Provided for non-commercial research and educational use only. Not for reproduction or distribution or commercial use.

This article was originally published by IWA Publishing. IWA Publishing recognizes the retention of the right by the author(s) to photocopy or make single electronic

copies of the paper for their own personal use, including for their own classroom use, or the personal use of colleagues, provided the copies are not offered for sale and

are not distributed in a systematic way outside of their employing institution.

Please note that you are not permitted to post the IWA Publishing PDF version of your paper on your own website or your institution’s website or repository.

Please direct any queries regarding use or permissions to wst@iwap.co.uk

2210 © IWA Publishing 2013 Water Science & Technology | 67.10 | 2013

Dynamic coupling of near field and far field models

for simulating effluent discharges

Robin Morelissen, Theo van der Kaaij and Tobias Bleninger

ABSTRACT

In many cases, (processed) wastewater or thermal effluents are discharged into the marine

environment, rivers or lakes. To accurately determine the dispersion, recirculation and

environmental impacts of outfall plumes, it is important to be able to model the different

characteristics of the outfall plume in detail – from the near field (metres around the outfall) to the far

field (up to kilometres away). The solution for engineering practice is to combine different types of

models (near and far field models) that each focus on specific scales, with corresponding optimised

resolutions and processes. However, to adequately describe the hydrodynamic processes on these

different scales, it is essential to couple these models in a dynamic and comprehensive way. To

achieve this, a dynamic coupling between the open-source Delft3D-FLOW far field model and the

CORMIX near field expert system is proposed. This coupled modelling system is able to use the

computed far field ambient conditions in the near field computations and, conversely, to use the

initial near field dilution and mixing behaviour in the far field model. Preliminary results are presented

to provide a first indication of the potential of the method for modelling the complete trajectory of

effluent outfall plumes, allowing an accurate assessment of the environmental effects and the design

of possible mitigating measures.

doi: 10.2166/wst.2013.081

Robin Morelissen (corresponding author)Theo van der KaaijHydraulic Engineering Department,Deltares,Rotterdamseweg 185,2629 HD Delft,The NetherlandsE-mail: Robin.Morelissen@deltares.nl

Tobias BleningerEnvironmental Engineering Department,Federal University of Paraná,Brazil

Key words | dynamic model coupling, far field model, near field model, outfall plume

INTRODUCTION

General background

In many cases, wastewater (e.g., sewage) or thermal coolingwater effluents are discharged into the marine environment,rivers or lakes. Obviously, effects of such discharges on the

environment should be as small as possible and the naturalsystem should be able to cope with them. It is thereforeimportant to be able to assess the expected behaviour (i.e.,mixing, spreading and dilution) of the wastewater outfall

plume resulting from such discharges, under the influenceof currents, density differences and other processes.

In order to accurately determine the recirculation,

spreading and resulting environmental impacts of outfallplumes, it is important to be able to model the character-istics of the outfall plume in detail at various mixing

stages. This is particularly the case in weak dynamic systems(i.e., low ambient flow velocities), such as calm coastalwaters, within estuaries or lakes. Those cases are evenmore critical for discharges of large volumes, such as

thermal discharges, where the discharge-induced flowsmay considerably influence the ambient flows (e.g., coastal

circulation), whereas this effect is usually of less importancefor wastewater discharges with smaller volume flows.

The processes dominating the plume dynamics occur on

significantly different spatial and temporal scales and aretypically characterised by three zones defined along theplume trajectory: (1) initial, active mixing zone, the nearfield; (2) intermediate zone (or mid field); and (3) passive

mixing zone, far field. The near field is a region where theoutflow characteristics (i.e., fluxes, geometry and orientationof outflow) dominate the plume behaviour. The far field

region is where the ambient flow conditions dominate thebehaviour of the plume. The intermediate zone is the tran-sition region from the near field to far field processes. Due

to the large differences in scales and processes involved,different types of models are typically utilised for the simu-lation of near field and far field processes. Existing modelsthat can theoretically cover this entire range of temporal

2211 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

and spatial scales in one integral computation (unsteady,

baroclinic, non-hydrostatic models) are very computation-ally expensive and are not yet usable for most practicalengineering applications.

Offline coupling approach

To be able to model the entire trajectory of the plume from

the initial metres to its effects several kilometres away, sep-arate near and far field models must be used in combination.To describe near field behaviour of a plume, expert systems

like CORMIX (Jirka et al. ) or jet-integral models likeCorJet (part of CORMIX, Jirka , ), VISJET (Leeet al. ) or Jet3D (Delvigne ) are often applied. To

assess the far field behaviour of a discharge plume, hydro-static far field simulation programs like Delft3D-FLOW(Lesser et al. ) or POM (Blumberg & Mellor ) are

used. The most common engineering approach is to ‘trans-late’ results of near field simulations into input (sources offresher/brine/warm water) for the far field model. Differentmethods exist to couple these models, though mostly offline,

i.e., not dynamic and one-way, and often in an arbitrary way.A change in ambient conditions as a result of the near fieldmixing of the plume is not accounted for. In addition, the

dynamics of the ambient conditions are only representedin a very limited way in the near field model. A first steptowards a more sophisticated, but still one-way and offline,

coupling between a near and far field model (CORMIXand Delft3D) was developed by Bleninger ().

Online coupling approach

To represent near field behaviour in a far field simulation,the far field three-dimensional (3D), baroclinic, hydrostatic

simulation program Delft3D-FLOW is coupled online anddynamically to different near field models. In this context,online means that during a far field simulation the compu-

tational results are used to define the input for the nearfield simulation and that near field results are used in thefar field simulation. Dynamically means that transferring

near field results to the far field simulation and vice versaoccurs at a time interval sufficiently small to account for achange in ambient conditions. With this type of coupling,near field effects will be accounted for as much as possible

in the far field model and changes in the far field will befed back to the near field.

Initially, a dynamic coupling between Delft3D-FLOW

(far field) was set up with the near field jet-integral modelJet3D and the CorJet module of the CORMIX system. The

jet-integral near field models are, however, often restricted

to single port diffuser type discharges (except CorJet) and,more critical for this study, assume unlimited ambientwaters, thus do not include boundary interaction (surface

or bed) of the plume. Therefore, subsequently, an onlinecoupling between Delft3D and the CORMIX expert systemis presently being developed. CORMIX is a much morecomprehensive system that includes different outfall con-

figurations and flow classes and can assess boundaryinteractions of the plume, resulting in a more realistic coup-ling between the near and far field models.

In this paper, first the different near and far field modelsare described, followed by the description of the coupledmodelling system. The first, preliminary validation results

are subsequently presented followed by the conclusionsand discussion.

COUPLING CORMIX TO DELFT3D-FLOW

The development of the coupling is presented in two steps:first, coupling a near field, jet-integral model to the farfield Delft3D model online and, second, coupling the

CORMIX system to Delft3D online. In this online coupling,the dilution (by entrainment of ambient water) of the jet inthe near field region is accounted for in the far field model

by using the Distributed Entrainment Sinks Approach(DESA) (Choi & Lee ). By following this approach,the near field behaviour of the plume is represented as accu-rately as possible in the far field model, which is particularly

important in a weakly dynamic environment.

The Delft3D-FLOW far field model

The hydrodynamic module Delft3D-FLOW simulates two-dimensional (2D, depth-averaged) or 3D unsteady flow andtransport phenomena resulting from tidal and/or meteorolo-

gical forcing, including the effect of density differences dueto a non-uniform temperature and salinity distribution(density-driven flow). The flow model can be used to predict

the flow in shallow seas, coastal areas, estuaries, lagoons,rivers and lakes. Delft3D-FLOW solves the unsteady shallowwater equations in two (depth-averaged) or in three dimen-sions. The system of equations consists of the horizontal

equations of motion and the continuity equation. Besidesthe hydrodynamic equations, the computational core alsoincludes an advection/dispersion equation describing the

transport of substances, including heat and/or salinity. Theimpact of horizontal density differences resulting from

2212 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

salinity and/or temperature differences is taken into account

in the hydrodynamic part of the simulation program. Theequations are presented in Lesser et al. () and are omittedin this paper for the sake of readability. One of the assump-

tions in deriving the shallow water equations is that verticalaccelerations are small compared to gravity. This results in avertical momentum equation that reduces to the well-knownhydrostatic pressure assumption. A drawback of this hydro-

static assumption is that the near field region of the plume isnot described correctly since vertical accelerations resultingfrom buoyancy differences are not accounted for.

Vertical exchange of momentum and matter followsfrom a k–ε turbulence closure scheme (Rodi ). Two sep-arate equations for the transport of turbulent kinetic energy

(k) and the dissipation of turbulent kinetic energy (ε) aresolved. From k and ε vertical viscosity and diffusivity arecomputed.

Delft3D-FLOW is being used extensively in consultancy

outfall studies regarding the large-scale (environmental)impact of wastewater discharges, heat discharged bypower plants or brine discharged by desalination plants.

The near field outfall plume behaviour is traditionally deter-mined by performing a number of offline near fieldsimulations and ‘translating’ the results (often arbitrarily)

into sources for the far field model.The Delft3D-FLOW model is now open-source, which

facilitates the development and use of the model.

The CORMIX expert system

CORMIX (Jirka et al. ) is a USEnvironmental Protection

Agency supported mixing zone model and decision supportsystem for environmental impact assessment of regulatorymixing zones resulting from continuous point source dis-

charges. The system emphasises the role of boundaryinteraction to predict steady-state mixing behaviour andplume geometry. The CORMIX methodology contains sys-

tems to model single-port, multiport diffuser discharges andsurface discharge sources. Effluents considered may be con-servative, non-conservative, heated, brine discharges or

contain suspended sediments. CORMIX contains a rigorousflow classification scheme developed to classify a given dis-charge/environment interaction into one of several flowclasses with distinct hydrodynamic features (Doneker &

Jirka ). Furthermore, it contains a jet-integral model(CorJet), and is capable of assessing plume boundary inter-actions and subsequent spreading of the plume in the

‘intermediate zone’, based on analytical and empiricalformulations.

CORMIX’s jet-integral model is based on the conserva-

tion principles for volume (continuity), momentumcomponents in the global directions x, y and z, state par-ameters, and scalar mass. The equations for a typical

buoyant jet with self-similar Gaussian type cross-sectionaldistributions are presented in Jirka () and have beenomitted in this paper for the sake of readability. Theseequations are solved along the jet trajectory in 3D within

the CORMIX expert system for a given ambient flow anddensity field, as well as a given discharge structure. Inaddition to other jet models, CORMIX includes furthermore

length-scale based classification systems combined withsimilar integral approach solutions to account for boundaryinteractions, buoyant spreading motions and near field

instabilities (Jirka , ).Since CorJet (as other jet models) are restricted to

unlimited ambient water, thus do not include boundaryinteraction (surface or bed) of the plume, an online coupling

between Delft3D and the CORMIX expert system (Doneker& Jirka ; Bleninger & Jirka ) is being developed.

Based on discharge characteristics (initial discharge

amount, initial density of the discharge), outfall character-istics (orientation of the outfall nozzle, diameter of thediffuser nozzle) and ambient conditions (vertical distri-

bution of ambient horizontal velocity, vertical distributionof ambient density), CORMIX computes different character-istics along the jet trajectory, including:

• the plume centre line trajectory (x-,y-,z-coordinates);

• a typical plume width, related to the assumed Gaussian

velocity and concentration distribution;

• the plume centre line jet velocity;

• the plume centre line dilution.

These output quantities need to be translated to ‘input’for the far field simulation in such a way that the impactof near field behaviour of a plume is accounted for in the

far field simulation as much as possible.

The coupled modelling system

The coupling of Delft3D-FLOW with CORMIX combinesthe benefits and efficiency of both models, which thereforeresults in a more comprehensive coupled modelling systemthat can also handle boundary interactions and provides a

more flexible and physically more appropriate location forthe coupling.

The development of the dynamic, two-way (online)

coupling between CORMIX and Delft3D-FLOW is carriedout in a collaboration between MixZon (developers of

2213 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

CORMIX), Deltares (developers of Delft3D) and the Federal

University of Paraná (Tobias Bleninger, initiator of the coup-ling scheme). In both modelling systems adaptations weremade to be able to carry out autonomous simulations of

the online coupled near and far field models, which are pre-sently under further development.

The coupling approach is presented below in the form ofthe two main steps in which the coupling is developed. The

first step presents the coupling method that is used for thenear and far field models, initially applied for the develop-ment of an online coupling of jet-integral near field models

to Delft3D. Theses cases were used to verify numerical stab-ility and to validate the coupling scheme. The second stepshows the coupling of the more comprehensive CORMIX

system to Delft3D by means of the same coupling method,but then including boundary interactions, and releasingthe limitations of jet-integral models.

Step 1 – coupling a jet-integral near field model toDelft3D-FLOW

In the first step, the dynamic coupling was developed for asteady-state, jet-integral near field model, such as Jet3D orCorJet (one of the modules in CORMIX), which are typically

restricted to single port diffuser type discharges (exceptCorJet) and do not include boundary interaction (surfaceor bed) of the plume. The online and dynamic coupling

between Jet3D/CorJet and Delft3D-FLOW was developedbased on the DESA by Choi & Lee (). They rightfullystate that traditional one-way ‘source only’ couplingmethods do not account for near field mixing induced by

plumes. To account for this mixing, they propose to positiona number of sinks in the far field hydrostatic model alongthe plume centre line trajectory, which is computed by

also taking into account the ambient conditions (e.g., cur-rents, stratification). The magnitude of each sink shouldcorrespond to entrainment of the plume in the spatial sec-

tion it represents. At the predicted near field terminal levelof the plume, a source corresponding with the dilutedplume is introduced. Due to entrainment of ambient water

into the plume, the discharge of the diluted source is anorder of magnitude larger than the original source and theconcentrations (pollutant, salt, temperature) are correspond-ingly lower. The discharged ‘mass’ at the predicted near field

terminal level of the plume is conserved and corresponds tothe mass discharged at the outfall point (plus the potentiallyentrained mass in case of e.g., temperature and salinity). A

more elaborate description of this method can be found inChoi & Lee () and in Morelissen et al. ().

In the implementation of this coupling method in

Delft3D-FLOW, a wide array of potential sinks andsources are defined in the horizontal grid cells and verti-cal layers in the Delft3D-FLOW model around the outfall

location. The coupling procedure determines the plumetrajectory in the far field model domain and assigns themagnitudes of the sinks and sources in a time dependentway to the appropriate cells. The unused defined sinks

and sources in the Delft3D-FLOW model will have a dis-charge equal to zero but will be activated when theplume enters those cells. This allows the outfall plume

to move in the far field domain under the influence ofe.g., tidal forcing or wind or density changes withoutthe need to define new sink/source terms every coupling

time step.

Coupling time interval and location

The coupling time interval is predefined in the Delft3Dmodel input files and is based on expert judgement and bytaking into account the variability in the ambient system.

The time interval of the coupling should correspond to thetime-scale of the expected variations in the far field. It isfurther noted that this coupling time step will also depend

on the actual ‘travel time’ of the plume to reach the couplinglocation. An expert system-like flow classification approach(as included in the CORMIX system) is currently under

development to improve the definition of characteristic tem-poral and spatial scales, based on the length-scaleclassification scheme (Bleninger et al. ) and proposedscales by Bleninger ().

The location of coupling is defined at the ‘end of thenear field’. Since the definitions of ‘near field’ and ‘farfield’ are somewhat ambiguous, the actual location of the

coupling is also not fixed. In general with the use of jet-inte-gral models, the end of near field is defined as either (1) thelocation where a boundary interaction occurs (plume reach-

ing bed or surface), since these models do not include thistype of interaction or (2) when the momentum of theplume itself is smaller than the momentum of the ambient

flow. It is noted that especially for the boundary interactioncases, a transition between the near and far field exists thatcan be important in the spreading of the plume. In jet-inte-gral models, this interaction, e.g., the lateral spreading of a

thermal plume along the water surface, is typically notincluded and assumptions need to be made for this inter-mediate zone, and the plume behaviour therein, for the

proper coupling of the near and far field models (seeFigure 1).

Figure 1 | Typical plume behaviour including boundary interaction and lateral spreading along the water surface. The typical and preferred locations for coupling the near and far field are

indicated.

2214 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

Furthermore, in order to avoid direct numerical dilutionin the far field model at the location of the coupling, the

selection of the coupling location should also take intoaccount the local far field grid cell size. This resolutionshould be able to resolve the dimensions of the plume at

that location.The optimal coupling location and coupling time step can

thus be determined dynamically based on the physical prop-erties of the plume, the local far field model grid size, and

the associated travelling time of the plume to that location.The refinement of this method and implementation in thecoupled modelling system will be part of future research.

Verification and validation of the implementedmethodology

It is noted that fresh wastewater discharges in a salineenvironment (i.e., coastal area) show similar behaviour

(i.e., rising plume) as thermal discharges (e.g., coolingwater) and brine discharges (e.g., from desalinationplants), although the latter shows a sinking plume instead.

Due to the availability of measurement data, some of thecoupling tests presented in this article relate to a thermal dis-charge and some to a freshwater discharge.

To demonstrate the impact of the coupling method-ology, a schematic case was modelled consisting of a largerectangular tank of stagnant water with a depth of 5 m.

The water within the tank is saline (31 psu) with a tempera-ture of 15 WC. Hence, the density of the water within the tankis 1,021.7 kg/m3. Near the bed in the centre of the tank, athermal jet was injected through a horizontal pipe with a

diameter of 2 m. A discharge of 5 m3/s, a salinity of 31 psuand a temperature of 25 WC, a temperature increase relativeto the water within the tank of 10 WC, was applied. The com-

bination of temperature and salinity of the discharged waterresults in a density of 1,018.7 kg/m3.

The jet was schematised in the far field model of thetank in two ways: (1) the traditional, typical method of coup-

ling by inserting the (undiluted) source at the water surfaceand (2) by means of the DESA method and thus includingsinks along the (vertical) jet trajectory and discharging the

diluted source at the surface (i.e., end of the near field).The purpose of this test case is to demonstrate the impactof the coupling methodology (proof-of-concept).

It is noted that in the first method, the source was

included as an undiluted source in order to be mass conser-ving. If this source would have been implemented as adiluted source (e.g., based on offline jet computations), the

mass would not be conserved since the entrainment of ambi-ent water that is responsible for this dilution is not includedin this method of coupling. Figure 2 shows the resulting flow

patterns in a vertical cross section through the tank. The leftplot shows the results of the traditional method of couplingthe near and far field model with the undiluted source

inserted at the surface and the right plot shows the resultsof the DESA coupling method. The difference in flow pat-terns can clearly be observed and the right plot with theDESA coupling method shows a much more realistic flow

pattern, considering the jet trajectory, in which verticalflow cells are formed due to the momentum and entrain-ment of the jet.

The first validation case of the Jet3D–Delft3D couplingagainst measurements consists of the reproduction of a lab-oratory test in which the effect of a freshwater jet on

stratification in a stagnant saline water body is tested. Thesemeasurements were carried out by Eysink (). In thistest, a linear density profile was applied to the test tank andsubsequently a freshwater jet was inserted from the bed

(Δρ¼ 12 kg/m3,Doutfall¼ 0.01 m,U0¼ 0.49 m/s). The densityprofile was measured again after 400 s, at the end of the test.

Along its trajectory, the jet entrains more dense water

from the ambience resulting in an increase in density ofthe jet. When the density of the jet approximately equals

Figure 2 | Comparison of flow fields (vertical cross section) in a schematic tank between the traditional, typical method of coupling (left) and the DESA coupling (right). The dashed vertical

arrow in the centre shows the jet centre line that is schematised with the two different methods in the model.

2215 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

the ambient density, the jet reaches its ‘terminal’ height andthe discharged water spreads horizontally. The density pro-file below the terminal point results from the entrainment

of the plume. Above the terminal point the density profileresults from the horizontal spreading and resulting flow pat-terns. As such, the test demonstrates the impact of the jet

characteristics (near field) on ambient conditions (far field).The laboratory test was modelled with the coupled

Jet3D–Delft3D models and compared to the measurements.The Delft3D model consists of 20 equidistant vertical layers

and has a typical horizontal grid size of 0.02 m. The coup-ling time step between Jet3D and Delft3D was set to 0.6 sand the Delft3D-FLOW model used a time step of 0.012 s.

The simulations were carried out using the two differentcoupling methods: (1) traditional method by inserting thesource at the end of the near field as computed by Jet3D

(note that this is about halfway in the water column due tothe linear density profile) and (2) by using the DESAmethod. The results are presented in Figure 3.

These results show that in the case of the DESA coup-ling method (right plot) there is a fair agreement betweenthe computed density profile after 400 s and the measureddensity profile. Compared to the results of the computation

without the online dynamic coupling (left plot), the coupledDESA method shows significantly better results.

In addition, another laboratory case was tested. The

experiment was performed by the authors using an existingexperimental set-up, which was used to study the impact ofcombined water/air jets on the salinity exchange when

opening a sluice separating fresh and saline water. Onlyone single experiment was conducted.

This flume experiment consisted of a discharge of fresh

water in a small flume with stagnant, saline water, in whichover time the salinity was monitored at different depths.The flume in this experiment was 14 m long, 0.5 m wide

and the water depth was 0.29 m. The outfall nozzle (diameter0.015 m, discharge 0.012 l/s on average) was positioned8.2 m from the far right side at the bottom of the flume andthe measurements were taken 0.8 m left of the nozzle, both

in the centre of the flume. The initial ambient salinity was20.5 psu uniform over depth and the discharged water 0 psu.

This experiment was subsequently modelled with a 3D

Delft3D-FLOW model in two ways (similar to the simu-lations in Figure 2): (1) by inserting the freshwater sourceundiluted at the surface (again to conserve mass) and (2)

by using the online, dynamically coupled Jet3D–Delft3Dmodel following the DESA method. The Delft3D modelconsists of 50 equidistant vertical layers and has a typical

horizontal grid size of 0.025 m. The coupling time stepbetween Jet3D and Delft3D was set to 0.06 s and theDelft3D-FLOW model used a time step of 0.06 s.

Figures 4 and 5 show the modelling results of the two

approaches in comparison with the measured salinities atdifferent depth levels over the time of the experiment. Thiscomparison shows that, although not perfect yet, the coupled

models produce a result that resembles the measurementsmuch more than the traditional modelling approach withan undiluted source at the surface. Figure 4 shows that in

Figure 3 | Comparison of the traditional modelling approach (left) and the DESA coupled Jet3D–Delft3D modelling results (right) with measurements of a freshwater jet in a saline tank with

an initially linear density profile.

2216 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

the traditional modelling approach, all fresh water remainsat the surface and does not reproduce the measured stratifi-cation pattern at all. The water at the surface has a salinitythat is much too low, while the other levels show a too

high salinity. Figure 5 shows that the eventual stratificationcharacteristics are much better reproduced, although it stillshows significant differences. The stratification in the

measurements is somewhat smaller than in the modellingresults, resulting in somewhat higher salinities at the surface(i.e., 0.015 m below the surface) and a similar salinity in the

next depth level (0.030 m below the surface). The modelresults show lower salinities in these levels and a largerdifference between these levels. On the other hand, the sal-inity at 0.060 m below the surface is computed higher than

measured. At 0.084 m below the surface (deepest level), the

Figure 4 | Freshwater discharge flume test: salinity at different depth levels when modelling t

measured and computed salinity are very similar. Further-more, the measurements show that the onset of change insalinity in the top level (indicated by the arrow) significantlyprecedes the changes at the other levels, while this is not

obvious from the model computations. This difference intiming can be caused by the initial lateral spreading of theplume along the surface (when the stratification is not yet

built up (i.e., larger density difference). This effect may beless well reproduced by the model, possibly due to missingthe simulation of boundary interaction in the near field

model, the coupling time step or other model settings. How-ever, in practical applications, usually these initial effects areof less importance than the eventual operational (or steady-state) situation. At steady state, the modelling results have

improved significantly in the online, dynamically coupled

he discharge as an undiluted source at the water surface.

Figure 5 | Freshwater discharge flume test: salinity at different depth levels when modelling the discharge with the DESA coupled Jet3D–Delft3D models.

2217 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

method compared to the traditional method, but further

improvements can still be made.

Step 2 – coupling the CORMIX near field expert systemto Delft3D-FLOW

As presented in the previous section, the inclusion of near

field entrainment effects in the far field model by means ofthe online, dynamic coupling (DESA), results in a morecredible representation of the plume than traditional

methods. However, by the use of only jet-integral models,the boundary interactions of the plume are still notdescribed well, due to the absence of this capability in

most jet-integral models. This boundary interaction is, how-ever, important for the modelling of the full plumetrajectory. Many of these boundary interactions cannot bemodelled accurately in typical far field models either, there-

fore leaving a gap in the modelling of the full plumetrajectory. In addition, these boundary interactions canplay an important role in determining a proper schematisa-

tion of the plume dimensions and characteristics in the farfield model. Therefore, a further model development is pre-sently ongoing to overcome these issues.

The CORMIX expert system (Jirka et al. ) is capableof assessing this intermediate zone (including the near fieldzone), in which these boundary interactions commonly take

place. Apart from the improvements by including the nearfield behaviour in the far field model, a coupling ofDelft3D-FLOWwith CORMIX will result in a more compre-hensive coupling that will also be able to handle boundary

interactions and provides a more flexible and physicallymore appropriate location for the coupling of near and farfield models.

The approach that is followed is a coupling betweenCORMIX to Delft3D-FLOW. It is based on the proprietary

input and output files of the CORMIX system. Recent ver-

sions of CORMIX contain the CorTime module, in which itis possible to carry out a number of CORMIX simulationsin batch, based on an input file with different model set-

tings (e.g., outfall and ambient conditions) and a basic‘basecase’ file that contains the details of the modelled out-fall. From this batch, a series of CORMIX prediction files is

generated.In this extended coupling approach, Delft3D-FLOW

writes the CorTime input file at a specified time interval,based on the simulated ambient flow conditions. The

Delft3D-FLOW simulation will pause to allow a CORMIXsimulation to run. Subsequently, Delft3D-FLOW will readand interpret the produced CORMIX prediction file and will

include the computed plume characteristics in the far fieldmodel, following the DESA method (see above). At theproper coupling location, the plume characteristics (e.g.,

dimensions, dilution) will be translated to source terms in theDelft3D-FLOWmodel, to allow for an accurate coupling.

The present status of the coupling development is that

Delft3D-FLOW is capable of writing the CorTime inputfiles and is able to read and interpret the CorTime/CORMIX prediction (output) files. The latter isimplemented for a selected number of CORMIX flow classes

and is set up in a modular way that allows for easy extensionof other flow classes in the next phase of the development.The extension of the CorTime module to allow it to wait

for the Delft3D-FLOW computation has been implemented.With these developments completed, the coupledCORMIX–Delft3D modelling system is able to run autono-

mously in dynamic interaction with each other. Thepresent coupling is operational for a limited selection ofCORMIX flow classes, but its range of applicability isbeing extended in ongoing research. At present, one of the

implemented flow cases in the coupling is a single port

2218 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

outfall with a thermal (positively buoyant) jet in a small

cross-flow, which has a boundary interaction at the watersurface (similar to the situation sketched in Figure 1, butwith an initially horizontal discharge). The initial results of

this coupled model system are presented below.A 20 km long, 600 m wide and 5 m deep channel with a

buoyant (thermal) discharge at about 10 km and locatedabout 80 m from the south bank was modelled. The

Delft3D-FLOW grid size is about 25 × 25 m near the outfalllocation and has an equidistant layer distribution of 10% per(sigma) layer. A constant cross-flow of about 0.16 m/s was

applied to the channel. There is a horizontal discharge1 m above the bed with a diameter of 1 m and a dischargerate of 2 m3/s with an excess temperature of 15 WC above

the background temperature in the channel (15 WC), i.e.,the discharge temperature is 30 WC.

The coupled simulation was run for 1 day to allow asteady-state situation to form. The coupling time step was 6

hours (i.e., every 6 hours a CORMIX simulation was carriedout with updated ambient flow conditions). It is noted thatin e.g., tidal cases, this coupling time stepwill bemuch smaller

(e.g., about 30 minutes or even less). In this simulation, thecoupling location from near field to far field was selected tobe just beyond the indicated ‘end of near field’ by CORMIX,

similar to the preferred location indicated in Figure 1. Atthis location, about 50 m from the discharge, the computeddilution and dimensions of the plume are taken from the

CORMIX results and ‘mapped’ to the appropriate cells andlayers of the Delft3D-FLOW model. In this particular case,the width of the plume (∼20 m) at that location still fallswithin one Delft3D-FLOW grid cell, but the thickness of the

plume (∼2.5 m) is divided over five layers (each layer corre-sponds to 0.5 m).

The initial results are presented as a vertical cross section

in Figure 6. This figure also shows the initial plume trajectory

Figure 6 | Initial results of a coupled CORMIX–Delft3D-FLOW computation. The contours in thi

after coupling to CORMIX beyond the near field region (coupling location at vertical d

are compared to the CORMIX results in the derived 15.5W

C contour line (white line

computed by CORMIX (black line) and the coupling location

(vertical dashed line). This figure shows that the dilutedplume is indeed released over about 2.5 m of water depth inthe far field model. The model results also show some

additional diffusion below and upstream of the defined dis-charge. This is explained by the momentum of the dischargein the far field that (presently) works in all directions.

For this stationary case, CORMIX produces a far field

estimate of centre line dilution and thickness of the plume.These CORMIX results are compared with Delft3D resultsbeyond the coupling location. The far field module of the

CORMIX model uses a top-hat concentration profile,which makes a direct comparison with Delft3D modelresults less straightforward. Therefore, the 15.5 WC contour

line was computed, based on a transformation of the top-hat profile to a Gaussian concentration distribution andtaking into account the dilution along its centre line (the sur-face in this case). This contour line is presented in Figure 6

as the white line and should be compared with the 15.5 WCDelft3D contour line. Although the contour lines of theCORMIX and Delft3D models are not at the exact same

location, the trends are very similar. The parameter settingsof the Delft3D model are still very preliminary and muchcan be improved, but these initial results are very promising.

CONCLUSIONS AND DISCUSSION

In order to accurately determine the dispersion, recircula-tion and environmental impacts of outfall plumes, it is

important to be able to model the different characteristicsof the outfall plume in detail; from the near field (smallscales, metres around the outfall) to the far field (large

scales, up to several kilometres away). To properly describethe hydrodynamic processes on these different scales, it is

s vertical cross section along the plume show the temperatures derived by Delft3D-FLOW

ashed line). The near field plume trajectory is presented in black. The Delft3D-FLOW results

).

2219 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

essential to couple near field and far field models in a

dynamic and comprehensive way.In this research, the first developments were made to

couple the near field CORMIX expert system to the far

field Delft3D-FLOW model, in order to include the plumebehaviour in the near and intermediate zone (between thenear and far field) in the coupled modelling system. Forthis coupling, the DESA method by Choi & Lee ()

was used that includes sinks along the jet trajectory in thefar field model to also represent entrainment effects in thefar field model.

The coupled modelling system was developed and testedfrom which the following conclusions can be drawn:

• The online, dynamically coupled near and far fieldmodels produce a more realistic flow field around the

jet than the traditional method of coupling with anundiluted source at the end of the near field.

• This coupled modelling approach results in a better

resemblance with measurements for stagnant, verticallystratified situations in laboratory experiments, where afreshwater jet is inserted in a saline environment, which

was tested against measurements for two laboratorycases.

• A more comprehensive modelling approach to model the

full plume trajectory is obtained by coupling Delft3D-FLOW with CORMIX, in which the plume behaviourin the intermediate zone is also included in contrast tocoupling only a jet-integral model that does not include

plume boundary interaction. This new coupling allowsfor modelling the near and far field in a physically moreaccurate way and allows for coupling at a more suitable

location along the plume trajectory.

• CORMIX is able to cover many different flow classes andis therefore applicable in many cases. The present devel-

opment of the coupling to Delft3D-FLOW results in avery comprehensive and practical modelling system forapplication in many different outfall situations.

Based on the present study, a number of points for dis-

cussion and further research were identified:

• The coupling between CORMIX and Delft3D-FLOW ispresently still under development. The initial results,based on a limited number of implemented flow classes

in this coupling, are presented in this paper. Other flowclasses need to be implemented in this coupling toextend its applicability in different practical cases.

• The results of both the coupling of jet-integral models andCORMIX to Delft3D-FLOW need to be further validated

against laboratory and field data. This is especially useful

for stagnant or weak-dynamic situations.

• The (physically) optimal location and time step for coup-ling CORMIX and Delft3D-FLOW, taking into account

the travelling time of the plume under different con-ditions, is subject to further research.

When the development of the coupling betweenCORMIX and Delft3D-FLOW is fully completed, this willresult in an accurate and comprehensive method for model-ling the complete trajectory of an effluent outfall plume,

allowing an accurate assessment of the environmentaleffects and the design of possible mitigating measures.

REFERENCES

Bleninger, T. Coupled 3D Hydrodynamic Models forSubmarine Outfalls: Environmental Hydraulic Design andControl of Multiport Diffusers. Thesis, Institute forHydromechanics, University of Karlsruhe, Karlsruhe.

Bleninger, T. & Jirka, G. H. Modelling and environmentallysound management of brine discharges from desalinationplants. Desalination 221, 585–597.

Bleninger, T., Herlina, H., Weitbrecht, V. & Socolofsky, S. Mixing and transport processes in environmental fluidsystems: Gerhard Jirka’s scientific achievements. In:Environmental Fluid Mechanics – Memorial Volume inHonour of Prof. Gerhard H. Jirka. IAHR Monograph (W.Rodi & M. Uhlmann, eds). CRC Press/Taylor & FrancisGroup, Boca Raton, FL.

Blumberg, A. F. & Mellor, G. L. A description of a three-dimensional coastal ocean circulation model. In: Three-Dimensional Coastal Ocean Models, Vol. 4 (N. Heaps, ed.).American Geophysical Union, Washington, DC, p. 208.

Choi, K. W. & Lee, H. W. Distributed entrainment sinkapproach for modeling mixing and transport in theintermediate field. J. Hydraul. Eng. 133 (7), 804–815.

Delvigne, G. Round buoyant jet with three-dimensionaltrajectory in ambient flow, Paper presented at the 18thCongress of the IAHR, Cagliari, Italy, September.

Doneker, R. L. & Jirka, G. H. CORMIX-GI systems formixing zone analysis of brine wastewater disposal.Desalination 139, 263–274.

Eysink Verticale stralen in een Medium metDichtheidsgradient (Vertical jets in a medium with a densitygradient). Waterloopkundig Laboratorium Delft, S74.

Jirka, G. H. Integral model for turbulent buoyant jets inunbounded stratified flows. Part 1: Single round jet. Environ.Fluid Mech. 4 (1), 1–56.

Jirka, G. H. Integral model for turbulent buoyant jets inunbounded stratified flows, Part 2: Plane jet dynamics resultingfrommultiport diffuser jets.Environ. FluidMech. 6 (1), 43–100.

Jirka, G. H., Doneker, R. L. &Hinton, S.W. User’sManual forCORMIX: A Hydrodynamic Mixing Zone Model and

2220 R. Morelissen et al. | Dynamic coupling of near and far field models for simulating effluent discharge Water Science & Technology | 67.10 | 2013

Decision Support System for Pollutant Discharges intoSurface Waters. US Environmental Protection Agency, Tech.Rep., Environmental Research Lab, Athens, Georgia, USA.

Lee, J. H. W., Cheung, V., Wang, W. P. & Cheung, S. K. B. Lagrangian modeling and visualization of rosette outfallplumes. Proc. Hydroinformatics University of Iowa, CedarRapids, USA, 23–27 July.

Lesser, G. R., Roelvink, J. A., van Kester, J. A. Th. M. &Stelling, G. S. Development and validation of a

three-dimensional morphological model. J. Coast. Eng. 51,883–915.

Morelissen, R., van der Kaaij, T. & Bleninger, T. DynamicCoupling Of Near field And Far field Models For SimulatingOutfalls. IDAWorld Congress 2011, Perth, Western AustraliaIDAWC/PER11-014.

Rodi, W. Turbulence Models and their Application inHydraulics – A State of the Art Review. InternationalAssociation of Hydraulic Research, Delft, The Netherlands.

First received 24 February 2012; accepted in revised form 4 January 2013