value varies

greatly, depending upon the intensity of fluid turbulence. D can not only be anisotropic, but may also

vary with time and location.

Even if a fluid is entirely stationary and without obstructions, chemicals will still move from regions of

higher concentration to regions of lower concentration, due to the ceaseless random movement of

molecules. Mixing resulting due to this process is called molecular diffusion. Flux density due to

molecular diffusion is also described by Fickโ€Ÿs first law and D in this case is called the molecular

diffusion coefficient. Unlike the turbulent diffusion coefficient, molecular diffusion coefficients can be

estimated without reference to a particular situation, because they depend primarily on the size of

the molecules that are diffusing. At environmental temperatures, molecular diffusion coefficients in

water for most chemicals are of the order of 10-9 m2/s (Hemond and Fechner-Levy, 1999). Molecular

diffusion increases in magnitude at higher temperatures and for chemicals with smaller size

molecules or particles.

Dispersion also includes the spreading caused by velocity shear which arises due to velocity

gradients within the flow. This phenomenon is known as the shear effect and flows with velocity

Page 29: NIO final report

28

gradients are often referred to as shear flows. When approximating three-dimensional flows as two-

dimensional, the depth integration filters out the vertical velocity profiles which are responsible for

additional spreading in the direction of flow. This effect is accounted for by including a โ€žshear

viscosityโ€Ÿ coefficient in the overall dispersion coefficient. For the large scales usually considered in

surface water quality modeling, molecular and turbulent mixing effects are practically negligible in

comparison to mixing due to the shear effect (DHI, 2009c).

At this point, it is important to note that diffusion and other processes are actually advective

transport processes at their own length scales. However, since the spatial resolution of the model

grid is much greater than these length scales, such advective processes are not resolved by the

model. It is because of this that their contribution to chemical mixing has to be accounted for using

Fickโ€Ÿs law. Continuing further, it follows that as the grid size is increased, higher and higher length

scales would count as non-resolved and would thus require inclusion in the dispersion coefficients.

Thus, as โˆ†x increases, the dispersion coefficients should also increase. It thus becomes natural to

relate them in some way to โˆ†x and โˆ†t, the spatial and temporal resolutions of the flow model. This

can be done using several possible forms, such as k.(โˆ†x2/t), k.(โˆ†x.u) and k.(โˆ†t.u2), where u is the

depth-integrated current velocity and k is a dimensionless constant. However, the scale above

which such relationships can be assumed to hold true must first be known. It has been found that

this lower limit occurs when โˆ†x is approximately equal to h. At this scale, the largest non-resolved

motions are due to the shear effect (DHI, 2009c).

The advection-dispersion equation used by the AD module in MIKE 21 is described below. This is

the equation for advective-dispersive transport of dissolved or suspended substances in two

dimensions.

๐œ•

๐œ•๐‘ก(๐‘•๐‘) +

๐œ•

๐œ•๐‘ฅ(๐‘ข๐‘•๐‘) +

๐œ•

๐œ•๐‘ฆ(๐‘ฃ๐‘•๐‘) =

๐œ•

๐œ•๐‘ฅ ๐‘•๐ท๐‘ฅ

๐œ•๐‘

๐œ•๐‘ฅ +

๐œ•

๐œ•๐‘ฆ ๐‘•๐ท๐‘ฆ

๐œ•๐‘

๐œ•๐‘ฆ โˆ’ ๐น. ๐‘•. ๐‘ + ๐‘† 13)

Here, h is the water depth, u and v are current velocities in the x and y directions, respectively, c is

concentration of the substance under investigation, Dx and Dy are the dispersion coefficients in the x

and y directions, respectively, F is a linear decay coefficient for non-conservative substances, and S

is given by:

๐‘† = ๐‘„๐‘  ร— ๐‘๐‘  โˆ’ ๐‘ (14)

where, Qs and cs are the source(sink) discharge rate and chemical concentration, respectively.

Page 30: NIO final report

29

3.3 Processes affecting DO levels

3.3.1 Reaeration

Reaeration is the process describing the exchange of oxygen between the water and the

atmosphere. The concentration of oxygen (or any gas or volatile substance) in surface water that is

equilibrium with the atmosphere above it, is given by Henryโ€Ÿs law as:

๐ถ๐‘’๐‘ž = ๐ถ๐‘Ž๐‘–๐‘Ÿ /๐ป (15)

where, Cair is the concentration of oxygen in air (usually expressed as a partial pressure), and H is

the Henryโ€Ÿs law constant of oxygen. If the concentration in water (Cw) differs from Ceq, exchange of

oxygen between air and water takes place. The flux density of this exchange is directly proportional

to the difference between Cw and Ceq:

๐ฝ = โˆ’๐‘˜๐‘ค ๐ถ๐‘ค โˆ’ ๐ถ๐‘’๐‘ž (16)

where, J is the flux density and kw is the gas exchange coefficient or reaeration rate. kw depends

upon the nature of the water flow and air movement above the water. Ceq is also known as the

saturation concentration and it depends upon the salinity and temperature of water.

3.3.2 Photosynthesis and respiration

All aquatic plant forms including attached and floating aquatic weeds, algae and phytoplankton

generate oxygen through photosynthesis. Photosynthesis can result in the supersaturation of water

with oxygen, and DO levels as high as 150 to 200% are not uncommon (Thomann, 1987). The

photosynthesis reaction can be represented as:

6CO2 + 6H2O โ†’ C6H12O6 + 6O2 (17)

Because photosynthesis is dependent on sunlight, the production of DO through photosynthesis

occurs only during daylight. DO levels in natural waters show a diurnal variation, with minima at pre-

dawn and maxima at afternoon. Apart from solar intensity, the rate of photosynthesis depends upon

the depth at which photosynthesis takes place and on the availability of nutrients. Photosynthetic

rates can be estimated through several methods, one of which involves relating primary productivity

to observed chlorophyll-a levels (Benrenfeld and Falkowski, 1997).

Concurrently with photosynthetic DO production, aquatic plants and other aquatic life forms also

consume oxygen during respiration. Net O2 production is equivalent to the gross O2 production

minus the oxygen consumed by autotrophic and heterotrophic respiration (Dickson et al., 2001).

Page 31: NIO final report

30

3.3.3 BOD exertion

The oxygen demand exerted by effluents is generally classified according to the nature of the

compounds that give rise to such demand. Biochemical oxygen demand or BOD is a measure of the

amount of oxygen required to completely oxidize all dissolved and suspended organic matter

present in a volume of water. BOD is thus an indirect measure of the organic content of water. BOD

is further differentiated as Carbonaceous BOD (CBOD) and Nitrogenous BOD (NBOD), based on

whether it originates from the presence of carbonaceous matter or nitrogenous matter (Thomann,

1987). CBOD is exerted by heterotrophic organisms capable of deriving energy by oxidizing organic

carbon substrates to CO2 and H2O. As an example, the aerobic oxidation of glucose (C6H12O6) is

presented below:

C6H1206 + 6O2 โ†’ 6CO2 + 6H2O (18)

NBOD exertion results from the oxidation of ammonia present in water to nitrite and then to nitrate

nitrogen. Nitrifying bacteria break down proteins, amino acids, urea and other nitrogenous

compounds present in sewage into simpler compounds like ammonia. This process, known as

deamination, is followed by nitrification, in which ammonia is oxidized to nitrate.

NH4 + 1.5 O2 โ†’ 2H+ + H2O + NO2โˆ’ (19)

NO2โˆ’ + 0.5 O2 โ†’ NO3

โˆ’ (20)

The BOD exertion phenomenon shows a lag phase, followed by a rising curve terminating in a

plateau, yielding an ultimate carbonaceous BOD (Swamee and Ojha, 1991). For substrates rich in

nitrogenous compounds, there is a further increase in the curve, which saturates at the second

stage ultimate BOD value (Figure 11). Nitrogenous BOD has not been considered in this study.

Figure 11: BOD exertion curve (schematic)

Page 32: NIO final report

31

In simplistic models of BOD exertion, like the Streeter-Phelps model (Metcalf and Eddy, 1991), BOD

is assumed to decay at a first order rate:

๐‘‘

๐‘‘๐‘ก ๐ต๐‘‚๐ท = โˆ’๐‘˜๐ต๐‘‚๐ท ร— ๐ต๐‘‚๐ท (21)

kBOD is typically of the order of 0.2/day (Hemond and Fechner-Levy, 1999), and depends upon

temperature and the concentration of dissolved oxygen remaining in the water.

3.3.4 SOD exertion

Sediment oxygen demand (SOD) or Benthic oxygen demand arises due to sediment organic

material which requires oxygen for stabilization. Dead rooted aquatic plants and settled

phytoplankton may constitute such benthic organic matter. SOD may also be due to settled sludge

banks below a waste outfall. These deposits may build up over time and remove oxygen from the

overlying water. The oxygen utilization therefore depends on the extent of organic material and the

nature of the benthic community. Estimates of SOD range from 0.07 gO2/m2/day for mineral soils to

10 gO2/m2/day for municipal sewage sludge in the vicinity of an outfall (Thomann, 1987). The rate

of SOD exertion depends upon temperature, DO levels in water, and the depth of the deposit in the

case of sewage sludge deposition.

3.4 Model setup

3.4.1 ECO lab

BOD-DO modeling has been carried out using the ECO Lab module in the MIKE 21 Flow Model.

ECO Lab is a numerical lab for ecological modelling. It is an open and generic tool for customizing

aquatic ecosystem models to describe water quality, eutrophication, heavy metals and ecology

(DHI, 2009e). The module can describe chemical, biological, ecological processes and interactions

between state variables, and physical processes, such as the sedimentation of various components.

Apart from BOD and DO, other variables which ECO Lab can model include ammonia, nitrite,

nitrate, phosphorus, fecal and total coliforms, and one or more user-defined pollutants. As ECO Lab

is integrated with the Advection-Dispersion (AD) module in MIKE 21, the physical transport of state

variables due to advection-dispersion processes can be described, based on the hydrodynamics of

the system being modelled (DHI, 2009f).

In ECO Lab, one can choose from several pre-defined templates, which contain mathematical

descriptions of various ecosystems. These can be modified to suit user requirements. It is also

possible to develop custom templates using in-built constants and functions (DHI, 2009e).

Apart from the data required to run the HD module, ECO lab requires additional data in order to

correctly simulate the processes in the template used. Initial and boundary conditions for the state

Page 33: NIO final report

32

variables being modelled have to be provided. Also needed are estimates of the values of the

coefficients governing various process rates. As the AD module is integrated with ECO Lab,

dispersion coefficients for each modelled component have to be specified. Additional forcings such

as salinity and temperature have to be specified over the model domain. Finally, if any sources and

sinks are included, component concentrations for these have to be specified.

3.4.2 Wastewater disposal scenario and outfall parameter values

As mentioned before, a single outfall at Worli has been considered for this study. Outfall parameters

relevant to model input have been taken from NEERI, 1995. Discharge rates vary widely, ranging

from 4m3/s in the dry season to 24 m3/s during rainy seasons. Outfall parameters are summarized in

Table 2 below.

Table 2. Outfall parameters

Outfall location and position in model domain Worli (406, 163)

Length of outfall 3.5 kms

Alignment (direction of discharge) 290o (clockwise from North)

Discharge rates 10 m3/s, 15 m3/s

Discharge velocity 1 m/s

The outfall is located 3.5 km away from Worli at a point where the water depth is 8.5 m. Through

this outfall, BOD is introduced into the model continuously, starting from the first time step.

Simulation of BOD and DO has been carried out for two different discharge cases:

Case 1: Effluent BOD is 40 mg/L and the outfall discharge rate is 10 m3/s.

Case 2: Effluent BOD is 50 mg/L and the outfall discharge rate is 15 m3/s.

The BOD throughout the model domain has been initialized to a very small value (10-6 mg/L). This

has been done to separate BOD dispersal patterns around the outfall from the decreasing trend in

background BOD (found in trial runs that high background BOD values decayed rapidly to zero). To

distinguish water from land, the initial BOD value was kept positive, since the model takes BOD on

land points to be equal to zero.

Dissolved Oxygen was initialized to a value of 6 mg/L throughout the model domain. This value was

chosen within the range of saturation oxygen concentrations, considering the variations in surface

temperature and salinity of water.

Temperature forcing was given as a time-varying input. Data on sea-surface temperature (SST) for

both periods of study was obtained from the Multi-sensor Improved Sea Surface Temperatures

Page 34: NIO final report

33

(MISST) for the Global Ocean Data Assimilation Experiment (GODAE) database (MISST, 2010).

The original SST data with a resolution of 0.088ยฐ was interpolated to the model grid following the

same procedure as for the wind data. Salinity was taken to be constant over the whole domain, with

a value equal to 35.2 psu (practical salinity unit).

Dispersion coefficients for the AD module were chosen following the guidelines given in the

scientific documentation for the MIKE 21 AD Module (DHI, 2009c). As the grid spacing (โˆ†x = โˆ†y =

500m) in this case is much greater than the range of depth values observed in the model domain

(10 to 100 m), it was decided to specify the dispersion coefficients as kโˆ†xu, where k is a constant in

the range 0.01-0.2 and u is the current velocity (DHI, 2009c). Since data for calibrating these

coefficients was unavailable, the value of k has been taken as 0.1.

3.4.3 Description of chosen template and process rate estimates

The ECO Lab template โ€žWQsimpleโ€Ÿ has been chosen for the purposes of this study. This allows

modeling of Biochemical oxygen demand (BOD) and Dissolved Oxygen (DO) in natural waters.

Three sinks, namely BOD, SOD and respiration, and two sources, namely photosynthesis and

reaeration, of dissolved oxygen are included in this template. The following expressions describe

the rate of change of BOD and DO, due to the combined action of included sinks and sources:

d

dt(BOD) = โˆ’bodd (22)

d

dt(DO) = reaera + phtsyn โˆ’ respT โˆ’ bodd โˆ’ sod (23)

The mathematical formulations of the processes included in these equations, as given in the

WQsimple template, are described below:

Reaeration

reaera = k2 Cs โˆ’ DO ; for DO > 0 k2Cs ; for DO < 0

(24)

Here, Cs represents the saturation concentration of oxygen at the given temperature and salinity,

DO represents dissolved oxygen variable and k2 is the reaearation constant. An expression for k2 is

included, which is as follows:

k2 = 3.93Vsp0.5/dz0.5 + 0.728Wsp

0.5 โˆ’ 0.371Wsp + 0.0372Wsp2/dz (25)

Here, Wsp represents the wind speed, Vsp the current speed, and dz the depth of the layer, which in

this case is equal to the water depth, as there is only one vertical layer in the model.

Page 35: NIO final report

34

Photosynthesis

The photosynthetic oxygen production is described by:

phtsyn = Lambert_Beer1 Pmax ., dz, 2.3

SD ร—suninp

depth ; if SD > 0

0 ; if SD < 0

(26)

Here, SD refers to Secchi disk depth, depth refers to water depth, Pmax refers to maximum primary

production at noon. โ€žLambert_Beer1โ€Ÿ and โ€žsuninpโ€Ÿ are functions which are defined below.

Lambert_Beer1 Pmax , dz, 2.3SD = Pmax ร— eโˆ’dz ร— 2.3

SD (27)

The above function returns the light intensity at the top of a layer, given the light intensity at the top

of the overlying layer. Secchi disk depth has been taken as a constant, with a value equal to 10m

(Suresh et al., 2006).

suninp = cos

2ฯ€t24

ฮฑ

0 ; during night

; during day (28)

where, t is relative daytime in hours. โ€žsuninpโ€Ÿ varies the light intensity returned by โ€žLambert_Beer1โ€Ÿ

according to the hour of the day. The primary productivity value for the Arabian sea has been taken

as 1.7 gO2/m2/day (Babu et al., 2006). Dickson et al. (2001) have reported remarkable similarity in

production rates in the Arabian sea during the southwest and northeast monsoon seasons.

Respiration

respT = resp ร— ฮธ tโˆ’20 ร— DO

DO +mdo ร—

1

depth (29)

Here, resp is the respiration rate of plants, t is temperature in degree Celsius and ฮธ is the Arrhenius

temperature coefficient for respiration rate. โ€žmdoโ€Ÿ is the half-saturation concentration of oxygen for

respiration. It is the DO level at which the respiration rate becomes equal to half its maximum value.

The respiration rate, โ€žrespโ€Ÿ, has been assumed to be 25% of the primary productivity value (Babu et

al., 2006).

Page 36: NIO final report

35

BOD decay

bodd = k3 ร— ฮธ tโˆ’20 ร— BOD ร—DO

DO +hdobod (30)

Here, k3 is the BOD decay rate, t is temperature in degrees Celsius, ฮธ is the Arrhenius temperature

coefficient which describes the temperature dependency of k3, and hdobod is the half-saturation

oxygen concentration for BOD. It is the DO level at which the BOD exertion rate becomes equal to

half its maximum value. k3 has been taken equal to 1/day, based on values given the literature for

the Mumbai coastal region (Gupta et al., 2006).

Sediment Oxygen Demand (SOD)

The rate of oxygen consumption due to sediment oxygen demand is given by:

sod = B1sed ร— ฮธ tโˆ’20 ร—DO

DO +mdosed ร— 1/dz (31)

Here, B1sed is the Sediment oxygen demand, t is temperature in degrees Celsius, ฮธ is the

Arrhenius temperature coefficient, dz is the water depth and mdosed is the half-saturation oxygen

concentration for SOD. It is the DO level at which the SOD exertion rate becomes equal to half its

maximum value. B1sed has been taken equal to the default value of 0.5g/m2/day.

3.5 Final simulations

For the final simulations, the HD time step was doubled to 60 s to reduce the total computational

time of the simulations. This gave a maximum Courant number of 3.86, well within the guideline

value (CR=5) prescribed by the HD reference manual (DHI, 2009a). The update frequency for the

ECO lab module was chosen equal to 5 AD time steps. Results were written to the output file every

360 time steps, i.e. BOD and DO outputs were obtained at 6-hourly intervals. The period of

simulations was reduced to 15 days, from the originally planned 30 days, in order to further reduce

the total computational time of the simulations.

Table 3 summarizes the values of ECO lab process constants used for the final simulations.

Table 3. ECO lab process constants

Parameter Value Units

ECO Lab template WQsimple.ecolab -

First order decay rate for BOD 1 day-1

Temperature coeff. for BOD decay 1.07 Dimensionless

Half-saturation O2 conc. for BOD 2 mg/L

Page 37: NIO final report

36

Max. O2 production at noon 1.7 gO2/m2/day

Secchi disk depth 10 M

Time correction for noon 0 Hours

Respiration rate for plants 0.425 gO2/m2/day

Temperature coeff. for respiration rate 1.08 Dimensionless

Half-saturation O2 conc. for respiration 2 mg/L

Sediment oxygen demand 0.5 gO2/m2/day

Temperature coeff. for SOD exertion 1.07 Dimensionless

Half-saturation O2 conc. for SOD 2 mg/L

Page 38: NIO final report

37

4. Results and discussion

4.1 Southwest monsoon season

As mentioned before, model runs were carried out for two discharge cases. BOD maximum and DO

minimum obtained at the outfall for both the cases are given in Table 4. Figures 12a and b show

the variation of BOD and DO at the outfall location in both discharge cases.

(a)

(b)

Figure 12: Variation of (a) BOD and (b) DO at the outfall location during the SW monsoon

Page 39: NIO final report

38

Table 4. BOD maximum and DO minimum at outfall location during SW monsoon

BOD maximum (mg/L)

(initial BOD โ‰ˆ 0 (10-6 mg/L))

DO minimum (mg/L)

(initial DO = 6 mg/L)

Discharge case 1 0.43 5.55

Discharge case 2 0.81 5.38

It can be seen that BOD and DO values oscillate in phase with the tidal variations. For an outfall

discharge rate of 10 m3/s and effluent BOD of 40 mg/l (Case 1), BOD at the outfall increases upto

0.43 mg/L from the initial level of 10-6 mg/L. When BOD loading is increased to 50mg/L, at a

discharge rate of 15 m3/s (Case 2), BOD levels at the outfall also show a corresponding increase,

reaching upto 0.81 mg/L. The drop in DO level is also more for the second case (0.62 mg/L) when

compared to that for the first case (0.45 mg/L).

The spread and concentration of BOD around the outfall for both cases after 72 hrs of discharge are

shown in Figures 13a and b. A plume of higher BOD value is seen to develop along the southwest-

northeast direction, parallel to the coast.

For case 1, approximately 50 km2 area around the outfall has a BOD value greater than 0.05 mg/L.

For case 2, the same area increases to more than 120 km2. Higher values of BOD near the outfall

for case 2 can also be clearly seen from these figures.

(a)

Page 40: NIO final report

39

(b)

Figure 13: BOD distribution after 72 hours of discharge for (a) case 1 (b) case 2

Plume pattern is found to vary in phase with tidal variations. Figures 15a and b show the BOD

plume (for the case 2 scenario) during typical flood and ebb tides in the SW monsoon, along with

the corresponding current patterns. During flood tide, the currents are alongshore and moving

northwards, whereas during ebb tide they are seen to be moving southwards. One can see a clear

change in the shape and direction of the plume during the two phases of the tide.

Figure 14 shows the relative magnitudes of the physical and ecological processes contributing to

the dissolved oxygen balance in water that have been considered by the model. Degradation of

Figure 14: Relative magnitudes of processes

Page 41: NIO final report

40

(a) (b)

Figure 15: Tidal variations in plume pattern: (a) flood tide (b) ebb tide

BOD from the sewage outfall consumes dissolved oxygen at the fastest rate, followed by aquatic

respiration and benthic (sediment) oxygen demand. The curve for BOD degradation mirrors the

curve for BOD (Figure 12a) since the rate of BOD decay is directly proportional to the amount of

biochemical oxygen demand present (Eqns. 21, 22 and 30). Replenishment of dissolved oxygen

takes place mostly through phytoplankton photosynthesis, as reaeration contributes little to this side

of the DO balance. Photosynthesis however, occurs only during daytime, and moreover, peaks only

at noon (12:00 h), as can be made out from the curve for photosynthesis rate. This explains why DO

values at the outfall exhibit a continuous decrease (Figure 12b) even though two inputs of DO are

present.

4.2 Northeast monsoon season

Figures 16a and b show the comparison between BOD and DO levels at the outfall location during

the two study periods under consideration. The curves are based on the results for discharge case

2 (effluent BOD load=50mg/L and outfall discharge rate=15 m3/s). The x-axis represents the time

elapsed in days since the starting of the simulation and outfall BOD loading. The BOD level appears

to be higher for the October-November NE monsoon period, but this difference may be attributed to

Page 42: NIO final report

41

the difference in tidal phase between the starting points of the two periods. Dissolved oxygen levels

though, do show a significant difference, with DO at the outfall being considerably higher for many

(a)

(b)

Figure 16: Comparison between (a) BOD and (b) DO levels at the outfall during the two study

periods, viz. SW monsoon (Aug) and NE monsoon (Oct-Nov) seasons

days in case of the SW monsoon period. However, outfall DO levels become almost equal after 14

days of continuous discharge in both periods, and thereafter DO level during the SW monsoon is

seen to fall below that for the NE monsoon period.

0.00E+00

1.00E-01

2.00E-01

3.00E-01

4.00E-01

5.00E-01

6.00E-01

7.00E-01

8.00E-01

9.00E-01

0 2 4 6 8 10 12 14 16

BO

D (

mg/

L)

Days elapsed since t=0

Aug_BOD

Oct_BOD

5.3

5.4

5.5

5.6

5.7

5.8

5.9

6

6.1

0 2 4 6 8 10 12 14 16

DO

(m

g/L)

Days elapsed since t=0

Aug_DO

Oct_DO

Page 43: NIO final report

42

Table 5 presents the maximum BOD and minimum DO obtained at the outfall during the NE

monsoon period, for both discharge cases. It can again be seen that increasing the effluent BOD

load and outfall discharge rate increases the BOD level and decreases the DO level at the outfall. A

comparison with corresponding values for the SW monsoon (Table 4) yields no significant

difference in values. BOD maximum and DO minimum values are nearly equal in both the seasons.

Table 5. BOD maximum and DO minimum at outfall location during NE monsoon

BOD maximum

(initial BOD โ‰ˆ 0 (10-6 mg/L))

DO minimum

(Initial DO=6 mg/L)

Discharge case 1 0.43 5.55

Discharge case 2 0.80 5.39

This lack of variability can be explained by the fact that the currents along the west coast of India

are primarily tide-driven during the northeast monsoon season (Tomczak and Godfrey, 1994).

Winds have a negligible impact upon the flow dynamics off the Mumbai coast during this period

(Vinodkumar, 2010). Tidal forcings play a major role and the hydrodynamics of the study region is

influenced mainly by them.

Page 44: NIO final report

43

5. Summary and conclusion

This study examines the water quality around an existing marine outfall located off the Mumbai

coast near Worli, Mumbai. The hydrodynamics of the Mumbai coastal region is simulated first, since

this is a prerequisite to the water quality investigation. Results obtained using the Hydrodynamic

(HD) module of the MIKE 21 Flow Model matched very well with measurements. Hence, the MIKE

21 model was further used along with its in-built ECO Lab module, for simulating Biochemical

Oxygen Demand (BOD) and Dissolved Oxygen (DO) concentrations around the above mentioned

outfall. Simulations were carried out for two periods, viz. the southwest monsoon season and the

northeast monsoon season, as these are known to display contrasting met-ocean conditions.

Further, two outfall discharge cases were considered, with a higher BOD loading and outfall

discharge rate being provided in the second case.

The study found no significant difference in maximum BOD values and minimum DO values at the

outfall during the two periods under investigation. However, as expected, values of these statistics

were seen to increase appreciably in discharge case 2, i.e. upon increasing the BOD input to the

coastal waters. A plume of higher BOD was found to develop in a direction parallel to the coast at

Worli. The plume pattern showed a periodic directional change, which was in phase with the tidal

reversal of off-shore currents.

This study has provided a rough estimate of the level of BOD increase and DO depletion that can

be expected to occur in ocean waters off the Mumbai coast during the operation of a single outfall. It

has also helped in visualizing and understanding the formation of the effluent plume pattern, and its

shifting with tidal reversals in coastal currents. This work can now be taken further to develop an

understanding of near-shore water quality for actual field conditions, where usually several

wastewater outfalls, with different effluent loadings and discharge rates, operate simultaneously.

Page 45: NIO final report

44

References

Babu, M.T., Vethamony, P., & Desa, E. (2005). Modelling tide-driven currents and residual eddies in

the Gulf of Kachchh and their seasonal variability: A marine environmental planning perspective,

Ecological Modelling 184: 299โ€“312.

Babu, M.T., Das, V.K., & Vethamony, P. (2006). BODโ€“DO modeling and water quality analysis of a

waste water outfall off Kochi, west coast of India, Environment International 32: 165-173.

Behrenfeld, M.J., & Falkowski, P.G. (1997). Photosynthetic rates derived from satellite-based

chlorophyll concentration, Limnology and Oceanography, Vol. 42, No. 1: 1-20.

Bentamy, A., & Croizรฉ-Fillon, D. (2006). Near real time blended surface wind characteristics,

Mercator Newsletter, No. 22, <http://www.mercator-ocean.fr/documents/lettre/lettre_22_en.pdf>,

accessed 19th July, 2010.

CERSAT (2010). <http://cersat.ifremer.fr/data/discovery/by_product_type/gridded_products/mwf_

blended>, accessed 19th July, 2010.

Cushman-Roisin, B., & Beckers, J.M. (2010). Introduction to Geophysical Fluid Dynamics: Physical

and Numerical Aspects, Academic Press, New York.

DHI (2009a). MIKE 21 Flow Model: HD Module, User Guide and Reference Manual, Danish

Hydraulic Institute, Horsholm, Denmark.

DHI (2009b). MIKE 21 Flow Model: HD Module, Scientific Documentation, Danish Hydraulic

Institute, Horsholm, Denmark.

DHI (2009c). MIKE 21 Flow Model: AD Module, Scientific Documentation, Danish Hydraulic

Institute, Horsholm, Denmark.

DHI (2009d). MIKE Zero: Creating 2D Bathymetries, Bathymetry Editor and Mesh Generator:

Scientific Documentation, Danish Hydraulic Institute, Horsholm, Denmark.

DHI (2009e). MIKE 21 Flow Model: Ecolab Module, User Guide, Danish Hydraulic Institute,

Horsholm, Denmark.

DHI (2009f). MIKE 21 Ecolab: Water Quality templates, Scientific Documentation, Danish Hydraulic

Institute, Horsholm, Denmark.

Dickson, M.L., Orchardo, J., Barber, R.T., Marra, J., McCarthy, J.J., & Sambrotto, R.N. (2001).

Production and respiration rates in the Arabian Sea during the 1995 Northeast and Southwest

Monsoons, Deep-Sea Research II 48: 1199-1230.

Page 46: NIO final report

45

Donelan, M.A., Haus, B.K., Reul, N., Plant, W.J., Stiassnie, M., Graber, H.C., Brown, O.B., &

Saltzman, E.S. (2004). On the limiting aerodynamic roughness of the ocean in very strong winds,

Geophysical Research Letters 31: L18306.

Gupta, I., Dhage, S., Jacob, N., Navada, S.V., & Kumar, R. (2006). Calibration and validation of far

field dilution models for outfall at Worli, Mumbai, Environmental Monitoring and Assessment 114:

199-209.

Gupta, I., Dhage, S., & Kumar, R. (2009). Study of variations in water quality of Mumbai coast

through multivariate analysis techniques, Indian Journal of Marine Sciences, Vol. 38(2), 170-177.

Haidvogel, D.B., & Beckmann, A. (1998). Numerical models of the coastal ocean. In The Sea,

Volume 10. Edited by K. H. Brink and A. R. Robinson. 457โ€“482. New York: John Wiley and Sons,

Inc.

Hemond, H.F., & Fechner-Levy, E.J. (1999). Chemical fate and transport in the environment. 2nd

ed. Academic Press, New York.

Hunt, C.D., Mansfield, A.D., Mickelson, M.J., Albro, C.S., Geyer, W.R., & Roberts, P.J.W. (2010).

Plume tracking and dilution of effluent from the Boston sewage outfall, Marine Environmental

Research 70: 150-161.

James, I.D. (2002). Modelling pollution dispersion, the ecosystem and water quality in coastal

waters: a review, Environmental Modelling & Software 17: 363-385.

Kumar, R., Subramaniam, J., & Patil, D. (2000). Water quality modeling of municipal discharges

from sea outfalls, Environmental Monitoring and Assessment 62: 119-132.

Lambkin, D. (2010). A Review of the Bed Roughness Variable in MIKE 21 FLOW MODEL FM,

Hydrodynamic (HD) and Sediment Transport (ST) modules,

<http://faq.dhigroup.com/images/FAQ164/Bottom_roughness_parameter_study.pdf>, accessed 19th

July, 2010.

Metcalf and Eddy, Inc. (1991). Wastewater Engineering: Treatment, Disposal, and Reuse. 3rd ed.

McGraw-Hill, New York.

MISST. (2010). Multi-sensor Improved Sea Surface Temperatures (MISST) for the Global Ocean

Data Assimilation Experiment (GODAE), <http://www.misst.org/>, accessed 19th July, 2010.

Moon, I.J., Ginis, I., & Hara, T. (2004). Effect of Surface Waves on Airโ€“Sea Momentum Exchange.

Part II: Behavior of Drag Coefficient under Tropical Cyclones, Journal of the Atmospheric sciences

61: 2334-2348.

Page 47: NIO final report

46

NEERI (1995a). Final Report, Bombay Sewage Disposal Project: Marine Outfalls, <http://www.world

bank.org>, accessed 19th July, 2010.

NEERI (1995a). Executive Summary, Bombay Sewage Disposal Project: Marine Outfalls,

<http://www.worldbank.org>, accessed 19th July, 2010.

Powell, M.D., Vickery, P.J., & Reinhold, T.A. (2003). Reduced drag coefficient for high wind speeds

in tropical cyclones, Nature 422: 279-283.

Sasamal, S.K., Rao, K.H., & Suryavansi, U.M. (2007). Sewage and industrial pollution in and

around Thane Creek, Mumbai using high resolution IRS data, International Journal of Remote

Sensing, Vol. 28, No. 19, 4391-4395.

Sawant, S.S., Prabhudessai, L., & Venkat, K. (2007). Eutrophication status of marine environment

of Mumbai and Jawaharlal Nehru ports, Environmental Monitoring and Assessment 127: 283โ€“291

Sharif, J.K.K. (2009). Numerical simulation of flow off Ratnagiri, west coast of India, M.Sc. Thesis

dissertation, Department of Physical Oceanography, School of Ocean Science and Technology,

Cochin University of Science and Technology, Cochin.

Stewart, R.H. (2008). โ€œAtmospheric Influencesโ€, Introduction to Physical Oceanography, p. 48,

<http://oceanworld.tamu.edu/home/course_book.htm>, accessed 19th July 2010.

Suresh, T., Naik, P., Bandishte, M., Deas, E., Mascaranhas, A., & Matondkar, S.G.P (2006). Secchi

Depth Analysis Using Bio-Optical Parameters Measured In The Arabian Sea, Proc. of SPIE Vol.

6406 64061Q-1.

Swamee, P.K., & Ojha, C.S.P. (1991). Modeling of BOD exertion curve, Water Research, Vol. 25,

No. 7, 901-902.

Tomczak, M., & Godfrey, J.S., 1994. Regional Oceanography: An Introduction, Pergamon Press,

New York.

Thomann, R., & Mueller, V. J. A. (1987). Principles of surface water quality modeling and control.

New York, NY: Harper and Row.

Vijay, R., Sardar, V.K., Dhage, S.S., Kelkar, P.S., & Gupta, A. (2010). Hydrodynamic assessment of

sewage impact on water quality of Malad Creek, Mumbai, India, Environmental Monitoring and

Assessment 165: 559โ€“571.

Vinodkumar, K. (2010). Simulation of hydrodynamics at the central west coast of India during

cyclone Phyan, M.Sc. Thesis dissertation, Department of Physical Oceanography, School of Ocean

Science and Technology, Cochin University of Science and Technology, Cochin.

Page 48: NIO final report

47

Vyas, L., & Vyas, S. (2007). Simulation Models for the dispersion of sewage outfalls along the west

coast of Mumbai, India, Proc. 7th WSEAS International Conference on Simulation, Modelling and

Optimization, Beijing, China, 502-507.