EPA-R2-73-162
MARCH 1973             Lnvironmental Protection Technology
Numerical Thermal  Plume

Model  for  Vertical Outfalls

in Shallow Water

                                  Office of Research and Monitoring

                                  U.S. Environmental Protection Agency

                                  Washington, D.C. 20460

-------
            RESEARCH REPORTING SERIES
Research reports of the  Office  of  Research  and
Monitoring,  Environmental Protection Agency, have
been grouped into five series.  These  five  broad
categories  were established to facilitate further
development  and  application   of   environmental
technology.   Elimination  of traditional grouping
was  consciously  planned  to  foster   technology
transfer   and  a  maximum  interface  in  related
fields.  The five series are:

   1.  Environmental Health Effects Research
   2.  Environmental Protection Technology
   3.  Ecological Research
   4.  Environmental Monitoring
   5.  socioeconomic Environmental Studies

This report has been assigned to the ENVIRONMENTAL
PROTECTION   TECHNOLOGY   series.    This   series
describes   research   performed  to  develop  and
demonstrate   instrumentation,    equipment    and
methodology  to  repair  or  prevent environmental
degradation from point and  non-point  sources  of
pollution.  This work provides the new or improved
technology  required for the control and treatment
of pollution sources to meet environmental quality
standards.

-------
                                                       EPA-R2-73-162V
                                                       March 1973
                 NUMERICAL THERMAL PLUME MODEL

            FOR VERTICAL OUTFALLS IN  SHALLOW WATER
                               By

                         Donald S. Trent
                         James R. Welty
                        Project 16130  DGM
                         Project Officer

                      Mostafa A. Shirazi
                Environmental Protection Agency
            National Environmental Research Center
                    Corvallis, Oregon 97330
                          Prepared  for

               OFFICE  OF RESEARCH  AND MONITORING
             U.S. ENVIRONMENTAL PROTECTION AGENCY
                    WASHINGTON, D.C.  20460
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402
              Price $4.80 domestic postpaid or $4.26 OPO Bookstore

-------
                          EPA Review Notice
This report has been reviewed by the Environmental Protection Agency
and approved for publication.  Approval does not signify that the
contents necessarily reflect the views and policies of the Environ-
mental Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
                                  ii

-------
                            ABSTRACT







      A theoretical study of the heat and  momentum transfer resulting




from a flow of power plant condenser effluent discharged vertically to




shallow, quiescent coastal receiving water is presented.  The complete




partial differential equations governing steady,  incompressible,  tur-




bulent flow driven by both initial momentum  and buoyancy are solved




using finite-difference techniques to obtain temperature and velocity




distributions  in the near field of the thermal discharge.




      Turbulent quantities were treated through the use of Reynolds




stresses with further simplification utilizing the concept of eddy




diffusivities  computed by Prandtl's mixing length theory.  A Richardson




number correlation was  used to account for the  effects of density




gradients on  the computed diffusivities.




      Results were obtained for over  100  cases,  66 of which are reported,




using the computer program presented in this manuscript.   These  results




ranged from  cases of pure buoyancy to pure momentum and for  receiv-




ing "water depths from 1  to 80 discharge diameters deep.  Various  com-




puted gross aspects of the flow were compared to published data and




found to  be in excellent agreement.   Data for  shallow water plumes




and the ensuing lateral spread  are not readily available; however,  one




computed surface  temperature distribution was  compared to proprietary




data and found also to be in  reasonable agreement.




      This report was submitted in fulfillment of Grant No.  16130-DGM




between  the Environmental Protection Agency and the Department of




Mechanical Engineering,  Oregon State University.






                                  ill

-------
                         TABLE OF CONTENTS
                                                                  Page

CHAPTER 1.  INTRODUCTION                                            1

       1.1  Objectives                                              3

       1.2  Summary                                                 4

CHAPTER 2.  DISCUSSION OF THERMAL PLUMES AND PROBLEM
            DESCRIPTION                                             7

       2.1  The Nature of Thermal Plumes in Marine
            Surroundings                                            7

            2.1.1  Discharge Magnitude                              8

            2.1.2  Outfall Configuration                           10

            2.1.3  Hydrodynamic Regimes                            12

            2.1.4  Oceanographic Effects                           15

                   2.1.4.1  Density Stratification                 17

                   2.1.4.2  Effect of Currents                     18

                   2.1.4.3  Ocean Turbulence                       21

                   2.1.4.4  Air-Sea Interactions                   22

       2.2  Plume Analysis State-of-the-Art                        23

            2.2.1  Submerged Outfalls                              25

            2.2.2  Horizontal  Shoreline Outfalls                   27

       2.3  Work Description                                       28

CHAPTER 3.  TRANSPORT EQUATIONS - GENERAL THEORY                   30

       3.1  Coordinate System                                      30

       3.2  Conservation Laws                                      31

            3.2.1  Continuity                                      33

            3.2.2  The Equations of Motion for Turbulent Flow      34

       3.3  The Boussinesq Approximation                           36

-------
       3.4  The Pressure Equation                                  39
       3.5  r Transport                                            40
            3.5.1   Transport of Heat, Salinity and Buoyancy        40
       3.6  The Equation of State for Sea Water                    45
       3.7  Vorticity Transport                                    47
       3.8  Non-dimensional Form of the Equations of Motion        50
       3.9  Further Comments on the Concept of "Eddy Viscosity"    53
       3.10 Two-Dimensional Forms of the Transport Equations
            in Rectangular and Axisymmetric Coordinates            56
            3.10.1  Two-Dimensional Transport Equations in
                   Rectangular Geometry                            58
            3.10.2 Two-Dimensional Transport Equations in
                   Axisymmetric Coordinates                        62
CHAPTER 4.  PLUME THEORY - SIMILARITY SOLUTIONS                    65
       4.1  General Description                                    65
       4.2  Simplified Equations for a Vertical Plume              67
       4.3  Radial  Velocity and Temperature Profiles               71
            4.3.1   Zone of Established Flow                        71
            4.3.2  Zone of Flow Establishment                      74
       4.4  Zone of Flow Establishment                             76
       4.5  Governing Differential Equations                       77
            4.5.1   Initial Condition                               80
            4.5.2  Evaluation of Terms Involving K and x           81
            4.5.3  Homogeneous Receiving Water                     82
       4.6  Lateral Velocity, u                                    83
CHAPTER 5.  FINITE DIFFERENCE MODELS                               84
       5.1  Physical System for the Vertical Round Port            85
                               VI

-------
       5.2  Governing Differential Equations                       87

       5.3  Vorticity Equations                                    90

       5.4  Dimensionless Forms                                    92

       5.5  Coordinate Transformation                              94

       5.6  Finite Difference Grid System                          96

       5.7  Difference Equations                                  TOO

            5.7.1  Stream Function and Velocity                   100

            5.7.2  Transport Equations                            103

            5.7.3  Summary of Required Difference Equations       108

       5.8  Boundary Conditions                                   109

       5.9  Rectangular Coordinates                               132

            5.9.1  Governing Differential Equations               132

            5.9.2  Rectangular Difference Equations               135

            5.9.3  Rectangular Boundary Conditions                138

CHAPTER 6.  CODE DESCRIPTION AND ORGANIZATION                     141

       6.1  Computational Procedure

       6.2  Executive Program and Subroutine Description          143

       6.3  Flow Charts                                           151

CHAPTER 7.  CODE VERIFICATION AND NUMERICAL EXPERIMENTS           163

       7.1  Deep Water Plumes                                     164

            7.1.1  The Momentum Jet                               171

                   7-1.1.1  Centerline Velocity and Concentra-
                            tion for Momentum Jets                172

                   7.1.1.2  Spread of the Momentum Jet            189

                   7.1.1.3  Radial Distribution of Vertical
                            Velocity, Concentrations and
                            Vorticity for the Momentum Jet        192

                               vii

-------
                   7.1.1.4  Distribution of Radial  Velocity
                            for the Momentum Jet                  201
                   7.1.1.5  Typical Contours and Three-
                            Dimensional  Plots for a Momentum
                            Jet                                   207
            7.1.2  Two Cases of Pure Buoyancy                     207
                   7.1.2.1  Center!ine Velocity and Temperature   207
                   7.1.2.2  Spread of the Pure Buoyant Plume      218
                   7.1.2.3  Radial Distribution of Vertical
                            Velocity, Temperature and  Vorticity
                            for Pure Buoyancy                     220
                   7.1.2.4  Radial Velocity and Entrainment
                            for Pure Buoyancy                     228
            7.1.3  Mixed Flow - Forced Plumes                     237
                   7.1.3.1  Center!ine Velocity and Temperature
                            for Forced Plumes                     237
                   7-1.3.2  Rate of Spread and Entrainment        251
       7.2  Transport Coefficients                                253
            7.2.1  The Radial  Transport Coefficient, er           266
            7.2.2  The Vertical Transport Coefficient, EZ         284
       7.3  Numerical Stability and Convergence                   298
            7.3.1  Numerical Stability                            299
            7.3.2  Convergence                                    304
CHAPTER 8.  NUMERICAL EXPERIMENTS FOR SHALLOW WATER CASES         319
       8.1  Modeling the Vertical Eddy Diffisivity Multiplier,
            FZ                                                    319
       8.2  Results for Homogeneous Receiving Water 10 Port
            Diameters Deep                                        325
       8.3  Results for Homogeneous Receiving Water 5 Port
            Diameters Deep                                        340
       8.4  Results for Two Different Methods of Computing FZ     346
                                viii

-------
       8.5  Numerical Experiments Involving Ambient
            Stratification                                        355

       8.6  Discharge at Very Shallow Depth                       376

       8.7  Comparison with Field Data                            382

CHAPTER 9.  CONCLUSIONS                                           384


BIBLIOGRAPHY                                                      389


APPENDIX A - CONVECTIVE TRANSPORT DIFFERENCE APPROXIMATION        400

APPENDIX B - FINITE DIFFERENCES FOR IRREGULAR NODE SPACING        415

APPENDIX C - COORDINATE TRANSFORMATION                            419

APPENDIX D - SOME RELATIONSHIPS BETWEEN TIME DEPENDENT
             AND STEADY STATE NUMERICAL METHODS IN HEAT
             TRANSFER AND FLUID FLOW                              423

APPENDIX E - LISTING OF SYMJET COMPUTER PROGRAM - 1108 VERSION    434
                                ix

-------
                           LIST OF TABLES

Table    	Title	.   Page

2.1      Summary of Work Pertinent to Ocean Outfall Plume
         Analysis                                                 24

3.1      Differential Equations Required for Velocity-Pressure
         and Vector Potential-Vorticity Methods in Two and
         Three Dimensions                                         57

4.1      Values of Terms Involving K and x                        81

7.1      Summary of Momentum Jet Verification Cases
         (FQ + -)                                                166

7.2      Summary of Pure Buoyant Plume Verification Cases
         (FQ = 0)                                                168

7.3      Summary of Mixed Flow Verification Cases                169

7.4      Comparison of the Spreading Constant Reported by
         Various Investigators                                   191

7.5      Correlation of the Vertical Diffusion Coefficient
         e  with the Local  Richardson Number, RI                 287

7.6      Values of Vertical Eddy Viscosities in the Sea          290

7.7      Convergence Behavior, 40x33 Grid                        308

7.8      Convergence Behavior, 31x34 Grid                        312

7.9      Convergence Behavior, 26x25 Grid                        314

-------
                 LIST OF ILLUSTRATIONS
Figure                                                       Page

2.1    Condenser coolant flow rate as a function of temper-
       ature rise and plant electric generating capacity
       (fossil fired plant)                                    9

2.2    Condenser coolant flow rate as a function of tem-
       perature rise and plant electric generating
       capacity (nuclear plant)                               10

2.3    Vertical thermal plume in deep water, illustrating
       possible flow regimes                                  13

2.4    Vertical thermal plume in shallow water, illus-
       trating continual transition of the flow field         16

2.5    Possible configuration of a vertical  buoyant plume
       in stratified receiving water                          19

2.6    Possible configuration of a buoyant plume in
       stratified receiving water with cross-current,
       u                                                      on
        00                                                     ^(J

3.1    Rectangular coordinate system                          30

3.2    Relationship between the buoyancy parameter, A,
       and density disparity, A?                              45

4.1    Zone of flow establishment for plumes with large
       and small densimetric Froude numbers, F                66

4.2    Coordinate system for axisymmetric vertical plume      68

4.3    Typical velocity profile in the zone of flow
       establishment for a momentum jet                       75

5.1    Physical system for axisymmetric vertical plume
       where the bottom boundary is some distance z,  t 0
       above the outfall port                                 86

5.2    Physical system for shallow water, axisymmetric,
       vertical plume                                         88

5.3    Computational grid for difference equations            97
                              XI

-------
Figure

5.4    Typical  finite difference cell  illustrating indices
       for v,  n,  r,  U and V                                   99

5.5    Typical  sea  surface boundary and interior cells      115

5.6    Typical  centerline boundary and interior cells       118

5.7    Typical  inflow boundary and interior cell  (deep
       water only)                                           120

5.8    Typical  inflow boundary and interior cell  (shallow
       water case only)                                      122

5.9    Typical  vertical  port side boundary and interior
       cell (shallow water case only)                        125

5.10   Typical  bottom boundary and boundary cell             127

5.11   Typical  inflow-outflow boundary and interior
       cells                                                129

5.12   Physical system for line plume  issuing to flowing
       receiving water                                      133

7.1    Computational grid for the stream function, 4*,
       illustrating the effect of the  sinh (c) trans-
       formation (A? = .14690, AZ = 1.0)                    170

1.2    General  features of momentum jet centerline
       velocity (based on Albertson's  data)                 173

7.3    Comparison of experimental data and similarity
       solution with computed results  for a momentum
       jet.  Centerline velocity and concentration for
       case 2                                               175

7.4    Comparison of experimental data and similarity
       solution with computed results  for a momentum
       jet.  Centerline velocity and concentration for
       case 4                                               177

7.5    Computed centerline velocity and concentration
       for momentum jet, case 5                             178
                              xii

-------
Figure                                                      Page

7.6    Comparison of experimental data and similarity
       solution with computed results for a momentum jet.
       Centerline velocity and concentration for case 6      179

7.7    Computed centerline velocity and concentration for
       momentum jet, case 7                                  181

7.8    Computed centerline velocity and concentration for
       momentum jet, case 8

7.9    Computed centerline velocity and concentration
       distribution for momentum jet, case 9                 184

7-10   Centerline velocity distributions for cases 4, 7,
       and 9, normalized to V  = 1.0                         186
                             o
7.11   Centerline velocity and concentration distribution
       for case 10 (includes effect of large vertical eddy
       diffusivity)                                          188

7.12   Shape preserving of velocity profiles computed for
       an inviscid, rotational fluid (reference case 2)      189

7.13   Computed rate of spread of the momentum jet half-
       radius, r1  ,„                                          190

7.14   Radial distribution of normalized vertical  velocity
       for case 2                                            194

7.15   Normalized radial distribution of axial velocity,
       momentum jet case 4                                   195

7.16   Normalized radial distribution of axial velocity
       case 4                                                196

7.17   Radial distribution of axial  velocity at various
       elevations  case 4                                     197

7.18   Normalized  distribution of axial  velocity case 6      198

7.19   Normalized  radial concentration distribution,
       type 1 boundary condition case 2                      199
                              xiii

-------
Figure
7.20   Normalized radial  concentration  distribution,
       type 2 boundary condition,  case  4                     200
7.21   Radial vorticity distribution for momentum jet
       type 2 boundary condition,  case  4                     202
7.22   Radial vorticity distribution for momentum jet
       at Z = 15.  A comparison between type 1  and 2
       boundary conditions, and the Gaussian distribution    203
7.23   Normalized radial  velocity  distribution  for
       momentum jet                                          204
7.24   Vertical distribution of stream  function at R^,
       case 6                                                206
7.25   Streamlines for case 6 -- momentum jet                208
7.26   Isopycnals for case 6 -- momentum jet                 209
7.27   Vorticity level lines for case 6 -- momentum jet      210
7.28   3D illustration of stream function -- psi,
       case number 6                                         211
7.29   3D illustration of stream function -- psi,
       case number 6                                         212
7.30   3D illustration of buoyancy distribution - A,,
       case number 6                                         213
7.31   3D illustration of fluid vorticity - omeqa,
       case number 6                                         214
7.32   Computed centerline velocity and temperature excess
       for case 13.  Pure buoyancy, F  = 0
7.33   Computed centerline velocity and temperature excess
       for case 14.  Pure buoyancy, FQ = 0
7.34   Computed rate of spread of half-radius, r,/?/D.
       Pure buoyancy, case 14 (D = 2 r )         '
219
                          XIV

-------
Figure                                                     Page

7.35   Normalized distribution of computed  axial  velocity.
       Pure buoyancy, case 13                                222

7.36   Normalized radial distribution of axial  velocity.
       Pure buoyancy, case 13                                223

7.37   Radial distribution of axial  velocity in pure
       buoyancy, case 13                                     224

7.38   Normalized radial distribution of axial  velocity.
       Stronger source, pure buoyancy, case 14                225

7.39   Normalized distribution of computed  radial  tempera-
       ture excess.  Pure buoyancy,  case 14                  226

7.40   Radial distribution of vorticity.  Pure  buoyancy,
       case 14                                    '           227

7.41   Normalized radial velocity distributions for pure
       buoyant plume, case 14                                229

7.42   Vertical distribution of stream function at R^.
       Pure buoyancy, case 14                      °°        230

7.43   Streamlines for case 14, pure buoyancy                231

7.44   Isotherms for case 14, pure buoyancy, AT/AT           232

7.45   Vorticity level lines for case 14, pure  buoyancy      233

7.46   3D illustrations of stream function  - psi, case
       number 14                                             234

7.47   3D illustration of temperature field - AT, case
       number 14                                             235

7.48   3D illustration of fluid vorticity - omega.
       Case 14, pure buoyancy                                236

7.49   Centerline velocity and buoyancy for cases 15, 16
       and 17                                            '    240

7.50   Centerline buoyancy distribution for cases 17  and  18  242

-------
Figure                                                      Page

7.51    Centerline velocity and  buoyancy for  cases
       17, 19 and 20                                         243

7.52   3D illustration of vorticity -  omega.   Case 17,
       buoyant plume with running  calculation  of half
       radius                                                245

7.53   3D illustration of vorticity -  omega, case
       number 21                                              246

7.54   Centerline velocity and  buoyancy for  cases
       22 and 23                                              247

7.55   Comparison between computed results and simi-
       larity solution for F  = 1.0                          249

7.56   Comparison between computed centerline  distribu-
       tions of velocity and buoyancy  for f  = 1000
       and F  •* »                                            250
            o
7.57   Comparison of half-radius spread for  various
       densimetric Froude numbers                             252

7.58   Effect of the eddy Prandtl  number on  half-radius
       spread                                                252

7.59   Entrainment trends in mixed flows                     254

7.60   Streamlines for case 22 - mixed flow, FO =  46         255

7.61   Isotherms  for case 22 - mixed flow,  FO  = 46           256

7.62   Vorticity level lines for case  22 - mixed flow,
       FO = 46                                               257

7.63   3D illustration of stream function -  psi.
       Case 22 -  deep water buoyant jet                      258

7.64   3D illustration of temperature  field  -  T.
       Case 22 -  deep water buoyant jet                      259

7.65   3D illustration of fluid vorticity -  omega.
       Case 22 -  deep water buoyant jet        '             260
                              xvi

-------
Figure                                                      Page

7.66   Regional  specification for turbulent eddy
       coefficient modeling                                  267

7.67   Computed values of FR for a momentum jet              270

7.68   Computed radial eddy diffusion factors,  FR
       for deep water plumes at various densi-
       metric Froude numbers                                 271

7.69   Comparison of computation using constant and
       variable radial eddy transport coefficients           273

7.70   Concentration distribution in the zone of
       flow establishment                                    275

7.71   Computed potential core and half radius
       (FQ = 46)                                             277

7.72   Centerline velocity and temperature distribu-
       tion for 44 diameter deep outfall                     279

7.73   Computed centerline velocity and temperature
       excess.  Cases for 10 diameter deep water             281

7.74   3D illustration of temperature field                  283

7.75   Dependence of e   on sea state                        288

7.76   Correlation of e  with density gradient                290

7.77   Observation of flow patterns past the end of
       a cylinder                                            300

7.78   Computed flow patterns past the end of a
       cylinder                                              300

7.79   Convergence behavior, 40 x 33 grid                    309

7.80   Convergence history of V and  r at selected
       cells, momentum jet, 40 x 33 grid                     311

7.81   Convergence history of U, V and A, at
       selected cells 31 x 34 grid       '                    313

7.82   Iteration history for one cell of case 2              316
                             xvii

-------
Figure

8.1    Computed center!ine velocity and temperature
       excess for intermediate depth, cases 48
       through 51 (10 diameters deep)                         326

8.2    Surface distribution of radial velocity, cases
       48 through 51 (see Table 8.1)                          327

8.3    Distributions of radial velocity case 50               329

8.4    Maximum radial velocity profiles^ cases 48
       through 51                                             33°

8.5    Radial velocity profiles at r/D=7.32, cases 48
       through 51                                             331

8.6    Surface temperature excess distribution* cases 48
       through 51 (see Table 8.1)                             332

8.7    Vertical temperature excess distributions for
       various radial positions, cases 48 and 50              333

8.8    Streamlines for case 48 - buoyant discharge,
       FO = 100                                               333

8.9    Isotherms for case 48 - buoyant discharge,
       FO = 100                                               333

8.10   Vorticity level lines for case 48 - buoyant
       discharge, FO = 100                                    333

8.11   3D illustration of stream function -- PSI.  Case
       No. 48, intermediate water outfall, surface 10
       diameters above port, FO = 100                         334

8.12   3D illustration of temperature field -- AT.
       Case No. 48, intermediate water outfall, surface 10
       diameters above port, FO = 100                         334

8.13   3D illustration of fluid vorticity - OMEGA,
       Case no. 48, intermediate water outfall, surface
       10 diameters above port, FO = 100                      334

8.14   Streamlines for case 49 - buoyant discharge,
       FO = 25                                                334
                             XVlll

-------
Figure                                                       Page

8.15   Isotherms for case 49 - buoyant discharge,
       FO = 25                                                335

8.16   Vorticity level lines for case 49 - buoyant
       discharge, FO = 25                                     335

8.17   3D illustration of stream function -- PSI.
       Case No. 49, intermediate water outfall, surface
       10 diameters above port, FO = 25                       335

8.18   3D illustration of stream function — PSI.
       Case No. 49, intermediate water outfall, surface
       10 diameters above port, FO = 25                       335

8.19   3D illustration of temperature field -- AT.
       Case No. 49, intermediate water outfall, surface 10
       diameters above port, FO = 25                          336

8.20   3D illustration of fluid vorticity - OMEGA.
       Case No. 49, intermediate water outfall, surface
       10 diameters above port, FO = 25                       336

8.21   Streamlines for case 50 - buoyant discharge, FO = 5    336

8.22   Isotherms for case 50 - buoyant discharge,  FO = 5      336

8.23   Vorticity level lines for case 50 - buoyant
       discharge, FO = 5                                      337

8.24   3D illustration of stream function -- PSI.
       Case No. 50, intermediate water outfall, surface
       10 diameters above port, FO = 5                        337

8.25   3D illustration of temperature field -- AT.
       Case No. 50, intermediate water outfall,
       surface 10 diameters above port, FO = 5                337

8.26   3D illustration of fluid vorticity - OMEGA.
       Case No. 50, intermediate water outfall, surface
       10 diameters above port, FO = 5                        337
                              XIX

-------
Figure                                                      'Page
8.27   Streamlines for case 51  - buoyant discharge, FO = 1    338
8.28   Isotherms for case 51  -  buoyant discharge, FO * 1      338
8.29   Vorticity level lines  for case 51 - buoyant
       discharge, FO = 1                                       338
8.30   3D illustration of stream function — PSI.  Case
       No. 51, intermediate water outfall, surface 10
       diameters above port,  FO = 1                            338
8.31   3D illustration of temperature field — AT.
       Case No. 51, intermediate water outfall,
       surface 10 diameters above port, FO = 1                339
8.32   3D illustration of fluid vorticity - OMEGA.
       Case No. 51, intermediate water outfall, surface
       10 diameters above port, TO = 1                        339
8.33   Computed centerline dimensionless velocity and
       temperature excess for shallow water, cases 52
       through 55 (5 diameters  deep)                          341
8.34   Vertical distribution of radial velocity at
       various radial positions, case 52                      342
8.35   Vertical distribution of temperature excess at
       various radial positions, case 52                      342
8.36   Streamlines for case 52  - buoyant discharge, FO = 1    344
8.37   Isotherms for case 52 -  buoyant discharge, FO = 1      344
8.38   Vorticity level lines  for case 52 - buoyant
       discharge, FO = 1                                       344
8.39   3D illustration of stream function -- PSI.
       Case No. 55, very shallow water outfall, surface 5
       diameters above port,  FO = 1                            344
8.40   3D illustration of temperature field — AT.
       Case No. 55, very shallow water outfall, surface
       5 diameters above port,  FO = 1                         345
                               XX

-------
Figure                                                       Page

8.41   3D illustration of fluid vorticity - OMEGA.
       Case No. 55, very shallow water outfall,
       surface 5 diameters above port, FO = 1                 345

8.42   Computed radial velocity at surface,
       cases 58 and 59                                        348

8.43   Vertical distribution of radial velocity, I),
       cases 58 and 59                                        349

8.44   Vertical distribution of radial velocity, U,
       cases 58 and 59                                        349

8.45   Streamlines for an axisymmetric, vertical plume,
       confined by a free surface, case 58                    350

8.46   Streamlines for an axisymmetric, vertical plume,
       confined by a free surface, case 59                    351

8.47   Surface temperature excess, AT , cases 57, 58 and 59   352

8.48   Vertical temperature excess distribution, cases
       58 and 59                    '                          353

8.49   Vertical temperature excess distribution,
       cases 58 and 59                                        353

8.50   Isotherms for an axisymmetric, vertical plume,
       confined by a free surface, case 59                    354

8.51   Ambient temperature profiles for cases 60
       through 65                                             356

8.52   Vertical distribution of radial velocity, case 60      357

8.53   Vertical temperature excess distribution, case 60      357

8.54   Streamlines for an axisymmetric, vertical plume,
       confined by a free surface, case 60 - intermediate
       depth, homogeneous ambient, Mamayev, BETA = .4         358

8.55   Isotherms for an axisymmetric, vertical plume,
       confined by a free surface, case 60 - intermediate
       depth, homogeneous ambient, Mamayev, BETA = .4         358
                             XXI

-------
Figure                                                       Pa9e

8.56   Vorticity contours axisymmetric, vertical  plume,
       confined by a free surface, case 60 - intermediate
       depth, homogeneous ambient, Mamayev, BETA = .4         358

8.57   3D illustration of stream function — PSI, case
       no. 60                                                 358

8.58   3D illustration of temperature field -- AT,
       case No. 60                                            359

8.59   3D illustration of fluid vorticity - OMEGA,
       case No. 60                                            359

8.60   Vertical distribution of radial velocity, case 61      359

8.61   Vertical excess temperature distribution, case 61      359

8.62   Streamlines for an axisymmetric, vertical plume,
       confined by a free surface, case 61 - intermediate
       depth, with 2 degree thermocline, Mamayev              360

8.63   Isotherms for an axisymmetric, vertical plume,
       confined by a free surface, case 61 - intermediate
       depth, with 2 degree thermocline, Mamayev              360

8.64   Vorticity contours axisymmetric, vertical plume,
       confined by a free surface, case 61 - intermediate
       depth, with 2 degree thermocline, Mamayev              360

8.65   Vertical Distribution of Radial Velocity.  Case 61     362

8.66   Vertical Temperature Excess Distribution.  Case 61     363

8.67   Surface Temperature Excess, AT<~ for Cases 60 and  63    364

8.68   Streamlines for an axisymmetric, vertical plume,
       confined by a  free surface.   Case  63 -  inter-
       mediate depth  with 4 degree thermocline.  Mamayev     365

8.69   Isotherms for  an  axisymmetric,  vertical plume,
       confined by a  free surface.   Case  63 -  inter-
       mediate depth  with 4 degree thermocline,  Mamayev       365

8.70   Vorticity contours axisymmetric, vertical  plume,
       confined by a  free surface.   Case  63 -  inter-
       mediate depth  with 4 degree thermocline,  Mamayev       365
                              xxii

-------
Figure                                                       Page

8.71   3D illustration of stream function — PSI.
       Case 63 - intermediate depth with 4 degree
       thermocline, Mamayev                                   365

8.72   3D illustration of temperature field -- T.
       Case 63 - intermediate depth with 4 degree
       thermocline, Mamayev                                   366

8.73   3D illustration of fluid vorticity - OMEGA.
       Case 63 - intermediate depth, with 4 degree
       thermocline, Mamayev                                   366

8.74   Vertical distribution of radial velocity.   Case 64     368

8.75   Vertical, excess temperature distribution.  Case 64     368

8.76   Streamlines for an axisymmetric, vertical  plume,
       confined by a free surface.  Case 64 - inter-
       mediate depth, with 5 degree thermocline,  Mamayev      369

8.77   Isotherms for an axisymmetric, vertical plume,
       confined by a free surface.  Case 64 - inter-
       mediate depth, with 5 degree thermocline,  Mamayev      369

8.78   Vorticity contours axisymmetric, vertical  plume,
       confined by a free surface.  Case 64 - inter-
       mediate depth, with 5 degree thermocline,  Mamayev      369

8.79   3D illustration of viscous stream function -
       Case 64 - intermediate  depth, with 5 degree
       thermocline, Mamayev                                   369

8.80   3D illustration of temperature field -- AT
       Case 64 - intermediate depth, with 5 degree
       tnermocline, Mamayev                                   370

8.81   3D illustration of vorticity -- OMEGA
       Case 64 - intermediate depth, with 5 degree
       thermocline, Mamayev                                   370

8.82   Vertical distribution of radial velocity.   Case 65     371

8.83   Vertical, excess temperature distribution.  Case 65    372

8.84   Streamlines for an axisymmetric, vertical  plume,
       confined by a free surface, case 65 -  intermediate
       depth, with 5 degree thermocline, Mamayev              373
                            XXlll

-------
Figure                                                       Pa9e

8.85   Vorticity contours axisymmetric, vertical  plume,
       confined by a free surface, case 65 - inter-
       mediate depth, with 5 degree thermoeline,  Mamayev      373

8.86   Isotherms for an axisymmetric,  vertical  plume,
       confined by a free surface, case 65 - inter-
       mediate depth, with 5 degree thermoeline,  Mamayev      373

8.87   Streamlines for an axisymmetric, vertical  plume,
       confined by a free surface, case 65 - inter-
       mediate depth, continued iteration.                    373

8.88   Isotherms for an axisymmetric,  vertical  plume,
       confined by a free surface, case 65 - inter-
       mediate depth, continued iteration.                    374

8.89   Vorticity contours axisymmetric, vertical  plume,
       confined by a free surface, case 65 - inter-
       mediate depth, continued iteration.                    374

8.90   Surface radial velocity, case 66                       378

8.91   Surface temperature excess, case 66                    378

8.92   Vertical distribution of radial velocity at
       various radial positions, case 66                      379

8.93   Vertical distribution of temperature excess at
       various radial positions, case 66                      379

8.94   Streamlines for case 66 (1.0 dia deep) FO = .111       380

8.95   Isotherms for case 66 (1.0 dia deep) FO = .111         380

8.96   Vorticity for case 66 (1.0 dia deep) FO = .111         380

8.97   3D illustration of fluid vorticity - OMEGA
       Case 66                                                380

8.98   3D illustration of stream function -- PSI,
       Case 66                                                381

8.99   3D illustration of stream function -- PSI,
       Case 66                                                381

8.100  3D illustration of temperature field -- T.
       Case 66                                                381
                             XXIV

-------
Figure                                                       Page
8.101    3D illustration of temperature field — AT.
         Case 66                                              381
8.102    Comparison of computed surface temperature
         with field data                                      383
A-l      Finite difference grid system                        404
A-2      Values of X  and  4.  for a preferred
         difference scheme                                    408
A-3      Convective r flux for an infinitesimal
         axisymmetric volume element                          410
A-4      Axisymmetric finite difference cell, p, with
         the four immediate neighbor cells                    412
B-l      Irregular spaced grid                                415
B-2      Grid layout for vertical differences                 417
C-l      Ratio of actual to computed node spacing             422
D-l      Finite difference cell                               428
                              XXV

-------
                            NOMENCLATURE



      Variables which are not listed in  this  nomenclature are defined


at the appropriate location within the manuscript.   A  few variable


names have been duplicated; however, the definitions listed  below hold


throughout the text with duplications defined at the point of use.


Dimensions are given in the force-length-time system (F-L-T).



      A            Constant used in Chapter 7.

                                         p
      a            Thermal  diffusivity,  L /T


      B            Buoyancy parameter defined in Chapter  3.


      b            Slot width (slot plume), L


      C-, ,C2        Constants


      c            Concentration


      D            Diameter of outfall port,  L


      ei, e        Unit vector (i  = 1,2,3)


      E            Momentum parameter for plume  similarity


                   solution.


      f            Coriolis constant, 1/T


      FR           Radial eddy momentum  diffusivity multiplier


      FZ           Vertical eddy momentum diffusivity  multiplier

                                              p
      g            Gravitational constant,  L/T


      k            Kinetic  energy of turbulent motion, FL


      K            Entrainment parameter


      £            Characteristic length, L


      L            Liebmann acceleration constant


      M            Total momentum, FT
                               XXVI

-------
n            Unit surface normal vector


P            Pressure, F/L2


P°           Deviatoric pressure (defined in Chapter 3),  F/L

                                      2
P            Fluctuating pressure, F/L

                                      2
Q            Plume entrainment rate, L/T


r            Radial coordinate, L

                                        2
R..          Reynolds stress tensor, F/L
 ' J

S            Salinity, ppt


t            Time, T


T            Temperature, °C


u            Velocity, L/T


u'            Fluctuating velocity, L/T


ft,            Irrotational velocity vector, L/T


IL           Solenoidal velocity vector, L/T


v            Vertical velocity, L/T


x            horizontal coordinate, L


x-            General rectangular coordinate, L


Y            Space coordinate, L


z            Vertical coordinate, L
                            XXV11

-------
Greek
      a            Entrainment constant
      B            Constant used in vertical  eddy diffusivity model
      6            Convergence criterion
                                                   2
      e            Eddy diffusivity for momentum, L /T
                                                 2
      £.            Eddy diffusivity for matter,  L /T
      eu           Eddy diffusivity for heat, L2/T
       n
      e            Eddy diffusivity for density, L /T
       p
      c,            Variable defined in Chapter 3.
      0            Longitude
      K            Thermal  conductivity, F/LT°C
                   Dynamic viscosity, FT/L2
                   Kinematic viscosity, L /T
      n            3.14159
      P            Density, FT2/L4
      p1           Fluctuating density, FT2/L4
      T            Shear stress
      4            Latitude
      •
      4>            Source or sink term in r-transport equation, i/|_
                                      2
      $            Scalar potential ,  L /T
                                     2
      *            Stream function,  L /T
      £            Vector potential,  L2/T
      oj            vorticity,  1/T
      fi*           Earth rotation velocity, 1/T
                               xkviii

-------
Standard Dimensionless Parameters



                                     1    2
      EU           Euler number, AP/-^  pv!"



                                                    "o
      fn » FO       Densimetric Froude number,   	
       O                                         O  ~D
      Pr           Prandtl  number,  V/K



      PR           Eddy Prandtl  number, e/e,,



      SO,A         Eddy Schmidt number, e/e



                                    wfl

      Re           Reynolds number,	

                                         wo

      RE           Eddy Reynolds number,—-
RI
                   Local Richardson number,  —9-  (dP/dz)
                                              po
      RI1          Gross Richardson number,  - ^—  	^

                                               po  (AU/AZ)^


                   for the thermal layer.




Dimensionless Parameters Defined in this Manuscript
      C            Concentration, c/cn      _

                                     °/V Z\3

      E*           Momentum parameter,
                                        /K
      P*           Pressure, P°Pr/APQ
      R            Radial coordinate, r/rQ
                                       E*1/3
      R*           Density parameter,
                              xxix
                                       /K (1+x)

-------
t*           Time, tvQ/D

U            Radial  velocity,  u /VQ

V            Vertical  velocity, UZ/VQ

X            Space coordinate, x/x

Z            Vertical  coordinate,  z/rQ

Z            Vertical  coordinate,  z/D

r            Conservative constituent parameter
                                  Pr-p
A-,           Buoyancy parameter, (	)
                                  pr"po
                                            oo   .
A?           Density disparity parameter,  (	)
                                          pr"po
                                  Vs
             Salinity parameter, (7—^
                                   r b
             Eddy diffusivity for momentum, e/e0
                                     T -T
             Temperature parameter, (=—=F-)
                                      o" r
             Radial coordinate, sinh"' (R)
             Stratification parameter, poo(Z)/p
                                     2
             Stream function, y/r  v


             Vorticity, 
-------
Subscripts
      The following subscript definitions hold  unless  otherwise
defined in the text.
      b            Refers to slot jet width
      c            Refers to center, or core
      e            Value at end of zone of flow establishment
      E            Elliptic partial  differential  equation
      H            Refers to heat
      i            Tensor index
      j            Tensor index, also computational  grid  index  in
                   the horizontal (radial) direction
      k            Tensor index, also computational  grid  index  in
                   the vertical direction
      m            Value at jet centerline
      max          Maximum value
      p            Computational grid index
      port         Refers to conditions at outfall  port
      q            Computational grid index
      r            Refers to radial  direction,  or reference condition
                   for scalar quantities
      s            Refers to condition at surface
      T            Refers to turbulent conditions,  or  transport
                   equation
      x            Refers to x (horizontal) direction
      z            Refers to 2 (vertical) direction
                               XXXI

-------
      Greek Subscripts


      Y            Refers to a conservative constituent



      A            Refers to the buoyancy parameter



      p            Refers to density



      Y            Refers to the stream function



      n            Refers to vorticity




      Other Subscripts


      o            Refers to conditions of or at the outfall



                   Refers to conditions far removed from the outfall



      1/2          Refers to the half-width




Mathematical Notations




      D_            Substantial derivative

      Dt




      v^           Laplacian operator

      ->-

      v            Gradient operator, del



      A            Finite-difference operator



      £            Summation except where otherwise specified



      6..:          Kronecker delta function
       * J


      e...          Permulation tensor
       1 JK


      Log          Natural logarithm



       ||           Absolute value



                   Hat, unit vector



      -            Overbar, time or space averaging



      sinh, cosh,  Hyperbolic functions

      tanh, coth
                           xxxii

-------
        A NUMERICAL MODEL FOR PREDICTING ENERGY DISPERSION
      IN THERMAL PLUMES  ISSUING FROM LARGE, VERTICAL OUTFALLS
                      IN SHALLOW COASTAL WATER

                              CHAPTER 1
                             INTRODUCTION

      The growing demand for electric power in the United States has
set the stage for an  additional environmental concern; the enormous
quantities of waste heat discharged to our natural waterways by
existing and planned  large thermal power plants.  The concern, of
course, is the impact of the waste heat on the resident ecosystem.
The answer to the underlying question, "are thermal effects a detri-
ment  to the environment?", is largely a matter of philosophy since
certain species of the flora and fauna are apt to thrive under the
altered conditions whereas others would doubtless perish.
      The central issue is,  however, that these large quantities of
discharged waste heat will in fact alter the environment and certain
changes in the ecosystem will occur.  Just what changes will take
place and the nature  of the  shift in the ecosystem are open to numer-
ous questions.  Preservation of species, the impact on the overall
food chain, and the encroachment of undesirable species are certainly
compromising aspects.  These questions and many others of equal im-
portance are certainly not unattended, but the interaction of the eco-
system with the environment and the complexity of ecodynamics as
influenced by artificial shifts in the environment presents an ana-
lytical and empirical task to arrive at reliable predictive methods of
monumental proportions.

-------
      Although the ultimate concern  of  so-called  "thermal  pollution"
lies in the ecological  impact,  it  is necessary  as  a  first  step  to
assess the receiving water temperature  changes.  Prediction of  the
temperature distribution in natural  waters is in  itself a  formidable
task owing to the complexity of such natural  phenomena as  hydro-
dynamics, dispersion, and atmospheric interaction (transport process-
es).  To date, no analytical or empirical  tool  has been devised to
predict thermal distributions with any  degree of  confidence for gen-
eral situations.  The state-of-the-art  has been developed  along the
lines of applying the most appropriate  simplified analytical or empir-
ical model to an immediate situation.  Unfortunately, some situations
are complicated to the extent that simplified methods are  a hopeless
exercise and can lead to a valueless or grossly overrestrictive
assessment.
      Such complexities lead to methods involving more elaborate
numerical models or physical scale modeling.   In  this work, we  take
the former approach, that of numerical  modeling.
      As is pointed out in Chapter 2, previous analytical  plume
modeling efforts have dealt primarily in two  areas which are:
      «  The initial mixing zone where, in certain cases,  simil-
         arity solutions apply, and
      «  The far field where heat transfer is governed by  turbulent
         diffusion and atmospheric interchange.

-------
      The past research has largely neglected an area cf prime  impor-
tance, that being the near field of large, vertical  outfalls  in  shal-
low coastal waters.   This neglect is in part due to  the  complexity  of
the flow region in question and the fact that it is  a  new problem.
The near  thermal  field for such outfalls  is, nevertheless, a very
important  aspect  of  plume analysis, and is  in need of analytical
attention.
1.1  Objectives
      The  primary objective of the work contained in this thesis is
the investigation and application of finite-difference methods in ana-
lyzing the dispersion of thermal effluents  issuing from large single
port vertical outfalls in shallow coastal receiving water.  Such sys-
tems are  typical  of  several existing and/or  planned thermal power
plant reject-heat discharge systems.  This  analysis, constitutes
research needed for  future thermal discharge management.  Since we are
interested primarily in the hydrodynamics and energy transport for a
shallow water, vertically confined plume, simplified analytical
methods cannot be applied with confidence.   Physical modeling holds
some promise as an alternative to numerical  modeling, at  least in  the
near field and in the absence of stratification.  Since the numerical
modeling devised  in  this study was a considerable effort  in itself,
physical modeling was not attempted.  Verification of the numerical
techniques was rather carried out by testing the computer program  for
several cases that could be checked with  data published in the
literature.

-------
      The secondary objective of the work was to develop a computer
program for analytical  study of the above mentioned outfall  systems
which would also include use of similarly solutions where applicable,
along with the more elaborate numerical  techniques.

 1.2  Summary
       In the initial  scoping of the vertical  plume problem  it  was
 planned to investigate both the transient and steady  state  operation
 of the outfall  system.  Initially, several  transient  cases  were run
 which were academically quite interesting but it was  soon ascertained
 that the application  of steady flow techniques was more efficient  in
 obtaining the desired  results—the quasi-steady flow  distributions.
 Consequently, the transient techniques  were abandoned.   In  general,
 the scope of the study encompasses nearly all  of the  real quasi-
 steady flow complication expected  in actual  situations  which conform
 to axisymmetric assumptions.  The  most  notable complication is that of
 plume induced turbulence.
       One exception to the modeling of  observed phenomena was  the
 surface boil; the surface was assumed flat and free-slip in all
 instances.  This assumption averted the problem of modeling a  dis-
 torted surface which  is thought to be of small  importance to the
 overall plume characteristics.   Other complications accounted  for
 include the possible  existence of  a potential  core, ambient strati-
 fication, and non-homogeneous,  anisotropic turbulence in both  the
 vertical rise and lateral  plume spreading.   Flows for the entire

-------
•range of densimetric Froude numbers  were  investigated,  including
 cases of pure natural  convection.
       The solution method deemed most practical  for  purposes of this
 study was the stream function-vorticity,  finite-difference approach,
 in axisymmetric coordinates.  The  transport equations  were used in
 their conservative forms and special  upstream  differencing techniques
 were employed for the convective terms.
       The finite-difference computation  technique  verification study
 was carried out for three deep water flow categories:
       •  pure momentum jets,
       •  pure buoyant plumes,  and
       •  forced plumes where both  initial  momentum and  buoyancy
          play important roles.
       Results from this portion of the study were  compared to data
 reported in the literature or  valid  similarity solutions.  These  com-
 parisons involved:
       •  centerline distributions  of velocity   and buoyancy  (or
          temperature),
       •  spread of the half-radius,
       •  radial  distributions  of vertical  velocity and  buoyancy
          (or  temperature),
       •  radial  velocities,
       •  entrainment  trends, and
       *  eddy diffusivities.

-------
      The effects  of several  different  computational  aspects  were
included which involved effects  of the:
      &  boundary  conditions  and their  computation,
      ®  various models for eddy diffusivities,
      «  Prandtl (or Schmidt) number effects,
      e  Richardson number modification of vertical  diffusivities,
      «  potential core,
      •  ambient turbulence,
      »  vertical  turbulence within the plume, and
      e  various factors involving numerical  stability and con-
         vergence.
      The general  results of this portion of  the study showed excel-
  lent agreement with experimental data where the eddy diffusivities
  are well modeled.  Plume generated turbulence was modeled using
  Prandtl's mixing  length hypothesis in all cases.
      In Chapter 8  the plume model is extended to shallow water cases.
  Verification  is not presented  since there are no readily available
  appropriate or reliable data.*  Here we rely on the extensive veri-
  fication study of Chapter 7 mentioned above.
 Verification of the  surface  temperature  distribution  was  obtained
  for one case.   The data  is proprietary,  hence  no  details  of operating
  conditions  are  disclosed.

-------
                              CHAPTER 2
         DISCUSSION OF THERMAL PLUMES AND PROBLEM DESCRIPTION
      The dynamical behavior of heated water issuing to the marine
environment from an ocean outfall is influenced by a number of variables
which fall into two general categories.  The first of these categories
encompasses engineered variables such as outfall design, effluent temp-
erature, etc; and, the second, those variables which we cannot control,
such as the oceanographic and meteorlogical parameters.  In this chapter,
we  shall  illustrate  and  discuss  how ambient and  engineered  variables
influence the gross behavior of a thermal plume, briefly discuss the
analytical "state-of-the-art," and  qualitatively describe the problem
undertaken in this research.

2.1    The Nature of Thermal Plumes  in Marine Surroundings
      In the following discussion the terms jet flow and plume flow
will be used, and to avoid confusion it is  appropriate to outline the
meaning of each at this time.  A convective flow in a free environment
caused solely by buoyancy is commonly called a simple plume.  In this
case, the general pattern of motion is caused by a density disparity
between the flow and the surrounding environment.  Such instances are
atmospheric thermal and the smoke plumes generated by field fires.
A jet, on the other hand, is characterized  by source flow inertia where
the  flow may not involve a density  difference.
      The flow which is of primary  concern  in this discussion is a com-
bination of the above where both initial momentum and buoyancy have
significant influence on the flow behavior.  Such a flow might be termed

                                 7

-------
a forced plume.  However, in this work the flow field will be called a
thermal plume or plume.  Reference will  be made to jet flow from time
to time, which will imply that conditions near the outfall, where
initial momentum dominates the dynamic behavior,  is the subject of dis-
cussion or that the effluent is neutrally buoyant.
      A temperature difference is not the only factor which must be con-
sidered as a buoyancy source in a thermal plume.   Differential  salt con-
centration is certainly a factor.  Salinity differences must be consid-
ered if the power plant condenser coolant is drawn from an estuary and
rejected off-coast,in which case, the effluent would most likely be less
saline than the receiving water and contribute to the overall  buoyant
force.

2.1.1  Discharge Magnitude
      The volumetric flow rate required  by a thermal  power station
depends on plant size, steam cycle thermodynamic  efficiency,  and coolant
temperature rise.  Typical installations  range from 1000 to 2000 Mwe and
operate at a coolant temperature rise between 15  and 20 °F.  Plant
efficiency depends  largely on whether the heat source is nuclear or
fossil.  The steam  cycle thermodynamic efficiency for a typical  fossil
fired plant will be in the neighborhood  of 42% for optimum conditions,
whereas typical efficiency for a modern  nuclear plant operating under
similar conditions  is about 32%.  Hence,  the nuclear plant will  reject
about 50% more heat than a fossil fired  plant having the same net
electrical  output.
      The condenser coolant volumetric flow rate  required by power sta-

                                   8

-------
tions in the 1000 to  2000 Mwe range is impressive by  any standards,
regardless of whether the plant is nuclear or fossil  fired.  Figures 2.1
and 2.2 illustrate this  fact.  It is possible that in the future a
particular site  will  consist of a number of individual  units.  Thus
the cooling load on a certain ocean locale may result from the produc-
tion of perhaps  10 Gw .
 o
 o
 o
 o
 o;
      30
      20
      10
         (ELECTRIC)
         (ASSUMED  EFFICIENCY, 42%)
                   2345          10
                    TEMPERATURE  RISE, AT  (°F)
20    30
     Figure 2.1   Condenser Coolant Flow Rate as  a  Function
                 of Temperature Rise and Plant Electric
                 Generating Capacity (Fossil  Fired  Plant)

-------
 GO
 U_
 O

 CD
 O
 o
 o:
      30



      20
10  -
       5

       4
       3  -
       2  -
            (ASSUMED EFFICIENCY,  32%)
                    2345         10         20    30

                     TEMPERATURE RISE,  AT  (°F)
       Figure  2.2  Condenser Coolant Flow Rate  as  a  Function
                  of Temperature Rise and Plant Electric
                  Generating Capacity (Nuclear Plant)
2.1.2  Outfall  Configuration

      Condenser coolant may be rejected to the  ocean  either at the

shoreline or offshore through a submerged outfall.

      The shoreline discharge may be either by  canal  or  conduit.

Examples of such existing systems are the following fossil fired plants

owned by Pacific Gas and Electric [113].
                               10

-------
      1) Contra Costa, 1298 Mwg, rejecting heat to the San Francisco
         Bay Delta.
      2) Pittsburgh, 1340 Mwe, rejecting heat to the San Francisco  Bay
         Delta.
      3) Morro Bay, 1030 Mwe, rejecting heat to the Pacific Ocean.
Numerous other examples might be cited since the shoreline outfall
system has widespread use.
      Submerged, offshore outfalls may be designed in two general
fashions:
      1) a single port (dual in some cases) outlet situated either
         vertical or horizontal, or
      2) a diffuser section at the end of the pipeline consisting of
         numerous ports.  The diffuser is typical of municipal  waste
         outfalls.
Some examples of Targe vertical port outfalls are:
      1) Moss Landing fossil fired plant.  Reject heat from 1500 Mwg
         generation, discharged about 800 feet offshore.  Dual  ports.
      2) San Gnofre nuclear plant.  Reject heat from approximately
         450 Mw  generation, discharged through a 14-foot diameter
         pipe 2600 feet offshore, about 15 feet below sea surface.
      3) Redcmdo Beach fossil fired plant.  Reject heat from 1612 Mwg
         generation.  Two offshore outfall systems: a) two 10-foot
         diameter pipes discharging vertically about 2100 feet offshore;
         and b) a single 14-foot diameter pipe discharging vertically
         300 feet off, about 16 feet beneath water surface.
                                      11

-------
      4) El  Segundo fossil  fired plant.   Reject heat from 1020 Mwe
         generation.  Two offshore outfall  systems:  a)  two 10-foot
         diameter pipes, discharging 2100 feet offshore,  vertically,
         20 feet beneath ocean surface;  and b) two 12-foot diameter
         pipes, discharging 2070 feet offshore, vertically, 20 feet
         beneath ocean surface.
To this author's knowledge, no large power plant uses diffusers for off-
shore ocean discharge at present, although such a system  is proposed for
the Shoreham plant [95], discharging to  Long Island  Sound.

2.1.3  Hydrodynamic Regimes
      Experimental observations of forced plumes issuing  from submerged
ports have revealed the existence of four distinct flow regimes, as
follows (Figure 2.3):
      •  Zone of flow establishment (jet flow)
      •  Zone of established flow (mixed flow)
      •  Transition from established to  drift flow,  and
      •  Zone of drift flow.
The zone of flow establishment is in effect a transition zone from pipe
flow to an established forced plume.  Consider fluid issuing from an
outfall port of diameter D (Figure 2.3), to the surrounding ocean, with
a turbulent velocity profile.  For the sake of analysis, this profile
is usually assumed uniform with velocity v  .   Immediately  the velocity
begins to deteriorate at the flow boundary as  a result of  turbulent
mixing with the surrounding ocean water.  This region of mixing spreads
both inward toward the center of the plume and outward into the sur-
                                    12

-------
                     SEA SURFACE
                             VELOCITY
                              PROFILE
                                         NOMINAL PLUME
                                           BOUNDARY
                                  (v   >  v  )
                                  ^  me —  o'
                                     POTENTIAL CORE
            OUTFALL
              PORT
Figure 2.3  Vertical Thermal Plume in Deep Water,
           Illustrating Possible Flow Regimes

                        13

-------
roundings.  Within a short distance, ze, from the outfall port, the
interchange of momentum due to mixing has spread to the center of the
plume.  At this point, it is generally assumed that the plume vertical
velocity profile is fully developed, or established.
      In the zone of established flow, velocity profiles are approxi-
mately similar at all  axial locations and the driving force may be
either initial momentum, buoyancy, or both (mixed flow).  As distance
from the outfall increases, the effective width of the plume and the
amount of plume flow increases as a result of lateral mixing or turbu-
lent diffusion (commonly called entrainment).  Momentum of the plume at
successive cross-sections is changing according to the density differ-
ence between the plume and surroundings.  Maximum velocity, v , of the
plume will decrease, except if the buoyancy is large compared to initial
momentum, in which case the maximum velocity may increase momentarily
near the outfall.
      The transition from established flow to drift flow is caused by
the plume encountering the ocean surface or by the plume attaining a
neutrally buoyant condition in a density stratified sea.  Here velocity
profiles change drastically with essentially all mean vertical motion
vanishing.  The motion at the transition zone termination may be dom-
inated by prevailing ocean currents.
      In the zone of drift flow, prevailing ocean currents will generally
dominate the plume motion, although a lateral density flow will persist
if the plume is situated on the ocean surface with buoyancy.  Lateral
mixing is dominated by ocean turbulence, whereas vertical mixing depends
on both the plume and environment driving forces.

                                 14

-------
      Under certain conditions, all of the above hydrodynamic regimes
will not prevail.  For instance, in the case of a large diameter port
issuing in shallow water, the zone of established flow will  most likely
be absent.  This situation is usually termed a "confined plume"
(Figure 1.4) and the hydrodynamics are characterized by a continuous
transition from pipe flow to drift flow.
      An example of a typical confined plume is the thermal  effluent of
the Southern California Edison power plant located at San Onofre,
California, discharging approximately 15 feet beneath the sea surface.
The port is vertical and 14 feet in diameter.  Based on experiments by
Albertson, et al. [4] concerning neutrally buoyant jets, this depth is
less than the length for flow establishment.
      For shoreline outfalls, the same flow regimes exist.  However,
the zone of established flow may be less distinct depending on the
relative magnitudes of initial momentum and buoyancy (initial densi-
metric Froude number).  This zone will assert itself if buoyancy is
small or initial momentum is large.  In the case of small initial
momentum and moderate or large buoyancy, the initial mixing zone will
be a continuous transition from the outfall to drift flow without
established flow in the sense of similar velocity profiles.

2.1.4   Oceanographic Effects
      The nature of the surrounding ocean can have a dramatic effect on
the behavior of a thermal plume.  Probably the most influential of these
oceanographic variables are the following:
                                 15

-------
                                       SEA  SURFACE
                                                   POSSIBLE
                                                   VELOCITY
                                                    PROFILE
NOMINAL  PLUME
  BOUNDARY
                                                           POSSIBLE POTENTIAL  CORE
                     OUTFALL PORT
EXTENT  OF  TRANSITION
       REGION
                Figure 2.4  Vertical Thermal Plume in Shallow Water, Illustrating
                           Continual Transition of the Flow Field

-------
       •   density stratification,



       •   currents,  and



       •   turbulence.






2.1.4.1  Density Stratification



      In all discussions concerning ambient density stratification,



stable stratification is implied.   One effect of stratification  is



stabilization of the ambient flow field insofar as vertical  convection



and mixing are concerned.  However, the discussion in this chapter will



be confined to the direct effect of limitation of height of rise for



plumes issuing from submerged outfalls.



      The maximum height that the thermal  plane will  attain (and whether



the plume will reach the surface or not) depends largely on the  ambient



density structure.  Obviously, this discussion does not apply to con-



fined plumes, but to cases where the outfall port size is small  com-



pared to the ocean depth, as for example,  diffuser ports.  Both  theory



and experiment have shown that the plume will always reach the surface



if the ocean is homogeneous with respect to density.



      The ocean, however, is rarely homogeneous, except perhaps  in very



shallow coastal waters where good vertical mixing occurs.  The reason



that a thermal plume may not penetrate to the ocean surface in a density



stratified environment is that the plume entrains the heaviest water



nearest the outfall.  This water causes dilution to some  extent and  is



carried upward with the plume. As the  plume ascends, the  density dif-



ference  between the plume and surroundings  steadily decreases because



the flow is  being diluted and cooled through entrainment,  and because




                                  17

-------
the density of the surroundings is decreasing upwards.
      If the density stratification has sufficient magnitude (among
other considerations which will be discussed later), the plume will
eventually reach a level of neutral buoyancy some distance below the
water surface.  At this point the flow continues upward only by virtue
of the vertical momentum it possesses at that point.  As the plume con-
tinues upward, it continues to entrain liquid that is now less dense
than the plume flow; hence, the flow is negatively buoyant.   Eventually,
all upward vertical momentum is lost and, since the plume liquid is
denser than the surroundings at that depth, the pollutants will  cascade
downward around the upward flow.
      Small oscillations in the vertical  motion will follow and when
these oscillations vanish the plume is said to be "trapped"  (Figure 2.5),
At the trap level all mean motion is horizontal since the flow is
neutrally buoyant (assuming that environmental  isosteric surfaces are
horizontal).

2.1.4.2  Effect of Currents
      Currents have a dramatic effect on plume behavior in nearly all
flow regimes.  The types of currents that might have influence are tidal
currents, longshore currents, upwelling, wind driven surface currents,
and persistent currents that are peculiar to a certain locale.
      The zone of flow establishment is essentially unaffected by cross
currents; but, in the zone of established flow (deep water), a cross
current will cause the plume to be "bent-over" (Figure 2.6).  The most
significant effect of this bending is a decrease in the height of rise,
                                18

-------
also, the dynamics within the plume  are changed.

      When the plume is bent over, two distinct counter rotating vor-

tices are formed (Figure 2.6).   These vortices are quite apparent in

atmospheric smoke plumes discharging into a cross wind; the same

phenomenon occurs in the ocean.

      In the drift flow regime,  the  plume flow is carried along with

the ocean current nearly as  though it were the ambient water.   Thus,

ocean currents play a dominant hydrodynamic role on the eventual  fate of

the pollutant.  Upwelling causes  a persistent offshore surface current,



                             SEA SURFACE
                                       ^ESTABLISHED
                                           VELOCITY
                                            PROFILE
                                       OUTFALL  PORT
                                                BOTTOM

      Figure 2.5  Possible Configuration of a Vertical  Buoyant
                  Plume in Stratified Receiving  Water
                              19

-------
                              SEA SURFACE
ro
o
        (AMBIENT   .
         CURRENT)
                           VELOCITY PROFILE
   SECTION A-A

COUNTER  ROTATING
    VORTICES
                   Figure 2.6  Possible Configuration of a  Buoyant Plume in Stratified
                              Receiving Water with Cross-Current, u

-------
thus, a surface plume could be carried out to sea.  Wind driven surface
currents and tidal currents can cause the pollutant to be carried on-
shore or out to sea, and longshore currents can cause the pollutant to
be distributed along the shoreline.

2.1.4.3  Ocean Turbulence
      The origin  of oceanic turbulence is not fully understood,
although in the surface zone  it is probably caused mostly from winde
generated wave action.  As such, the turbulence is neither homogenous
nor  isotropic, and only the gross  behavior can be described.
      Ocean turbulence has some effect on all regimes of plume flow.
Turbulence scales that are on the  same size or larger than the plume
cross-section will have an effect  similar to a crosscurrent, and all
scales should have some influence  on the plume entrainment rate
(although it is thought that  the influence is small in all zones except
the  drift regime, since turbulence generated by the plume dominates
the  ocean turbulence).  In the zone of drift flow scales of motion
larger than the flow field result  in action similar to oceanic currents,
and  the pollutant field simply flows along with the turbulent motion.
Smaller scales of motion add  to the eddy diffusion of the pollutant;
thus, as the pollutant field  spreads, larger and larger scales of eddy
mixing come into  play.
      Another factor complicating  oceanic turbulence is that it is
highly anisotropic, at least  in the larger scales of motion.  Since
most oceanic waters are density stratified to some degree (except per-
haps in shallow water), vertical mixing is suppressed to a great
                                21

-------
extent.  Thus, a pollution field diluted by eddy diffusion will  spread
much more rapidly in the lateral direction than in the vertical.

2.1.4.4  Air-Sea Interactions
      Wind and heat transfer are the major air-sea interface phenomena
which may significantly affect thermal  plume dynamics.   Wind stress at
the sea surface causes two local effects which have previously been men-
tioned:  wind driven surface current, and turbulence.   And, on a  larger
scale, wind is responsible for coastal  upwelling.   We  will only point
out these wind stress effects here and  refer the interested reader to
such references as Neumann and Pierson  [63] or Wada [107]  for addi-
tional details and references.
      Heat transfer at the interface is carried on by  atmospheric con-
vection, radiation, and evaporation.  Evaporation is probably the most
significant of these modes and is materially affected  by the surface
temperature and conditions in the atmospheric boundary layer such as
temperature, humidity and turbulence.  Again, wind plays an important
role here through promotion of atmospheric turbulence  and  convective
currents.  Radiation heat transfer depends on the sea  surface tempera-
ture and albedo, atmospheric conditions such as turbidity, and position
of the sun.
      The effect of surface heat transfer is more complicated than
merely heating or cooling of the plume.  For instance, if heat is lost
at the surface, convective downcurrents of cooler water may occur,
tending to homogenize the plume vertically.  If heat is gained at the
surface, the plume will become more stable and suppress vertical mixing.
                                 22

-------
      Atmospheric heat transfer will affect the plume dynamics predomin-
ately in the drift flow regime when the plume is situated at the surface.
The area exposed to the atmosphere in the surface transition (zone 3)
is small on a comparative basis and will likely be unaffected by sur-
face heat transfer.

2.2  Plume Analysis State-of-the-Art
      There has been a great deal of theoretical and experimental work
carried out in the past 20 years or so dealing with the dynamics of
buoyant plumes.  Most of this work has dealt directly with either atmos-
pheric smoke plumes or ocean plumes caused by submerged offshore indus-
trial and municipal waste outfall systems:  (cf.  Baumgartner and Trent
[12]).  Much lesser and more recent efforts have treated horizontal  shore-
line discharges (cf. Stolzenbach and Harleman [94]).   More basic studies
concerned with turbulent transport quantities in jet flow have also
received much attention.
      In this section we will briefly outline the state-of-the-art and
past studies dealing with plume calculations.  Table 2.1 summarizes a
good share of the work related to plume investigations both theoretical
and experimental.  This table is by no means all inclusive and the
particular categories may not be completely descriptive of the work
accomplished in the cited references.  However,  it does serve to illus-
trate where research emphasis has been placed on problems which are
related both directly and indirectly to thermal  outfall analysis.
      A brief discussion of Table 2.1 will  be given separately for
submerged and horizontal shoreline outfalls.
                                  23

-------
TABLE 2.1.   SUMMARY  OF WORK PERTINENT TO  OCEAN  OUTFALL PLUME ANALYSIS















Principal
Investigator
Albertson
Albertson
Balnes
H1nze
Schmidt
Rouse
Priestley
Priestley
Morton
Morton
Aoranam
Abraham
Fan
Fan
Keffer
Cederwall
Brooks
Tomlch
Zeller
Jen
Tamaf
Hayashl
Sharp
Frankel
Saml
1 Stolzenbach
Hart
Bosanquet
Hoult
Wada
Wada
Manabe
Okubo
Okubo
Leenderste
Ramsey
Fay
Murota
Masch
Fox
Murgal
Hirst
Schmidt
Hirst
Scorer
Morton
Csanady
Anwar
Abraham
Turner
Rawn
Harremoes
Tulln
Baumgartner















Ref.
4
4
8
41
85
31
73
74
60
58
1
j
26
27
50
17
16
99
112
48
96
38
88.89
30
83
94
37
14 •
45
106
108
56
64
65
53
75
29
62
57
28
61
44
86
43
87
59
22
5
2
102
76
36
101
11















Application
Sub. Jets
Sub. Jets
Sub. Jets
Sub. Jets
Thermals
Thermals
Plumes
Plumes
Thermals
Plumes
Waste outfall
Jets
Waste Outfall
Waste Outfall
Plume
Waste Outfall
Waste Outfall
Jets
Thermal Outfall
Thermal Outfall
Thermal Outfnll
Thermal Outfall
Thermal Outfall
Waste Outfall
Jets
Thermal Outfall
Waste Outfall
Waste Outfall
Plumes
Thermal Outfall
Thermal Outfall
Thermal Outfall
Dispersion
Thermals
Tidal Hydraulics
Heated Jet
Plumes
Jets
Tidal Hydraulics
Plumes
Thermals
Plumes
Smoke Plumes
Plumes
Plumes
Plumes
Plumes
Waste Outfall
Waste Outfall
Waste Outfall
Waste Outfall
Waste Outfall
Waste Outfall
Waste Outfall



Geometry




c
3
S
4, iS«
u *-*"5
•o a o a:
C (/> J=
§t/i o
- -V
°^ 'to *n ~~*
Q C C >
£S8S
tjrt.=
X L
X L
X L
X L
L
L
X L
X L
L
X L
XX L
L
XX L
X L
X L
XX L
X L
X L
X
X
X
X
X L
X I
X
X
X L
X L
X L
X
X
X
NA
L
NA
X
X L
X I
NA
X L
L
X X L
X L
XX L
X L
X L
X L
XX L
X L
X L
X L
X L
X L
X L


Type of
Flow








^ r- f—
IO-Q-
C r—  t-
c C c t_>
01 4> C I/I
o *0 0* t/i
X X
X X
X X
X X
X X
X X
XXX
X X
X X
X X
A A A
X X
XXX
X XX
X XX
XXX
X X
X X
X X
X X
X X
X X
X X
X X
X X
X XX
X X X
XXX
X XX
X X
X X
X
X
X X
X
X X
X XX
X X
X
XXX
X X X
X X X X
XXX
X X X X
X X X
XXX
X X X
X X
X X
X X
X X

X X X X
X X
nncipai
Zone
Inves-
1 gated










r-  ^
C C C C
o o o o
Mt-JMM
X
X
X
X
X
X
X
X X
X
X
A A
X
X
X
X
X
X
X
X
X
X
A
X
X X
X X
XXX
X
X
X X
X X
X
X
X
X
X
XXX
X X
XXX
X
X
X
X
X
X
X
X
X
X
X
XXX
X

X X
X








+J
«
o
^
u
o
C «J
I £
i- X
£•=
X C
UJ <
X X
X X
X
X
X X
X X
X X
X X
X
X
A A
X
X X
X X
X X
X X
X
X X
X X
X X
X X
X X
X
X
X
X X
X
X X
X
X
X
X
X
X
X X
X
X
X
X X
X X
X
X
X
X
X
X
X
X X
X
X X
X

X
X X


Solutlor
let hods'
I/I
rt
s.
is
c
o
c
g
o5
° 1.
• 4-1 O
t— ia ^.
O 3 -0
(/i a* c
UJ ro
4J C 1—
T -2 3
f2 -3 T
"ifi O ul
X
X
X

X
X
X

X
X
A
X
X
X
X
X
X X

X
X
v
X
X
X

X X

X
X X



X
X


X
X

X
X
X
X X
X X
X
X
X
X
X
X
X

X
X





-D
1n1te
Iff.
c
o
^
o
c
u.
gi
Is
S x
£
(0 •«-
II

















X











X
X
X


X



X















     *The L/D ratio applies only to submerged outfalls: S, I/O <5; I, 5
-------
2.2.1  Submerged Outfalls
      For submerged outfalls the depth of discharge dictates the method
of analysis.  Deep water cases are substantially simpler to analyze
than the shallow water counterparts (at least in the absence of cross
currents) which is a result of the applicability of similarity solutions.
Similarity  analysis has expedited  the theoretical analysis in this zone
and  resulted in mathematical models that are sufficiently accurate for
engineering calculations.
      Zone  1 has received substantial attention but is of minor impor-
tance in deep water analysis because it is a relatively short-distance
effect (approximately six port diameters or less).  Most of the work
involving this zone has been carried out in the absence of buoyant forces.
Abraham [1] presents a mathematical model for cases where buoyant forces
have a significant affect on the zone length.  Recently, Hirst [43]
has  presented a more thorough analysis.
      There has been essentially no theoretical work done for zone 3 of
the  deep water plume (i.e., near the surface or in the region of the
maximum height of rise).  It is generally assumed that the similarity
solutions of zone 2 hold in zone 3; but, this is a very poor assumption.
Frankel and Gumming [30] have shown through experiment that this is the
case.  Sharp [88,89] has experimentally investigated the surface spread
of a hot water plume, and Murota and Muraoki [62] have investigated the
effect of a free surface on plume  hydrodynamics.
      Very  little theoretical work has been done on deep water plumes
in the presence of a crosscurrent.  This lack of effort is undoubtedly

                                   25

-------
a result of the solution difficulty since similarity principles are not
strictly valid for this case.  However, Fan [26] has treated the cross-
flow problem for a vertical plume using similarity assumptions and
obtained reasonable results.  There are serious theoretical questions
concerning the use of similarity profiles in the presence of a cross-
current.  Hirst [44] presents analysis for crosscurrents which includes
a stratified ambient medium.  Various experimental studies coupled with
dimensional analysis have  been carried out for the crossflow problem,
but as yet no generally proven computational model has been published
which relates details of the plume dynamics.
      Deep water plume analysis is particularly applicable to waste out-
falls having small ports,  common to diffuser systems.  Typical submerged
thermal outfalls such as those off the Southern California coast cited
by Zeller and Rulifson [113] utilize very large, vertical single ports.
The amount of receiving water between the port and sea surface may be
on the order of 1-3 port diameters.  No published theoretical studies
have treated plumes with such L/D ratios.  In this case zone 2 does not
exist and there is no delineation between zones 1  and 3.  All that may
be said is that the flow undergoes a transition from pipe flow to drift
flow.
      The following general conclusions are made concerning submerged
outfall state-of-the-art computational models.
   1. Acceptable computational models are available for deep water
      plumes except for;
      •  Zone 3, the surface or maximum-height-of-rise transition
         zone, and
                                26

-------
      •  plumes  issuing in crosscurrent (existing models  to be
         proven).
   2.  There is  no  acceptable computation model  or technique available
      for shallow  water plumes such as those typical  of large thermal
      power plant  outfalls.

2.2.2  Horizontal  Shoreline Outfalls
      Horizontal shoreline  discharge is also utilized by a number of
thermal power plants throughout the United States.  Table 2.1 illustrates
that there has been only modest effort made to analyze this problem.
From a mathematical modeling standpoint the horizontal surface discharge
of a thermal plume is extremely complex since the phenomena involved
are inherently three-dimensional (the same is true for horizontal  sub-
merged ports in shallow water, and the case of a crosscurrent in deep
water).
       In spite of the three-dimensional aspects of the shoreline plume,
various solutions have been formed using similarity princples (e.g.,
Zeller [112], Jen, et al. [48], Hayashi, et al. [38], Tamai, et al.[96]
and Stolzenbach, et al. [94].   Except for the work of Stolzenbach,
none of these methods are, in this author's opinion, acceptable for
engineering computations.  Before a completely acceptable model is con-
structed for general application, three-dimension flow characteristics
will need to be accounted for in some manner along with crosscurrent
effects.
                                27

-------
2.3  Work Description
      The previous section delineates several  areas of outfall analysis
which need attention.  As a practical matter it is not feasible to
incorporate all of these areas into a general  mathematical model which
would apply to all outfall configurations and oceanographic conditions.
      The scope of this manuscript is limited to vertical plumes.
We are primarily  interested in large single port vertical thermal
outfalls issuing  in shallow water (Figure 2.4).  Typical existing
configurations are those located at Moss Landing, San Onofre and
Redondo Beach, cited earlier.   However, the ultimate objective of the
work is to provide a complete program which mathematically models the
temperature and velocity distribution in a vertical thermal  plume, from
outfall port to the drift flow regime (zone 4), regardless of ocean
depth.  The transition region, as defined here, refers to any part of
the flow field for shallow water plumes.  This region is the portion of
the program which must be treated by finite-difference techniques and
constitutes the principle effort of this work.
      In addition we also set down the difference equations appropriate
for a line plume, but do not include these in the modeling program.
      In summary, the work covered by this manuscript deals with the
problem of mathematically modeling velocity and temperature distribu-
tions in the locale of vertical thermal outfalls.  The techniques for
analysis are as follows:
      •  Shallow  water plumes  : finite-differences
                              28

-------
      •  Deep water plumes
   1.  Zone 1  : existing empirical
   2.  Zone 2  : similarity solution
   3.  Zone 3  : finite-differences
      The primary task described in this manuscript is the finite-dif-
ference application to the confined plume and computation of the entire
flow field dynamics for zones I, II, and III.  The circulation of the
ambient is also included.  Although there have been various related
studies, none have dealt with the  numerical  solution of a confined,
vertical plume and radial surface  spread.  Tomich [99] numerically
modeled the compressible free jet problem, Ma and Ong [55] investigated
an impulsively started momentum jet, but paid little attention to the
more complicated features of the dynamics.  Recently, Pai and Hsieh [68]
have carried out numerical work with laminar jets.
                                 29

-------
                              CHAPTER 3

                 TRANSPORT EQUATIONS  - GENERAL  THEORY

      In this chapter the fundamental  laws  and  equations  which  govern

marine hydrodynamics and energy transport are set  down. We  begin  by  con-

sidering the fundamental equations  for laminar,  incompressible  flow  and

modify these equations so they are  appropriate  for marine considerations.

      These equations are written in  various forms which  are appropriate

for later discussion concerning theory review,  similarity solutions, and

numerical considerations.


3.1  Coordinate System


      The governing differential equations  are  given in Cartesian

tensoral form with coordinates x- (Figure 3.1).  For analysis of  the

local  sea, the geopotential  surface is assumed  to  be flat.
                                  (EARTH  ROTATION)
                                     VERTICAL:
                                                 EARTH
                                                           Figure  3.1

                                                           Rectangular
                                                           Coordinate
                                                           System
                              30

-------
3.2  Conservation Laws




      The differential equations governing the heat and momentum trans-



port of a thermal plume in the oceanic environment may be derived from



the following physical laws:



      •  Continuity (conservation of mass)



      •  Newton's Second Law (conservation of momentum), and



      •  The first law of thermodynamics (conservation of energy)



In addition, an appropriate equation of state is needed to relate sea



water density in terms of local temperature and salinity.
                                                              i


      Detailed derivation of the primitive conservation equations will



not be discussed here but may be found in such texts dealing with fluid



dynamics (cf. Bird, Stewart and Lightfoot [13], Welty, Wicks and



Wilson [115], Hinze [40]). A few modifications of the standard form of



the conservation equations must be made so that they apply in general



to a thermal plume in the sea.  These modifications are chiefly con-



cerned with turbulent approximations, incorporation of coriolis effects,



and the Boussinesq approximation concerning small density variations.



Additional  detail concerning these approximations may be found in



standard references dealing with marine hydrodynamics (cf. Hill [39],



Phillips [70]) and the general subject of turbulence (e.g. Hinze [40]).



      The primitive equations appropriate for our analysis are pre-

                                 i
sented in Cartesian tensoral form  as follows:
^insteinian notation is used where repeated indices imply summation

 over all three index values (i = 1,2,3).
                                31

-------
      Continuity:
      The operator D/Dt in the above equation  is  the  substantial  deriva-


tive and has the usual  meaning:
           D_  _  3_

           Dt     3t

where t is time and u.  is velocity along the j    coordinate.   In Equa-
                     J

tion (3.1) the quantity p is density.


      Momentum:


           Du •                     sp             3Tii

                                                +                <3-2'
where n* is the component of planetary angular velocity along the j
       J

coordinate, P is pressure, g is the local  gravitational constant and T-.
                                                                      ' J

is the fluid molecular stress tensor.   The symbol  e- ..  is the usual


cartesian permutation tensor which takes values of zero if any two of


the three subscript are identical, +1  for even permutations and -1 for


odd permutations.  The symbol 6.. is the Kronecker delta which is equal
                               ' J

to 1 when i = j, and otherwise 0.  Coriolis effects are incorporated

                                                *
into the momentum equation by the term e- .,2pu^n-  and, according to



the specified coordinate system,  (Figure 3.1) gravitational forces act


only along the x7 direction; hence, 6.. = 6.,.
                •j                    i j    i j


      In any fluid dynamic system, variations of density may cause



fluid motion due to the action of gravity.  In the ocean, these density



variations may be caused by temperature differences and variation  of



local salt content, or concentrations of other materials whether  in
                                32

-------
solution or not.  Hence, in lieu of the heat transport equation we will


consider at this point a transport equation for a general  scalar


quantity, r, where r may be heat, salinity or other dilute transferable


constituents.  The r transport equation is:



            Dr  _  9   /   3T \  .  :                             /., ,x

           ~Dt  - a*7  (KY^7J      '                            (3'3)


Constituent sources, sinks and dissipative mechanisms are  incorporated


in the term $ and the symbol  <  is the molecular diffusion coefficient


for the r quantity.


3.2.1  Continuity


      In the ocean, and especially in the case of the thermal  plume,
                                                         -»

the density field, p, varies  with both space and time,



            P  = p(xist).                                        (3.4)



However, essentially all density variation is caused by distributions


of heat content, salinity, etc., as opposed to compressibility effects


(i.e. high speed compressible effects).  The local density anomally is


very small compared to the local value of density, and the conservation


of mass (Equation 3.1) may be approximated with sufficient accuracy by


the volume continuity equation


            9U.
              l  -  0.                                           (3.5)
We point out here  that although f^- = 0 may  be an acceptable approxi-


mation with regard to mass conservation, this quantity cannot be
                              33

-------
ignored in the momentum equation (see Section 3.3), and is precisely
the coupling between momentum transport and r transport.
3.2.2  The Equations of Motion for Turbulent Flow
      Within the framework of assumptions concerning continuous fluid
properties, constant gravitational force, and negligible earth curva-
ture, the momentum transport equations (3.2) are valid regardless of
the nature of the flow or fluid.  The usual additional assumptions in
hydrodynamics are that the fluid is Newtonian, incompressible and that
Stokes viscosity relationships are a valid description of the fluid
stress rate-of-strain (cf. Welty et al.).  Thus, the stress terms
(Equation 3.2), T.., may be replaced by
                 ' J
             3U,
               J
where y is dynamic viscosity.
      For the purpose of treating turbulent flow, it is assumed that
the velocity components, ui ,  pressure, P, and density, p, are composed of
mean or average parts and superimposed random fluctuating parts
(cf. Hinze [40]).  Symbolically,
        ui ="i  +">
        P  = P"+ P',  and
        p  = p  + p',
where the everbar represents mean of values and the prime, random
values.  These definitions are substituted into the equations of
motion and the result is time averaged term-by-term over a sufficiently
long period of time to obtain
                               34

-------
    Du .    a
    or   —.  u'iu    +  eijk«juk   = - 37:  -'96i3  +  ^r-   (3.7)
                                           I
which is seen to be identical  in the mean  motion  with  Equation (3.2)
except for the appearance of the term
A new quantity is now defined:
which is called the Reynolds  stress.   Finally  the complete equations
of motion in the rotating Earth  reference  frame are written as
for the mean flow.   Here the  overbars  denoting average quantities have
been omitted since  mean, or average, quantities are implied.  The tur-
bulent stress terms may be  related  to  mean flow quantities through the
Prandtl mixing length  theorem (cf.  Neumann and Pierson [63]) to obtain
 Terms envolving  fluctuations  of  pressure and density have been ignored,
                               35

-------
         R.. = -pu'.u'.  =  PC.:; 37-  •                           (3.10)
           1J       1  J            j

Hence, using Equations (3.6)  and (3.10)  in (3.9)  yields
                                                                 (3.11)
                                                                     i
  /Du.                \       ap                 s    -  •           3Ui-
 . I  	L  +   ?p    0*11     =   - —-—
 PI   n+-   T   teT i lx"iul/  /       av
where e.. is the eddy diffusion coefficient for momentum, a second order
       ' J
tensor, and v is kinematic viscosity.
      In the case of a thermal  plume,  e. -»v except where velocity
                                       ' J
gradients are small and the flow has strong stratification.  We will
assume that e.. includes molecular viscous effects and write the momen-
             ' J
turn equations as
    Du,
    DT
3.3  The Boussinesq Approximation

       In this work, four quantities of density are defined as follows:

       •  P =  p(x.,t), the density at a point in the thermal plume.
       •  P =  P00(x3), the density distribution which would exist in
              the local sea in the absence of the plume.
       « pr =  Constant, a reference density for the receiving water.
       « pQ =  Constant, the density of the effluent issuing  from
 _ the outfall port. _
1 The  summation convention for repeated tensoral indices  does not  apply
  to underscored indices in this text.
                                 36

-------
The density distribution of the reference ocean,  P^xOj  is  assumed to

be independent of time and vary with x, alone.

      Buoyant forces on a fluid element are established  by the  density

difference

            Ap =  p - p^ .                                        (3.13)

So that,

            P = PTC + Ap.                                         (3.14)

      According to the Boussinesq approximation,  (cf.  Phillips  [70])

when density variations, Ap, are small,(i.e.  |Ap/p|«l)  these  varia-

tions may be ignored as they influence inertial  and viscous  terms  in

the equations of motion, but must be accounted  for in  the gravitational

term.  In view of Equation (3.14), the equations  of motion may  be

written
Du-
Now, let P° be the pressure difference between a point in the plume

and outside the plume located on the same geopotential surface, so that
                            .  L.  . g

                     
-------
Here, we have assumed that the pressure distribution in the reference


ocean is hydrostatic.  Hence, Equation (3.15)  may be reduced'to:
 Du.
                                                 JL   e..Zl)  (3.17)

                                                 3xj  \
Equation (3.17) is the so-called "advective" form of the equations of


motion.  This name has become popular among oceanographers and meterol'


ogists and is so called because the convective terms are expressed in


the form u.au./ax..
          J  '    J

      The convective terms may be written in slightly different form


by noting that


      au .u.          au •           au .
       ax.
        J
             =  u.
J ax
u .   — d-
 1   ax.
      J
 However,  by  Equation  (3.5)
       3u.
     j  ax..
         J
              =   0,
so that for an incompressible flow
      J ax.
         J
Thus, Equation (3.17) may also be expressed as
     at
                                       9PC
                                                 au.
                            'i3
                                                                 (3.18)
                                 38

-------
 which is  called the  "conservative"  form of the equations  of motion.

 3.4  The  Pressure  Equation

       Equation  (3.17),  or (3.18)  contains  four unknown quantities;
 u-pUpjUoj and P°.   Since only three scalar equations  are  involved,  an
 additional  relationship is  required.
       An  equation  for pressure,  P°, may be derived by taking the diver-
 gence of  Equation  (3.17).  This  operation  yields:
   __
   9t
            Q  3(p«,-p)
   3X.3X.
                                  LI,
               J  9X:
         9     f 9        9U1 1
        ax.    [ax.   eij  3x.J
                                                    =  0.
                                                            311.
                                                                  (3.19)
By continuity
      ax.
                0,
so that Equation (3.19) is reduced to
     92P°
     9X.9X-
9U-J
  vJ
                 9X.
                   J
    ax.  /  \ 9x.
      1  '  x   J
j 9x.
 where B is the buoyancy parameter, defined as
                               ax,
                                              (3.20)
                                                                  (3.21)
                                                                  (3.22)
                                 39

-------
For the case where coriolis forces are neglected and quantities involv-
ing derivatives of eddy viscosity are small  compared to other terms,
the presssure equation is
                                                                 (3'23)
3.5  r Transport
      The r transport Equation (3.3)  may be modified for turbulent flow
by considering the transported quantity, r, to be composed of a mean
part, r, and a fluctuating part,  r',  or
            r = f + r' .
Then in a manner analogous to the method applied to the equations  of
motion, the turbulent r  transport equation becomes
            or  _   s    /      sr\   ,   •
            Dt  '  8x7  1 £Yj  377 1      $  '                        (3.24)
Where e •  is the eddy diffusion coefficient and  is  assumed to include
       i J
molecular effects.
3.5.1  Transport of Heat, Salinity and Buoyancy
      Letting r = T, in Equation (3.24),  where T is temperature, the
heat transport equation is
                 -   8       9T
                          Hi 8y
                          HJ  X
                                                                 (3.25)
                               40

-------
In this case * corresponds to heat sources and sinks and/or viscous
dissipation.  Since none of these effects are significant in an ocean
       •
plume, * is neglected.  The quantity eH- is the turbulent heat diffusion
coefficient and is assumed to include molecular effects.  For salt trans-
port, we let r = S, when S is salinity; hence,

            DS  _3/3S\                               / •} ne\
            Dt  -  W.  (tSjJT}'                               (3'26)
                     J         J
Salinity is a conservative property, thus i is omitted.  The quantity
e^j is the combined molecular and turbulent mass diffusion coefficient.
      The equations for heat and salinity transport are coupled to the
Equations of motion (3.17) or (3.18) through the buoyancy term
(pep-p)/Pr.  For that matter, any r constituent, which when transported
in the system of interest causes density variations to occur, is
coupled in the same fashion.  Thus,  it is not the absolute value of
temperature, salinity, etc., which is important to the system dynamics,
but resulting density variations in a lateral plane caused by the trans-
port of these quantities.  For this reason it is necessary only to deal
with the transport of buoyancy 1n analyzing the dynamical behavior of
the system.  However, we must solve the equation for heat or salinity
transport (in a system where differences of salinity and temperature
are the causes of density variations) in order to establish the magni-
tude of temperature and salt content, and to treat certain boundary
conditions.  Once the density and temperature (or salinity) distribution
is known, salinity (or temperature) may be calculated from an equation
of state for sea water.
                                41

-------
      A "density transport"  equation may  be  derived  by combining  Equa-
tions (3.25) and (3.26)  [assuming  that  an equation of state,  p = p(S,T)
holds] after the independent variables, T and S, have been  changed  to p
Hence,
 fi.  =  _L_  (.   to- )  + .          '    -                (3  27)
fit     ax.  U. ax, '    £P  pax I  lax.)               {6^J}
                   P. ax
where
                                                               (3.28)
 A buoyancy  parameter may be defined as
             Pr  " p
             pr  ' po
 and the  appropriate  transport equation for AI  is
                                                               (3'29)
      nt         aV   ^t^-i 1w^ '      t  -il av  I   I 3^ /
      Ut         dX-     pj  dX-        Pj\c'x-i/   \   i '

 A second parameter  A2  which  incorporates p^ may  be defined as

      A2   =   A!  -  (   r  _   °°  )  =  —	   '                      (3.30)

 The transport of  A    is described by
                               42

-------
            u
                       --             __
      Dt      3 3X3     ax.    pj  3X.  ' " 3X3    p3  3X
                         J        J                3
where  p*   =  pM/(pr - PQ)





If density  is a linear function  of both temperature  and salinity,  that




     p  - PO = _a (T - TQ) + b(S-So),




then ^ = constant and H/3x.  =  o.
                           J
Equations  (3.27), (3.29) and (3.21)  become
                       3A1
,   and
respectively.
                             43
    _     _ —   c  .  -   ,  an                            /o oo\
    Dt   =   ax.  VEPJ  ax.  ;                                  (3.33)
               J         J

-------
      The quantity c, is seen to be a correction term which accounts for
nonlinearities in the equation of state, p = p (S,T).  As it turns out,
sea water density does vary approximately linearly with salinity
(See Section 3.6) so that p = f (T) for. constant S.
      In the remainder of this manuscript, A-|  and A2 will be referred
to as
            pr - p
         -  — ! - '   buoyancy parameter,
            pr - po
                   - P
   • A2  =  -   =  AJ -  AI   ,  density disparity parameter.
                  -
               Pr - PO
The motivation for defining two buoyancy quantities is that it is more
convenient to use A-, in the numerical  analysis (Chapter 5), whereas Ap
is convenient for similarity analysis.  For consideration of salinity
transport, a third buoyancy term is defined as
                  S  - S
                  Sr    ID'

where Sr and SQ are the reference and  outfall  effluent salinities,
respectively.
      Figure 3.2 illustrates the relationship  between the quantities
and A-| and A2 at elevation X3 = constant.
                              44

-------
                                                                   CO
                                                                   UJ
                                                                   UJ
                                                                   Qi
                                 X2
 Figure 3.2  Relationship Between the Buoyancy Parameter, A-,  and
             Density Disparity, Ap
3.6  The Equation of State for Sea Water
      The density of sea water is a function of pressure, temperature
and salinity, in the absence of other pollutants.  Hence, the equation
of state has the form
            P = p(P,S,T).                                        (3.35)
Since we are dealing only with rather shallow water on an oceanographic
scale, pressure effects are negligible; therefore,
            P = P(S,T).                                          (3.36)
                              45

-------
If other contaminants, having concentration, C,  are present,  then
            P = p(S,T,C)  .                                        (3.37)

In this work, we will  deal  only with Equation (3.36).
      Since density variations are small  in the sea, oceanographers
deal with a modified density called sigma-t, defined as
            ot = (p-1) x  1000,
which has cgs units and is  a measure of the deviation in density from
1.0 gm/ml .   The equation  of state in general use by oceanographers may
be found in U.S. Navy Hydrographic Office publication number 615 [103]
(or in Hill [39]) and has the form:
       °t  = Et + (ao + J324) [1  - At + Bt (00
where
It  •
                -  (T - 3.98)    „   T + 283
                    503.370          T + 67.26
        At  =   10" T(4.7867  -  .098185T +  .0010843J2)
                 _6
        Bt   =   10~~T(18.030  -  .8164T +  .01667T2)

            =   -.093 +  .8149S  -  .000482S2 +  .0000068S2,
                              46

-------
      In the above equations,! is in degrees Celsius, and salinity in


parts per thousand.  The quantity a  is the density of sea water, in


sigma-t units at zero pressure and temperature,  a  is usually expressed


in terms of chlorine content instead of salinity, S, but for purposes


here, salinity will suffice.



3.7  Vorticity Transport - An Alternate Approach



      In dealing with geophysical fluid dynamic problems it is frequently


difficult, if not impossible, to set realistic boundary conditions


required for the solution of Equation (3.21).  Pressure, and consequently


associated boundary conditions, may be eliminated entirely from consid-


eration by introducing the quantity, vorticity.


      A brief summary of the general theory will be presented here for


a homogeneous, isotropic turbulent flow field (i.e., e^- = e = constant)
                                                      * V

in three dimension.  Additional information concerning vorticity trans-


port may be found in Batchelor [10].


      As demonstrated by Batchelor, a conservative fluid velocity field


may be defined by vector addition of an irrotational contribution, Uj


and a solenoidal contribution us, or



            u = Uj + us.                                         (3.39)


The solenoidal part satisfies



            v • us = 0



whereas the irrotational part satisfies



            v x U  = 0.
                                 47

-------
In addition the irrotatlonal part of the velocity field, Uj, may be
described in terms of a scalar potential » so that
            UT = V*
and the solenoidal part in terms of a vector potential, $, or
            jj  = v x ^ .
Hence, the total velocity field is described by the vector and scalar
potential as
            u • 7* + 7 x  $.                                      (3.40)
Vorticity, u>, is defined  as
            U) = V X U.
Taking the curl of Equation (3.40) and use of the above expression for
vorticity, ytelds
            u • v x (7x$).                                       (3.41)
However, by vector identity
            v x (vxf) = 7(7-$) - v2? ,
which for an  incompressible flow gives
            v * " -w                                             (3.42)
since 7«^ = 0.
      Equation (3.42) is  a Poisson type partial  differential equation
relating the vector potential  to the distribution of vorticity in the
flow field.
                              48

-------
      The divergence of Equation (3.40) gives
             2       -»•
            7 4 • V- U
In view of the incompressibility condition,
            7 • u  =0,
and satisfies LaPlace-'s equation
            72$ = 0.                                             (3.43)
Hence, the velocity field may be established through solution of
Equations (3.42), (3.43) and (3.40).
      Hirasaki and Heliums [42] have shown that Equation (3.43) is
extremely useful for the purpose of prescribing inflow-outflow boundary
conditions in a three dimensional velocity field.  In fact, they have
demonstrated that the flux boundary condition may be prescribed
entirely by the scalar potential, * (velocity potential), or Uj.
Hence, one is permitted to set tangential components of $ =  0 and the
normal derivative of $ = 0 at all boundaries.  The utility of this
theory lies in the fact that vector potential boundary condition may
be intractable without consideration of the scalar potential, *.  One
exception is the case of flow in a closed system where the boundary
conditions on $ remain as described above and since there is no
boundary mass flux,7$ = 0 everywhere (cf. Aziz [7]).
      An equation for vorticity transport may be derived by taking the
curl of the Equations of motions (3.17) (after setting e.. = e).  This
                                                         ' J
operation yields
             ->•                     A                              i,
                                   49

-------
where e3 is a unit vector in the vertical direction.
      The vorticity transportation equation was simplified appreciably
by assuming a homogeneous, isotropic turbulence field.  If the turbu-
lence field were not treated as such, numerous terms involving the
gradient of &•• would appear.  These terms will be investigated in
             ' \j
Section 3.10, which covers two-dimensional flow fields.  The two
dimensional counterpart to Equation (3.44) is
            §| = V x Be3 + eV2u ,                                (3.45)
where one coordinate is vertical (x.,) and the other lies in the
lateral plane.
3.8 Non-dimensional Form of the Equations of Motion
      A non-dimensional formulation of the equations of motion permits
the investigation of the magnitude of the various forces exerted on a
fluid element in terms of similarity parameters.  The importance of
the various parameters may then be analyzed on an order-of-magnitude
basis and the results used to justify simplification of the governing
equations under certain flow conditions.  To this end, we define the
following dimensionless variables:
            ui   =  W
            P*  =  P°Pr/AP0,
            **       *
            t*  =  tv0/D,
            Xi  =  xi/D>
            ij  =  eij/eo-
                                                                  (3.46)
                                50

-------
 In  the  above,
        v  -  reference  velocity (for the thermal  plume  we  will  use  the
             effluent velocity  at  the outfall  port),
      AP  -  reference  dynamic  pressure  (may be  taken as  1/2  p  v  ^)
        f  -  characteristic  coriolis parameter
         D -  characteristic  length (may  be  taken  as  the outfall port
             diameter)
        e  -  characteristic  eddy diffusion  coefficient  for momentum
            (may be set to Cv D, where C is a constant).
 Substituting the set (3.46) into  Equation  (3.18)  yields,

       3Ui    8UjU'         /fo°\  **      / APo   ^   3P*

The dimensionless groups in Equation (3.47)  are:
         vn2
         YJ- =  Ro, Rossby number (ratio of inertial  forces  to coriolis
          o
                forces),
           2
       Prv
        P    =  Eu, Euler nipber (ratio of inertial  forces  to pressure
                forces),
      vn2    .
     —	    F ,  densimetric Froude number (ratio of inertial  forces
        \  n
        )gD
        /
.   ,        to  internal  buoyant forces),
                                51

-------
        VD
            =  Re-,-,  turbulent Reynolds number (ratio of inertial
        eo
               forces  to turbulent shear forces).
In terms of the above  similarity parameters Equation (3.47) becomes
                     J
            T_ 3P*   +  l_fi.    +   J	L_(e*
            Eu 3X-;      Fn   "i       Rej  3X-; \  la~---i
                 i       o            i    j    —   j
                                                               (3.48)
      Equation (3.48) represents a gross non-dimensionalization.
Ideally, we should treat each component of momentum separately and use
length scales which correspond to the particular coordinates.  However,
for purposes here the form of Equation (3.48) is sufficient.
      At middle latitudes, the characteristic coriolis parameter, f , is
approximately equal to 10, and vQ/D has magnitude on the order of 1
for a large outfall part.  Hence, the Rossby number for the thermal
plume is on the order of 10,000.  Where smaller ports are considered
vQ/D may be from 10 to 100, giving Rossby numbers from 105 to 10 .
The densimetric Froude number, F  , for a large thermal outfall will be
on the order of 10-50 and the reference Reynolds number ReT will be of
the same order.  Also, we cannot neglect pressure effects.  All other
terms are on the order of 1 except eddy coefficients in some portions
of the flow field.  Hence, it follows that for a thermal plume  and the
scales of motion to be considered here, the coriolis term is suffic-
iently small to neglect by virtue of the apparaent size of the  Rossby
                                   52

-------
number.  In consideration to follow we will deal with the equations of
motion in the general form of

          aUj  +  8UjUl   =      ]  ap*    ,

                +  Re?  a!j WSTjf*                           (3.49)
and dimensional  variations of the same.
3.9  Further Comments on the Concept of "Eddy Viscosity"
      In Section 3.22, we introduced velocity fluctuation, u., as a
means of describing turbulent flow.  Without the coriolis term,
Equation (3.9) is known as Reynolds' equation, after Osborne Reynolds
[78] who first expressed the turbulent equations of motion in this
fashion.  Reynolds' equation for the mean flow differs from the laminar
flow counterpart only by the Reynolds stress terms, R.^
                                                     1 J
      The Reynolds equation represents a vast simplification (at least
outwardly) of extremely complex flow conditions.  However, the task
still remains in relating the turbulent or "apparent" stresses to mean
flow quantities.
      Boussinesq (cf. Hinze [40]) was evidently the first to use the
concept of "apparent" viscosity, in his studies of two-dimensional flow.
He assumed that turbulent stress, T  could be expressed  in a manner
analogous to molecular viscous stress or

           i   =  -p u1 v"  = e -3—   •                           (3.50)

                               53

-------
In the above, e is the "apparent" or eddy viscosity, u1 and v' are x
and y components of the velocity fluctuation, respectively, and u is
the mean velocity in the x direction.
      Prandtl [72] introduced the concept of "mixing lengths" to
describe the turbulent exchange coefficient.  This idea was motivated
by the mean free path concept of molecular motion and has turned out
to be a fruitful hypothesis in spite of obvious physical questions.
      The idea of mixing length theory is that a small parcel of fluid
containing any transferrable property is transported, unchanged, by
velocity fluctuation from one position, a distance £ to a new position
where it is absorbed in the flow field.  The distance a is the mixing
length.
      Let u-|(x-| ,x2,x3) be the mean velocity at the origin of the
exchanged fluid parcel, and u-)(x1 + a^ ,x2 + £2»X3 + ^3) be tne mean
velocity at the absorbed position.  Then the velocity fluctuation is
(cf. Neumann and Pierson).
                   3U-,     3U-.       3U-,
                   oXn   t 0X0    O  dXn
                          du,
Then        u     = -£
            Ul(2) = - £2  a3cj                                  (3.51)
                          au1
            UK3) = - £3  9x7
                               54

-------
                                I
Here, the fluctuating velocity ui /.»  is shown as a second order ten-

sor where the subscript j indicates the particular turbulent component
    i
of u-,.   Hence, in a somewhat nebulous fashion:
                              —         9U-
                              ui>  =  eii air   •               <3-52)
                               J       U aXj

Mixing length theory is rather unsatisfying because of the physical

basis; nevertheless, it does accomplish the purpose of relating mean
flow behavior to the Reynolds stresses.  Actually, the concept of an

eddy viscosity requires a fourth order tensor quantity (Hinze, Pond

[71]) to satisfy theoretical treatment of the Reynolds stresses.

Such a quantity would be completely unmanageable from a practical

standpoint.  Even the second order tensor e.. is difficult, if not

impossible, to calculate from measurable quantities such as frictional

forces and velocity gradients.
      Hot wire and laser techniques offer a method for direct mea-

surement of the fluctuating velocities and hence correlation of the

Reynolds stresses through statistics.  However, statistical theory has

not yet provided a means for evaluating e-. in practical engineering
                                          ' J
calculations.
      As a result of our lack of understanding and inability to cal-r

culate or measure E.., further assumptions must be made.  In the ocean
                    ' J
we must deal with at least two values of eddy viscosity, a lateral

value and a vertical one.  Gross measurements have shown that these

two values are vastly different.  Fofonoff (cf. Hill) suggests using

a form from Saint-Guily which gives

                             55

-------
            Rij =  - '   -   ej^r  +  ei  a*?'            (3-53)
                                       J          '
where e- is the lateral  eddy viscosity  for i,j^3  and  the  vertical
       J
for i,j=3.
      For the work presented in this  thesis, we will  use  three
components  given by e..
                     J
3.10  Two-Dimensional  Forms  of the Transport Equations  in Rectangular
      and Axisymetric Coordinates
      In the previous sections of  this  chapter, the appropriate  differ-
ential equations for solving the thermal  plume  problem  in three-space
were layed out.  Ideally, we would prefer to solve the  plume  problem in
this manner since the nature of the flow is  distinctly  three-dimensional
However, computational  requirements necessary to  obtain proper  resolu-
tion of desired quantities in three dimensions  are prohibitive  from a
practical standpoint in view of available computer hardware and
economics.
      Two-dimensional considerations which demand significantly less
computation time and computer capacity, are  appropriate in cases where
flow symmetry is approximately realized.   Such  cases  are  the  vertical
plume and line thermal  investigated in  this  thesis.   Hopefully, compu-
tation economics will permit practical, three-dimensional engineering
calculations in the near future,  thus avoiding  certain  restrictions
inherent with two-dimensional approximations.  Table  3.1 gives a sum-
mary of general requirements for  two- and three-dimensional  forms of
the velocity-pressure and Vorticity-Vector potential  equations.
                              56

-------
    TABLE 3.1.   DIFFERENTIAL EQUATIONS REQUIRED FOR VELOCITY-
                PRESSURE AND VECTOR POTENTIAL-VORTICITY
                METHODS IN TWO AND THREE DIMENSIONS
                        Velocity-Pressure
                           Equation Set
Vector Potential-
    Vorticity
  Equation Set


ul
U2
U3
0
P
r

ul
(Oo
(i)-5
fl
\l/
m
$


Parabolic
Parabolic
Parabolic
Elliptic
Parabolic
(1 or more)
Parabolic
Parabolic
Parabolic
Elliptic
Elliptic
Elliptic
Elliptic
3-Dim. 2-Dim. 3-Dim. 2-Dim.
i=l,2,3 1-1,2 i=l,2,3 1=1,2
X X
X X
X
X X
XX XX

X
X
X X
X
X
X X
(X)*
Total of Required
Equations (minimum)
  7(8)
*Used only in the case of open boundaries.
                                  57

-------
3.10.1   Two-Dimensional Transport Equations in Rectangular Geometry



      The two-dimensional  rectangular coordinate system which will  be  con-

 sidered  in  this  study  is  defined  as  a  plane normal  to the geopotential

 surface  (Figure  3.1).   The  two  coordinates  are defined as x and z, where

 x  is in  the xl5x2  plane,  with no  particular orientation,  and z is aligned

 with the vertical  x3 axis.   Corresponding  velocity  components u and v are

 in the x and z directions,  respectively.


 Velocity-Pressure  Equations:

     The velocity-pressure  equations are  as follows.

      Continuity:

         9u   ,  9v    n
         9x     9Y  = U                                             (3.54)


      Momentum transport:

          x-direction,

          DU  _     9P°  ,  9  f  9ul .  3   f    9u]
          Dt  "   "ax    9X  leX3xJ   3Z~ [ez  3?J »                     (3.55)

          Dy_  _     9P°  + g + 3_ L ivj  +  3_ f   3v_j
          Dt       9z        9X  (  X  3xJ    9Z [ Z  9zl'                (3.56)


 In the  above momentum  transport equations,  ex is the  lateral eddy diffu-

 sivity  coefficient and ez is the  corresponding vertical  value.  The

 substantial derivative is given in  two dimensions as



                   T»  = "- +  V^-.
                   Dt     9X       9Z


      Constituent transport:

         P^T1      ^ll     ^iT1!      Cif     'NT1!
         L/l      O  I     Oi I   i^l     Ol !
         Dif  = "9x"(eYx^j   +  9! [EYZ ilj •                        (3.57)

                                        58

-------
Equations for the transport of specific constituents  such as  A,,  A2, S,

etc. will be developed where appropriate.


      The appropriate pressure equation may  be obtained from  Equation

(3.20) by letting i=2,3 and j=2,3.  Hence
       V2po=   (lii)2
       V K         ;
                                  IT
where  v2 = 	+ 	•
            9X2   3Z2
If turbulent contributions are neglected,



                   {<\iio       l^\i  Ci 11       si w i ? I     ^ R
                  / o U \ ^     o/^*\/^^\  _i  /   i  v  -i-  ^ ^*





Recall that by Equation  (3.16),



       P°  =  P/Pr  -  9 J2°Pco/Pr dz'



      The most notable work in obtaining  numerical  solution  to the


laminar form of the velocity-pressure equations given above  was per-


formed at the Los Alamos Scientific Laboratory by Welch and  colleagues


(cf. the "MAC Method" [109]).  Based on these  pioneering efforts at


LASL, numerous other investigations have  employed MAC techniques to


viscous flow problems [6, 23, 46].  Pagnani [67] applied the MAC



                              59

-------
techniques successfully to natural circulation in an enclosed  cell.
Stream Function - Vorticity Equations:
      An expression for the stream function in (x-z) coordinates may be
obtained by considering only the x3 component of the vector Equation
(3.43), or
            /* =  - u,                                        (3.60)
where: v (stream function) =  f- and w =   cn\
               Dt   =   - 97  +  £V w                            (3'64>
where  again we let  to =  u>3.

      However,  in general  we must  consider the two anisotropic, nonhomo-
geneous components  ex  and ez.  In this case, numerous terms involving
derivatives of ex and  e,,  appear.  The vorticity equations  are  derived
for  this case  by cross differentiating Equations (3.54) and (3.56),  ther
subtracting the latter result from the former to obtain
                              60

-------
             Dw_      !§.  +      92t»
             Dt     " 3X     GX  3X'
                                                       3e.
                    +
!!>L i!y.  + I!*, alii    IfiL i!y.    l!z.i!v
82 '9x2    3Z *3Z2 "  8X '3X2  " 3X '3Z2
3  f9ex   3U    _,   3eZ   3U
                      3Z [3X    3X        3Z    3Z

                      3  f9ex   3V    ^   9ez    3V
                      3X [3X    9X        3Z     3ZJ              (3.65)

If the structure of the turbulent field is  homogeneous, and isotropic,
Equation (3.65) simplifies to
                  =  .    +  e
              Dt       8x     X
      Stream function-vorticity transport solution methods have been
employed for a number of years by oceanographers in computing such geo-
physical phenomena as western boundary currents (e.g. the Kuro Shio and
the Gulf Stream,  cf; Neumann and Pierson).  But  these techniques have
become popular in engineering application only in the past few years, a
result due in part to the recognition that these methods are extremely
well adapted to problems involving natural convection.  Solution to the
laminar form of the stream function-vorticity equations given above
have been carried out by a number of researchers [7, 31, 82, 100, 104,
106, 108, 111].  The most notable work being carried out on the turbulent
form of the equations is at the Imperial College by Spa!ding and
                               61

-------
colleagues [69, 82, 90, 91, 92, 93].
3.10.2  Two-Dimensional Transport Equations in Axisymmetric Coordinates
      Again referring to Figure 3.1, the axisymmetric coordinate system
is oriented such that the radial  coordinate, r, may be considered a
rotating line in the x,, x2 plane.   The vertical coordinate, z, is again
aligned with the x3 direction, normal  to a geopotential surface.
Velocity-Pressure Equations:
      The velocity-pressure equations  are as follows:
     Continuity:
           1 9urr      av
          FIT   +  If  '  °'                                  <
          where u  is the radial  velocity.

     Momentum transport:
           r - direction,
                                          .
          Dt        3r     3r V r  9r )      9z  ^ z 2zj '       (3.68)

 The substantial  derivative in axisymmetric  coordinates  is:
           2_  =  JL+u   3_+  v 1-                              (3.69)
           Dt    at    r  ar     az
           z-direction:
           Dv       8P°  ,  n  ,  1  9   L   av]  .  a   f   av'i          fo
                        +  B + -TTT  re,- 7—  +  —  e, —-  ,        ^.
           Dt       9z  '  "  '  r 3r  lcr 9r  '  3Z  -z 3z
                               62

-------
In the above equations,  er is  the radial  eddy diffusivity  coefficient



for momentum.




     Constituent transport:



          Dr  _  1  8 (re    arl  ,  3  f

          Dt  -  ?37lr%r  arj  + al [EYZ




     where evr is the radial  coefficient  for turbulent  r diffusivity.
            1                                                         i



      The pressure  equation may  be derived  by differentiating  Equations



(3.68) and (3.70),  then  adding these two  results  to  Equation  (3.68).



Hence,


                             n       p      p

                           u]      faul    [3V|  + 2 au .  avl     (3<72)

                            j      (3rJ    [9zJ      3z   3rJ     v    7



         where the operator




           V2  =  ll.  +  ll_  +  iL.   .                      (3.73)

                  3r2     r 3r     9z2
    Vorticity in  (r-z)  coordinates  is  given  as
                         .
              3z       3r


    Also  we  define  a  stream  function  $  according  to



       u     =   _ 1  M.                                          (3.75)
        r       r  3z
   and
                                                               (3.76)
                                63

-------
 Substitution of Equations (3.75)  and (3.76)  into  Equation (3.74)
 yields
          £i_  l|t + l!i  =  .  ru                             (3.77)

for the stream function  *.   Note that Equation  (3.77)  is  not  the usual
Laplacian for (r-z)  coordinates  (e.g.  Equation  3.73).
      The vorticity  transport  equation is  derived  by cross-differ-
entiating Equations  (3.68)  and (3.70)  and  then  subtracting the  latter
result from the former.   This  operation leads to
     3u^ +  u^" + 3vai _     3B
     3t    3r    3z         3r
                          3Z2
       3z  Lr  3r    3r     3z    3z  J     3z    9r Lr  3r
                              r +  3v_    ;
             32^    dr Ldr    9r    9Z    3Z
  °-lL .  _r_  3_ T3v
  3z    372    3r L9r
. I!r .fiiy. +1 IV.T  ,  3ez  .  32v
  3r  'Lar2    r 3rJ
                            3r     aZ2                            (3.78)
If the turbulent structure of the flow field is  homogeneous,  and isotro
pic, derivatives of tr and ez vanish  and  the vorticity  transport
equation becomes
                           3B

3t    3r    3z        3r     r  3r
                                       r  3r  I   -*  9z2  •
                                                                (3.79)
                              64

-------
                              CHAPTER 4
                 PLUME THEORY - SIMILARITY SOLUTIONS
      As an integral part of the thermal plume dispersion program, this
chapter is concerned with flow regimes 1 and 2, which may adequately
be described by empirical correlations and similarity solutions.
4.1  General Description
      The zone of "flow establishment" (Figure 2.3) is a region of
transition from essentially a pipe flow at the outfall orifice to a
fully developed velocity profile some distance downstream.  This
situation occurs only in deep water, and when velocity profiles
become fully developed, the flow field is said to be "established."
This zone is characterized by velocity profiles which are very similar
in shape at each axial location.
      The zone of flow establishment is a region of intense turbulent
mixing between the plume flow and surrounding water.  The mixing
process which starts at the periphery of the outfall port spreads
inward  toward the center of the plume and outward into the surround-
ings.  Eventually mixing will spread to the plume centerline where
the centerline velocity will begin rapid diminution.  Upstream from
this point, flow in an approximate conical section is relatively
unaffected by the mixing process.  This zone is called the "potential
core" and is characterized by relatively flat velocity profiles at all
axial locations.
      Figure 4.1 illustrates a precise change from one flow regime to
the next.  In reality, however, before the velocity field becomes
                                 65

-------
  VELOCITY
vg   TEMPERATURE
                                                                v   >  v
                                                                m     o
                                                                     B. SMALL F
          A.   LARGE  F
Figure 4.1.  Zone of flow establishment for  plumes with large and small densimetric Froude numbers, FO

-------
fully established in the sense of similar velocity profiles, the
centerline velocity will begin to deteriorate giving a transition
zone between the two regimes.  This transition is apparent from the
data of Albertson et al. [4].  Although Murota and Muraoki [62] have
proposed a correlation  for this zone, according to Hinze [40] this
distance is relatively  short and is generally excluded from analysis.
      In the case of a  momentum jet (neutrally buoyant flow, or
FQ -»• «>) velocity in the potential core is that of the issuing jet and
analysis is based on the assumption that momentum is conserved at
each axial cross-section.  However, in the case of buoyant plumes,
momentum is generated by the density disparity and velocity will
actually increase in the potential core (as indicated in Figure 4.2B).
      As mentioned previously, the zone of established flow is
typified by velocity profiles which have nearly the same shape at all
axial locations.  For this reason similarity analysis has played an
important role in analysis of this flow regime.  Numerous experimental
and analytical studies  have been carried out for both the momentum jet
and buoyant plume in the absence of restraining boundaries.
      In this manuscript, the work of Albertson and Abraham [1] is
used for modeling Zone  1, and Abraham's work for the established flow
regime is extended for  the analysis of Zone 2.
4.2  Simplified Equations for a Vertical  Plume
      Governing equations for a vertical  plume issuing from a round
port are more convenient to derive in axisymmetric coordinates.  Thus,
with reference to Figure 2.3 and the coordinate system given in
                                   67

-------
                               NOMINAL
                                PLUME
                              BOUNDARY
Figure 4.2.  Coordinate  system for axisymmetric vertical  pi
ume.
                       68

-------
Figure 4.2, the following assumptions are posed:



      •  steady flow



      •  flow is axisymmetric



      •  coriolis effects are neglected


                                                       3D0
      .  flow field is assumed hydrostatic throughout:  if— = 0
                                                       a Z


      .  density difference between the plume and surroundings is



         assumed small compared to the density at any point in the



         flow field:  IP^-PI  « p



      •  plume is fully turbulent



      •  eddy transport of momentum and heat is only effective in



         the lateral direction (normal to jet axis)



      •  molecular heat conduction and viscosity are ignored.



      With the above simplifications and assumptions it is possible to



disregard a number of terms in cylindrical governing Equations (3.69)



through (3.73) and arrive at the following equation set:



      Continuity:
                 3r
      Momentum:




      Employing "order of magnitude" analysis common to boundary layer



theory (e.g.  Schlichting [84])and incorporating previous assumptions,



we see a need for the z-direction momentum Equation (3.72) only. This
                               69

-------
equation reduces to
where trz is the turbulent shear stress.


      Energy transport may be accounted for by the appropriate


axtsymmetric form of the density transport Equation (3.34) or


                                                                (4.3)
         9A      9A        9p*
     u
      r                            ar



For salinity we use Equation  (3.71),  with  r  =  A,
                                                 \ .
                                                 /
                               9  fe    9A3  I                  (4.4)
      r  3r         3z      r  3r \  Sr  3r



with the buoyancy parameter, A,, defined as'


             - S
     A, -^
      3   Sr - SQ   •                                            (4.5)




Using the continuity relationship Equation (4.1), Equations (4.2),


(4.3) and (4.4) may be rearranged to yield the following:
              1 -i rr     )  -   v _ _  1 _L J E   !_21»  (4.7)
              r  9r  ^ ur  2'        3z       r 3r  1  Hr 3r
                                70

-------
     fi <"3>  + F  If  K 43'  - F  If | ESr  *T \ ,          (4.8)
respectively.



4.3  Radial  Velocity and Temperature Profiles



      A large amount of experimental work has been carried out in the



past concerning radial  velocity and temperature profiles for free


jets.  Earlier work was concerned primarily with momentum jets.


Schmidt [85] in 1941 was evidently the first to consider the mechanics


of convective plumes, such as convective currents over fires, etc.


Schmidt's work was reported in the German literature, and apparently


because of the war, went unnoticed until Rouse et al. [81] carried out


similar work in the early 1950's.  Since then a number of researchers


[8, 26, 41,  77, 83] have investigated velocity profiles and associated


transport coefficients  for both momentum jets and buoyant plumes.



4.3.1  Zone of Established Flow




      The experimental  studies have established that velocity and


temperature profiles are approximately similar at all axial locations


in the zone of established flow for all vertical plumes in a stagnant,


free environment.  Also, profiles are nearly Gaussian and may be


adequately described by the normal distribution curve:



                     1 (L\2
       v(r,z) = vme~ 2  V                                       (4.9)




for velocity, and
                               71

-------
                      .
       e(r,z) = eme" 2  V                                     (4.10)

for the temperature distribution.   In the above equations the sub-
script m refers to condition at the plume center!ine, A is the eddy
Prandtl number, and a is the standard deviation.
      The standard deviation has been found to relate to the vertical
coordinate, z, by
                 2
            2  K
       a2 =i  |                                               (4.11)
where K is an experimental  entrainment parameter.   Hence,
                    K^2
       v(r,z) = vme'K (z>                                       (4.12)

and
                    K x n2
       e(r.z) = eme-K x V.                                   (4.13)
It is important to remember that these profiles have no theoretical
basis and are merely the result of curve fitting.
      The values K and x must be determined by measurement and have
been found to depend on the extent of buoyancy.  For instance, in
the case of a simple plume  (pure buoyancy, FQ=0) Schmidt found that
       K = 48
       x = 1.2 .
The data of Rouse yields
       K = 96
and
       x = .74
                               72

-------
for a buoyant point source.   Abraham in his analysis of a simple
plume used values
       K = 92
and
       A = .74.
For the momentum jet case, (neutral  buoyancy) Albertson found
       K = 77-
Abraham used
       A = .80
for this case.
      Baines [8] observed in his investigations that the initial
Reynolds number affected the results.  He found the following best
fit for his experimental results:
                       r N
       v(r,z) = vme'K (I}                                      (4.14)

where K = 43.3 and N = 1.82 for Rfi = 2.1xl04, and K = 64.4 and N = 1.84
for Rfi = 7x1O4.
      Where values for  K and A are needed  in the present work, the
following are used:
       simple plume (pure buoyancy,  or FQ^0),
       K = 92
       A = .74, and
       momentum jet (neutral buoyancy, or FQ -> °°),
       K = 77
                               73

-------
       A = .80


4,3,2  Zone of Flow Establishment



      Figure 4.3 illustrates a typical velocity distribution in


this zone.  Albertson estimated this distribution for a momentum jet


by assuming a flat profile across the potential core and a Gaussian


distribution for the mixing zone.  Albertson derived an integral


expression for momentum flux across a lateral plane in this zone by


integrating Equation (4.6) with  p  = p, for r = 0 to r -»• », or
                  /CO



                    v!
       M  „    o	  =                                    (4J5)
       M      ./  A
        0     V   ft
               00


The quantity M is total momentum flux crossing a plane normal  to the


mean flow and A is cross-sectional area.  Thus, Equation (4.15) states

                                    2
that momentum is conserved with M =v A  the momentum source strength.


By letting C, = a/z, the momentum flux relationship above yields



       ^e  _   1                                                (4.16)
       D   ~  2CTJ~



where C-j is an experimental constant.  By approximating the potential


core diameter, D , according to



       !!c  - I      1                                          (4.17)

       D


the mean velocity distribution in this region takes the form
                               74

-------
              
-------
       V
Equation (4.18)  above will  be used in  the following work when a
velocity distribution near the outfall  port is  required.  Note that
this equation is not correct for buoyant plumes since density dif-
ferences have been ignored.   However,  very near the outfall  port (say
one port diameter downstream), inertial  effects are assumed  to domi-
nate the flow behavior regardless of the degree of buoyancy.   Evalua-
tion of the empirical constant C, and  the length z  are dealt with in
the next section.
4.4  Zone of Flow Establishment
      For a plume issuing from a small  diameter port in deep water
the length for flow establishment, z ,  has relatively small  influence
on conditions far downstream except as  it enters in the established
flow solutions as a boundary condition.   On the other hand,  for large
outfall ports, the theoretical zone may extend  over a good portion
of the flow field, or even to the ocean surface.  In this section, we
will discuss methods for evaluating z   in deep  water for both the
neutrally buoyant and buoyant cases.
      Many experiments have been carried out by various investigators
in an effort to establish the length of the potential core for  turbulent
round jets issuing into stagnant fluids.  Good  reviews  of this work
are given by Hinze [40] and by Gaunter, Livingwood, and Haycak [32].
Gaunter et al. in their review, state that values for z  /D vary from

                                 76

-------
about 4.7 to 7.7.   For Instance, Albertson et al. found that z = 6.2
for their work.  Baines reports that jet Reynolds number had
considerable effect on z /D for his experiments.   In fact for Re  =
1.4xl04, ze/D = 5  and for ReQ = 105, ze/D = 7.
      Where buoyancy affects the potential core length, Abraham bases
z  on the concentration distribution.   Hence, z  for concentration is
given by
where x and K take values .8 and 77, respectively.  The limiting value
of z /D in Equation (4.19)  for F  -> °° is approximately 5.6.   The value
of z  /D for concentration profile establishment is about 10% less than
the value of 6.2 for velocity profiles found by Albertson.
4.5  Governing Differential  Equations
      To derive the equations governing the dynamics of a vertical
plume in Zone 3 we integrate Equations (4. 6), (4. 7) and (4.8), in a
lateral  plane, from r = 0 to r -> °°.  Thus, the following expressions
apply as indicated,
Vertical momentum transport:
•/
                                                              (4-20)
                                77

-------
Density disparity transport:
                            *
                           9p
                vA,rdr  -  —^- J     vrdr  =  0             (4.21)
            o     Z        3Z     o
Salinity or concentration transport:
        _
       dz /    vA,rdr = 0 .                                    (4.22)
         •r  0    »J

Equation (4.20) may be written in terms  of A_  by  rearranging the  gravi

tational contribution to yield,
 fi*>   r.
A  /
                rdr ' 9   "             'rdr                 (4-23)
Integration of Equations (4.21)  through  (4.23)  may be  completed  by

utilizing profiles given by Equations  (4.12)  and(4.13)  (4.13) for  A,

and A2-   Hence, the resulting expressions  are

Vertical momentum:
             3 3
            K

Density disparity:
                   ,-V2   ^                               (4.25)
                      K     dz
                              78

-------
Salinity or  concentration:
       K(A  +  1)        4"



Cast in dimension! ess form, the above equations become
where,



       Z  =   z/D
       V  =   v  /v
        m    nr o
              (v)
                                                            (4.26)
                           »*•
       dR*

       dT
       A    _           	T_                              (4.29)

        1m     —4      £J/3Z
E* =   -_ni_'                                          (4-30)
              E*l/3


       R* =    _ A2mZ , and                                (4.31)


               VK (1+A)




      Alo=   K
                           79

-------
4.5.1  Initial Condition
      The solutions of Equations (4.27) and (4.28) are begun at
1=1 , or in the beginning of the established flow regime.  Abraham's
relationship (4.19) may be used to evaluate this distance for the
entire range of densimetric Froude numbers, F .   Once Zg is known,
the initial values E*fi and R*e may be established.  We assume that
ambient stratification may be neglected over Z   then
      A
       2
      dp *
        oo
      HZ"
 =   1,  and
    =  c  .
"e
Hence, by Equations (4.30) and (4.31)
      R *  -
      Ke   "  4-  •                                             (4.32)
The initial value of E* may be found by considering Equation (4.30).
For large initial Froude numbers (F  -*• °°) V   •> 1 , so that
                                   0       IDS
                                                              (4.33)
However, for low FQ, Vme is typically larger than 1 and unknown.  To
avoid estimation of Vme, we use Equations (4.29) and (4.30) with
A2me = lj to obtain
                               80

-------
       Ee*  •
                     e
vaiiA)                                   (4.34)
For large F ,  Equation (4.34)  reduces  to

       Ee*  'T  ;                                           (4'35)
in which case  Zg = 5.6.   This  result agrees with Equation (4.19).

4.5.2  Evaluation of Terms Involving K and A

      Listed in Table 4.1  below are limiting values of K and A as
suggested by Abraham along with limiting  and mean values of terms
involving K and A.
          TABLE 4.1   VALUES OF  TERMS INVOLVING K AND X
Term
K
A
1+A
Momentum
Jet
(F -*• col
ir0 + »;
77
.80
.256
Simple
Plume
(FQ - 0)
92
.74
.245
Mean
Value


1/4
Max.
Error
J%L

2A%
                   .114             .104             .109       4.8%

                   .253             .239             .245       2.9%
      v/K(A+l)
Using convenient values  for the  above  terms,  the governing equations
and initial  conditions  are:

       dE*   _  3^_  Z  R*.                                    (4.36)
                  o
                                81

-------
      dR*
               .nzE*1/3
                         ~dT
                                                           (4.37)
                                                           (4.38)
        im
               Z E*
                    1/3
and the  initial conditions  are:
               64
       F  *   =  —o
       Ee        ,
      Re*   =  1/4 .
4.5.3  Homogeneous Receiving  Water
                                                           (4.39)
                                                           (4.40)
      For  the case of homogeneous receiving water the  above equations

may be solved analytically since dp^/dZ = 0.  Therefore,  from Equa-

tion (4.37)
      R*     =  1/4,

and Equation  (4.36) becomes
      dE*       32
      dT~    "  T6F '
                  o

Equation (4.42) may be integrated immediately to yield
                                                           (4.41)
                                                            (4.42)
      '
              (5*
            2 21
           -ze]
}   •
(4.43)
Centerline  concentration is then  given by Equation (4.41)  as

                                              i i r\
                73
       im
            _/Z3
            •[64
64
                                    2   -, 2,
                                      - Ze J
                                        (4.44)
                               82

-------
Apparently, the stratified case must be solved numerically.
4.6  Lateral  Velocity, u
      Once the plume centerline velocity, v, has been calculated,
and the lateral distribution of axial  velocity has been established,
it is a simple matter to calculate u  from the continuity equation,
      r     8r          9r       *                              (4.45)
Since v(r,z) is known, Equation (4.45) may be written as
          [urr]  =  rf(v)                                        (4.46)
or
      ur   =   -  f  f(v) xdx.                                 (4.47)
                 J(\
                              83

-------
                              CHAPTER 5
                       FINITE-DIFFERENCE MODELS
      The finite-difference models  developed in this chapter are applic-
able to the following two situations:
      .  Vertical  round ports issuing  into  quiescent receiving
         water, and
      •  Line plumes which may include ambient current effects.
From a practical standpoint, the vertical  round port in shallow water
is of foremost importance because this configuration is typical of pres-
ent and planned installations.  The line thermal  model would find appli-
cation in analyses of the plume which develops over a diffuser line once
the individual round plumes have interferred with one another.
      The numerical models are formed in two dimensions for steady flow
conditions.  In the case of a vertical round plume, a two-dimensional
model will not accommodate any ambient cross flow which would destroy
the plume symmetry.  Hence, the solution is strictly valid only during
slack tide conditions in the absence of prevailing local currents.
However, cross currents, tidal or otherwise, have little effect  on the
initial mixing  (near-port  locale) of plume flow from large outfalls in
shallow water.  The reason for this is that the effluent momentum
dominates the ambient flow.  At the San Onofre outfall, data show that
isotherms in the near vicinity of the outfall are reasonably concentric
even in the presence of tidal currents [24].  In view of available data
it appears that a two-dimensional axisymmetric model for the vertical
round plume should give adequate results for the initial mixing region,
                                  84

-------
in spite of ambient cross flow.
      The line thermal model may accommodate ambient flow perpendicular
to the plume since in this case the phenomenon remains two-dimensional.
End affects are, of course, ignored in this case.
      Difference models are based on the vorticity-stream function equa-
tion described in Chapter 3.  Where the finite-difference solution is
started some distance above the outfall port, boundary conditions are
obtained from available data or similarity solutions as described in
Chapter 4.  As indicated by Table 3.1, the minimum number of equations
required is three.  We will also consider salinity transport so that
four partial differential equations are required, these being one
Poisson type equation for the stream function and a total of three
transport equations for vorticity and two r constituents.
5.1  Physical System for the Vertical Round Port
      The physical system of primary concern is a large, single port,
submerged vertical thermal outfall issuing to stagnant receiving
water.  Figure 5.1 illustrates this system in axisymmetric coordinates
(r, z).  Later, conditions for a line plume will be discussed in an
appropriate cartesian coordinate system.  The receiving water has
depth, L, and is assumed stratified with density pro(z).  Flow enters
the system along the bottom boundary (z = zb) with some known velocity
and temperature distribution.  In all cases to be analyzed the  inflow
will occur only over a small portion of this boundary, which extends
from the plume centerline to a point rb, the nominal plume boundary.
For the shallow water cases, rfc = RQ the outfall port  radius
                                85

-------
                            SEA SURFACE
                      RECEIVING WATER  P°°
       z   INFLOW
         BOUNDARY
                                                      O
                                                      CO
                                                      O
                                                      I
                              NOMINAL PLUME BOUNDARY
Figure  5.1  Physical System for Axisymmetric Vertical Plume
           Where the Bottom Boundary is Some Distance
           zb t 0 Above the Outfall Port
                         86

-------
(Figure 5.2).  It is assumed that no flow crosses that portion of the
bottom boundary extending from r.  to r^.
      The plume centerline and ocean surface form no-flow boundaries,
or a reference streamline.  A free-slip condition is assumed at the
ocean surface, but this surface is not allowed to distort vertically.
The flow boundary condition at r = r^ is free except that streamlines
are assumed to have constant slope.  Flow will both enter and exit
over portions  of this boundary; the exact distribution is a part of
the numerical  computation.  The mean velocity might be assumed wholly
horizontal since r^ is a large distance compared to r, , and since
density stratification will impede vertical flow.  This assumption
would lead to  level streamlines.
      For shallow water geometry, (Figure 5.2) the ocean bottom is
assumed flat and z. = 0.  The port side and ocean floor are assumed
no-slip boundaries.
5.2  Governing Differential Equations
      For incompressible, turbulent flow in axi symmetric coordinates,
the differential equations describing continuity, linear momentum
and buoyancy transport were given in Section 3.10.2 and are reiterated
below.
Continuity:
1  1ML  + 9v
r  ar     3z
                                                               (5.1)
                                87

-------
PLUME CENTERLINE
=  L
            INFLOW
          BOUNDARY
I    I
t    4
                               SEA  SURFACE
                                   RECEIVING WATER

                                       Pro (z)
                         OUTFALL PORT
                          OCEAN FLOOR  (z b  =  0)
                                                          o:
                                                          <
                                                          o
                                                          o
                                                         'CO
                                                        U- O
                                                         •o
                                                           i
                                                              UJ
                                                              on
      Figure 5.2  Physical System for Shallow Water, Axisymmetric,
                Vertical Plume

-------
where  u and v are radial and vertical  velocity components, respectively
(note  that u is used instead of u  as  in  Section 3.10.2).
Momentum transport:
r = direction,

         ft'-ir  *
z = direction,



In Equations (5.2)  and  (5.3) above,  derivatives of tf and ez have
been ignored.
Buoyancy  transport:
     In lieu of the energy equation, the  transport equation for AI  is
considered,

         DA,    e             3A,         32A,
         	L  -  Pr   £_ t~  	L\  j.  c   	L                 (e; A}
         W  ~  r    3r  (r  3r '     £Pz  3Z2   '              {    '


where again derivatives of the eddy  buoyancy diffusivities, Epr and t^
have been ignored.   The buoyancy parameter, A-J , as defined in chapter  3
is
         A  =   r
               pr '  po
                                   89

-------
5.3  Vorticity Equations

      For the problem at hand, it is more convenient to deal with

vorticity transport rather than linear momentum transport.  In dealing

with vorticity, we need not be concerned about pressure and need to

consider one less partial differential equation.  The appropriate

vorti city-stream function equations were given in Section 3.10.2 and

as a matter of convenience are listed below.


Stream function, Y:

          2           2
         3 *   1 3* , 9 * _                                       (5 z\
             ~       —  ~ ~ ru                                  P-3;
         3r

Vorticity,
_           3Vu) _   3lB +    3_ / J_ 3gjr_x +   3 (u        (3.79)
                        E                e
_
at   3r    3z      3r    r 3r   r 3r
                                                 +
                                                   ez  2
where vorticity is defined as

               3U    3V
 Once  having  solved  for  the  stream  function  distribution  (Equation 5.5)

 the velocity field  is found by  the relationships,


                    ll.                                            1C C\
                                                                  (5.6)
 and
          v = - ^
              r 3r
 In the remainder of this work we will consider only steady flow.   Hence,

 the vorticity  transport Equation (3.81) has the form
                                90

-------
                           r, /      \                        O
                            ^p°°~ p'  ,     9/1  9^r\ ,    3 o
                              a? — +  Er a? (  FaF-} +£z ~
                         0                               o Z
where B has been replaced by the definition Equation (3.22).  Steady
flow transport of the buoyancy parameter A-, is given by
                                        (r     ;   e
          r    3r       3z      r    3r  v   3r  '    pz   2         /5 gx

The convective terms in Equation (5.9) are in "conservative" form
which was obtained from Equation (5.4) through the use of the contin-
uity Equation (5.1).
      In summary, the equations to be solved for the axisymmetric
plume dispersion are (5.4), (5.8) and (5.9) along with (5.5) and (5.6).
Equation (3.76) will be considered to evaluate vorticity boundary con-
ditions.  To account for salinity transport (if applicable) a second
Equation (5.9) will be solved with A-J defined as a salinity parameter,
A3,where
             3   Sr - SQ   '
Temperature distributions may be calculated from the Equation of State
(3.38) once A, and A~ have been established.  Hereafter only the con-
              •      0
servative form of the transport equations will be considered.  Although
the pressure distribution is not considered in this work,  it could be
calculated through Equation (3.74).
                               91

-------
5.4  Dimensionless Forms


      To cast the governing equations in dimensionless form consider

the following dimensionless variables:


      R  =  r/r0,

      Z  -  z/rQ, X

      U  -  u/v0,

      V  =  v/v0.
and, the dimensionless parameters:
            r v
      RE  = 	  ,  (radial,  turbulent Reynolds  number)
            r v
      RE  = 	  ,  (vertical,  turbulent Reynolds number)
            er
      PR  = •;—   ,  (radial, turbulent Prandtl  number)
        r   Kr
      PR, = rr-   , (vertical, turbulent Prandtl number)
        2   Kz
      FQ = — — - -   (densimetric Froude number).
                   2 r g
 Note that a second dimensionless vertical distance is used in this
 manuscript defined as Z = z/D and should not be confused with Z.



                                92

-------
      In the above definitions r  is the outfall port radius and v



is the effluent velocity issuing from the port.



      With these variables, the system of governing equations is written



as



      stream function:



             2      ,           2
            3 f     1   W  ,   3 y  _   Do                       /c lnx
            — 9  "  D"  W     — 2"  = - Rfi.                      (5.10)

            8R2     R  9K     3Z



      vorticity:



Note that in Equation (5.8) the Boussinesq term may be rewritten as
                                   (Pr"p) +
             PO   3r          PO     9r      pQ     9r
since p  is a function of Z alone.  Hence,
            -f (un) + rf-  (Vn) = - J	p1
            dr\        d Z.           L.\ ^   t\
                  2              \

                   fi,l3fifi\,

                    "         ~ ~
                 8R            R       Z  3Z
      buoyancy parameter:
                       2              \              2

                      3 Ai      ,   8Ai \      i       3 AI

                          11111       -7-         (5.12)
            REfPRz  I 9R?      R   3R   \    REZPR2
                                 93

-------
along with
and
            u = -ff!i   •                                       (5J3)
                  t\ oL
            "'IT  Iff    '                                       (5'14)
5.5  Coordinate Transformation
      When solving partial  differential  equations  numerically,  it  is
desirable to have fine grid space  resolution where large derivatives
of the dependent variables  are  expected.   In the present problem,  a
fine grid spacing is  needed in  the  radial  direction near the outfall
port and plume centerline.   At  large  distances from the centerline,
large grid spacing may be used  since  radial changes in the dependent
variables are expected to be small.   To  this end,  a non-linear  trans-
formation is employed on the radial coordinate, of the form
            R  = sinh 5 .                                        (5.15)
This transformation has the desirable properties:
            R  =  5,   AR =  A5   for  small  R,
and
            R  =  75- e?, AR  * |5- e5   for  large R.

In terms of transformed coordinates,  the governing equations are:
      stream function:
   sech
- (tanh £ +coth  f)  ~
                                 94

~ = - sinh £ n   .   (5.16)

-------
     vorticity:
sech

             se
                             RE
                                           sech'

                       2    tanh
                                     Ml _ o coth2
                            Re.
                                                  (5.17)
       buoyancy parameter:
|sech_U.l_
\sinh 5J K
sinh n  UT
                     sech
               3Z
                     REr PRr
sech   .  3r
tanh £ '  3?
                                               3Z
                                                 2
                                                  (5.18)
 Transformed expressions for velocity are given  by:
          U  =
               _^_
               sinh  g    3Z
                                                   (5.19)
         V = sec.h € . 91
             sinh     9
                                                   (5.20)
                               95

-------
Finite-difference calculations will be based on even increments of the
transformation coordinate, 5.
      In the vertical direction, fine resolution is needed in the
region where the plume spreads laterally.  In all  thermal plume cases
of interest, this region is in the vicinity of the receiving water sur-
face.  However, for other pollutants, such as municipal and industrial
wastes, lateral spread may take place below the surface and pollutant
concentration information is needed in the vicinity of this plane.
Since methods presented here are also applicable to these pollutant
plumes, a fine grid arrangement near the surface is not specified as a
general case.  Rather, the vertical grid spacing will  be treated as
node-wise variable and exact specification left to the discretion of
the computer program user.
5.6  Finite-Difference Grid System
      The finite-difference grid layout consists of two grid systems.
One grid is usedto calculate the stream function,  y, which provides
information to compute velocity components, U and  V.  This system
coincides with the physical boundaries and is illustrated by the wider
lines on Figure 5.3.  The stream function is calculated at the interior
intersection points designated by the solid round  symbols.  Solid box
symbols represent boundary points.
      Velocities are not calculated at these same points.  The U com-
ponents are computed at vertical midpoints which are designated by open
circle symbols; whereas, the V components are computed at horizontal
midpoints U coordinate) and designated by open box symbols.  In this
                                   96

-------
            COMPUTATIONAL  POINT LEGEND
             • -V;  X-A, ft;  O-U;  D-V
              BOUNDARY  VALUE LEGEND:
                    • -V;  O-A,  ft
                                          / / f/f/ttf/ff/
Figure 5.3  Computational Grid  for Difference Equations
                         97

-------
manner, the stream function grid layout defines a system of cells with
the stream function, f, computed at each corner point (or set by
boundary conditions, as the case may be) and velocities defined at the
center of the cell face (see Figure 5.4).
      The second grid system is used to calculate vorticity, ft, and
buoyancy parameter, A,, (also A3) and is illustrated in Figures 5.3
and 5.4 by the narrow lines.  This layout completely overlaps the
grid (and physical system) with interior intersection points centered
in the cells defined by,the f grid system.  These interior grid points
are indicated by crosses with boundary values at cross-and-box points.
      The reason this staggered grid system is used is for computa-
tional convenience in treating boundary conditions and to permit con-
vective transport terms to be evaluated at cell faces.
      In Figure 5.3, the Y grid system is sized by NJ and NK grid
points in the £ (or R) direction and vertical direction, respectively.
The ft,  A-| system has size NJ + 1 and NK + 1 in the respective direc-
tions.  Points on the f grid are indicated by j, k, whereas points on
the n, A-| grid are indicated by p, q.  In this figure, Z.  defines the
bottom boundary of the stream function grid  (physical boundary) and Zh
the top (sea surface).  Vertical spacing for the system is defined by
AZ^ and may be variable.  Grid spacing along the ? coordinate is even,
designated by A£.  System boundary points for the n, A-, grid are
located at Zfa - ^ AZ2 for the bottom \ * \  AZNK-1 at the top>  "I A?
on the left and ^ + ^ A£ at the right boundary, where ^ is the
assumed right hand physical boundary.  Figure 5.4 also illustrates
                                98

-------
q+1
               COMPUTATIONAL  POINT LEGEND:

              •  - V         x - r, Q

              O-U          D-V
j-l
                                        j
             P+l
<>
/
-c
\
)
-c
\
)
V L J
^ V.
,
} \
AZk
, ,
1 '
^ r
r
\ J ' [
P [
)J •*•! " N
)
1 [
\ N
^ w *
f A s
S
i \f
\ J,K
J 1
C C
3 f
^ f
r
i j'k c
5 L
\ J' N
J /
J.k-1
' L
^ ^
V '
X
k
3-
/
V
3-
^_
V
k-1
q-1
               1-1
               R.
                j-l
           R
             P
R.
        Figure  5.4  Typical Finite Difference Cell
                   Illustrating Indices  for 
-------
indices, computed quantities,  cell  size and radial  distances for a

typical  interior cell.


5.7  Difference Equations


      Standard difference representation is used wherever possible in

this work.   Central  differences are used for both first and second

partials except for convective terms where a special donor-cell method

is  used.  Techniques for uneven spacing are used for the vertical

differences.


5.7.1  Stream Function  and Velocity


      Consider the stream function  grid system illustrated in

Figures 5.3 and 5.4.  The finite difference representation of  Equa-

tion (5.16) based on central  differences for both first and second

partials is as follows:
9
t_
sech 5- •,
V |
? r
sech" £_•
+

4.


J r

/
j
2 r
sech C-
j „
2
AC
1 1 +^nl>
i tanr
\
                                ?,  +  coth  cj  • r
                                 j         j/   ^
                           tanh £.  + coth r,l  .  ^k
                                 J         ^J/    2
w
 j+l,k
                                                              (5.21)
                                  100

-------
In the above difference equation, the quantity fi- .  is the average
                                                J j K
value of a at point (j,k), hence the overbar.  This average value must
be used since n    does not lie on the ¥ computational grid points.
               p »q
Vorticity is averaged for the four cells neighboring point (j,k)  as
follows:
where
and
      Velocity is calculated in first-order manner as
and
                    sech Cn
            V    = —=—;—2-  (y. ,-y. ,  ,)                    (5.24)
             j.k    sinh £_AE v  j.k  j-1,k'
      Thus far, we have discussed differencing the governing equations
only in transformed radial  coordinate, ?.   To permit more versatile
computation we also include provision for  calculation directly in
(R.Z) coordinates.  This is easily done by collapsing the hyperbolic
functions so that
                               101

-------
            sinh 5 + £ =  R,  (also set tanh 5=0)
and
            cosh £ -»• 1
giving
            A? = AR.
      Hence, for linear radial  coordinates  Equations (5.21), (5.22),
(5.23) and (5.24) collapse to

                                             i-l,k
                                AZk(AZk+l +AZk)   ^^

                             +   JAR n,  .                       (5.25)
                                    J , N
      Velocity:
            R-component
            Z-component
               V
                J.K   l]^2   "

                               102

-------
      In Equations (5.25) through (5.27) sinh ?. is replaced by JAR
                                               J

and sinh ?  by (j - -~) AR.



5.7.2  Transport Equations



      Except for the convective transport terms, central differences


are used to approximate all derivatives in the transport Equations


(5.17) and (5.18).  Special consideration is given the convective


terms which involves basing numerical approximations on transport


integral techniques (see Appendix A).


      Referring to the p,q  grid system illustrated in Figures 5.3 and


5.4 the difference representation of the steady flow vorticity trans-


port Equation (5.17) is written as (after collecting terms)
                                                        coth
2
REzAZk
1
[AZk+AZk+l
, 1
AZk+AZk-l_
|Sech ?p
RErA£2
            sech
                     v. , I  + v. ,  +
                                                        p.q
sech E
P
2AE
I
|U. , , + U. , ,
1 1 -1 V 1-1 K
J 1 » l» J ' J ^
sech 5p
9
^ A?sech Si
1 2 tanh £
                                                                 p-l.q
                                 Equation  (5.28)  continued on next page.


                                103

-------
sech
                              sech
                                           AZk+AZk-l
                                                    'p.q-1
    2AZ,.
    sech
         IE .
               A1p+l,q"Alp-l,q
                                                    Wl      (5.28)
The turbulent Reynolds numbers, RE  and RE ,  in the above difference
equation are point variables of the form RE (p,q) and RE  (p,q).
Derivatives of these quantities are neglected in the above equations
but are accounted for in the computations.
      Equation (5.28) may be collapsed to radial coordinates in the
same fashion as illustrated in Section 5.7.1.  Hence, in non-
transformed radial coordinates the vorticity transport difference
equation, after collecting terms, is (note that  numerically p = j-1/2)
                                   104

-------
          I
   [2AZk     j.k-l     j,k-l      REzAZk
Z   (|Vj,kl  '  Vj,k)  +REZ" (AZ           n
                                                   p,q+l
             (Vl.q  "  Vl.q)    '                             (5-29)
The convectlve terms  are  formed  In a manner such that vortldty con-
vected out of cell  (p,q)  has the value n_ _ and vortldty flowing Into
                                       P«H
the same cell  is  convected  in with the value of the cell where it
originated, regardless  of the directional sense of fluid motion.  This
character of convective transport is essential in properly conserving
                            105

-------
the transported quantity and in avoiding  certain  computational
difficulties.
      The difference formulation of the buoyancy  equation  after  col
lecting terms  is written as
|PRzREzAZk
                  1
                              1
              Azk+l+Azk
2 sech^ r    sinh r  . sech
	IP A       J	
PRrRErZ
                      sinh  £. i  •  sech  £
                    + 	J"'	E.
                          O A >• "  • " I ^
                          2A? sinh
    2AZ,
                                         .
                                              'p.q
     sinh5, , • sech. £
    	'"^nhtp
                                                  sech
           sech
     1  -
         2   tanh
                                 sinh £.  • sinh.  5
                                       2  tanh
                                                'P J
                                                      'p+l,q
                                  PR2REzAZk
                                              AZk-l+AZk
                      'p.q-1
     I2AZ,
                               PRzREzAZk
                                             AZk+l+AZk
                                                              (5.30)
                              106

-------
       In  linear  radial  coordinates,  Equation  (5.30)  reduces  to
   PRrRErAR'
  PRzREzAZklAWAZk   '   AZk-l+AZk
                                         j-l
                                            p.q
2pAR
                                r  r
                                                     -i»q
                                                'p+i.q
                            PRzREzAZk
                                                    I
'p.q-l
                                             A!
                                               p»q+i .
                                                               (5.31)
      The A., transport difference equation corresponding to Equations
(5.30) and (5.31) are obtained simply by replacing A-, with A3 and noting
that the eddy Schmidt number, SC, should be used in the case of material
transport, instead of the eddy Prandtl number, PR.  Materials other than
salt may be treated in a similar fashion.
                                  107

-------
5.7.3  Summary of required difference equations

       The difference equations to be solved are:
            Transformed coordinates (?, Z),
                 f - Equation (5.21)
                 a - Equation (5.28)
                A-j - Equation (5.30)
                A3 - Equation (5.30)
                 U - Equation (5.23)
                 V - Equation (5.24)
            Linear Coordinates (R, Z)
                 V - Equation (5.25)
                 n - Equation (5.29)
                A-| - Equation (5.31)
                A- - Equation (5.31)
                 U - Equation (5.26)
                 V - Equation (5.27)

5.7.4  Vertical Grid Space Restrictions
       Although the vertical grid spacing is variable, there are three
locations where an exception is expedient for the treatment of bound-
ary conditions (Section 5.8).  These exceptions are as follows:
      1.  At the grid system bottom boundary AZ~ = AZ-,
      2.  At the sea surface
where  Zh = ]T  AZK
                      NK
                      ]T
                      k=2
                             108

-------
      3.   At the level  of the plume inflow boundary
            AZ
              Kp

          where KP is the grid boundary location.
These exceptions place no serious limitation on vertical grid spacing
and are incorporated only to expedite computer bookkeeping in treating
the various boundary conditions.
5.8  Boundary Conditions
      Attention is now focused on evaluation of boundary conditions
necessary to carry out solution of the equation sets summarized in
Section 5.7.3.
      Referring to Figures 5.1 and 5.2, the sea surface (Z = Z, ) is
considered a free-slip boundary which is vertically rigid.  A specified
flow enters the bottom inflow boundary where R <_ R .   Depending on the
water depth, this boundary may constitute the outfall port orifice
(shallow water case, see Figure 5.2) or an arbitrary lateral plane
through the plume (deep water case, see Figure 5.1) at elevation
Z = Z, .  In the former case, the port geometry must be considered along
with the ocean floor.  The radial velocity distribution, VQ, depends
on R  and the port side and ocean floor are no-slip surfaces.  In the
latter instance, the velocity distribution is obtained either directly
from data (hydraulic model or prototype) or calculated by the similarity
techniques described in Chapter 4.  Outside the plume nominal boundary
(Figure 5.1) the bottom boundary is assumed slip-free.
                             109

-------
      Surface heat transfer is neglected in this study since the sea
surface area is relatively small and surface heat exchange will have
very little effect on the overall temperature distribution.  Boundary
condition sets a and b given below refer to the physical systems
shown in Figures 5.1 (deep water) and 5.2 (shallow water) respectively.
To eliminate confusion, the boundary conditions are stated in terms
of R (in lieu of the transformed coordinate, ?)•
      1.  Sea Surface ( OfR^, Z = Zh)
          a.  41 = Constant = ^ ,
              n = 0                                            (5.32)
           3A,
           -T—  = 0 (adiabatic condition)
           3A,
           b.  Same as above.
      2.  Plume Centerline
          a.  R = 0, Zb <_ Z <_ Zh
              v =  Constant =  Y,
          3A,
          XT   -  °
          b.  R = 0, ZQ £ Z £ Zh
              Same as above.
                           110
                                                               (5.33)

-------
3.  Inflow Boundary
    a.  Z = Zb, 0 <_ R <_ Rfa
                 /-R
          = *1 +J   V(R,Zb)RdR                           (5.34)
          = Alb
    b.  Z = Z . 0 < R < Rrt
             o           o
        * = f1 +7   V(R,ZQ)RdR                          (5.35)
                  o
        _ _ 9U.   9V_
        " " 8Z " 3R
4.  Port Side (R = RQ, Zb = 0 <. Z <. ZQ)

    a.  Not applicable
                 VR
        Where the reference velocity, V  = 1
          = .   V  (no siip)                              (5.36)
                          m

-------
          3A,

            -  -
          3A,
      5.   Bottom Boundary


          a.   Z = Zb, Rb < R < Ro
                       f "b
              ¥ = V]  + J   V(R,Zb)RdR = *2                     (5.37)
              ft =  0


          3A,
          3Z

          3A
            3-   =0
          b.  Z = Zb = 0, RQ <. R <
              n -  |y  (no slip)                               (5.38)



          3A,
          3Z




      6.  Inflow-Outflow Boundary (R = RM, Zfc <^ Z <_ Zj



      The distance to the inflow-outflow (or free-flow) boundary, R^,


must be chosen in advance and this distance must be large enough such


                                112

-------
that boundary conditions listed below prevail approximately.


      a-   ||=0; or  4  =0
           8R          8R

Meaning that streamlines  are level, or the streamlines do not change

slope, respectively.


          « • S  - &                                        (5.39)


              P  - P^,
         A-, =	      (Ambient condition),


              S  - S
              °v«   °00
         A3 = S—^~T~      (Ambient condition)
               r    o

      The conditions on A, and A., are valid so long as convection

dominates the transport at the boundary and upstream differencing is

used.

      Now consider the difference form of these equations.  Again,

refer to Figure 5.3 and note that boundary values for the (j,k) grid

(vgrid) fall on the boundary of the physical system; whereas, on the

(p,q) grid (grid for n, A, and A3) the boundary cells are fictitious

in that they fall  outside of the physical system.  These cells are for

the purpose of obtaining specific conditions at the real boundary.

Again conditions a and b refer to cases given in Figures 5.1 and 5.2,

respectively.  The difference forms are given in terms of the trans-

formed variable, 5, for computer application.
                                113

-------
1.   Sea Surface (k = NK,  q = Nq + 1)




     a.  Deep Water (Refer to Figure  5.5)





         Velocity:



          Uj,NK+l  =Uj,NK  (^ee Slip)                        (5.40.1)





          Vj,NK=0                                          f5'40-2>




         Stream Function:




         Vj NK = 1  (Arbitrary)                              (5.40.3)





         Vorticity:




          Let  fi  be the  vorticity at point  (p,  Nk).   By the free



          slip velocity condition above and  the  fact  V.  ...  = 0,
                                                      J »INK
        Hence,  H  is the nodewise average value at Zh>  or
         so  that



                                                              (5.40.4)
                                 114

-------
•C
1
U.M,NK.
s


boundary
!.i-l .NK r
interior
V

f 1
r
r S
cell
j V.i.NK ,
cell
r
p,Nq
f
, VJ,NK-1 t
\ \

q - Nq+1
I^.i.NK k = NK
(sea surface)
u .
^ ' q = No
V .

Figure 5.5  Typical Seo Surface Boundary and  Interior  Cells
             [r    indicate any of  the cell  centered
              quantities n, A-, and A-,.]
                         115

-------
         Buoyancy:



         Since  the adiabatic condition prohibits heat transport



         across  the  surface,
                 1    =0,

          Hence,
             1   .                  .

           AL7.7 ^Alp,Nq+l  "  Alp,NqJ  "   '                        (5.40.5)
             NK
          or
          Alp,Nq+1  = Alp,Nq.



          Salinity:



          Likewise,
                   - '3p,Nq  •




b.   Shallow Water



     Same as deep  water case  above.
                                 116

-------
2.   Plume Center! ine (R =  0)




     a.   Deep Water Case (Refer to  Figure  5.6)



         Velocity:




          U, .  = IL .  (velocity gradient vanishes)             (5.41.1)




          Vljk=0.                                           (5.41.2)



         Stream Function:




          Y, ,  = 1   (Must  be  consistent with  condition  l.a).   (5.41.3)




         Vorticity:



          From the  conditions on velocity  given  above,  the centerline



          vorticity, ^_ -  0,  or averaging  across  the  centerline,




             V 1/2 ( "l,q  + "2,q)  =0"



          Hence,



            nl.q =  - n2,q-                                   (5.41.4)




         Buoyancy:




          At the centerline,  the buoyancy  gradient  must vanish.



          Hence,





             77 
-------
                 R = 0,   £ = 0
                .,k
      boundary cell
Y^
                           2 k
                            *
                    interior cell
                                       2,k-l
                              k
                              .K
                                                           k-1
P = 1
j = 1
                       p = 2
                                               j = 2
Figure 5.6  Typical Centerline Boundary and Interior Cells
            [r    indicates any of the cell centered
            quantities, n, A-, and A,.]
                          118

-------
        Salinity:


         Since the same conditions  hold for salinity and buoyancy


         transport at the centerline,


           A3    = A3                                        (5.41.6)
            Jl,k    J2,k




           b.   Shallow  Water Case


               Same as  deep water case above.



3.   Plume  Inflow  Boundary



     a.  Deep  Water Case  (Z = Zb, 0 ^ R •_ Rb;


        (Refer to Figure 5.7)
         Velocity:
                         \ AZ^                              (5.42.1)
            Calculated by  methods  in Chapter 4.


         Vj., -vu.zb).


            Data function, or  calculated by methods in


            Chapter 4.


         Stream Function:
                                 sinh  -cosh  C  .             (5.42.3)
                    P=2
                               119

-------
          )t-x^
                   interior cell
',1-1.1    r^LL
                  boundary cell
   JM,1
                                              k = 2
                                              q = 2
                                              k = 1
                                              q = 1
         j-l
Figure 5.7  Typical  Inflow Boundary  and  Interior  Cell
            (deep water case only).

            [rp,q indicates any  of the cell  centered
            quantities  n,  AI  and A.,.]
                           120

-------
Vorticity:


Vorticity at the inflow boundary, n  b is  calculated from,
Hence,


  "n i = - °n 9 + 7T- [U4 o-U, i  + U. ,
   p,l      p. 2   AZ    J,2  j,l     .1-1,
Buoyanpy:



AIP,I    = Al^J'1/2^5» Zb'1/2 AZ1)]-               (5.42.5)


 Data, function, or calculated by methods in Chapter  4.


Salinity:

A,    « Ao[(M/2)A£, Z.-1/2 AZ,)]                   (5.42.6)
 3p,l    3             b       '

 Data, function, or calculated by methods in Chapter  4.




b.   Shallow Water Case (Z=ZQ,  OfRfRQ;  Refer to Figure 5.8)


    Velocity;


    Uj)Kp = 0                                         (5.43.1)
    Vj,KP = Vo = Constanti  or»  vj,kp = V(C.Zb).      (5.43.2)


    Stream Function:
            1  +  A5Z  V(n,KP)  sinh  ?_ cosh £_        (5.43.3)
                  n=2                P       P
                        121

-------
Uj-l,KP+2

f_2_^
tLr
. 4fj-l,KP+2 lVj,KP+2

interi
y»
i — T I^P+T
K 	 1
r
or cell
LVJ,KP+1
Jp,QP+l
, 4/.i.KP+2
UJ,KP+2
V.
,UJ,KP+1
Uj-l,KP
 interior  cell  for  A-|&A.J
[boundary  cell  for
  )
  J-l.KP
                       U
                       'j.KP
       boundary  cell


            |FP.QP
UJ,KP
                                               k = KP+2
                                               q = QP+2
                                               q = QP+1
                                               k = KP
                                              (port orifice)
                                                 = QP
  Figure  5.8   Typical  Inflow Boundary and Interior Cell
             (shallow  water  case  only).

              trp,q  indicates  cell  centered  quantities

              ft,  A-|  and  A,.]
                            122

-------
Vorticity:



Since in  one  case, V. •, = V_ is assumed constant over the port
                     J > '    o


radius, we choose to evaluate vorticity at QP+1 , instead of at



the port orifice. HD QD+, will then become the boundary value.



Convenience is the primary reason for doing this, because to



remain consistent with V  = constant, n. np is impossible to
                        0              J ,ljr


define correctly at the port edge.  This procedure is also help-



ful in using power law profiles for V(s,Zu) (see Chapter 7).



Vorticity at a point (p,k) is given by





    «p,k ' 1 «Wl+flp.q>   *




Hence,




    °p.q = • <


at q = QP + 1,
                        [U j,KP+2+U j-l,Kp+2"U
                   "  A?cosh5p  ^Vj+l,KP+rVj-l,KP+lJ


                                                       (5.43.4)



  Buoyancy:



  A-,     = Constant = A,                               (5.43.5)

    'p.QP                ]o




  Salinity:



  A-     = Constant = A-                               (5.43.6)
                           123

-------
4.  Port Vertical Side

     a.  Deep Water Case - Not applicable

     b.  Shallow water case (R = RQ,      0 <_ Z <_ ZQ; Refer to

         Figure 5.9)


         Velocity:

         UNp>k=0                                            (5.44.1)


         VNP,k = " VNP+l,k (N°-s1iP condition).               (5.44.2)


         Stream Function:
                         . NP
         f     = 1 +   A?     V(n,KP.)  sinh  £    cosh  5.        (5.44.3)
                         n=2
                         P=2
                                                   1    2
         Although the exact value of 4*   ,  = 1 +   j R   V  , the
                                      lir j K           0   0

         difference approximation will lead to a slight deviation.


         Vorticity:
          Hence,


           "MP.q  = -  "MP+l,q  '  ACcosh      {VNP+1 ,k+VNP+l .k-
          Buoyancy:

          A,      = A,        (Adiabatic condition)              (5.44.5)
           'MP.q    'MP+l,q
                                  124

-------
                        Port side, R=RQ, 5=
-


boundary cell
-
_[
.rMP,q (
1
,VNP,k-l .

V V
NP,k lvnPH,k
r^
interior cell
,UNP'k .

LfNP.k-1 r

FMP+I.Q e
f L
SfNP+Lik-1.

V+l.k


[WLik_

JNP+TJJ^I

                                                     k-l
p=MP       j=NP
                                        j=NP+l
Figure 5.9  Typical Vertical Port Side Boundary and Interior Cell
            (shallow water case only).

            Tr    indicates cell centered quantities n,
            1 p»q
            A-J and A.J.]
                             125

-------
         Salinity^
5.  Bottom Boundary

     a.  Deep Water Case (R.  <_ R <_ R^, Z = I^\ Refer to Figure 5.10)

         Velocity:

         U. i = U. j,  (free-slip condition)                   (5.45.1J
          J 5 *    tj 9 *~

         V. , =0  (level stream line condition)              (5.45.2)
          J > '

         Stream Function:
                      NB
         *. , = 1+A5V   V(n,l)sinh Sn   tosh E .            (5.45.3)
          J » i         £->               P         P
                     n=2
                     p=2

        where  NB is the number of inflow cells to the nominal plume

         boundary.

         Vorticity:

         Qn i = ~ nn 9 (free-slip condition)                   (5.45.4)
          [J 9  I      (J 9 C.

         Buoyancy:

         A,    = A,                                            (5.45.5)
          'p,l     'p,2

         Salinity:
                                                               (5,45.6)
         A     = A
                               126

-------
                                    11*2.
                                                k =  2
                        •P.2
                  interior  cell
            J-1.1
                 boundary  cell
                                    UJ,1
                                                q  =  2
                                                k  =  1
                                              (boundary)
                                                q  =  1
Figure 5.10   Typical Bottom Boundary and Boundary Cell

             [r    indicates cell centered quantities

              n, A, and A,.]
                        127

-------
      b.   Shallow Water  Case  (RQ  <. R i R^,  Z  = Zfa = 0;



                              Refer to Figure 5.10)




     Velocity:


     U. , = - U. 9 (No-slip condition)                         (5.46.1)
      J >i      J >^


     V, , = 0                                                  (5.46.2)
      J »i


     Stream Function:


     y    = i + 1  R 2v                                         (5.46.3)




     Vorticity:



     n    = " n  2 + AT"  ^U> 2 + U'-l 2^' (No's11P Condition) (5.46.4)




     Buoyancy:


     A,    = A,    (Adiabatic condition)                       (5.46.5)

      Vl     'p.2



     Salinity:


     A,    = A,                                                (5.46.6)

      dp,l     P»2




6.  Inflow-Outflow Boundary



     a.  Deep Water Case (R = R^  , Zfa <_ Z <_ Zh;  Refer to  Figure  5; 11)


         j = NO, p = Np



         Velocity:


         U k. t i  = "•   ---••..-..-- . ,.^.— [OJ     _ m       \               /r>lTT\
          NJ.k      Sinhf,N,AZ,  v  NJ.k    NJ,k-r               (5.47.1)



                                  128

-------
                             .R=R  , g = NJAC


J-l,k



w V 1 w
.NJ-l.k !vNJ.k l¥NJ,k J

i *
interior
^NJ-l.k-l p
1
FNP,q
cell
,VNJ,k-l .
'
UNJ,k
,VNJ+l,k

_rNP+l ,q
T
boundary cell
/NJ.k-l
T 1 1
VNJ+l,k-l
f
j=NJ-l      p=NP
j=NJ
                                                     k-1
p=NP+l     j=NJ+1
  Figure  5.11    Typical  Inflow-Outflow Boundary
                 and Interior Cells

                 ["iv  „ indicates cell  centered
                 L P»q
                 quantities n, A,  and  A^.J
                        129

-------
VNJ+l,k  = VNJ,k                                       (5-47


Condition   5.47.2  results  follow  from  the  stream  function


condition given  below.



Stream Function:
 *NJ,k  =  *NJ-l,k  (Level  Stream  11nes)                 (5.47.3)



 Vk =  2*NJ-l,k - *NJ-2,k (N° Chan9e of


 Vorticity:

      av
 Since |jj- =0  (Equation 5.47.3)


 the vorticity, nNp+] k is given by
   0      -  o."
   "NP+l,q   9Z
                NJ,k
Note that (3U/8Z)NJ k has been replaced by a central difference

form using even spacing of AZ.  For the more general case of


uneven AZ,  , refer to Appendix B   •



Buoyancy:

            pr ~pn
A-,        =  (-—^ )                                 (5.47.6)
  NP+l,q    pr  po  Ambient



Salinity:

              sr - s

A3        =  ^ S—:S  ^                               (5.47.7)
  NP+l,q      ^r   ^o Ambient
                        130

-------
6.  Shallow Water Case
      Same as above.

      A number of assumptions and restrictions are involved with the
above boundary values.  For instance, the sea surface is restricted
to remain flat, although visual observation indicates that a slight
"boil" will occur at the plume center!ine,  At the bottom boundary
of the deep water case, beyond Rb, it is assumed that there is neither
a vertical component of mean velocity nor any change in the horizontal
velocity profile.  Additionally it is assumed that neither A,  (nor r)
is diffused across this boundary.
      Transported quantities are assumed constant at R = RM.  Within
the framework of the difference scheme, this is a perfectly valid
assumption if convective terms, acting normal to this boundary,
dominate the diffusion terms.  Any quantity convected into the system
is assumed to have the ambient value.  Stream lines are assumed flat
or having constant slope at this point and recirculation of flow out
of the system is prohibited.
      Many of the above assumptions are a result of ignorance with
regard to processes outside the chosen system boundary.  Since there
is no way of regulating these processes, assumptions based on physical
insight are the only viable alternative.  Fortunately, the more
nebulous assumptions occur at points far removed (at R = Rro and
bottom boundary) from the region of prime interest.  And there is some
recourse, in that numerical experiments are possible which give in-
sight to the importance and effect of these assumptions.  Results
                                 131

-------
given in Chapter 7 reveal that the boundary specifications at R^
have little influence on the numerical solution as long as R^ is a
reasonable distance from the port (approximately two plume diameters).
5.9  Rectangular Coordinates
      The previous sections have dealt exclusively with (R,Z)
coordinates or transformed, (£,Z) coordinates.  In this section we
treat the governing differential and difference equations in rectang-
ular (X,Z) coordinates.
      Detailed derivation of these forms are omitted; only the
results are presented.  In contrast to previous considerations neither
transformed coordinates or unequal grid spacing will be considered.
      The physical problem which we wish to analyze is a two-
dimensional line plume that forms over a multiport diffuser line.
This condition is approximately realized once the flows from a series
of single round ports spread and interfere with one another parallel
to the diffuser line.  In dealing with the single round port, we were
restricted to stagnant environment because any cross-current would
destroy the problem symmetry and require a three-dimensional analysis.
In the line plume case we may consider environmental velocity com-
ponents which fall in the (X,Z) plane.  Figure 5.12 illustrates the
physical system for the line plume considered here.
5.9.1  Governing Differential Equations
      Differential equations for  the  X,Z coordinate system comparable
to Equations  (5.10),  (5.11), (5.12),  (5.13),  and  (5.14) given  in
                                  132

-------
          SEA SURFACE
RECEIVING
  WATER
                     PLUflE  IN
                                                       z=z,
                                        NOMINAL
                                         PLUME
                                       BOUNDARY
o

3
O
Figure  5.12  Physical System for  Line Plume Issuing
            to Flowing Receiving Water
                      133

-------
Section 5.4 are:
      Stream Function:
            
-------
            n  = <"/(v0/b0)  ,
where VQ and bQ are reference plume velocity and width, respectively.
      Dimensionless parameters are:

                   £x
             F  -
defined as in Section 5.4, keeping in mind that the radial direction
is simply the "X" direction in this section.
5.9.2  Rectangular Difference Equations
      Rectangular difference equations are formulated on a grid iden-
tical to that illustrated in Figures 5.3 and 5.4 with the corresponding
change from the £ coordinate to the X coordinate.
      Here we consider only a regular grid, which has spacing AX and
AZ.  Difference equations are given below.
                               135

-------
      Stream Function:
                           +  v.
               AZ
                                              +_L\AZ
                                         AX     AZ  7
                           "j.k
                              AZ
                                                              (5.55)
Vorticity,  n.  k,  is  the  average value for the four surrounding cells
(see Figure 5.4)  and given  as
Velocity is calculated  by
            U
             j,k
             j,k
and vorticity by
REVAX      RE AZ'
                                p»q
                                   l
                                                .q+0
(5.56)
                                                  (5.57)
                                                              (5.58)
                           136

-------
              2F7X
The buoyancy  parameter, A],  is  calculated by,
              RE PR AX2    RE PR AZ2 J  Vq
(IU.  ,
\  J'
                              u.
                                      REPR  AX
                                        X  X
          i  k)  + — ] — z- 1
           ,K;   R^PR^ J
                                              Ai
                                                '
                            137

-------
      The salinity or A3 transport equation is given exactly by
Equation (5.60) with PR replaced by SC, the eddy Schmidt number.

5.9.3  Rectangular Boundary Conditions

      Boundary conditions for the rectangular problem are substantially
the same as in the axisymmetric problem.  Notable differences are pro-
vision for crossflow and lack of problem symmetry.
      Referring to Figure 5.12 boundary conditions are as follows:

      1.  Sea surface (OfXfX^, Z=Zh)

            y  =  constant = y,
            n  =  0 (free slip condition)
           8A-j
           -ry  =  0 (adiabatic condition)
           3A,
2.   Inflow boundary (X=0, Z^

                           dz
                        r
            *  =  fi  -  I
                         Zh
                   3U
            8  •
                   3Z
                  D  ~ 0
                   r    °°
                  pr- po
                   .
           .    _   r
           43  '
                           138

-------
3.  Bottom boundary  (O^UXC  - ^ Xb, Z=Zb)

                 f Zb
      *  =  v-i  -  /      U   (Z) dZ = constant
               \     "
      n  =  0 (free  slip condition)
   3A-,
      1   =  0
    3Z
4.  Plume inflow boundary  (Xc  - ^ X^X^ + ^ X., Z=Zb)

    Assume that V,  A-,  and  A3 are  known from data or empirical
    relationships.
                   X
                          VbdX
                  Xc  ' 7 Xb
       n =   iy..  M.
5.  Bottom boundary  (X   + i X.< X
-------
    3A,
    3Ao
6.  Inflow-outflow  boundary (X=Xoo, Z,
-------
                              CHAPTER 6
                  CODE DESCRIPTION AND ORGANIZATION
      The computer program described herein obtains the solution of
the transformed difference Equations (5.21), (5.23), (5.24), (5.28),
and (5.30) for the quantities f, U, V, n, and A-| , (or r), respectively.
Through input option one may also obtain these solutions in ordinary
radial coordinates (see summary Section 5.7.3).  A program which
obtains the solutions through the use of the density disparity param-
eter A2 (as opposed to AI) has been used but is not presented in this
manuscript.
      The program consists of 20 subroutines and/or functions which in
part are managed by an executive routine called "SYMJET".  Initially,
the code was set up for the Oregon State University CDC 3300 time
sharing system.  This system, although extremely handy for program
development, is too small in terms of available core and too slow for
economically treating large problems.  The code version presented
here is adapted to the Computer Science Corporation Univac 1108
located in Richland, Washington.  This version of the code has also
been successfully executed on the Control Data Corporation 6600
located in Palo Alto, California, and on the CDC 6400 system at the
Battelle Memorial Institute in Columbus, Ohio.
6.1  Computational Procedure
      The primary task at hand involves the simultaneous solution of
one elliptic partial differential equation for the stream function, y,
                                141

-------
(Equation 5.21) and two parabolic transport equations for the
vorticity, n, and the buoyancy, A-J , [Equations (5.28) and (5.30),
respectively].  Equations for U and V (5.23 and 5.24, respectively)
may be considered as auxiliary, but are, nevertheless, essential and
need to be solved along with (5.21), (5.28), and (5.30) during itera-
tion.  In the case of neutral buoyancy, only Equations (5.21) and
(5.28) need to be solved simultaneously.
      The iterative procedure is built about the equations for Y, n,
and A-,. The technique used in the Gauss-Seidel method for all quanti-
ties defined by second order partial differential equations.  Liebmann
acceleration is employed with the alternatives of both under and over
relaxation.  Assuming all boundary conditions are set and pertinent
variables are initialized, the procedure is as follows:
      1.  Compute A,  _ and r    using Equation (5.30) based on
                    1K»H      PJ^
          previously calculated values of U. .,V. .  and appropriate
                                           J »K  J ,K
          transport coefficients.
      2.  Compute n    using Equation (5.28) and the previously
                   V »H
          computed values of U. .  ,V. ,., A,    and appropriate trans-
                              j »*  j»K-    i p »q
          port coefficients.
      3.  Update necessary boundary values  for Apr, and n.
      4.  Use the newly computed values of  n to compute the  stream
          function distribution from Equation  (5.21).  One or more
          iterations may be required to arrive at a satisfactory
          solution for v.  Compute  a new  velocity field V.  . and U.  ,.
                                                         J »K      J ,K
          from the  newly calculated f distribution.
                             142

-------
      5.   If the  eddy  transport terms are not constant, compute multi-
          pliers  FR  and  FZ  from new velocity field (for definition of
          the FR  and FZ  multiplier, see Chapter 7).
      6.   Repeat  Steps 1  through 5 until a preset convergence criter-
          ion is  satisfied  or a specific number of iterations has been
          completed.
6.2  Executive Program and  Subroutine Description
      As  mentioned previously, the computer code consists of an execu-
tive routine called  "SYMJET" and 20 subroutines and/or functions.   The
following discussion relates the primary duties served by each of
these routines.
SYMJET
      Executive routine
      1.   Reads case header and integer case set-up information.
      2.   Reads alphanumeric data for line printer output array
          option, plot tape options, isoline  interpolation options,
          and  program  control.
      3.   Calls subroutines for data input, problem set-up and
          initialization, and problem execution.  The subroutines
          called  are (in the calling sequence):
          •  INPUT
          .  READY
          .  PLABAK
          •  STREAM  (for inviscid flow  solution)
          .  SSCOMP
                           143

-------
          .  INTERP
      4.  Performs other miscellaneous tasks such as clock initial-
          ization, tape rewind,  presetting variables,  etc.
SUBROUTINE INPUT
      General  data input routine
      1.  Reads restart tape if  required.
      2.  Reads remaining input  data from  cards.
      3.  Converts portions of input data  to appropriate quantities
          and  units (e.g.,  temperature data to density data).
      Subroutine is called  once  during execution.
SUBROUTINE READY
      Problem  set-up routine
      1.  Sets all computed constants.
      2.  Sets constant boundary conditions.
      3.  Presets turbulence multipliers.
      4.  Option to call SUBROUTINE SIMJET.
      5.  Option to call SUBROUTINE GAUSS.
      Subroutine called once during execution.
SUBROUTINE PLABAK
      General  information and debug output
      1.  Writes to line printer various computed and input supplied
          variables and the operation modes of current case.
      2.  Writes to line printer constant arrays used in the difference
          equation computations.
      Subroutine is called once  or not at all at the user's option.
                             144

-------
SUBROUTINE STREAM (IT, NSKIP)
      Solves for stream function, Y
      1.   Computes the viscous or inviscid stream function (Equation
          5.21) by Gauss-Siedel iteration.  When called, this  sub-
          routine iterates on y (PSI) "IT" times.
      2.   Upon completion of "IT" iterations the velocity components
          U. ,, and V. b are computed by the auxiliary Equations  (5.23)
           j »*•      J »K
          and (5.24).
      For an inviscid flow computation (the inviscid flow solution
      may be called for the purpose of initializing the viscous  flow
      computation if desired) STREAM is called and returns control  to
      the executive routine.  When STREAM is called from SSCOMP,
      which computes the viscous flow field, control is then returned
      to SSCOMP.  Subroutine STREAM constitutes what is referred to
      in this manuscript as the "inner iteration loop" (subroutine
      SSCOMP constitutes the "outer iteration loop") and is called  at
      least once for each "outer iteration".
SUBROUTINE SSCOMP
      Computes steady flow solution of all transport equation
      1.   Solves transport equations for
          .  A1
          •  r and
          •  ft,
          using Gauss-Siedel iteration with Liebmann acceleration
          (deceleration).
                                  145

-------
      2.   Updates boundary values of L,  r,  and :<-.
      3.   Computes convergence rate information and the cell
          indices having the slowest convergence.
      4.   Calls subroutine STREAM to compute velocity field.
      5.   Calls subroutine EDDY to compute  eddy transport multipliers
          as required.
      6.   Writes out monitor node values.
      7.   Calls subroutine OUTPUT for either interim or final  array
          output.
      8.   Generates plot data tape.
      9.   Computes surface area above Tamb  in 1 °C increments.
     10.   Performs a Gamma constituent balance error, (rin~rout)/r- »
          for the overall  system and then  returns  control to the
          executive routine.
      This subroutine is referred to as  the "outer iteration loop" and
      is  called but once during a case execution.   The code spends
      the majority of the execution time in this routine.
SUBROUTINE EDDY (M)
      Computes eddy transport multiplier FR and FZ
      1.   Computes potential core.
      2.   Computes plume half radius, RI .^  and nominal plume boundary,
          R405» at each vertical grid point.
      3.   Computes FR from mixing length theory.

                    FR=Vmax  '  Rl/2
                            146

-------
      4.   Computes FZ based on mixing length theory and  incorporates
          Richardson number modification (computes  point Richardson
          number,  RI, and calls function RCHMOD for modifier).
      If  eddy multipliers are computed based on the velocity  distri-
      bution, this subroutine is called once during each "outer
      iteration".   Either FR, FZ or both may be computed selectively.
      (Parameter M in the call list specifies the option).  Details
      of  the particular eddy transport models used  and regions of
      applicability are discussed in Chapter 7.  Also, this sub-
      routine may be bypassed a set number of iterations for  computa-
      tion stability purposes (discussed in Chapter 7).
SUBROUTINE OUTPUT (MODE)
      Primary line printer output call routine
      1.   The primary purpose of this routine is to call selectively
          the output array writer subroutine, AROUT, based  on the
          alpha input read in through the executive routine.   The
          arrays and array header Holleriths are aligned in the call
          list of AROUT.  This subroutine may be called  selectively
          for array writing through the input Fortran variable NOUT.
          That is, every time that the "outer iteration" number is
          divided by NOUT and yields a whole number, the array
          writing  routine is called.  The parameter, MODE,  is an out-
          put option.
      2.   The secondary purpose of subroutine OUTPUT is  to  write out
          selectively the convergence rate information computed in
                               147

-------
          subroutine SSCOMP, that is, maximum changes in *, A  and n
          and the nodal  location of these changes,  during successive
          iterations.   The iteration numbers selected for output are
          specified by the input Fortran variable NTTY,  in the exact
          manner that NOUT is used in 1. above.
SUBROUTINE AROUT (list)
      General array writer
      This subroutine is used to write out all  computed  arrays speci-
      fied for printing.  The appropriate array,  header  and grid
      coordinates are aligned in the call list at subroutine OUTPUT.
      Miscellaneous computations are also performed here as necessary.
      For instance, if normalized arrays are desired, these are
      normalized in AROUT and if temperature arrays are  required the
      buoyancy parameter (A-|) array is converted  to a temperature
      array through successive calls to function  TEMP.
SUBROUTINE INTERP
      Calling routine for isoline interpolation
      The only job performed by this subroutine  is  selectively setting
      up arrays to be interpolated by the general interpolator
      routine, IS06EN.  Selection is made through input  of the Fortran
      alpha array TERP during execution of the executive routine.  The
      particular array,  header and other appropriate data are aligned
      in the call list of ISOGEN.  This subroutine is optionally
      called through the executive routine following execution of
      SSCOMP.
                              148

-------
SUBROUTINE ISOGEN (list)
      General isoline interpolator
      The function of ISOGEN is to interpolate a given array,  aligned
      in memory through the subroutine call list, for isolines whose
      values are selected at input and specified by the Fortran
      array, ISOLN. For a specific array (say the stream function
      array) the coordinates of an isoline (streamline) are quad-
      ratically interpolated and coordinates printed.  Contouring may
      be accomplished by hand plotting the results.  Automated
      plotting of the computed points would be quite difficult since
                                 i
      the points are not ordered.
SUBROUTINE GAUSS (N)
      Optionally computes Gaussian distributions for inflow
      This subroutine computes Gaussian boundary distributions for V,
      A,, and r in either the zone of flow establishment or the zone
      of established flow.   The particular option is determined by the
      parameter, N.  These computations are based on the Albertson
      et al. [4] data and theoretical results given by Abraham [1].
      The routine is called once from subroutine READY.
SUBROUTINE SIMJET (list)
      This routine computes the center!ine distributions of V, Ag and
      r from the similarity solutions of a vertical plume given in
      Chapter 4.  For the homogeneous problem, V is calculated from
      Equation (4.43) and A2 from (4.4).  In the case of stratification
 Automated contouring is accomplished using a special contouring
 routine.
                              149

-------
      these quantities are computed from Equations  (4.36),  (4.37),
      and (4.38) using the fourth order Runge-Kutta technique.
      Results from this routine may be used  for inflow boundary
      information in the more elaborate finite-difference method for
      the confined plume.  Calling is through subroutine READY and
      is performed at most once.
FUNCTION SIGMAT (SAL, T, N)
      Given the salinity, SAL, and temperature, T,  this function com-
      putes Sigma-t (at, see Section 3.6) based on  algebraic equations
      given in the U.S. Navy Hydrographic publication  number 615 [103]
      or as given in Hill [39].
FUNCTION TEMP (SALT, SIGMA)
      Given the salinity, SALT, and the density in  Sigma-t  units,
      SIGMA, this function solves the equations referenced  above for
      the temperature in degrees centigrade  by the  Newton-Raphson
      method.  The function SIGMAT (SAL, T,  N) is repeatedly called
      during the iteration process.
FUNCTION SANK (X, N)
      Hyperbolic sine coordinate transformation function which yields
      Sinh (X) for N = 1 and X for N = 0 (linear radial coordinates,
      no transformation).
FUNCTION CASH (X, N)
      Hyperbolic cosine transformation function which yields COSH (X)
      for N = 1, and 1.0 for N = 0.
                           150

-------
FUNCTION RCHMOD (N, RICH)
      Computes Richardson number (RICH) modification of the vertical
      eddy viscosity coefficient by one of five different models
      (option given by N).  These models are given in Chapter 7
      (cf. Table 7.5).
6.3  Flow Charts
      Detailed flow charts of all subroutines in the SYMJET computer
code would require an extensive amount of space.  For this reason
only the main subroutines and the executive program will be illus-
trated.   The charting of these will also be somewhat abbreviated.
A partial bibliography of the computer variables may be found in the
program listing (Appendix E).
                               151

-------
           SYMJET  FLOW  CHART

          (Executive  Routine)
     Read alpha case header

    Read integer set-up data
/  Read alpha TLIST option
|    Set write option arrays
 AOUT,  PLOT, TERP, AND CONT
     Set auxiliary Indices
 Initialize arrays and  constants
   Compute monitoring  arrays
      Generate array
    Instruction vectors from
    TLIST Options: NRITE(J
  N3DPT(J),  fSOPT(J) CONTR
                152

-------
     CALL INPUT:   Reads
    main data file from
    cards and optionally
Initializes arrays from tape
   CALL READY:   Completes
 Initializations,  computes
    constants,  and sets
   fixed boundary  values.
   Positions  output tape.
        CALL  PLABAK:
    Write out computed
  and  supplied constants,
      and debug arrays.
   CALL STREAM:  Compute
   1nv1sc1d flow solution
                 153

-------
 CALL SSCOMP :  Compute
 solution to transport
and auxiliary equations
CALL  INTER?  : Compute
 contour coordinates
   REWIND output
        tape
                154

-------
      SUBROUTINE  INPUT  FLOW  CHART
        READ LUN 7
  UNO. n, a,, U, V, y  and r
Data from previous computation
    for Initialization or
     continued iteration.
       READ  data card
      DATA.  JI, KI,  NI
 GO TO (N,,  N2—N12), NI+1
   N.  : Replace appropriate
     variables with DATA
                           155

-------
       SUBROUTINE READY  FLOW  CHART
     Set up various computed
Constants: e.g., SC(J,L),  SZ(K,L)
 Preset variables: e.g.,  FZ(J,K)
             FR(O.K)
 Set  up coordinate systems: Z(K),
 ZC(K), X(J),  XR(J), R(J), RC(J)
      Set inflow boundary
     velocity according  to
           "INMODE"
                           156

-------
Compute and set all
fixed or Initialized
boundary condition not
treated above


157

-------
       SUBROUTINE  STREAM  FLOW  CHART
       SET STREAM FUNCTION
      Inflow-outflow boundary
       condition. * (NJ.K)
       Compute vortlclty, n
     at cell  corners (OMEGA 3)
         Compute f (J.K
         Equation (5.27
    Accelerate  (or decelerate)
        Solution, v (J.K)
     Compute  U(J,K) and  V(J.K)
from * (J,K), Eqs. (5.23) and (5.24)
                           158

-------
     SUBROUTINE  SSCOMP FLOW  CHART

Compute
Compute
i

^ (p.q).
r (P.q).

Eq.
Eq.

(5
(5

.30)
.30)
  Accelerate (or decelerate)
    (p.q) and r (p,q)
   Set boundary values for
next Iteration on AI, r and n
                                          Reset iteration
                                             limits;
                                          initialize eddy
                                          factors,  FR & FZ
                    159

-------
• Compute n (p,q) by
Equation (5.28)
• Compute maximum change 1n
n (p,q) and value
of p and q for location
• Accelerate or decelerate
solution n (p.q)


Compute updated boundary
conditions for n (p,q)
(

Write out monitor
node Information
      CALL  EDDY
    Computes  eddy
transport multipliers
 FZ(J.K)  and  FR(J.K)
                       160

-------
Write to LUN 8 (MAG. TAPE)
 ITNO, n, A, U.  V.  v,  r
                                             CALL OUTPUT (1)
                                            Calls array writer
                                             CALL OUTPUT (2)
                                         Call Intermediate  output
                              161

-------
 Create plot files
  LUN 8 according
    to N30PT(J)
   Compute surfacs
    Isotherms 1n
  Increment of 1 °C
    Perform GAMA
   Sum convergence
       check
Print GAMA Sum error
              162

-------
                              CHAPTER 7
             CODE VERIFICATION AND NUMERICAL EXPERIMENTS
      In this chapter we are concerned with verification of the
numerical model.  Ultimately, the program is to be used in describing
the plume resulting from large vertical thermal outfalls in shallow
water, and, as previously mentioned, published field data concerning
velocity and temperature distributions along with other pertinent data
needed for evaluation or verification are essentially non-existent for
these cases.  Even laboratory data from hydraulic models are scant and
steady flow experiments to model quasi-steady oceanic conditions with
stratification are essentially impossible.
      Verification of the numerical techniques will be carried out by
using the code described in Chapter 6 to simulate various problems
which have been well  studied, both experimentally and analytically,
and for which much information has been published in the literature.
One such problem which the code can easily handle is the deep water
momentum jet.  In this case much knowledge has been compiled concern-
ing velocity distributions, concentrations, and turbulence parameters.
The computer code can easily handle interacting buoyancy for the same
geometry.  Although there is a lesser amount of experimental data pub-
lished in the open literature for buoyancy cases, especially on turbu-
lent parameters, there is enough information for meaningful comparisons
with the numerical model.
                           163

-------
      Once the computer program is verified using this published infor-
mation, the program can be applied with confidence to conditions of
more interest and practical value, such as shallow water and stratified
ambient cases.  Having checked the program against experimental  results
for simple cases, we know at least that the numerical procedures are
working correctly, although auxiliary models (e,g., turbulence)  may not
be entirely correct.
      Also presented in this chapter are some of the code operating
experiences, turbulence modeling, solution convergence and stability,
and discussion of some of the more troublesome boundary conditions.
7.1  Deep Water Plumes
      By deep water plumes we are implying that the effluent is  dis-
charging into a semi-infinite water body, although as a practical
matter computational boundaries must be finite.  For program verifica-
tion, we use the following deep water flow categories:
      •  Momentum Jet - the fluid motion is induced entirely by  the
         effluent initial momentum.  Buoyancy is also calculated but
         is decoupled from the momentum equation and may be used as
         a measure of concentration.  This case is indicated by
         FQ + ».
      •  Pure Buoyant Plume - in this instance there is no effluent and,
         consequently, no initial momentum.  The driving force is pure
         buoyancy caused by a source of heat located in the position
         of the outfall port. An arbitrary reference velocity is used
                               164

-------
         along with a length scale that corresponds to a port radius.
         This case is indicated by F  =0.
      •  Mixed Flow - both initial momentum and buoyancy have varying
         degrees of importance.  In this case 0 < FQ < <*>.
Various cases of the above categories have been checked against avail-
able experimental data and similarity solutions.  These cases are
itemized in Tables 7J, 7.2, and 7.3.
      Four different effluent velocity profiles and concentrations (or
temperature) have been used in this work which are:
      •  Type 1     : Gaussian profiles, established at 4.5 diameters
                      from the port exit,
      •  Types 2, 3 : Power law velocity profile at the port exit  with
                      a constant radial concentration (or temperature)
                      distribution, and
      •  Type 4     : Constant radial distribution of all  quantities
                      at the port exit.
Equations for these profiles are given in Table 7.1.
      Figure 7.1 illustrates a typical grid system in R-Z coordinates.
Note  the effect of  the  hyperbolic sine transformation in  stretching
the cell widths as the distance R is increased.  The computation grid
(C,Z-coordinates) has uniform radial cell widths as illustrated in
Figure 5.3.
                               165

-------
TABLE 7.1.  SUMMARY OF MOMENTUM JET VERIFICATION CASES (Fc
                    PR,, = .80,  PR, = .80
Case
No.
1
2
3

4
5
6
7
8
9
10
11

12
Grid
Size
26 x 40
35 x 40
26 x 40

40 x 33
40 x 33
40 x 33
40 x 33
40 x 33
40 x 33
40 x 33
30 x 26

30 x 26
^
.2
.1
.2

.12591
.12591
.12591
.12591
.12591
.12591
.12591
.2

.2
AZ
2
2
2

2
2
2
2
2
2
2
2

2
Z
(Surface)
43.5
43.5
43.5

64
64
64
64
64
64
64
43.5

43.5
R
00
74.2
14.96
74.2

67.85
67.85
67.85 '
67.85
67.85
67.85
67.85
74.2

74.2
Boundary
Type
1
1
1
2
N=7
3
N=7
N=10
2
N=10
3
N=7
2
N=6.6
3
N=10
1

1
er
Type2
3
1
1

3
3
3
3
4
3
3
Inviscid
Test
Creeping
Test
ez
.0001
.0001
.0001

.0001
.0001
.0001
.0001
.0001
.0001
e =e
z r




-------
       Inlet velocity profile type:
        .  V(R,Z) = V(0,Z)
       2. .v(Rf0) = Ai-JLLj2N+ll


                        2N2




       3.  V(R,0) = (1-R)1/N
-^      4.   V(R,0) = Vrt = Constant
CT>                    0
      2

       Radial  eddy viscosity calculation type:



       1.   er = .0295 rQv0 = Constant




       2.   e  = .0256 r,/?v :  Prior specification of r1 ,2 from Gaussian distribution of velocity,


                              v  calculated iteratively.
                               m


       3.   er = .0256 r1/2vm:  Iterative calculation of both r]/2 and vm.




       4.   er = .0263 r1/2vm:  Same as Type 3.

-------
cr>
CO
            TABLE 7.2.   SUMMARY OF PURE BUOYANT  PLUME VERIFICATION CASES (FQ = O)


                                  PRf = .714,  PRZ =  .714



                                                                                      Heat
Case     Grid                                   Z          R             er          Source
 No.     Size           Ag          AZ       (Surface)       °°           Type        Condition


 13    40 x 33        .12591         2           64        67.85            3             1


 14    40 x 33        .12591         2           64        67.85            3             2
            i
            Reference  densimetric  Froude number is not zero but based on a reference velocity since there
            is  no  inflow  at  the  source.


            2See Table  7.1.

            3
            Heat Source Type:

            1.  Weak Source:  Simulated heated plate maintained at AT = 25 °C.
                Heat transferred to fluid by conduction alone over range 0
-------
                                  TABLE 7.3.   SUMMARY OF MIXED FLOW VERIFICATION CASES
VO
Case
No.
15
16
17
18
19
20
21


22

23

24

25

26

27
i
See
Grid
Size
26 x 40
26 x 40
26 x 40
26 x 40
26 x 40
26 x 40
26 x 40


40 x 33

40 x 33

^40 x 33

40 x 33

40 x 33

40 x 33

Table 7.1.
M.
.2
.2
.2
.2
.2
.2
.2


.12591

.12591

.12591

.12591

.12591

.12591


AZ
2
2
2
4
2
2
2


2

2

2

2

2

2


#e = 1.0, Gaussian distribution
Z
(Surface)
43.5
43.5
43.5
82.5
43.5
43.5
43.5


64

64

64

64

64

64



R
00
74.2
74.2
74.2
74.2
74.2
74.2
74.2


67.85

67.85

67.85

67.85

67.85

67.85


e" A ^Zs
p Boundary
o Type
52
52
52
52
35
106
52


45.5

45.5

1

1000

45.5

45.5


rZ)
1
1
1
1
1
1
1

2
N=7
2
N=7
2
N=7
2
N=7
2
N=10
2
N=IO



Type
1
2
3
3
3
3
3


3

3

3

3

3

3



ezQ
.0001
.0001
.0001
.0001
.0001
.0001
#See
Below

.0001

.0001

.0001

.0001

.0001

.0001



PRr
.714
.714
.714
.714
.714
.714
.714


.714

.80

.80

.80

.714

.714



PRZ
.714
.714
.714
.714
.714
.714
.714


.714

.80

.80

.80

.714

.714




-------
£t-
22-

?fi

18-

Ifi-

M
» Id
g"~
i—
co 12
S "'
D-
_l
S in
x IU
<
oo
m «
uj o-
o
h— 1
^ 6
£ 6
i—i
o
4_

2 _

0-


ca -
23 -

91 .

1Q .

17.

1 c;

13 -
It
K
11 -

Q .

7 .

5 -

3 -

1 .
1
0
























10
2






































































































1

























5
4



















































e


















































20
E

























\

























i
10

























12

























14
























2i
16

























18

























1
20
                   DIMEiMSIONLESS RADIAL POSITION,  R

Figure 7.1.   Computational  Grid  for  The Stream  Function,  v,
             Illustrating  the  Effect of the  Sinh  U)
             Transformation (A?  =  .14690,  AZ =  1.0)
                           170

-------
7.1.1   The Momentum Jet
      A vast amount of information has been gathered concerning the
dynamic behavior of momentum jets dating back to Tollmien's [98] work
of 1926.  Hence, there is sufficient data reported in the literature
to check all of the gross aspects of the jet structure computed.  In
verifying the computational technique with the published data we use
the following jet characteristics:
      •  Centerline velocity and concentration,
      •  Radial distribution of axial velocity and concentration,
      •  Rate of jet spread, and
      •  Radial velocity.
Although there is a vast amount of published data available for veri-
fication, the primary data used is from Albertson, et al. [4], Baines
[8], Abraham [1] and information obtained from several researcher's
published in Chapter 24 of Schlichting's text "Boundary Layer Theory"
[84].  Additional information is obtained from reviews by Gauntner
et al. [32] and Chapter 6 of Hinze's text "Turbulence" [40].
      Some of the relevant restrictions in this section are:
      •  Vertical turbulence is negligible; one case is run to
         verify this fact.
      ,  The computational grid system has an impermeable upper
         boundary.  Hence, velocity profiles begin to "feel" the
         boundary some distance before it is reached.
Aside from the quantitative verification mentioned above, illustrations
of streamlines, concentration, and vorticity contours, and three-

                             171

-------
dimensional plots of the same information are provided for additional
qualitative assessment.  Table 7.1 summarizes the momentum jet cases
run.
7.1.1.1  Centerline Velocity and Concentration for Momentum Jets
      A similarity solution for vertical plumas was given in Chapter 4
as
            E* = ^  + Jp-  (Z2 - Z2)                        (4.43)
                 Z3    3ZFQ         e

In the case of a momentum jet F  -»• » so that,

            E*=  ^   •                                       (7.1)
                  Je
Then by definition

            ^•f                                          (7,,
            /FT    Ze

where  again Vm is the centerline velocity, K is related to the plume
entrainment (see Table 4.1), numerically equal to 77, and Z  is the
potential core length based on concentration (cf. Abraham [1]).
      By Equation (7.2)
                                                              (7-3)
According to Abraham I  ^ 5.6; hence,
            \ - 6-2/2                                         (7.4)
which is also the result obtained by Albertson.
                            172

-------
      Equation (7.4) implies that a plot of the dimensionless center-
line velocity, Vm, versus axial distance in port diameters Z has slope
of -1 when plotted to Log-Log scale, and has an intercept of 6.2 on the
^-coordinate when Vm-= 1.  Experiments carried out by Albertson are
probably the most frequently quoted data bearing out Equation (7.4).
Various other researchers have carried out similar experiments (e.g.,
Baines, Tollmien and Reichardt [77]).  Although there seems to be
general agreement that V  ^ Z~ , there is some disagreement on the
potential core length  (hence, the constant of proportionality), or the
Log-Log plotted  intercept value mentioned above.  A review of a portion
of this work is  given  by Gauntner.   It is noteworthy to point out here
that the potential core length (see  Figure 4.1) is assumed to be the
centerline velocity plot intercept  (Vm=l, Z=6.2; see Figure 7.2),
                           AXIAL DISTANCE,  t
         Figure  7.2.   General  Features  of Momentum  Jet  Centerline
                      Velocity (Based on Albertson's  data)
                                 173

-------
although the actual potential core length may be somewhat smaller.
For instance, Albertson measured an actual length of approximately  4.5
whereas their similarity solution is based on 6.2.   The reason for
using the value 6.2 is that it is more representative of downstream
data than 4.5.  As a matter of fact, similarity solutions are not valid
out to approximately 10 to 12 diameters.   In Figure 7.2 , the distance
Z = 4.5 is the approximate distance where deterioration of centerline
velocity is first apparent.
      Figure 7.3 illustrates centerline velocity, V , and concentration,
Cm, comparisons for
      •  Similarity theory
      •  Experiment, and
      •  The present computational technique.
The similarity theory concentration distribution along the centerline
is
            Cm = 5.6/Z                                          (7.5)
as given by Abraham.
      Figure 7.3 indicates remarkable agreement between the computed
and measured centerline velocity distribution.  Concentrations agree
with the similarity curve almost identically past t ^ 20.  These results
are based on the Type 1 boundary conditions (Section 7.1).  Computa-
tional runs 1 and 3 also use the Type 1 boundary condition, for differ-
ent water depths and node spacing; although these cases are not plotted,
centerline distributions nearly identical to those depicted in Figure
7.3 were obtained.  The only deviation found between experimental and
                               174

-------
    U
en   
1.0
.5
.4
.3
.2
0-1








-
I









i
VELC
• EXF
• CO
°IK
A
)CIT
=»ERI
MPL
/IILA
-
















Y
MENTAL RESULTS ALBERTSON ET AL. [4]
JTED
R SOLUTION V- ^p
. v CONCENTRATION
>
' i








^








\
\
4







i\
V
1







\
^





o CO
	 SI
AE
MPUTED
MILARITY SOLUTION c^
JRAHAM [1] C - *%
\ TYPE 1 BOUNDARY CONDITION
•xW v(R A *>) •=. c*~TT (.%/*)

Vx




\
V
i
23*5 10 20 3


I


1












-
-
-
0 4-0 SO 1C
                                           AXIAL DISTANCE,  2

              Figure 7.3.   Comparison  of  Experimental Data and Similarity Solution with  Computed
                           Results for a  Momentum  Jet. Center!ine Velocity and Concentration  for  Case  2.

-------
computed centerline velocity in these cases is that a very slight dif-
ference in slope was noted, whereby the computed slope was very
slightly less steep than -1.
      Similar results for Case 4, which uses the Type 2 boundary condi-
tion, are given in Figure 7.4.  Note that the 1/7 power velocity pro-
file gives a centerline value of 1.22 for an average jet exit velocity
of VQ = 1.  These centerline velocity results are somewhat higher than
Albertson's data, but agree well with the data obtained by Baines for
an initial Reynolds number of 7 x 10 .   Baines contends that there is
a Reynolds number effect on the potential core length and offers data
which apparently substantiates his assertion.  According to Gauntner,
this facet of jet theory is apparently still unresolved.
      The computed data for this case reveals the relationships:
            Vm * 7/Z.                                           (7.6)
and
            Cm2i5.1/Z.                                         (7.7)

Again, the computed velocity distribution is very slightly less steep
than a slope of -1.
      Figures 7.5 and 7.6 illustrate centerline velocity and concentra-
tion distributions for Cases 5 and 6.  Both of these cases again use a
Type 2 boundary condition with the inflow velocity distributions given
by
            V(R,0) = (1-R)1/N.                                  (7.8)
Case 5 uses N equal to 7 whereas N in Case 6 is equal to 10.
                            176

-------



1
1*0 1

E
U
i= :>
It"''
LU t J .•?
U 0
z: _i
0 U
u >

0 (
VELOCITY
• EXPERIMENTAL DATA FROM BAINES [8] REYNOLDS NO.-7*I04
• COMPUTED
•
»• i






-
-

-

-
i


<











1
<













i



(










i

v
\'
[k










I

\

v
V
<











\
i '
ss
i X












\
i
s
i








CONCENTRATION



\
•
N.,
'







o COMPUTED
r _ 5.1
m Z
\ ™
KV(.


VVI
\ N
\
\



i
PE 2 BOUNDARY CONDITION
R,0)» 1.22(1 -R)'/r

\

*X ^>s
^V i
V

i




\
\
,





^
^o
i





\
? f






\















-

•~

-
















2345 10 20 30 40 50 100
                                 AXIAL DISTANCE,

Figure 7.4.   Comparison  of  Experimental Data and Similarity Solution with Computed Results
             for a  Momentum Jet.  Centerline Velocity and Concentration for Case 4.

-------
   1 .0
o
o
o
.5
    .3
    .2
    . 1
VELOCITY:
   •   COMPUTED
  	Vn  =  5.8/2
CONCENTRATION:
   o   COMPUTED
  	 c   =  5.1

TYPE  2 BOUNDARY CONDITION
   V(R,0) = (1-R)1/7
                               5           10
                               AXIAL  DISTANCE, 2

         Figure 7.5.  Computed Centerline Velocity and Concentration
                    for Momentum Jet, Case  5
                                                                          50

-------
vo
     £
     U

     Z
     o
l-
z
(J
       u
       o
     U >
1.0
.2
O.I
VELOCITY
• EXPERIMENTAL RESULTS ALBER^
•COMPUTED
— SIMILAR SOLUTION Vm a ^f




-
-
-
-
<






1




__..


i

4





i
\
N
,





i
\
V
\
i





\
\
\ I
\








V
(






\
K



CON(
o rn


V 1 YPt
^%
\

rSON ET AL.
:ENTRATION
[4]
MPUTED
'••¥
E 2 BOUNDARY CONDITION
t

\


1


1^












-
-
-
2345 10 20 30 40 50 10
                                            AXIAL  DISTANCE, 2

          Figure 7.6.   Comparison of Experimental  Data  and  Similarity Solution with Computed Results
                       for a Momentum Jet.   Centerline  Velocity and Concentration for Case 6.

-------
According to Schlichting, these profiles correspond to pipe Reynolds



numbers of 1.1 x 105 and 3.2 x 106, respectively.  The computing tech-



nique shows a marked difference between the asymtotic centerline veloc-



ities for these two cases, that is, for large 2,



            Case 5:  Vm ^ 5.8/2                                 (7.9)



            Case 6:  Vm ^ 6.2/2                                 (7.10)



Although the slope is still  approximately -1  and the asymtotic concen-



tration for both cases is given by,



            Cm^5.1/Z.                                         (7.11)



Note that



            V(0,0) = 1



which results in an average inflow velocity less than unity.



      From these results it is tempting to conclude that since the



inflow velocity profile has an effect on the -1  slope intercept,



a Reynolds number effect on the potential core length is demonstrated.



However, it is felt that the lack of finite difference resolution and



shortcomings in modeling turbulence in the zone  of flow establishment,



are sufficient to shadow such a conclusion.  Comparing Case 4 where,




            V(R,0) = 1.22 (1-R)1/7                              (7.12a)



and Case 7 (Figure 7.7) where




            V(R,0) = 1.155 (1-R)1/10                            (7.12b)



reveals asymtotic velocity profiles,



            Vm-7/Z                                            (7.13)



and concentration



            Cl 5.1/2.                                         (7.14)
                           180

-------

                                                                                        I   I  I
          1 .0
oo
           .5
            .1
VELOCITY:
   •   COMPUTED
 	  Vm  =  7/2
CONCENTRATION:
   o   COMPUTED
 	  Cm  =  5.1/2
TYPE 2  BOUNDARY CONDITION
 V(R,0)  - 1 .ISSO-R)1/™
                                          1

                        Figure 7.7.
                   5           10
                       AXIAL DISTANCE, 2-
             Computed Center-line Velocity and Concentration
             for Momentum Jet, Case 7
                                                                                             100

-------
It is important to note that the jet exit average velocity in both
Cases 5 and 6 is unity whereas it is 1.22 in Case 4 and 1.155 in
Case 7.
      In all computer runs cited thus far, the radial  eddy viscosity
has been computed from Prandtl mixing length theory.  This particular
aspect of the work is discussed in more detail in Section 7.2.
Essentially, the eddy viscosity is calculated by

           er = c vmax rl/2>                                    (7J5)
where vmav is the centerline velocity, r,/0 is the jet half radius and
       ITiaX                              I / L.
c is a constant having the value .0256 for an axisymmetric momentum jet
(cf. Schlichting [84], p.  699); all  cases thus far use c = .0256.
Case 8 (see Figure 7.8) uses c = .0263 (picked quite arbitrarily and
as a fraction is 1/38 = l/REr) and is to be compared to Case 5,
Figure 7.5.  The net effect of this change is a slight shift in the
velocity slope toward -1 (difficult to see slope shift from compared
figures, but numerical results bear out the change).  Although the
higher value of c appears  to yield a velocity slope nearer -1, the
value c = .0256 is used for all following computations in this
manuscript.
      Case 9 (Figure 5.9)  represents an additional case using Type 2
boundary conditions with a velocity profile at the jet exit given by
                                1
            V(R,0) = 1.24 (1-R)6'6   .                          (7.16)
Note that all cases (4, 7  and 9) use the boundary velocity profile
                           182

-------
                                        	T
       1 .0
CO
oo
         .5
      E
         . 1
VELOCITY:
  •  COMPUTED
- Vm  =  5.7/Z
CONCENTRATION :
  o  COMPUTED
                   TYPE  2  BOUNDARY CONDITION
                     V(R,0)  = (1-R)1/7
                          j	i
                                                   10
                                            AXIAL DISTANCE,
                                                                       TOO
                         Figure 7.8.   Computed Centerline Velocity and Concentration
                                     for Momentum Jet, Case 8

-------
         1 .0.
oo
-p.
          .5
          .1
                    VELOCITY:
                       •   COMPUTED
                    CONCENTRATION:
                       o   COMPUTED
                     -- Cm =  5.1/Z
                    TYPE  2 BOUNDARY  CONDITION
                     V(R,0) = 1 .24(1-R)
                                         5           10
                                            AXIAL  DISTANCE,  Z-

                         Figure 7.9.   Computed Centerline Velocity and Concentration
                                     Distribution for Momentum Jet, Case 9
                                                                                       i	i  i
100

-------
            V(R.O) =   N+12N+1)   (1_R)N                        (7.17)
                          2fT
for the jet.   In all of  these  cases the asymptotic centerline velocity
profiles are essentially identical and represented quite accurately by
            Vm^7/2                                            (7.18)
and concentration given  by
                                                                (7.19)
      Figure 7.10  illustrates  these cases where the distribution is
normalized by dividing each value  by the corresponding value of V(0,0).
The net result of  this operation is that the solution collapses to the
cases using corresponding values of N and where V(R,0) is set by
Equation (7.8).  Although this result was certainly expected, it serves
to illustrate that the computer program is functioning correctly in
this sense and to  bear out again the velocity profile effect on the
asymptotic centerline velocity distribution (Figure 7.10).  Computa-
tionally, this condition is apparently caused by the differences of the
jet exit vorticity distribution.
      Vertical eddy diffusion, which should be of minor importance in
the jet mainstream, has also been  ignored in cases cited to this point.
By ignored, it is meant that the value has been set to compare with
molecular viscosity which is perhaps three orders of magnitude smaller
than the jet induced eddy viscosity.  The primary reason for vertical
diffusion being set to a very small value in these verification studies
                                  185

-------
                     1 .0
oo
CT>
                      .5
                                         CASE  9
                                                                    I	I
                                               1 0

                                               AXIAL  DISTANCE,  Z-
100
                         Figure 7.10.   Centerline Velocity Distributions for Cases  4,
                                       7,  and 9, Normalized to  VQ = 1.0

-------
is so that vertical entrainment near the surface where the jet is
spreading laterally will be minimized.
      With a large value of vertical diffusion, in   nonstratified
media, streamlines outside the jet would be distorted upward because of
the vertical entrainment in the lateral spread and would not be a
realistic representation of deep water conditions.
      In Case 10 (Figure 7.11) the vertical eddy viscosity has been
accounted for by setting
               = er
      Figure 7.11 is to be compared with Figure 7.5 (Case 6).  Case 10
shows a slight increase of centerline velocity over Case 6 which is an
effect to be expected if vertical diffusion has any importance, since
the shape-preserving vorticity will be transported downstream at a
slightly higher rate.
      As further discussion of the above statement, Case 11 has been
run where the fluid was considered as inviscid, although rotational.
The numerical fluid reacted in a manner such that the jet exit velocity
profile was completely shape preserved until the surface effects were
encountered (see Figure 7.12).  Considering the opposite extreme of
a hypothetical fluid where vertical diffusion completely dominates
radial transport, the same shape preserving nature would exist.  Case
11 also served to illustrate the computational stability of the differ-
encing technique used for cases where Rer = Rez -*• ».
                                  187

-------
         1 .0
                                                                   I    t
oo
oo
          .5
          .3
          .2
          .1
VELOCITY:

  •   COMPUTED



CONCENTRATION:

  o   COMPUTED

__ c   =5. 1/2-
      m                                   i /in
TYPE  2  BOUNDARY  CONDITION  V(R ,0) = (1 -R) '' 'U


	I   .... I
                                         5            10

                                             AXIAL  DISTANCE,  Z-
                                                                                   I   ''  r
                                                               50
               Figure 7.11.  Center-line Velocity  and Concentration Distribution for  Case 10
                            (Includes effect of  large vertical eddy diffusivity.)

-------
                    linn/
                                  VELOCITY
                                  PROFILE
                                     OUTFALL
                                     PORT
           Figure  7.12.
Shape Preserving of Velocity Profiles
Computed for  an  Inviscid, Rotational
Fluid (Ref. Case 2)
7.1.1.2  Spread  of  the Momentum Jet

      The rate of spread of the half radius,  ry2  is  illustrated in

Figure 7.13-A and compared to measurements in Figure  7.13-B.  The com-

puted rate of spread  is given by
where
            rl/2  =  C1Z
            C1  =  .0875.
                                      (7.20)
      For the several momentum jet computations carried out, the above

equation holds.   Table  7.4 compares some of the reported  values of (^.
                                189

-------
o
^.
 JP
               10
                                            SLOPE =  .0875
                                               I
    20         30         40
      AXIAL POSITION,?
                    50
                       60
 JP
S-
        • MEASUREMENTS BY TAYLOR  ET  AL. [97\
        —REICHARDT [77],
        o COMPUTED
                     8
       12
16
20
24
                                    B
28
        Figure  7.13,
        AXIAL POSITION,  =t
Computed Rate  of Spread of the Momentum
Jet Half-Radius, r,/2
32
                              190

-------
   TABLE 7.4.  COMPARISON  OF THE  SPREADING CONSTANT REPORTED BY
               VARIOUS  INVESTIGATORS
Investigator Comment
Albertson et al . [4]
Baines [8] Reynolds Number
7 x ID4
Baines [8] Reynolds Number
2.1 x 104
Reichardt [77]
Taylor et al. [97]
Corrsin and Uberoi [20]*
Keagy and Weller [49]*
Present numerical computation
Cl
.095
^ .085
^ .095
.0848
.0854
.0814
.0888
.0875
      *Based on momentum measurements.
      As Table 7.4 indicates, there is no universal  agreement  of  the
value for C, among the cited investigators.  These discrepancies  are
possibly due to measurement methods and/or flow condition dependence.
Again, Baines offered data which tends to confirm the role of  the
latter.  Hence, the computed value of .0875 seems to be a realistic
value in view of reported measurement, but cannot be compared  as  an
absolute because of experimental discrepancies.  Variations in the
half-radius may also be observed from Figure 7.12.
                            191

-------
7.1.1.3  Radial Distribution of Vertical Velocity, Concentrations
         and Vorticity for the Momentum Jet
      The radial distribution of vertical velocity for a momentum jet
is essentially Gaussian.  For instance the data obtained by Albertson
is adequate expressed by

            V = Vm e   I     ,                                  (7.21)
where
            K = 77.
Likewise, concentration distributions are adequately given by

                    -AK(i)
            C = Cm e    Z                                       (7.22)
where X is the eddy Schmidt number and equal  to .8.   The coefficient
K will vary from experiment to experiment similar to the variation in
data measured to establish the length of the  potential core.  As given
in Chapter 4, Baines found

                    -64.4(4) '
            V = Vm e      L                                     (7.23)
for a Reynolds number of 7 x 10  and
                          R J-82
                    -43.3(f)
            V = Vm e      L                                     (7.24)
for a Reynolds number of 2.1 x 104.  Gortler  [34] found K = 100.  For
a summary of additional experimental data on  the value of K one may
refer to Abraham [1],
                            192

-------
      One should bear in mind that the use of the Gaussian distribution
has no theoretical basis, but is a result of curve fitting.   Figures
7.14 through 7.18 all illustrate the vertical velocity profiles plotted
against different coordinates.  Figure 7.14 illustrates the distribu-
tion of computed velocity for comparison with the data of Albertson
for Case 2 which uses the Type 1 boundary condition.   Figure 7.15
relates this same type of information for Case 4 compared to the data
of Reichardt (cf. Schlichting).  Figures 7.14 and 7.15, along with
Figure 7.16 (Case 6) provide a comparison with the Gaussian distribu-
tion.  Computed information shows excellent agreement with the data and
essentially the same deviation from the Gaussian curve.  Unfortunately,
correct numerical modeling at the jet boundary is practically unob-
tainable because of numerical smearing and inability to correctly model
turbulence at the jet boundary.  These facets account for deviations
at the boundary and the fact that the computed velocity does not
attain zero at a finite radius.
      Figures 7.16 and 7.18 also bear out the similarity of the com-
puted velocity profiles whereby the computed velocity at elevation
I = 10 shows the only appreciable deviation from complete similarity.
Baines' data is also illustrated in Figure 7.15.  The various other
momentum jet case runs showed, upon spot check, that these curves are
typical of all  cases run with similar assumptions.
      Typical computed concentration profiles are shown in Figures  7.19
and 7.20.  Again, as in the case of velocity, striking similarity is
evidenced in the radial distributions at all elevations.  One noticable
fact is the deviation from a Gaussian distribution is more pronounced

                           193

-------
                            COMPUTED  AT  ELEVATIONS
                               •2=5.5
                               • =t =  10.5
                                   =  15.5
                                   =  20.5
                                 GAUSSIAN,      „
                                                     °D  ,  ••
 04
.08
12
. 16
.20
.24
                          R/Z
Figure 7.14.   Radial Distribution of Normalized Vertical
             Velocity for Case 2

-------
10
en
         1.0
o
_J
LLJ
           .6   —
           .4   —
           .2   —
                                                                 EXPERIMENTAL DATA OF REICHARDT  REPORTED
                                                                 IN SCHLICHTING [84]
                                                                 COMPUTED BY PRESENT TECHNIQUE
                                                                 •  2 = 20
                                                                 o  f = 30
                                                                 D  2 = 40
                                                                 CURVE FIT OF COMPUTED POINTS
                                                                 GAUSSIAN,  i
                                                      1 .0
                                                 RADIAL  POSITION,  R/R.
                                                                   1 .5
2.0
                             Figure 7.15.
                                     Normalized  Radial  Distribution of Axial Velocity,
                                     Momentum Jet  Case  4

-------
CTi
1.0

 .9


 .7

 .6


 .4

 .3

 .2
 .1
  0
                                                         COMPUTED AT ELEVATIONS:
                                                           o  2 = 10
                                                           •  2 = 20
                                                           A  I = 30
                                                           A  2 = 40
                                                           -I-  EQUATION 7.24
                                                           x  EQUATION 7.23
                                                                             BAINES [b]
                                                       I
                                                       x
                                                                                 •  fc
                                    02   .04   .06   .08  .10   .12   .14   .16  .18   .20   .22
                                                                 R/Z
                                     Figure  7.16.   Normalized  Radial  Distribution  of
                                                    Axial  Velocity Case 4

-------
(£>
                                         COMPUTED  AT  ELEVATIONS

                                           o   2 = 4

                                           •   2 = 10

                                               2 = 20
                                                       RADIAL POSITION, R
                                    Figure  7.17.   Radial  Distribution  of  Axial  Velocity
                                                  at Various  Elevations Case 4

-------
10
00
                    o
                    o
1.0



 .9



 .8



 .7







 .5



 .4



 .3



 .2



 .1



 0 *
                                                          COMPUTED AT  ELEVATIONS:

                                                            o   =t = 10

                                                                2 = 20
                                                                                 A*  O
                               02   .04   .06   .08  .10   .12  .14   .16   18   20   22   24

                                                          R/Z



                                  Figure 7.18.  Normalized Distribution of Axial
                                               Velocity Case 6

-------
.   4
COMPUTED AT ELEVATION:
  o  2=5
  •  Z = 10
  A  Z = 20
  T  E = 30
  A  2 = 3!5
                                           V   Z  =  25
                                                 GAUSSIAN:
                                                               _
                                                                 e
                          -68.1
                                                             m
    0   .02   .04   .06   .08
          .16
          n 17
          K, L
24   .26   .28   .30   .32
                       Figure 7.19.  Normalized Rad'al Concentration Distribution,
                                     Type 1  Boundary Condition Case 2

-------
ro
o
o
                         o
                         a:
                         o
                         o
 1.0



  .9



  .8



,£.7



  .6



  .5



  .4



  .3



  .2



  .1
                             0
                                                       COMPUTED AT ELEVATIONS:


                                                         o  2=11


                                                         •  Z = 21



                                                         A  Z = 31
                              0  .02  .04  .06  .08  .10 .12  .14  .16 .18  .20  .22  .24


                                                             R/Z



                               Figure 7.20.  Normalized Radial Concentration Distribution,

                                             Type 2 Boundary Condition Case 4

-------
for these profiles.  As in the case of velocity, concentration is
smeared to some extent across the jet boundary.
      Figure 7.21 illustrates the vorticity profiles at several  loca-
tions and Figure 7.22 compares the computed vorticity to the Gaussian
vorticity at elevations Z = 11, 31 and 41.  Note that the computed
vorticity maxima occur nearer the jet centerline than similar maxima
for the Gaussian velocity profile.  This fact is also revealed by the
experimental velocity data presented in the literature (cf.  Figure 7-15).
7.1.1.4  Distribution of Radial Velocity for the Momentum Jet
      A typical normalized distributional of radial velocity is  illus-
trated in Figure 7.23 (Case 6).  The solid line represents the Albertson
et al. theory and the dashed line represents an approximate envelope  of
their experimental data.  Albertson was unable to resolve clearly the
difference between the theory and his data.  Misinterpretation of the
collected data may have been the cause of such a large discrepancy for
it hardly seems logical that his theory (based largely on empirical
results) could be so far in error.  The radial velocities computed in
this study show good agreement with Albertson's empirical model, at
least over the range of positive velocities.  Again, Albertson's data
shows gross disagreement with computed and experimental results  for the
distributions of vertical  velocity.  The effect of this discrepancy
should be revealed most clearly along the jet centerline which is not
apparent from results (cf. Figure 7.6).
      Figure 7.23 also reveals the similarity of radial velocity.  It
is difficult to compare computed entrapment rates with the result

                             201

-------
    ,20
    .15
>-
I—

o
a:
o
    .05
                        COMPUTED AT ELEVATIONS:

                          •  ?  =  11

                          o  Z  =  21

                          o  2  =  31
                                6810

                                  RADIAL POSITION, R
                                  12
16
       Figure 7.21.
Radial Vorticity  Distribution for Momentum Jet
Type 2 Boundary Condition Case 4
                              202

-------
.15
                                       •  CASE  4
                                       o  CASE  1

                                       — GAUSSIAN:
                                             .  144  Vm R
 10  —
,05
    01234567
                          RADIAL  DISTANCE, R

 Figure 7.22.   Radial  Vorticity  Distribution for Momentum Jet at
               Z  =  15.   A  Comparison Between Type 1 and 2 Boundary
               Conditions, and the Gaussian Distribution.
                         203

-------
                              COMPUTED AT ELEVATIONS

                                         * 2=11
                                         ° Z=21
                                         • Z = 31
                                         o Z=41
R/Z
     .3
           f *
     .2
                                                        I   APPROXIMATE BAND
                                                           OF DATA REPORTED
                                                             BYALBERTSON
      -0.1
0.1
0.2
0.3
                                                                            0.4
                                         UZ
              Figure 7.23.  Normalized Radial  Velocity Distribution
                            for  Momentum Jet
                                 204

-------
given in the literature because we have typically assumed a  jet nozzle
extending into the fluid, whereas reported data is usually for wall
flush jets.  In Case 6, this distance is four port diameters.   Typical
experimental data may be correlated by
where C-| is an empirical constant, Q is the total vertical  flow at
elevation Z, and QQ is the jet flow.  Albertson gives C1 as .32.
      Equation (7.25) indicates a constant entrainment rate for momen-
tum jets, or

            §=C,Q0.                                          (7.26)

Figure 7.24 is a plot of the computed stream function vertical distri-
bution at the inflow-outflow boundary (i.e., ^(R^.ZjJfor Case 6.  By
definition the differential stream function along this vertical plume
is a measure of the entrained flow; that  is,
            A4- = - URAZ.                                        (7.27)
The total flow through the plane  Z = 4  is given  by
                4) -  
-------
      The straight  line  fit of the computations illustrated  in
Figure 7.24 is
               ,  Z)  =  .233  (Z-Zport) + 1.75.                    (7.29)
Then, based on the  intercept with 2 = 4,
               =  -33  ("*  •                                (7-30>
Hence, using QQ as  the  total of the jet effluent plus entrainment
from below the port gives a  lateral entrainment rate comparable  to  the
reported work where the fluid  issues from a wall-flush jet.
          12

          10

           8
     r+J
      2    6
      —*
      ^
           4
                j	I
I
                   10     20     30     40
                        AXIAL  DISTANCE,
                     50
60
            Figure 7.24.   Vertical  Distribution of Stream
                          Function V , Case 6
                            206

-------
7.1.1.5  Typical Contours and Three-Dimensional Plots for a Momentum
         Jet
      Additional information may be obtained by inspecting the level
lines and distribution surfaces of the stream function, concentration
and vorticity.  The centerline and surface streamlines are set at
¥ = 1.0.  This information is illustrated in Figures 7.25 through 7.31.
The three dimensional plots (Figures 7.28 through 7.31) have been arbi-
trarily scaled to fit a prescribed data box and are valuable for quali-
tative reasons alone.
7.1.2  Two Cases of Pure Buoyancy
      To check the computer program and computational techniques where
buoyancy is the sole driving force, t.wo cases were run where the out-
fall port or jet was replaced by a heat source (see Table 7.2).   In the
case of pure buoyancy, we are checking the same general features of the
plume as in the case of the momentum jet.  However, there is much less
information published.  Here we check the computed
      •  Centerline velocity and temperature,
      .  Radial distribution of axial velocity and temperature,  and
      •  Rate of plume spread
for a very weak and intermediate strength buoyant source.  Both cases
are well within the validity of the Boussinesq approximation.  Solution
restrictions are the same as those pointed out in Section 7.1.1.
7.1.2.1  Centerline Velocity and Temperature
      For a purely buoyant source  (and also for effluent cases where
F0 = 0) it has been established by Rouse et al. [8]] and Schmidt  [85]

                             207

-------
ro
o
c»
                         S-OD
                                          1O.OO              I5-OQ

                                            RRDIRL D1RECTIQN.  R/D
                                                     ZO.QO
         FIGURE 7.25
STRERMLINES F9R  CRSE  6  --  MOMENTUM JET

-------
ro
O
                         5-oo
                                            10.DO               15-DO

                                            RRDIRL  DIRECTION. R/D
ZD.OQ
        FIGURE 7.26.  ISQFYCNRL5   FQR  CRSE 6 -- MOrlENTUM JET

-------
                 5-DO
                                   10.00               15-DO
                                   RRDIRL DIRECTION.  R/D
za.aa
FIGURE  7.27.   VQRTICITY   LEVEL LINES FOR  CRSE  6 -- MOMENTUM JET

-------
ro
FIGURE 7.28
                         ILLU5TRRTI0N  9F  STRERM FUNCTIQN -- P3I.    CflSE NO. 6

-------
ro
r>o
        FIGURE 7.29.  3D ILLUSTRRTIQN  0F  STRERM FUNCTIQN -- PSI.  CASE NO.  6

                     CRSE    - DEEP WRTER  MOMENTUM JET -  VfR.OJ =  (1-R)••(1f\ 0 )

-------
IN3
CO
FIGURE 7.30.   30 ILLU5TRRTION 0F  BUOYANCY DISTRIBUTION -
                                                                          CRSE NQ.6

-------
ro
        FIGURE 7.31.  3D ILLUSTRRTI0N 0F  FLUID  VQRTICTY  -  QMEGR.     CRSE  NQ.6

-------
that


            V'V?                                              / "7 O 1 \
             m —                                                (7.31)

and


            ATm - Z"                                            (7.32)



In the case of an effluent with little initial momentum and strong


buoyancy, Abraham [1] gives
            Vm = 4.4(FQZ)-                                      (7.33)
           ATm = 9'5 Fo    Z"                                   (7-34)


based on Rouse's data.



      Figure 7.32 illustrates the center! ine velocity and temperature


for Case 13.  In this case, the source is very weak and gives a maxi-


mum fluid temperature rise of only .95 °C.  The maximum velocity is  a


little above .09 ft/sec occurring at an elevation of about seven


source diameters above the source.  The flow apparently does not


become established until an elevation of 15 to 20 diameters has been


reached.  Above this approximate region the computed center! ine


velocity shows decay very closely approximating the -1/3 law given by


Equation 7.31.  Velocities computed above Z = 50 (surface at Z = 64)


show influence of the free surface.

      Temperature decay, on the other hand, begins to follow Equation
                                                                      s

(7.32) at approximately Z = 10 and computed values are extremely close


to a - 5/3 slope.  However, there is no apparent surface effect on


temperature, whereas Case 14 (Figure 7.33) reveals noticeable change



                             215

-------
1.0
 .5
 .1
.05
.01
                                SLOPE = -5/3
           SLOPE = -1/3
CENTERLINE VALUES OF:

  •VELOCITY, Vm

  0 TEMPERATURE EXCESS, AT

I     I   .  I   ... I I	,
                                       m
                     5       10

                  AXIAL DISTANCE,
                                   50
Figure 7.32  Computed Center!ine Velocity and Temperature
           Excess for Case 13.  Pure Buoyancy, F  =  0.
                  216

-------
1.0
 .5
 .1
.05
.01
    0
               T	T
                                   T	1	'  |  I  I I I.

                                     *»*•
                             SLOPE = -1/3
                                 v
                            \
                              \  SLOPE = -5/3
                               \
CENTERLINE VALUES
 •VELOCITY, V
              m
                                    \
               TEMPERATURE EXCESS,  AT
                                       m
                                             \
                .   .  i  .... I
         5     10
       AXIAL DISTANCE,-Z
50      100
Figure 7.33.  Computed Centerline Velocity and Temperature
            Excess for Case 14.  Pure Buoyancy, FQ = 0.
                 217

-------
in slope near the surface.   It is felt that continued iteration would



have shown somewhat larger deviation from the -5/3 slope near the sur-



face in both of these cases.



      Figure 7.33 (Case 14) illustrates similar results for a situation



where the fluid directly in contact with the heat source was maintained



at a 25 °C temperature rise.  Under these conditions the maximum



velocity was about 0.8 ft/sec occurring at approximately 8 diameters



above the source.  The shape of the center!ine velocity distribution is



very nearly the same as in Case 13 and achieves the -1/3 slope at



approximately 20 diameters above the source.  The temperature distribu-



tion, however, shows some differences in that the -5/3 decay is not



attained until about 20 diameters and, as mentioned previously, there



is demonstrated a marked surface effect.  Results for both of these



cases could be improved somewhat by continued iteration in the vicinity



of the surface.  Convergence was slow in this region for both runs, but



temperature changes indicated an increased surface effect.  Another



aspect is that vertical turbulence has been essentially neglected, a



poor assumption in the surface effects region.  A realistic approxima-



tion of vertical turbulence here would also tend to increase the sur-



face temperature.




7-1.2.2  Spread of the Pure Buoyant Plume




      The rate of spread of the half radius, r, ,„, for pure  buoyancy  is



demonstrated in Figure 7.34 for Case 13.  Case 14 was found  to  be



essentially identical to Case 13. Based on Rouse's data, Abraham ascer-



tained that the half radius is approximated by





                             218

-------
o
<
00
O
=     20  -
x
<
        0
       Figure 7.34.
                  '1/2
                   D
Computed Rate of Spread of Half-Radius,
Pure Buoyancy, Case 14  (D=2rQ)
              1/2
              D
  L69
   K
Z = .0866 Z
(7.35)
   where K = 92.
         The data obtained by Rouse revealed K = 96, at least for the
   selected curve fit.  Abraham's theory and experiments yield K = 92,
   and according to him, no major discrepancy in results is obtained in
   either case.  Figure 7.34 reveals a computed spread of approximately
                            219

-------
.092 I.  Not only is this rate of spread different from the rate based
on a Gaussian profile, but the rate is greater than in the case of
pure momentum (r1/2/D ± was computed).  Gaussian profiles show the
opposite to be true.  The reason for these discrepancies has not been
completely resolved.
      Barring difficulties with the computer code, which has been
checked, the discrepancy may be caused by incorrect modeling of the
turbulence in the presence of buoyancy.  It is also possible that the
data obtained from flame sources in air may be significantly influenced
by effects not accountable through the Boussinesq approximations.  That
is, the Boussinesq approximation would not be valid for modeling plumes
over diffusion flame plumes because of the large density variations
compared to the reference density, even though temperature will decay
quite rapidly.  In both Cases 13 and 14, the density variations may
influence the rate of spread and explain the present discrepancy.
Additional data for a low Froude number flow case is presented in
Section 7.1.3.2.
7.1.2.3  Radial Distribution of Vertical Velocity, Temperature and
         Vorticity for Pure Buoyancy
      The data obtained by Rouse and Schmidt demonstrate  that the
normal distribution curve again fits the buoyant plume radial profiles
quite well.
      In this case, data obtained by Rouse gives
                           220

-------
                      ^2
                    VI  \
                   ""  »"T~'
            V = V P                                             / •? i/- ^
                vm                                              (7.36)
where K = 96, and
            AT =  Tme                                           (7.37)
where
            A = .74.
However, the Gaussian curves used for comparisons here will  be based  on
Abraham's value of K = 92 which yields xK = 68.1.  As in the case of
the momentum jet, these distributions have no theoretical  basis,  but
are a result of curve fitting.
      Radial distributions for Case 13 arc illustrated in Figures 7.35,
7.36 and 7.37, for various elevations.  Computed results show excellent
similarity at all  elevations except near the source (Figures 7.35 and
7.36).
      Figure 7.37 shows the velocity profiles for Case 13 as computed.
Figure 7.38 again shows excellent similarity at all elevations except
near the source for Case 14.
      A normalized temperature profile is illustrated in Figure 7.39
and vorticity at various elevations is plotted in Figure 7.40.  One
notable feature revealed by Figure 7.39 is that the temperature distri-
bution is in considerably closer agreement with the Gaussian curve in
the case of a momentum jet (cf.  Figure 7J
                            221

-------
ro
IN}
ro
              O
              o
1.2


1.0


 .8


 .6


 .4


 .2


  0
                        -A.
                       0
COMPUTED AT ELEVATIONS:

  •  2=20

  *2=30        GAUSSIAN:
                      R 2
  A  -2 = 40    V     -92(f)

             V"=e   Z  ;
              m

             ABRAHAM [1]
                5          1.0         1.5

                   RADIAL POSITION,  R/R1/2
                         2.0
                    Figure 7.35.  Normalized Distribution of Computed Axial Velocity.
                                Pure Buoyancy, Case 13

-------
r\>
ro
CO
                      1 .0
                   E   .8  ~
.6  —
                  
-------
ro
ro
CJ

CD
         .10
          08
06
          .04
          02
—
_
••°°o
_ • o
• 0
LODD8°D8
AA A A A A A A 8
_ Q ft A
A
• DA
~ * o a
• _ a
__ • o
a
• o
~ • °

o

•
HEAT SOURCE •
•


COMPUTED AT ELEVATIONS:
• ? = 4
o Z = 10

a 2 - 20
A I = 30




A
A

A
D A
a A

D
o a
0 D
0
o
o
• I • 1
1 • • • 1
—
^_

	 	

_


««•

—

, — _



A



a —

o

                                                           4

                                                           R
                                                                                                    8
                     Figure 7.37.  Radial Distribution of Axial  Velocity  in Pure Buoyancy, Case 13

-------
ro
IN3
en
o
o
               1 .0
                .8  -
                .6  -
                 4  -
            X
            •=c
                                I    I    I    T    i    \    I    I    I    I     I    I
                                                  COMPUTED AT  ELEVATIONS:
t =  10
Z =  20
Z =  30
* =  40
GAUSSIAN:
y =  v e-92(R/Z)2
     m
   ABRAHAM [1]
                                                          I	I
                                                        • ma

                                                      J	I
                                                                                 o-
              J	I
                           .04      .08      .12      .16      .20
                                         RADIAL POSITION,  R/Z
                                                         .24
                .28
                       Figure 7.38.   Normalized Radial Distribution of Axial Velocity.
                                    Stronger Source, Pure Buoyancy, Case 14.

-------
ro
ro
CTi
X* w

! AT =e L '
\S \° ABRAHAM [1]
, 1 , I>H , 1 , "
.12 .16 .20 .24 .2
                                                              R/Z
                      Figure 7.39.  Normalized  Distribution  of  Computed  Radial  Temperature Excess.

                                    Pure  Buoyancy,  Case  14.

-------


.20




.15

G
>-"
o
£• in
o •"
>

.05
0 i
0
1 ' 1 ' 1 '
*\
(?>
o
~ o —
o COMPUTED AT ELEVATIONS:
o
o 2 =11
0 -, 2 01
0 * Z = 21
* 2 =31
o
0
0
^»_
w®
0 • • _
• «
• o
9
O A •
X*°*
• / »
-• a o *
° » A
-)• A
J A 0 • A
o • A
• A ° ^ A
AA o • A
? , , , 1° «. o' ,' • 1
4 8 12 1(
R
Figure 7.40.
Radial Distribution of Vorticity.
Pure Buoyancy, Case 14.
                 227

-------
7.1.2.4  Radial Velocity and Entrainment for Pure Buoyancy
      The normalized distribution of radial  velocity for Case 14 is
given in Figure 7.41.  As opposed to momentum jet results (Figure 7.23),
similarity of the radial flow is not apparent using the coordinates
R/Z and UZ.  Also note that, compared to the corresponding momentum
jet data, the magnitude of negative radial  flow is somewhat larger,
indicating an increased radial  entrainment  rats.  Although it has not
been plotted, the radial flow below about six source diameters is
negative over the entire flow field.
      From similarity theory it has been established (cf. Abraham)
that
            §  =  W-                                       (7-3S>
By Equation (7.31)
                      2/3
            —*•  =  C^
Then integrating Equation (7.39) yields
            Q  =  C3 Z5/3.                                      (7.40)
The values C-j,  C2 and C3 are appropriate constants; magnitudes are
unimportant since we are interested only in how Q varies with Z.
Figure 7.42 illustrates the value of ¥(Roo,Z) as a function of Z, and
since i^R^.Z) is directly proportional to the entrainment Q, this plot
reveals the variation of Q with Z for pure buoyancy.  The computed data
in Figure 7.42 is obviously represented by a functional relationship
more complicated than Equation  (7.40).  At lower elevations  (Z^ 6 to  15)

                            228

-------
   .6
   .5
Ivl
•^
ct
   .4
   .3
   .2
               T	1
         COMPUTED AT  ELEVATION:
        -.8
-.6
                           -.4
                  -.2
.2
                               U2
Figure 7.41.  Normalized  Radial  Velocity Distributions
               for Pure  Buoyant Plume, Case 14
                      229

-------
       TOO
   r+j
        50
         10
                                            I    I
                                       SLOPE = 5/3
                               SLOPE  = 1
                                                I   I  I  I
            10
                  50
AXIAL  DISTANCE,  Z
100
       Figure 7.42.   Vertical Distribution of Stream Function
                     at  R  .  Pure Buoyancy, Case 14
            Q -\, Z.                                              (7.41)
Thus, in this range  the  plume entrains ambient fluid proportional  to a
momentum jet.  The  5/3  slope is  never indicated by the data, but
Figure 7.42 shows that  the  entrainment data would apparently approach
a 5/3 slope  asymtotically for sufficient depth.
      Figures 7.43  through  7.45  illustrate streamlines, isotherms and
vorticity level lines for Case 14.  As in all cases reported, the cen-
terline  value of the stream function is  1.0.  Three-dimensional illus-
trations of the same information is displayed in Figures 7.46 - 7.48.
                            230

-------
ro
CO
                                                    in.an                 IS.QQ
                                                       RRD1RL DIRECTION.  R/D
                                                                                             ZO.DD
                                  Figure 7.43.   Streamlines for Case  14, Pure Buoyancy

-------
ro
co
       rvi
                                  .005
                                                    Isotherms not labeled (from heat source)

                                                        No.      AT/AT0
                                                                  .75
                                                                  , 50
                                                                  .25
                                                                  .20
                                                                  .15
                                5-aa
la.aa                   iS-aa
RflDIBL  DIRECTIQN. R/D
                                                                                                     zo.oa
                            Figure 7.44.  Isotherms for Case 14, Pure Buoyancy,  AT/ATQ

-------
r-o
CA>
co
         PJ
                                                                                JS-00

                                                          RRDIRL DIRECTION. R/0
                                                                                                      za.oa
                                Figure 7.45.  Vorticity Level  Lines for Case 14, Pure  Buoyancy

-------
ro
CJ
                           Figure 7.46.  3D Illustration of Stream Function - PSI, Case No. 14.

-------
ro
CO
en
                   Figure 7.47.  3D  Illustration of Tenrperature Field - AT, Case No. 14.

-------
Figure 7.48.  3D Illustration of Fluid Vorticity - Omega,   Case 14,  Pure Buoyancy.
              Temperature Difference of Source Maintained  at 25 oc.

-------
7.1.3  Mixed Flow - Forced PI
umes
      In Sections 7-1.1 and 7.1.2 we have checked in some detail  the
computed flow characteristics at both ends of the dynamic spectrum--
pure momentum and pure buoyant flows.  This section deals with flows
having dynamic characteristics of both which are appropriately classi-
fied as "forced plumes' as coined by Morton [58].  Cases used to com-
pare with similarity solutions and experimental data are summarized in
Table 7.3.  To this end, a variety of effluent boundary conditions have
been investigated.
      The cases here are too numerous to treat each in full detail so
that only the general characteristics of
            «  Centerline velocity and temperature, and
            •  Rate of spread and entrainment
will be illustrated, along with selected contour and three-dimensional
plots.  The similarity solution discussed in Chapter 4 will be used for
comparison.
7.1.3.1  Centerline Velocity and Temperature for Forced Plumes
      In Chapter 4 the following similarity solution was given for
vertical forced plumes:

                                                                (7'42)
                  e
and
                                             I-  1/3
                                                                 (7.43)
                         237

-------
the variable
                    V_Z
             *
            E   =   -—    ,                                   (7.44)
                      ~
and Z  is based on Equation (4.19).
      The above equations, except for (7.44),  do not reveal  variations
in the values of K and \.   These values and their effect on  the govern-
ing equations have been discussed in Chapter 4 and are summarized in
Table 4.1.  The largest error in velocity is seen to be introduced by
1//K (4.8% deviation from the mean value) but  is absorbed in E*.
      Equations (7.42) and (7.43) reveal  the use of simple fractions
which simplify the equations and are very close to the mean  values
given in Table 4.1.  Since these variations are small, and in view of
experimental data scatter,  it does not seem justified to use more
complicated relationships for K and x as did Abraham, at least for the
vertical plume.  At any rate, the subject equations yield results that
are in good agreement with Abraham's computations and yield  excellent
agreement with Fan's [27] data concerning the maximum height of rise
where stratification is of concern (cf. Baumgartner and Trent [12] )•
Thus Equations (7.42) and (7.43) will be used to compare with the finite
difference results.
      Cases 15, 16 and 17 compare the effect of three different methods
of computing the radial component of eddy momentum diffusiviity, e .
In all cases er is computed from
                  -  -0256 VmaxRl/2 ' -0256
                             238

-------
however, different methods for computing FR are used.  A detailed dis-
cussion for this computation is given in Section 7.2.
      Case 15:
            FR = constant = 1.178
      Case 16:
            FR = .180 VmZ.
            where Vm is the currently ...aiculated value of center!ine
            velocity at elevation Z.
      Case 17:
            FR = VmRl/2
            with running calculation of both V  and R-i/2>
            all other conditions for these cases remain fixed.
      Figure 7.49 illustrates the centerline velocity, V , and  buoyancy,
A, , for these three cases.  The significant feature of results shown
in this figure is that using a constant value for er (Case 15)  gives
results with appreciable error in buoyancy (or temperature).   The use
of a pre-calculated half-radius (mixing length) based on a Gaussian
velocity distribution gives somewhat better results (Case 16).   The
similarity solution is found to give quite accurate results for I > 15-
20 and Case 17 shows buoyancy results in excellent agreement with the
similarity solution, although the velocity distribution shows a sizable
difference.  The large discrepancies in both velocity and buoyancy at
lower elevations (Z ^ 10) are expected since similarity solutions are
not valid in this range.
                            239

-------
         1 .0
ro
-P.
o
.5



.4




.3






.2
          .1
                   CASE    15    16
                   '1m
                   m
A




A
          17

          a
                  SIMILARITY  SOLUTION
                             I
                                              ... I
                                        5          10


                                           AXIAL DISTANCE, I
                                                                   50
                                                              100
                  Figure 7.49.  Centerline Velocity and Buoyancy for Cases 15, 16 and 17

-------
      These three cases also represent progressively more difficult
computational problems owing to the non-linearity of the eddy
diffusivity.
      Case 15, where a constant value of e  is used, caused no compu-
tational difficulties and is of course the fastest with regard to com-
puter time.  This problem is quite similar to the laminar flow plume
problem, but er can be several orders of magnitude larger than the
counterpart molecular momentum diffusivity.  Case 17, where V  and
RI12 are computed iteratively is the most difficult and requires the
most computer time.  The computational difficulty stems from the fact
that velocity profiles at the initiation of the FR computation cannot
be too far in error or a numerical instability will result.  In addi-
tion, the convergence rate is slowed by continuous updating of FR.
      Returning to the discussion of momentum jets (Section 7.1.1),
only Cases 2 and 3 used e  = constant, all other cases used FR calcu-
lated as in Case 17.  However, in the case of a momentum jet, FR is
indeed constant so that any of the three methods for computing er
should yield essentially identical results (see Section 7.2).  Only in
the case where buoyancy is present will variations in FR become appar-
ent, and for this reason, demonstration of results was deferred to
cases dealing with mixed flow.
      Case 18 is identical to Case 17, except the vertical grid spacing
has been doubled giving an overall depth of 82.5 port diameters.
Figure 7.50 illustrates centerline buoyancy for the case compared with
the results of Case 17 along with the similarity solution.  Slightly
higher values for buoyancy were calculated in Case 18 compared to

                              241

-------
Case 17,  an  effect of doubling the vertical  grid  spacing,
  o:
  <
  D_
 o
 rj
 CO
      1.0
       .5
       .1
       .05
        01
 •  CASE  18
 •  CASE  17
— SIMILARITY
   SOLUTION
                                10
                       AXIAL  DISTANCE, Z
                              100
            Figure 7.50.  Centerline Buoyancy  Distribution
                         for Cases 17 and  18
      Figure  7.51  illustrates the centerline  velocity and buoyancy
distributions for  Cases 17, 19 and 20 where the  densimetric Froude
numbers are 52,  35 and 106, respectively.   All other variables are
fixed for these  cases.  Case 21 is identical  to  Case 17  except the
vertical  eddy momentum diffusivity, EZ, was assumed to have the form
            ez  =  ezoe
                    -A2(ZS-Z)2
                                      (7.46)
                           242

-------
1.0
   10
20                   30
    AXIAL DISTANCE,  2
  Figure  7.51.   Centerline Velocity and  Buoyancy for Cases 17, 19 and 20
                               243

-------
where A is a constant, Z  is the surface elevation and e   is a
                                                         o
reference eddy diffusivity.  The objective of this case was to illus-
trate the effect, on the plume flow, of substantial eddy diffusion
confined near the surface.  Although the exact values of e   and A are
                                                          zo
of little importance to this end, they have values: e   = 1 and A = .2.
                                                     zo
The only significant effects caused by this treatment of e  are in the
radial spread and vertical diffusion of vorticity and radial  velocity
at the surface.  In the case of negligible vertical momentum diffusion,
vorticity tends to accummulate in the surface nodes and the mass tends
to spread frictionlessly within these surface nodes at high velocities.
The presence of significant vertical eddy transport diffuses the
vorticity and velocity further downward into the ambient fluid.
Figure 7.52 shows a vorticity ridge near the surface for an essentially
frictionless flow (Case 17), whereas Figure 7.53 (Case 21) illustrates
considerable mitigation of this ridge through vertical diffusion.
      Cases 22 and 23 differ from the preceding mixed flow computation
in that a Type 2 boundary condition is used with N equal to 7 (refer to
Equation 7.17).  Various other differences are noted from Table 7.3
(e.g. the Froude number and finite-difference grid).  These two cases
are identical to one another except the eddy Prandtl number in Case 22
is .714 whereas in Case 23, .80 is used.  These computations were per-
formed primarily to determine the effect of the Prandtl number on the
rate of spread (Section 7.1.3.2).  However, an appreciable effect is
also noted on the centerline buoyancy distribution  (Figure 7.54),
whereas little difference was found in centerline velocity for the two
                             244

-------
ro
-P.
en
                          FIGURE  7.52.   3D ILLUSTRflTION 0F V8RTICITY	OMEGfl


                                        CASE 17	BUGYflNT  PLUME WITH RUNNING CflLCULflTION OF HRLF  RROI'JS

-------
ro
-pa
en
                        FIGURE  7.53.   30 ILLUSTRRTION  QF  VORTICITY --- GMEGfl        - CflSE NO. 21

-------
        1 .0
ro
          .1
          .04
                                 g
                                 CASE 22,  PR   =  .714
                             BUOYANCY:
  D   CASE 23, PRr =  .800
•      SIMILARITY SOLUTION
VELOCITY:
  •   CASE 22
  •   CASE 23
—  SIMILARITY SOLUTION
                                                  ,  ,  1
                       Figure 7.54.
                         10
                  AXIAL  DISTANCE, Z
        Center-line Velocity and Buoyancy for Cases 22 and 23
                                                                                           100

-------
cases.  Since the power law effluent velocity profile is used, with N
equal to 7, the maximum centerline velocity is approximately 1.2.
This boundary condition indicates much better agreement with the simi-
larity solution for downstream velocity than was obtained using the
Gaussian profile (Type 1  boundary condition) in preceding mixed flow
cases.
      Figure 7.55 shows centerline distributions for Case 24 which is
identical to Case 23 except the densimetric Froude number is 1.0 as
opposed to 46.  Unfortunately, the eddy Prandtl number for this case
was not reset to .714 (.8 was used).  This error was not discovered
until the contents of the restart tape were destroyed; hence, for
economic reasons the case was not rerun (cases for very low Fn are
slow in converging).  However, the slope of the buoyancy curve is
essentially identical to  the similarity solution and, borrowing the
trends of Cases 22 and 23, the buoyancy curves would nearly coincide if
PRr equal to .714 had been used.  Also, from Figure 7.54 we would
expect no appreciable change in the velocity distribution of Figure
7.55.
      Figure 7.55 illustrates that for low F , the velocity initially
increases due to the large relative buoyancy, reaching a maximum at
about 5 diameters downstream.  The velocity distribution then tends to
a - 1/3 slope as in the case of pure buoyancy.  Likewise, the buoyancy
distribution tends to a - 5/3 slope as in purely buoyant plumes.
      Figure 7.56 illustrates centerline distributions for F  = 1000
(Case 25) compared to computed results for a momentum jet (Case 4).

                            248

-------
 10
  1.0
  .1
  .001
                                                    SIMILARITY SOLUTION
                                                    COMPUTED  (PRr • .80)
            VELOCITY. V
                            5         10
                             AXIAL DISTANCE. Z
                                                             50
                                                                      100
Figure  7.55.  Comparison  Between Computed Results and Similarity
               Solution for  FQ = 1.0
                           249

-------
ro
en
           1 .0
             .5   -
             . 3
             .2
            . 1
                                                           10

                                                  AXIAL DISTANCE,  I
50
100
                         Figure 7.56.   Comparison Between Computed  Center!ine Distributions
                                       of Velocity and Buoyancy for FQ = 1000 and FQ -> ~

-------
7.1.3.2  Rate of Spread and Entrainment
      Results from momentum jet computation revealed that the jet half
radius  spreads  according  to
            r]/2 = .0875 z.                                     (7.47)
and pure buoyant plume calculations yielded
            r]/2 ^ .092 z.                                      (7.48)
Although these results showed a reverse trend from experimental
observation, absolute values are not in large disagreement with experi-
ment.  Figure 7.57 illustrates the rate of half-radius spread for
FQ = 0, 1, 46 and ».  The effect of different eddy Prandtl numbers for
FQ = 46 is revealed by Figure 7.58 (Cases 22 and 23).  As pointed out
earlier, the case for FQ = 1 was inadvertently run using PRr = .8 and
Figure 7.57 shows that this case has the same spread rate as the case
where FQ -»• °°.  Thus, the fact that one case is dominated by initial
inertia and the other by buoyancy seemed to have no effect on the
half-radius spread rate.  This being a fact of the computational tech-
nique then explains why the plume has a larger computed spread rate
where PRr = .714 as opposed to PRr = .8.  It is expected that had
PR  = .714 been used in the FQ = 1 computation, the half-radius curve
would have coincided with the curve for F  = 0.  Case 22 where
F  = 46 shows that the half-radius begins to spread as a momentum
jet (I <_ 10-12), passes through a transition and then spreads at the
same rate as a purely buoyant plume, at far downstream points.  Case
23 begins to spread as a momentum jet, then passes through a
                            251

-------
en
ro
      rtj
      eC
      »—i
      X
          60
          50
          40
          30
          20
          10
                         ~\      I
(CASE  22)  -^
                                                               60
                         50
                         40
<"   30
i—i
Q
                     x   20
                     
-------
transition to a wider spread, and at far downstream points,  again
spreads like a momentum jet (but wider).
      Figure 7.59 shows the variation of ^(R^.Z),  a measure  of entrain-
ment, with elevation.  Again we cannot expect good correspondence  with
wall-flush jets at lower elevation since for the cases  illustrated the
outfall port has finite height above the bottom.  At higher  elevations
we note that for FQ = 1000 a slope of 1  is attained which is appropriate
for momentum jets.  The case for F  = 1  has obtained a  slope of
approximately 1.4 and is increasing.  Had the solution  been  carried to
higher elevations, the experimental value of 5/3 would  perhaps be
attained.  For F  = 46 we find intermediate values of \HR »2) with the
                o                                        °°
slope tending toward that of the case for F  = 1.   Again  the slope is
increasing and would perhaps attain the value of 5/3 as in pure
buoyancy, at increased axial distance.
      Figures 7.60 through 7.62 illustrate streamlines, isotherms  and
level lines of vorticity, respectively,  for Case 22. Figures 7.63
through 7.65 show this same information in three-dimensional plots.
7.2  Transport Coefficients
      In obtaining the results presented thus far, we have made use of
certain transport coefficient models which describe the required compo-
nents of radial and vertical turbulent diffusion.   This thesis, in the
main, is not a study of modeling these coefficients but through neces-
sity one must utilize reasonable methods for modeling these  quantities
if reliable results are to be obtained.  For the momentum jet issuing
to a semi-infinite medium, the important transport coefficient models
                             253

-------
    TOO
O
O
c
3
U_

E
fO
0)

-M
OO
     50   -
     10   —
5  -
                            AXIAL  DISTANCE,
           Figure 7.59.  Entrainment Trends in Mixed Flows
                          254

-------
IV)
on
cn
                                              10-00              15.00
                                        RRDIRL  DIRECTION,  R/0
20-00
           FIGURE 7.60.  STREAMLINES FOR  CRSE 22 -  MIXED FL.0W.  FO  = 46

-------
ro
01
O1
       ISOTHERMS  NOT LABELED:

        NO.       T, °C
                                                            30
                                                            28
                                                            26
                                                            25
                                                            24
         °o
        10.00
RRDIRL  DIRECTION.  R/D
           FIGURE  7.61
                                          15.00


ISOTHERMS  FOR CRSE 22 -  MIXED FLOW,  FO =  46
20.00

-------
I\J
in
LEVEL LINES NOT LABELED:

  NO.        ft
                                                                 .5
                                                                 .25
                                                                 .15
s.t
                                               10.00
                                       RflDIRL DIRECTIQN. R/D
                                                                  15-00
                                20.00
          FIGURE  7.62.  VORTICITY  LEVEL LINES  FOR CflSE  22 - MIXED  FLOW, FO  - 46

-------
en
CO
                                                                                 \
         FIGURE 7.63.  3D  ILLUSTRRTI ON  QF STRERM FUNCTION -- PSI
                      CRSE 22   -  DEEP  URTER BUOYRNT JET

-------
r>o
en
          FIGURE 7.64-  3D  IL.LUSTRRT I ON QF TEMPERRTURE  FIELD  --  T
                       CRSE  22 -  DEEP WRTER BUQYRNT  JET

-------
ro
en
o
R --
         FIGURE 7.65-  3D  ILLUSTRRTI0N QF FL.UID VQRTICTY  -  QMECfl
                      CRSE 22  -  DEEP KRTER BUQYRNT JET

-------
turn out to be trivial since they are constant.  However, where
buoyancy plays a role and the buoyant surface spread in stratified
media is of concern these models are quite complicated and in certain
instances (surface spread) the theoretical and experimental efforts
are sadly lacking.
      In this work it is necessary to model the momentum diffusion
coefficients for the radial and vertical directions, er and e , along
with the corresponding Prandtl (or Schmidt) numbers, PR  and PR .
Turbulence contributions may be considered to fall into the following
two categories:
      1.  that generated by the effluent stream, and
      2.  the ambient contribution which has origin from
            •  wind stress and wave action,
            •  shear flow at solid boundaries, and
            •  contributions depending on the local history
               and/or convection across system boundaries.
In general, the effluent generated turbulence will dominate the
ambient contribution within the plume except in the surface zone
where plume velocities may be low and wind and wave action under a
high sea state dominate the effluent induced effects.  However, in
the circulating portion of the flow field, ambient contributions will
dominate.
      The turbulence models used in the present work are  based on
Prandtl's second hypothesis which is appropriately modified to include
the influence of stratification.  Experience has  found that Prandtl's
                             261

-------
hypothesis may be applied with good results where mean velocity grad-
ients have reasonable magnitude and a mixing length may be easily
defined, but breaks down entirely, at least computationally,  where
velocity gradients are very small, or confused, and the mixing length
has dubious interpretation (e.g., the circulating flow).  Prandtl's
hypothesis, as stated by Schlichting [84], is
                      8V        . /         \ 3V
            T  =  pe  —  -  pC-, b(V   -V • ) —
where T is fluid stress, C,  is an empirical  constant, and b is the
width of the mixing zone.  The eddy diffusivity for momentum er is
then
            er  =  ClV                              <7-
where £ is the mixing length of an axisymmetric plume and assumed to
be the width of the half-radius in established flow.  An equivalent
relationship may be written for e , the vertical  component, in the
zone of surface spread.  In the mainstream of the plume, the usual
case is that-only one or the other of the transport coefficients will
have a significant effect on the flow dynamics.  For instance, in the
vertical rise, er is of utmost importance ,whereas e  may be neglected
as a practical matter.  However, EZ is included in the computations,
and may in fact be important near the surface where vertical velocity
may be small.  In the lateral spread, the opposite is true where e
has relatively small influence.  The value v_,  in the zone of plume
                                            max                r
rise is easily defined as the centerline velocity,  u    in the lateral
                                                     max
spread will occur at the surface for a buoyant flow in homogeneous
                             262

-------
surroundings.  In both cases the maximum velocity has sufficient magni
tude compared to velocities outside the plume so that v   »v .   and
                                        r              max   mm
                      vmax
umax>>umin-  Hence>
and
where £  is an as yet undefined vertical  mixing length in the vertical
direction.  Note, that Equation (7.51) includes no compensation for
stratified flow.
      Equations (7.50) and (7.51) are adequate for modeling the tur-
bulence inside the plume and are relatively convenient to use, but
only because we have prior knowledge of the plume geometry.  Outside
the plume, in the region of flow induced circulation, these expres-
sions are useless because we have no adequate criterion for mixing
lengths and, in fact, velocity gradients may have nothing to do with
the primary contribution to the field of turbulence.   Fortunately, for
the problem at hand, turbulence in the circulating field is of nominal
importance, and except for the fact that some degree  of viscosity in
this region helps to speed the numerical  computation, we could assume
the fluid as inviscid.
      It is recognized that Prandtl's second hypothesis has limited
application in the numerical computation of circulating and recircu-
lating fluid flow.  Prandtl recognized the shortcomings of this
hypothesis in that it could be applied with confidence only to rea-
sonably simple, steady-state flows.  Various other investigators also

                           263

-------
recognized that a more fundamental  approach needed to be employed.
Such an approach needed to consider such various aspects as
      •  convection,
      •  diffusion,
      ,  creation, and
      •  dissipation
of the turbulence which could be related in some fashion or another to
mean flow quantities.   Earlier models were based on the transport of
turbulent energy.  However, these models still  depended on the defini-
tion of a mixing length to relate the dissipation or decay.  Chou [18,
19] sought to overcome this difficulty by introducing a second trans-
port equation for decay scale.  Rotta [79,80] developed these ideas
even further and set down the transport equations for the complete
Reynolds stress tensor.
      Based on the pioneering work of Rotta, Spalding [92] and his
colleagues at the Imperial College in London, have had considerable
success in applying these ideas to generalized numerical computation
in recirculating flow fields.  Spa!ding's model for computing turbu-
lence quantities involves transport equations (cf. Reference [93])for
      •  k, the kinetic energy of turbulent motion,
      .  W, which may be considered as the average value of the
         fluctuations of the fluid vorticity, and
      •  g, the average value of the square of the fluctuating
         component of the mass fraction of injected fluid.
                           264

-------
      Spalding defines a length scale as:

            * = (k/W)1/2,                                       (7.52)
hence,
            e = C3pk1/2£.                                       (7.53)
      Thus, in addition to equations for the stream function, vorticity,
buoyancy, and/or other required constituents, transport equations of
the following types are also required:
             .. 3k.  ,     3k    ]_ 9_ /e   3kx
            pu 3X    pv 3r  " r ar ^k  3r;
                     I/?   II 2       k3/2
            -  C4  Pk1/2£ (^)  - C5 pf   •                      (7.54)

where the C's are constants defined by Spalding.  Similar equations
are required for W and g.  As testimony to these and similar methods
the reader is referred to the following work carried out at the
Imperial College:  Patankar and Spalding [69], Gosman, Pun, Runchal,
Spalding and Wolfshtein [35], Bradshaw and Ferriss [15], and Spalding
[89].
      Although solving additional  transport equation for turbulence
quantities, such as Equation  (7.54), appears to be a considerable
effort in itself, this approach offers a realistic and negotiable
compromise to otherwise unapproachable problems in turbulent flow.
These, or similar methods have not been employed in the present study,
but only because the flow field offered enough a priori knowledge to
justify and permit the use of simpler mixing length models.
                             265

-------
7.2.1  The Radial  Transport Coefficient, er
      To model  the plume and circulating flow fields, the radial
component of eddy diffusion must be modeled throughout the fluid
system.  To this end, four flow regions are defined which are illus-
trated in Figure 7.66.
      These regions are defined as follows:
            Region I:   Zone of.established plume flow
            Region II:  Zone of flow establishment
            Region III: Circulating ambient
            Region IV:  Lateral surface spread
Each of these regions has special  characteristics and must receive
special attention.
Region I
      Equation (7.50) relates the  radial component of momentum trans-
port as
            er - ClV vmax                                      <7-55)
For Region I (established flow),  C-j = .0256, *r = r1 /2,  the plume
half-radius, and vmax as the centerline velocity.  Tomich [99] used a
similar relationship for his analysis of a compressible free jet.  In
dimensionless form
       r^T   = '°256 Rl/2 Vmax                                 (7-56)
where
                             266

-------
ro
                   PLUME  BOUNDARY
                                                         VERTICAL  CUT-OFF
                                        RADIAL  CUT-OFF
                                 FLOW ESTABLISHMENT
                           OUTFALL  PORT
                                                    R
                            Figure 7.66.
Regional Specification for Turbulent
Eddy Coefficient Modeling

-------
and
            u    = Vmax
             max ~  v

As a reference value for er we set
           e   = .0256                                          (7.57)
            ro
^ere R1/2- RQ - 1. 0 and Vmax-V0- 1.0.
So that
            e
                      1
           r vn     RE
            oo       r
                       0
Hence, the reference radial Reynolds number is
            REr  = 39,                                          (7.58)
              ro
the value used in all computations except Case 8.
      To obtain the point Reynolds number RE/.. .<>  (the indices on REr
will be omitted hereafter with the point value always implied), we
define
           er = er0FRjk
where, er may be viewed as the point value of eddy diffusivity with
subscripts omitted.  With the definition Equation  (7.59)
and
            FRjk = Rl/2 Vmax«
            REr  = REr /FRjk  .                                  (7.61)
      For the momentum jet, in the zone of established flow (cf. Sec-
tion 7.1), it has been established that
            Vmax  =12'4/Z                                       (7-62)

                            268

-------
and at the half radius

             V
            \T—  = -5 = e                                      (7.63)
             max
or
                                •                               (7-64)


Using the value K = 77 from Abraham [1],


            FRjk = 1.176,                                       (7.65)


for a momentum jet (subscripts on FR will be omitted hereafter, with

the point value implied).

      Equation (7.65) represents an empirical value for FR.  Two

numerical experiments were carried out for the momentum jet, one case

where FR = 1.176 was held constant and the other where I-K was computed

according to Equation (7.60).  The centerline velocity distributions

for both cases were found to be essentially identical.  Figure 7.67

illustrates  the  result of iteratively computing FR = R1/oVm,v.  In
                                                       I / i. in a A

both of these cases a Gaussian profile at Z = 4.5 was used for the

inflow boundary condition (Type 1 boundary condition - Region II does

not enter into compution).  Figure 7.68  illustrates FR for cases having

varying degrees of buoyancy using the Type 2 inflow boundary condition

(power law velocity profile).  Note that in this instance FR ^ 1 for

the momentum jet (F  -»• ») and is owed to a slightly higher centerline

velocity, the ratio approximately equal  to the value of Equation (7.65),
                             269

-------
ro
--4
                          1 .2
1 .1
                      cc
                         1 .0
                         0.9
                                  EQUATION (7.65)
                                         I
                          I
I
                                        10         20         30

                                            AXIAL DISTANCE, I
                                               40
                           Figure 7.67.   Computed Values of FR for a  Momentum Jet

-------
ro
               o
               I—
               CJ
               O
               LlJ
=c

O
                   10   -
                    8   -
                   6   -
                    4   -
                    2   -,
                                                            EQUATION (7.67) (F  =  1 )
                                    10
                                   20             30

                                          AXIAL  DISTANCE,
40
                                                                                             50
60
                           Figure  7.68.  Computed Radial  Eddy Diffusion Factors, FR for Deep Water
                                        Plumes at Various Densimetric Froude Numbers

-------
      For the cases dominated by buoyancy, Equation (7.33)  gives
or
            Vmax = 4-4(2)(F0Z)'                              (7.66)
The radial velocity distribution is again given by Equation (7.63)  with
K = 92.
      In this instance
            FR = (2)1/34.4(FZ)-1/3 V       Z'
            FR =
            FR =
based on
                  Fo
            Z = Z/r0.
      Equation (7.67) is also plotted on Figure 7.68 for comparison of
the empirical approximation and computed value of FR for F  =1.
      Aside from merely illustrating how the radial  eddy transport
coefficient er varies as a function of the degree of buoyancy,
Figure 7.69 also reveals that the use of a constant  transport coeffic-
ient is untenable in buoyant plume flow computations and can lead to
order-of-magnitude errors.  A numerical experiment was carried out to
                            272

-------
         1.0
          .5
                                            I    r
                                       Fo  = 52
                                           FR  =  1 .176   1
                                           FR  =  V     R.
                                                  max   !
             10
                 50
AXIAL DISTANCE,  I
100
       Figure 7.69.   Comparison  of  Computation Using Constant and
                     Variable  Radial  Eddy Transport Coefficients

ascertain these differences  for  F   =  52 using the Type 1 boundary
condition.   Figure 7.69 illustrates that large errors will occur in
both the centerline velocity and buoyancy distribution if er = con-
stant is used.   The curves corresponding to Cases 15 (FR = constant)
and 16 (FR  = VmR1/9).   Also  refer to Figure 7.49.
              max
      In the early development  of  the computer program, how to effect
the variable transport computation iteratively was unknown and such
                           273

-------
attempts led to numerical instability.  Although this problem was sur-
mounted later (see Section 7.3 for stability problems related to var-
iable e  and ez),  as an interim step, the mixing length R    was com~
puted prior to computation from similarity assumptions for the given
Froude number.  With R, .^ fixed, FR was computed from Equation (7.60),
and eliminated this source of numerical instability.  However, the
solution was only nominally more accurate than using FR = constant.
Hence, this method was adjudged inadequate and, as mentioned earlier,
later abandoned.
Region II
      The zone of flow establishment is characterized by turbulence
regimes (see Figure 7.70), 1) the potential core, a roughly conical
region, where mixing is dictated by the convected pipe flow turbulence,
and 2) the zone of intense mixing lying outside the potential core,
spreading into the ambient, and created by the shear between the
effluent and the ambient fluids.  A mixing length, «, , may be philos-
ophically defined as being proportional to the width of the shear
region.  However, the geometry is difficult to define and the criterion
as to the width of the mixing zone is quite arbitrary.  Also the length
of the mixing zone, Ze, that is, the point where the zone of intense
mixing reaches the plume centerline, is also quite arbitrary and cer-
tainly is not defined by a sharp point as Figure 7.70 indicates.
Tomich [99] bypassed the mixing length problem in the region by
setting er =  .2 times the value in the established flow regime.  For  a
                              274

-------
momentum jet, this value was found to yield downstream results  in good
agreement with experimental results.
                                                    MIXING
                                                     ZONE
        POTENTIAL
          CORE
         Figure 7.70.   Concentration  Distribution in the Zone
                       of Flow Establishment
      In this study,  we have not followed  Tomich's method since we
deal with cases of high relative buoyancy  (low densimetric Froude
numbers).
      To set up a turbulence model  for  this  regime we need:
      1) a mixing length,  and
      2) a definition of the  region  of application:
            •  radial region
            •  vertical  extent.
                           275

-------
      To compute a mixing length, a reasonable criterion is

            *c = rl/2 - rc
where r  is the radius of the potential core and r1 ,2 is again the
half-radius.  The transport coefficient is then defined by,
            erc - .0256 (r1/2 - rc) v^.                      (7.68)
Physically, e   would apply over the region r.  - r , where r.  is the
mixing zone outer boundary.
      The next problem then is to define r  based on some relevant
mean flow quantity.  In the present work, the concentration profile
was used for such a criterion.  Velocity could not be used because of
power law boundary profiles and because buoyancy tends to distort the
velocity profile.  The criterion was set as
            rc - r>95                                          (7.69)
or r  extended to the point where the concentration was decreased to
95% of the centerline value.  The outer boundary was set as

            rc = r.05                                          (7-7°)
or where the concentration had decreased to 5% of the centerline value.
The length of the potential core was computed from the criterion
            zc = z.90                                          (7-71)
where the concentration at the plume centerline is reduced to 90% of
the initial value.  The numerical model does not account for
                             276

-------
derivatives resulting from variations in z  , and where convection
terms dominate transport, this deletion is valid.  However, in the flow
establishment region, this treatment can lead to large constituent
discrepancies if the gradients of er are not accounted for.  For this
reason and other computational difficulties, e  has been assumed
radially constant at a given elevation, laterally to the plume cut-off.
      Based on Equation (7.68) along with criterion Equations (7.69)
and (7.71) a typical computed potential core and half-radius is illus-
trated in Figure 7.71 for FQ = 46.  This method for computing the
                                     transport coefficient was felt to
                                     be unsatisfactory in that the
                                     computations were slowed down
                                     compared to preset specification,
                                     and definition of the potential
                                     core appears to have questionable
                                     accuracy.  However, one fact was
                                     established as a result of these
                                     experiments in that R, ,~ ^ 1.0
                                     for all cases run.  The method
                                     finally used was to define the
                  .5
                            1 .0
                    R
Figure 7.71.   Computed Potential
              Core and Half-
              Radius FQ = 46
                                     length, Z , based on a criterion
                                     similar to Equation (7.71), use a
                                     straight line fit between the
  Refer to Equations (3.73) and (3.80) and note the computer program
  does not contain any viscous terms envolving derivatives of er.  It
  may be shown that such terms are small except in Region II.
                             277

-------
points (0,Z ) and (1.0, Z ) to define the potential core, and set
R1/2 = 1.0 up to Zg (see dashed lines on Figure 7.71).  This procedure
was found to be satisfactory and added speed to the computation.
      The remaining problem, in computing quantities within Region II,
is that the computer model treats e   constant across a lateral  plane,
where, in fact, there is considerable variation.  Treating e   constant
in this fashion is to overestimate the diffusion coefficient within the
core since the value used is typical of the turbulent mixing region.
The net result of this procedure is to effectively reduce the computed
core length which can result in downstream errors.  One way to bring
the computed core length more in line with experimental results  con-
cerning the core lenth is to reduce the value of e  .
      One such model, which is based solely on numerical experiment is
given by,
            erc - .0256 (r1/2-rc)(r1/2-r0)/r1/2 vmax           (7.74)
which is the same as Equation (7.68) except for the multiplication fac-
tor (ri/2~rc)/ri/2'  This factor nas tne effect of reducing the eddy
diffusion, given by Equation (7.68), near the outfall and has decreas-
ing importance as the end of the potential core is approached.  This
model for radial eddy diffusivity gives good results over the entire
range of Froude numbers for deep water plumes (see Figure 7.72) and is
the preferred method of computing e .  All cases discussed earlier are
based on Equation (7.68) where applicable.  The case for F  = 1 illus-
trated in Figure 7.72 may be compared with Figure 7.55 (Case 24).
Cases displayed in Figure 7.72 were computed on a 26x25 grid.

                            278

-------
2.0
1.0
 .50
 .10
        CASE  F   Vm
        	  —o— —m
 .05
44
45
46
47
  1
  5   a
 25   *
100   v
           V(R,0) - 1.15
                  SIMILARITY  SOLUTION
                  FOR AT/AT
                        m0
                           1
 .01
                                       10
                                AXIAL DISTANCE, I
                                                               50
  Figure  7.72.
        Center!ine  Velocity and  Temperature  Distribution
        for 44 Diameter Deep Outfall
                                                                100
                             279

-------
      For the purpose of comparing results using the two different
methods for computing c  in the core, refer to Figure 7.73, which
illustrates the centerline temperature distributions for intermediate
water depth cases.  A summary of these four cases may be found in
Chapter 8, Table 8.1 listed as Cases 48, 49, 50 and 51  for Froude
numbers 100, 25, 5 and 1, respectively.  Note that Equation (7.68)
(Figure 7.73) yields much more rapid deterioration of the centerline
temperature than Equation (7.72) (Figure 8.1,  Chapter 8).
      Results from Figure 8.1 may also be compared to Figure 7.72.
Note that for the deep water cases at low Froude number, the centerline
temperature distribution again decays more rapidly near the source
than for corresponding cases at intermediate depths.  This discrepancy
is caused by lack of axial finite difference resolution in the core
region of the deep water results.
Region III
      In the region outside the plume, the value of er is set to a
reference constant that is descriptive of the ambient conditions.
Reasonable variations of this value have been found to have little
effect on the circulation patterns or on the plume computed quantities.
In fact, several early runs were made letting e  in the ambient take
the same value as computed within the plume.  Only slight differences
were noted in the plume size when the value of the ambient was set to
1% of the plume interior value.
      Most calculations and the present version of the computer program
use a "cut-off" point (see Figure 7.66) for e  at a point just outside
                            280

-------
1.0
0.1
                                    *  *"**—•-•
                            EDDY  COEFFICIENTS IN CORE
                            COMPUTED BY  EQUATION (7 . 70)
                                                  i v  i  (
i
1
                                                         10
                     AXIAL DISTANCE, Z
  Figure 7.73.  Computed Center!ine Velocity and Temperature
              Excess.  Cases for 10 Diameter Deep Water.
                       281

-------
the plume where radial  convective effects dominate radial transport
(again radial  derivatives in e  are neglected).  The first attempt to
establish a radial  cut-off was based on r (concentration) dropping to
5% of the center!ine value.  This seemed to be a reasonable criterion
but proved to be computationally unacceptable because oscillation of
the cut-off point position between nodes, near the plume boundary,
dramatically slowed convergence and grossly added to the computation
time.  The convergence  problem was eliminated by extending the cut-off
point two nodes beyond  the r = 5% criterion, but resulted in a
"ragged" plume edge, the raggedness being unrelated to flow physics
(Figure 7.74).  The next step was to preset an envelope in which the
plume would always  exist and e  could be held constant at a particular
elevation.  This envelope extends two to five nodes beyond the r = 5%
criterion but is computationally very attractive because convergence
is significantly speeded with no real loss of accuracy.
Region IV
      In the lateral surface spread, the plume boundary is defined by
the presence of the circulating or reverse flow field.  For a vertical
cut-off point, the  boundary is extended two nodes below this region of
negative radial velocity.  The value of er is set to the value com-
puted within the vertical rise region and being held radial constant.
For all cases run,  the  convective effects are reasonably large in this
region;  hence, e  is of minor importance.
                             282

-------
ro
CO
CO
                                FIGURE  7JI4 .   3D ILLUSTRflTION OF  TEMPERRTURE  FIELD--A T

-------
7.2.2  The Vertical Transport Coefficient, e.
      Referring to Figure 7.66, the unique regions of vertical eddy
diffusion computation are identical to those of the radial component.
However, it is generally true that for the present model only one of
the coefficients, e  and e , will  be of major importance in a given
region.  For instance, in Regions  I and II, e  was found to play a
major role in computing the plume  dynamics, whereas, for all intents
and purposes, e  may be ignored.   This statement is proved by numerical
experiment (Case 10, Figure 7.11)  where e  was set to er in the mixing
zone.  Only minor differences were noted between Case 10, and Case 6
where e  was set to a constant value of .001.  From our knowledge of
jet induced turbulence we expect that point-wise, e  and e, should be
                                                   i       b
nearly the same in Regions I and II.  (cf. Hinze [40]).  Some differ-
ences may be noted near the surface where larger vertical mixing
scales are suppressed.
      The fact that vertical mixing is of little importance in Regions
I and II may be ascertained on theoretical grounds by comparing the
order of magnitude of the various  vertical transport terms in the
Equations of motion (3.67).  Although the details are not presented
here, one finds that vertical convection dominates vertical diffusions
in these regions, an expected result, except near the surface where the
two transport mechanisms may play equally important roles.
      Hence, we may dispatch concern for EZ in Regions I and II remote
from the surface, without further investigation.  However, numerical
experiments have shown that e  is very important in Region IV and there
                            284

-------
is, nevertheless, incentive for extending the vertical  cut-off to the
plume centerline to overlap that portion of Region  II.
Region III
      The vertical  transport coefficient associated with  Region III  is
that of the ambient sea,  and as such,  e  depends on water depth,
currents, sea state and ambient stratification.   Extensive work has
been carried out by the Oceanographic  community  to  determine e  as
influenced by the above mentioned variables.   Summaries and discussions
of this work may be found in work by Koh and Fan [52],  and Wada [107].
      The presence of vertical  stratification can dramatically impede
vertical  mixing, whereas  shear  force tends to enhance this mixing.
Hence, the vertical mixing coefficient must depend  on,  in some fashion,
the relative importance of the  stabilizing effect of stratification
and the destabilizing forces of shear  flow.  The local  Richardson
number, RI, relates the relative importance of these forces through
the ratio
            RT  =   stabilizing forces
                   destabilizing forces   '
                      a  dp.
            RI  =  .  £_dz   .                                 (7.73)
In terms of the dimensionless quantities defined in this manuscript,
                           285

-------
            RI  =  9r   •       r    •                           \7 •
If RI < 0, the flow is obviously unstable.   Various researchers have
proposed methods for computing t  using a Richardson number correla-
tion.  The most notable of the efforts are  summarized in Table 7.5.
Note that in this discussion we are speaking of a general  vertical
eddy transport coefficient with no distinction between the transport
of material, heat or momentum.  Since any correlation for general
application is at best a rough approximation, we are assuming that  the
vertical Prandtl (or Schmidt) number is unity.
      The various correlations given in Table 7.5 are essentially
Richardson number modifications of the neutral diffusion coefficient
ez .  Thus, the first task lies in determining e   for a neutral am-
  0                                             zo
bient (RI=0).  Kent and Pritchard [51] give one such correlation for
the wave induced component, for the James river estuary, as
                            d  H
            ez  = .Old (1  - {4 f  e                            (7.75)
              o
where
            d = distance from the surface,
            L = depth of the water body,
            H = wave height,
            T = wave period, and
            SL = wave length.
      For a well mixed surface layer only,  Golubeva [33] and Isayeva
                           286

-------
    TABLE 7.5.   CORRELATION OF THE VERTICAL DIFFUSION COEFFICIENT
                ez WITH THE LOCAL RICHARDSON NUMBER,  RI

                (extracted from Koh and Fan [52])

Note:  e   =  e   at RI  = 0, i.e., the neutral case,  3:  proportionality
        o
      constant;  varies from case to case.
Rossby and Montgomery
      (1935)*

Rossby and Montgomery
      (1935)*


Holzman (1943)*


Yamamoto  (1959)*


Mamayev (1958)*
Munk and Anderson
      (1948)**
ez =
ez =
ez =
         (1+6 RI)
         (1  +  3 RI)
                                                    -1
                  -2
       0
         (1  -  3  RI)
         (1-3
''•V
          -3  RI
«'4
nr   1
ez =
         (1  + 3  RI)"3/2
 *As given by Okubo  (1962)

**As given by Bowden (1962)
                                  3 = 3.33 based upon data  by
                                           Jacobsen (1913)  and
                                           Taylor (1931)
                        287

-------
and Isayev [47] give
H
T
                                                               (7.76)
      Figure 7.75 (extracted from Reference  [52])  illustrates  the


relationship between e   and the local  sea state.
       400
  01
       300   -
       200   -
        100  -
                                           Figure  7.75.   Dependence  of
                                                         EZ  on  Sea

                                                         State
      For the case of tidal  currents,  Wada  [107]  gives
                   K2US /d" z
                  L/L log
                                                               (7.77)
                           288

-------
where K is the Karman constant, Ug is the surface current, L  is the
scale of the bottom roughness.  Where both components, tidal currents
and wind waves, are acting, Wada gives

                 K2(d+l_J  z2      /L IL       ftH  -2Trd/£
           e   =	\	                +^le            (7.78)
            zo        \-          Y^d + LQ Z    '

Various measured values of pe  are given in Table 7.6 (extracted from
Reference [107]).
      In the absence of ambient currents Harremoes [36] gives
                      •3  /  A i   X-2/3
                      3
                                         cm/sec
where z is in meters.  This correlation was obtained off the coast of
Denmark.  Koh and Fan have obtained the relationship
                                                               {7'80>
                   dz
where again z is in meters.  Data used in obtaining this result is dis-
played in Figure 7.76.
      Any estimate of e  or e   in the ambient sea has questionable
                              o
accuracy.  At best, these correlations, and measurements for that
matter, are accurate only for the observed conditions, conditions
which may change drastically with time and location.   Aside from this
complication, just how the researcher deduced the transport coefficient
value from physical measurements may shadow the validity of results
                            289

-------
ro
io
o
                 o
                 0)
                 to
                 LU
                 I—I
                 O
                 LU
                 O
                 •a:
                 o
                     10'
                     10
                           I    I  I I I    I   I   I I I    I   I  1  I I    I   1
                                  A  IIARREHOS (1967)
                                 ov  KOLESNIKOV (1961)
                                  •  JACOBSEN (DEFANT, 1961)
                                  a  FOXWORTHY (1968) PATCH
                                  •  FOXWORTHY (1968) PLIU1E
                                     FOXWORTHY (1968) POINT SOURCE
                      1     _
                      0.1   —
                      0.01
                         10
                                  •  •    x      \
                                          X a        N
                                                         \     -N
                              I    ill    I    II
-7
          10
-6
,-5
                            ,-4
               10           10  ^        10  J
             (-1/p) dp/dz,  DENSITY  GRADIENT,  m"1
Figure 7.76.  Correlation of ez with Density Gradient
                                             10
-2
                                             10
-1

-------
                             TABLE 7.6.  VALUES OF VERTICAL EDDY VISCOSITIES IN THE SEA
ro
      Current or Sea
         Region	
      All oceans
North Siberian Shelf
North Siberian Shelf
North Siberian Shelf

Schultz Grund
Caspian Sea
North Sea
Danish Waters
Kuroshio
Japan Sea
off San Diego
      a) W = wind velocity  in m/sec
      b) z = distance from  sea  bottom  in meters
      c) Very great  stability
Layer
Surface
0 to 60m
0 to 60m
0 to 22m

0 to 15m
0 to 100m
0 to 31m
0 to 15m
0 to 200m
0 to 200m
Near the
sea bottom
pt in y/ v*m/ acv.
a)pe_ = 1.02W3
(W 6m/sec)
= 4.3W2
(W 6m/ sec)
75-260
10-400
b) Z+Q i 3/4
385(~227p)
1.9-3.8
0-224
75-1720
c)1.9-3.8
d)680-7500
150-1460
e) 93-(z+0.02)
                                                             pe  Derived From
                                                             Thickness of upper
                                                             homogeneous layer
                                                             (wind currents)
                                                                   Tidal currents
                                                                   Tidal currents
                                                                   Wind currents
Reference
Thorade, 1914
Eckman, 1905
Sverdrup, 1926
Fjeldstad, 1936
Fjeldstad, 1929
                                                                                          Jacobson, 1913
                                                                                          Stochman, 1936
                                                                   Strong tidal currents  Thorade, 1928
                                                                   All currents           Jacobson, 1928
                                                                   All currents           Suda, 1936
                                                                   All currents           Suda, 1936
                                                                   Tidal currents         Revelle & Fleming
                                                     d) Very strong currents
                                                     e) z = distance from sea bottom in  meters

-------
and the application to numerical  modeling.   Generally,  these coeffi-
cients are deduced from concentration measurements and  back-calculated
through an analytical  diffusion equation.   Hence,  the values are valid
only for the diffusion equation used to calculate  them  in the first
place.  Just how appropriate these values  are as they enter into more
elaborate numerical computation is open to question in  this author's
opinion.  It is felt that the determination of ambient  diffusion coef-
ficients is an area that needs extensive research.
      In the present computer model  for Region III diffusion coeffi-
cients, various of the models discussed above were tried.  Very little
difference was noted in the Region III circulation patterns in any case.
Influence on the plume was noted only when the value of t  was unreal-
istically large, in which case the flow dynamics took on the charac-
teristics of a creeping flow.  For this reason e  was set to a value
                  -4      -3   2
on the order of 10   to 10   ft /sec in Region III for  succeeding
computation, a value corresponding to moderate stratification, low sea
state, and low ambient current.
Region IV
      In modeling the plume lateral  spread, the vertical turbulence
component is of utmost importance.  As the plume encounters the surface
and begins the radial  surface spread, plume induced turbulence dominates
the mixing phenomena.   At increased radial distance, the induced turbu-
lence decays and is suppressed by stratification.   Generation of tur-
bulent energy by virtue of the lateral shear flow is also declining
because of smaller velocity gradients.  At some larger  radial distance

                            292

-------
the field of turbulence will be dominated by ambient effects such as
sea state.
      We have just discussed the ambient contribution to e  and indi-
cated rough methods for such calculation.  The plume induced turbulence
in the zone of initial spread (or the transition zone) is the important
feature of Region IV.  Unfortunately there is very little data avail-
able in the literature which is directly applicable to the problem of
turbulence modeling in this zone.
      From a theoretical point of view, we assume that Prandtl's second
hypothesis holds, or that

            ez  =Cl*2umax                                    <7'81)
              o
for the neutrally buoyant case.  We also expect that a Richardson
number modification of Equation (7i81) would suffice for the case of a
spreading thermal layer, of the form

            ez = Cl*z "max
      For the neutrally buoyant situation we may gain some insight as
to how the produce £, u_,v behaves by assuming the flow can be approxi
                    Z  nloiX
mated by a radial jet similarity solution.  The appropriate similarity
equations for a radial jet following the methods devised by Morton, et
al. [60] for a vertical jet, are
      Continuity:

            3F < V> - <"um                                   (7'83)
                            293

-------
      Radial  momentum:

            3? (um rt)  = °'                                     (7'84)
      In the above equation  a "top-hat" velocity profile has been
assumed, where um is the mean radial velocity, T is the characteristic
thickness of the jet and a is the usual entrainment constant.  The use
of the top-hat velocity profile is entirely satisfactory for purposes
here, since we are only interested in the relative behavior of T and
u , which is insensitive to  the similarity profile used.
      Solving these equations, one finds
            u r = constant                                     (7.85)
and
             p
            UITT =  constant                                   (7.86)
Hence,
            (umr)(umT)  = constant,
and
            umT = constant.                                     (7.87)
      Equation (7.87) reveals that if the velocity field is approxi-
mately similar then the eddy coefficient ez must be constant in view of
Prandtl's second hypothesis  (a result identical to the axisymmetric
jet).  Hence
            e..  = C-,4, u _ = Ci-Constant                       (7 88)
             Z-      \ 2. max     '                                \/ .00;
The remaining problem lies in evaluation of C,  and (a u
      In the present work, C] is assumed to take the value .0256 as in
                                                       max
                             294

-------
the case of the axisymmetric flow region.
      The quantity &,um_v was treated by four different methods during
                    Z iTlaX
numerical experiments as listed below.

      Method 1 :
      Compute the value of a  from the local velocity profile based, on
the distance from the level of maximum lateral velocity to the level
where lateral velocity is 1/2 the maximum value.  That is,

            lz = zl/2'

This method is identical  to that used to compute FR, but in the instance
of lateral flow was found to be unsatisfactory because of numerical
instability.  All attempts to compute FZ   where

            FZo = Zl/2Umax'                                    <7'89)
iteratively from local information were found to be unstable and the
method was abandoned.

      Method 2:

      Use a constant value of Z-    based on the value of R-| ,^ at the
point of lateral spread.  This method proved to yield diffusivities
which were too large.

      Method 3:

      Use a constant value of FZ  = Z, ,2 Umax where Z-j ,2 and Umax for
the entire system are computed in the vertical plane where the maximum
lateral velocity occurs.  This method, based on the insight given by
the similarity solution, also yielded diffusivities which were
                           295

-------
too large.   This method was applied only to cases having buoyancy;
hence, the  failure may have been due to an inappropriate Richardson
number modification of FZ .
      Method 4:
      Use the method given immediately above,  except scale the result
by the local ratio (Umax): (Umax)system.   As in  the  two  methods
immediately above, this calculation proved to  be numerically stable
under all conditions once a reasonably realistic lateral velocity dis-
tribution was established.  But, unlike the above methods, local  dif-
fusivities  are computed which give more realistic velocity fields.
Hence,
            REZO = REz(ref)/FZo                                (7-9«>
and
            FZo = Zl/2 Umax
where Z,y2 is calculated at the system maximum lateral  velocity and
U    is the local maximum lateral velocity. The subscript o again
indicates the condition of neutral buoyancy.
      To account for local stratification, the local Richardson number
model due to Mamayev (cf. Reference [108]) was employed,
                   o                                           (7.91)
where RI is again the local  Richardson number as defined by Equation
(7.76) and $ is an empirical constant.  Wada [108] used Equation (7.91)
in his study of planar thermal  outfalls discharging horizontally, but
used a constant value of ezo.
                             296

-------
      Although there is no data known to the author relating point
eddy diffusivities to the point Richardson number in turbulent jets,
data has been obtained which relates the entrainment of such flow to
the overall Richardson number (cf. Ellison and Turner [25]).  Stolzen-
bach and Harleman [94] have illustrated that the data of Ellison and
Turner may be adequately represented by the form,
                                                              (7.92)
where a  and a   are the entrainment coefficients for buoyant and
               o
neutral spreading surface flows, respectively, and RI'  is the gross
Richardson number.  Stolzenbach also illustrates the relationship
between eddy viscosity and entrainment as
            ez   _  az
            EZn     "Zo
              0       0
      Thus, based upon the data of Ellison and Turner, and the func-
tional relationship, Equation (7.92), derived from this data, the
Manayev Equation (7.91) is apparently a credible method for modifying
point-wise neutral  eddy diffusion coefficients for application in
laterally spreading buoyant plumes.  In the computer program, we use
the form,
            e  • e_        + t  e-3RI                         (7.93)
                   ambient     o

      The computer program is also set up to use the various other
models given in Table 7.5.  These models have not been used owing

                             297

-------
primarily to lack of appropriate information concerning the empirical
constant B.
      The value of B (for Equation 7.93) used by Wada [108] was .8 for
momentum diffusivity and .4 for heat diffusivity based on ambient
conditions.  According to work done by Stolzenbach this value should
be appreciably higher for plume flow.  Computations using various
values of B for the present work are illustrated in Chapter 8.
      In the present work, another form of e  has been used, primarily
for starting solutions where Equation (7.93) results in numerical
instability.  This form is given by the equation
               = s7   e                                       (7.94)
             z    zref
where d is depth or distance from the surface.   The result is a
Gaussian depth decay of eddy momentum diffusivity from a surface
reference value.  Equation (7.94) is used in computation merely as a
computational aid and is abandoned in favor of Equation (7.93) once
reasonable velocity and temperature profiles are established, or a
numerically stable situation is attained.
7.3  Numerical Stability and Convergence
      During the course of this investigation various experiments were
performed dealing with solution stability and convergence.  For each
case run, at least five node points were monitored for convergence
rates of U, V and r.  Additionally, the program computes the maximum
change of ty, ft, and AI throughout the system at selected iterations,

                           298

-------
and an overall r balance error is computed at the end of each run.
Liebmann relaxation factors were employed to each of the equations  for
if;, n, A,, and r to either accelerate or decelerate solutions.
7.3.1  Numerical Stability
      To define what is meant by numerical stability in this manu-
script, we take the opposite view—that of numerical instability.   The
reasoning for this view is that it is entirely possible that the
system of buoyant fluid may have physical instabilities which are  not
divergent.  The solution which we are trying to attain may, in fact,
be physically unsteady, and may never be attained by steady flow
methods.  Since the Gauss-Seidel method with under/over-relaxation  is
not unlike certain transient methods (see Appendix E), continued
iteration may reveal a cyclic behavior of the computations.  This
situation cannot be termed a numerical instability.  It merely illus-
trates the inability of steady flow techniques to simulate transitory
flow physics.
      To demonstrate this idea, the computer program was set to a
different task, that of predicting the flow field past the end of  a
cylinder contained in a larger pipe.  From experiments we know that,
at low Reynolds numbers, streamlines past the end simply are distorted
toward the centerline, much as the case of irrotational flow (Figure
7.77-A).  At much higher Reynolds numbers, vortex shedding from the
end of the cylinder will occur and the flow field is termed unsteady
although patterns may be repeated in time or in a cyclic fashion
(Figure 7.77-B).
                            299

-------
                   PIPE WALL

//////////;//A////////////////////////////////////////////////////////////////////////
 CYLINDER
    A.   LOW  REYNOLDS NUMBER  (CREEPING FLOW)
                                 REGION OF
                              VORTEX SHEDDING
    B.   HIGH  REYNOLDS NUMBER

       Figure  7.77.   Observation of Flow Patterns
                    Past the End of a  Cylinder
    A.   LOW  REYNOLDS NUMBER  (CREEPING  FLOW)
        HIGH  REYNOLDS NUMBER
       Figure 7.78.   Computed  Flow Patterns Past
                    the End of a Cylinder
                   300

-------
      We expect that the steady flow computer program would converge
to a steady solution at low Reynolds number, and, in fact,  this was
the result as illustrated in Figure 7.78-A.   At a high Reynolds number,
however, a converged solution could not be attained.  Computed quanti-
ties demonstrated quite the same behavior that one would expect from a
transient solution to this problem.  Generally, as computation pro-
ceeded, a recirculation pattern formed behind the cylinder, grew by
elongation, and collapsed to nearly circular form, and elongated again
(see Figure 7.78-B).  This process occurred repeatedly as computation
continued.  Although it is impossible to quantify the physics from
these results, it is reassuring to know that the numerical  technique
will reveal the presence of a physical instability, or unsteady flow,
and not converge to an erroneous steady solution.
      Thus, it is entirely possible to have non-converging  (although
not diverging) solutions that are not associated with numerical
instability.  Hence, we define numerical instability as that situation
which upon repeated iteration leads to increasingly divergent and
physically ridiculous results.
      As a general observation, involving perhaps a hundred or more
computer runs of various duration, the numerical techniques used were
found to be unconditionally numerically stable provided that:
      .  All Liebmann acceleration factors were less than unity,
      •  All eddy diffusion coefficients were constant, or  the
         velocity field at the beginning of computation has at least
         reasonable similarity to the final  solution.

                            301

-------
      These observations are a result of flows  having  Reynolds numbers
from 0 to infinity and a variety of other testing  conditions.   Based
on these numerous experiments, difficulties encountered  by other
authors, the accuracy outlined in Section 7-1,  and comments made by
Spalding [91], the present difference formulations and grid system used
is extremely attractive.
      At an early date in this investigation it was discovered that
solutions invariably became unstable if the acceleration factor, Lj,
for the n and A-I  transport equations was greater than  unity.   However,
the value Lr = 1.6 was used for the stream function, y,  elliptic equa-
tion without difficulty.  Later, it was discovered that  under  some
flow condition, the value of Lp also needed to  be  less than unity to
avoid instability.  For cases involving constant eddy  coefficients,
only the transport equations needed to be decelerated.  After  these
initial investigations, the general rule used was  to decelerate all
equations or set L£ and LJ < 1.0.  The general  form of the decelerated
solutions is
                           rV                               (7'95)
where the subscript p indicates the nodal  point in question,  n is the
n   iteration and r  is the result of the  n+1  unaccelerated Gauss-
Seidel iteration.
      A value

                                                               (7.96)

                           302

-------
was found to be satisfactory for nearly all  cases.   In a few instances
of very shallow water and non-linear e  and e , values as low as
LT = .80 were used.  In all cases, the acceleration factor is applied
as soon asr1 is computed at a node.
      No attempt of a theoretical analysis of stability will be pre-
sented here since the presence of non-linear eddy coefficients negate
meaningful analysis and the case of constant diffusion coefficients
has been presented by various authors, [7,111], at least for time
dependent problems.  Some insight to stability of steady state computa-
tions is given in Appendix A.  Further insight into this question may
be gained by the analysis given in Appendix D which compares the Gauss-
Seidel iteration technique to an appropriate (similar) transient
solution.
      It was, perhaps, propitious that a superior grid system was
devised at the outset of this study (see Figures 5.3 and 5.4).  In a
recent publication, Spalding [91] points out that making vorticity
adjustments in a cluster of five adjacent points and the stream func-
tion at the central point has a striking effect on divergence removal
for reasons unknown.   Unlike the grid system to which Spalding refers
where vorticity and the stream function are computed at the same space
points, the present grid system is staggered.  The vorticity values
which interact as a source for the stream function elliptic equation is
averaged from the four adjacent neighbor points, which is closely akin
to the method referred to by Spalding and may be responsible in part
for the seemingly inherent stability of the present method.
                             303

-------
      Another aspect of the present computational  technique is that
linear gradients are always used for flux terms whether on the boundary
or in the interior, by the use of fictitious boundary cells.  Spalding
again points out that higher order methods for treating boundary con-
ditions may in fact lead to less accurate results  due to violation of
reciprocity and conservation principles at boundaries.   In the present
method, through the use of the correct conservative difference equa-
tions and fictitious boundary cells, quantities are identically con-
served.  This feature may also contribute to the success of the tech-
nique in avoiding instabilities propagated from system  boundaries.
7.3.2  Convergence
      The question of solution convergence has been partially answered
in the preceding section.  It is obvious that solutions which are
numerically unstable will not converge.  On the other hand, it is also
possible that a solution which is numerically stable will  not converge
as demonstrated in the example of flow past the end of  a cylinder at
high Reynolds number (Section 7.3.1).
      The condition for convergence used in this work is defined by
            fn+l _ -n
             P   "  P
               f
                P
                n+l
(7.97)
where 6f is the convergence criterion for the quantity f.  The sub-
script p again indicates the nodal point in question and n is the nth
iteration.  The condition for 6f approaching zero is not a sufficient
                            304

-------
condition to guarantee solution accuracy, however, since the numerical
procedure may in fact converge to an erroneous solution.  The method
used in this work to decrease the probability of erroneous solutions
was to check the continuity of matter by evaluating net flux of
matter at the system boundaries and selected interior planes.  This
check is subsequently referred to as the r-balance error (6r),r
referring to a conservative constituent.  This quantity is effectively
given as a surface integral for the system in the form of
            6F =
                 r  f r(tf-n) dS + /  (-Verr)-n
dS
                    ST             ST
                    Sin             Sin
                              dS + J   (-verr)-n) dS
                                    f,      1
      100%     (7.98)
where ST represents a vertical plane in the flow field and S  is a
       1                                                     in
radial plane at the inflow boundary extending to RQ.  Equation (7.98)
gives 6r as a percent error of the system inflow.
      Typical results showed the r-balance error to be on the order
of 1%.
      General observation of the various numerical experiments illus-
trated the following:
      •  The convergence rate decreased significantly with increased
         grid size.
      •  The stream function distribution converged with respect to
         6  more rapidly than vorticity and buoyancy (or r).
                            305

-------
Vorticity was the slowest to converge and also the most
erratic.
Convergence of all quantities near the outfall was much
more rapid than in the far field.  Thus, sizable errors in
the far field did not influence the validity of solutions
near the outfall.
The relative magnitude of buoyant fortes compared to inertial
forces played a significant role in the rate of convergence.
Highly buoyant effluents (low F ) converged much slower than
pure inertial flows.
Runs made with constant eddy diffusivities converged much
more rapidly than those runs using variable coefficients.
One inner iteration (stream function elliptic equation) was
sufficient.  Increasing the number of inner iterations
served to aggravate the convergence rate.
Deeply stratified cases (as opposed to surface layer strati-
fication) significantly aggravated the convergence rate.
This item is discussed further in Chapter 8.
Neglecting derivaties of EZ in the transport equations led
to r-balance errors on the order of 10-20% where variable e
was employed.
Beginning a solution from an irrotational flow solution as
opposed to zero velocity everywhere, appeared to have no
particular advantage, and in some instances tested, actually
slowed convergence.
                    306

-------
      Most computer runs were initialized from a restart tape gener-
ated by a previous case.  There was considerable economy in this
action since a solution would need to be started from a zero initial
velocity distribution (or irrotational distribution) only when the
grid layout was changed.  Unfortunately, from another aspect however,
not many solutions beginning at iteration number one and ending at
convergence are available for comparison.  To illustrate the conver-
gence behavior, some of the computational aspects will be compared for
identical grid layouts.  'This information is displayed in Tables 7.7,
7.8 and 7.9 for grid layouts of (JxK) 40x33, 31x34, and 26x25,
respectively.
      The four cases cited in Table 7.7 constitute the worst lot as
far as convergence lethargy is concerned.  The starting run was the
momentum jet case which took 800 iteration cycles to converge properly.
All succeeding cases used the momentum jet solution as initializing
information.  Of these succeeding cases, the run for F  = 1 (buoyancy
dominated) was the most reluctant to converge.  Convergence lethargy
in this lot is laid chiefly to grid size although there is some
suspicion that cell aspect ratio and position of the inflow-outflow
boundary also have some effect.  Figure 7.79 shows the convergence
history of Y, A-J and a for the 40x30 grid layout.  This illustrates
the behavior of &    for these variables where again the momentum jet
                 max
is used as a starting solution (first 800 iterations) for the succeed-
ing runs FQ = 46 and FQ = 1.  As noted previously, 6max is the maxi-
mum relative change in the entire system and does not always occur
                           307

-------
                                    TABLE 7.7  CONVERGENCE BEHAVIOR, 40x33 GRID
CO
o
OD

A 5
A Z
Fo
Start Variable
e and £7, Iterations
Total Iterations
Incremental Iterations
r - Balance Error
6 Stream Function
at Node
'max Vorticity
at Node
&m*v Buoyancy Parameter
fflclX
at Node
Starting
Case
.12591
2.0
00

75
800
800
- .1281
1.1 95x1 O"4
37,10
7.221xlO'3
32,6
3.210xlO~3
37,2
Succeeding Cases
A
.12591
2.0
46*

801
1100
300
.734
6.571xlO"5
20,37
3.191xlO"3
26,5
7. 380x1 O"4
2,33
B
.12591
2.0
1.0*

801
1400
600
- .4641
2.92xlO~5
14,32
1.466xlO~3
37,12
3. 345x1 O"4
2,33
C
.12591
2.0
1000*

801
1200
400
- .9133
6.989xlO"5
37,9
9.422xlO"3
32,6
1. 508x1 O"3
37,2
           indicates variable changed in restart case.

-------
    10
X
*t
   -1
   -2
   -3
   -4
   -5
              MOMENTUM JET

              STARTING SOLUTION
                         500
1000
1500
                                ITERATIONS

          Figure 7.79.   Convergence Behavior, 40x33 Grid
                          309

-------
at the same cell.   Figure 7.80 illustrates  the convergence history of
the starting solution for V at nodes  (2,20) and (2,30)  and r at node
(10,33).
      Table 7.8 illustrates similar data  for a 31x34 grid  layout.
Convergence in this lot is rather slow also.  Note that the values for
6 _w are considerably larger in this  lot  than in the lot given in
 max
Table 7.7, although the r-balance error is  about the same.  The explan-
ation is that &„ gives a relative change, and these changes are
               max
occurring where the absolute value of the quantity is very small.   For
instance, the maximum relative change of  vorticity in the  starting
case is .1595, whereas the value of vorticity at this point is
-9.76x10   ( the maximum value in the flow  field is 2.944).  Figure
7.81 shows convergence history of selected  data.
      Table 7.9 illustrates the convergence characteristics for the
26x25 grid.  Note that each case is not finely converged with respect
to  6may, whereas the r-balance error  is less than 1% in all cases.
Thus, this table illustrates that the system may be reasonably well
converged with regard to absolute quantities although relative changes
in  part of the system may be comparatively  large.  Also, only 150
iterations were required to obtain each solution using the starting
run initialization.
      For this case, the primary concern  was plume centerline condi-
tions.  Changes of velocity and temperature were occurring only in
the fourth and fifth significant figures  along the centerline, indi-
cating  that computation time may be saved by using a regional
                           310

-------
.4
.3
.2
             START MIXING LENGTH  CALC
                                     V(2,10)
                                     V(2,30V
                                     r(10,35)
                      500               1000

                           ITERATIONS
   Figure 7.80.   Convergence  History of V and r  at Selected Cells,
                Momentum Jet, 40x33 Grid

-------
                                     TABLE 7.8  CONVERGENCE BEHAVIOR, 31x34 GRID
co
ro

AS
AZ
Fo
Start Variable
e and £7 Iterations
Total Iterations
Incremental Iterations
r -Balance Error
 Stream Function
II id A
at Node
W Vorticity
at Node
6 , Buoyancy Parameter
at Node
Starting
Case
.14690
.2
51
150

600
600
- .4381
1.350 x 10~3
28,25
1.595 x 10~]
23,24
2.918 x 10"2
20,5
Succeeding Cases
A
.14690
.175*
105*
601

900
300
.2778
1.115 x 10"3
26,12
1.455 x 10"1
23,23
4.126 x 10"2
30,13















                          *Indicate changed variable

-------
oo
CO




Ui?
1 0

on



n_c\

OA


O.^i
A













•

a
•
0




rf
e
0
o
*


*



* (
a
« a
a
X
o
(




ff°

^
f
Jt
«




/•
a
i •
*
;
*
i




o <

X i






1 •


* '














• 1










' * :






• i








O i

•M :





' U I
1 • '




c
^
2
L
n















ii
'X
^
3^-
D^
•K
"•*

> i









\

)

./
j
W

/
/




















a

•







































































































































, « J
	 1

















	
L X .
1 	 	 	






















m ,









NODE
(2,11) U
. (2,21) V



/"p 31) v
l^->01/ »
^2 34) A
(£.tj"+j z\|



(2 1,34) A,


                           40      SO     120     160     200     240     280    320    360     400
                                                      ITERATION  NO.
                                      Figure 7.81.   Convergence History of U,  V  and
                                                    at Selected Cells 31x34 Grid

-------
                                    TABLE 7.9  CONVERGENCE! BEHAVIOR, 26x25 GRID

A?
AZ
Fo
Start Variables and e Iterations
Total Iterations
Incremental Iterations
r -Balance Error
<5m,v» Stream Function
Hi a X
at Node
W Vortldty
at Node
6 , Buoyancy Parmater
at Node
Starting
Case
.14690
.50
1.0
100
400
400
.8035
1. 644x1 O"4
(24,9)
1. 776x1 O"3
(20,14)
1. 976x1 O"3
(24,6)
Succeeding Cases
A
.14690
.50
5.0*
400
550
150
.3700
l.OllxlO"3
(24,10)
9.350"2
(20,20)
4. 683x1 O"2
(21,20)
B
.12591*
.50
25.0*
600
600
150
- .9936
2.135xlO"3
(24,12)
5. 345x1 O"2
(22,20)
4.850xlO"2
(24,3)
C
.14690*
.50
100*
800
750
150
- .3203
2t210xlO"3
(24,11)
5.435X10"1
(22,17)
3.440xlO"2
(24,5)
GO
-f=>
        indicates  changed variable.

-------
convergence criterion.  In the computer program one has some control
over this criterion by applying the convergence check only out to a
set radius.
      As a final illustration of numerical convergence behavior,
Figure 7.82 shows the iteration history of V and A, at one cell for
Case 2 (see Table 7.1).  The significance of this plot is that the
velocity initialization is the irrotational flow solution (for the
other cases cited, U and V are zero everywhere except the inflow
boundary).  Note that velocity V shows considerable oscillation.
      The theoretical development of difference equations in this
manuscript is based on Equations (5.8) for vorticity, Equations (5.9)
for the transport of buoyancy and equations similar to (5.9) for the
transport of materials.  These equations make no allowance for contri-
butions, or more accurately, corrections, issuing from variable eddy dif-
fusivities.   In  the instance of Equation (5.9),  these'corrections may be
made  rather  straightforwardly  by adding  the terms
            9AlV3evr
                  9r
However, in Equation (5.8) the appropriate correction terms add con
siderable complication as noted by comparing Equations (3.80) and
(3.81).  Fortunately not all of the terms involving derivatives of
and e  need to be incorporated into the numerical model either
because they are zero in accordance with assumptions concerning the
eddy coefficient model, or transported quantities are minute where
                            315

-------
CO
<4
3,


2.




C
£*»
k
• *
'•
:
-
•
• i
y
• V
> 1C
!*•
• •
* * * *
* • • **
• • t
• * ,
• v
\j *


X> 2(


t~'-- ,—^ ^,
J
X
K



)0 3


^^>^_J^t 	 . 	 ^^r-






00 4(


^V

A,




DO 5









00 6C
                                                ITERATION  CYCLES
                           Figure 7.82.  Iteration History for One Cell of Case 2

-------
the variations occur.   For instance, we may neglect all  terms involv-
ing 3er/3r since er is constant where diffusion is important., and
convective terms dominate the transport where step changes  in e
occur.  Likewise, other order-of-magnitude approximations may be made.
Having eliminated these second order factors one is left with the cor-
rection terms for vorticity of:

             1    „ 3FZ      3 U  ,   3U    3 FZ
            RF    *•  a7        2     a?       ?
            Kt7      dL      a7      9Z     s7
              L              Of-             OL
where FZ is again the vertical  eddy diffusion multiplier (cf.  Section
7.2).
      Similar approximation for Equation (5.9) yields the correction
term,
              1       3FZ   .   9A1
            REzPRz   3Z       3Z
with a similar correction for r transport.
      The importance of these terms was ascertained by the system r-
balance.  Without the corrections, the r-balance error ran as high as
20%.  For the same conditions, addition of the correction terms
reduced the error to less than 1%.
      Before closing the subject of convergences the author wishes to
note that in all cases run where the transport equations were decel-
erated and turbulence modeling did not lead to numerical instability,
the stream function convergence was extremely well behaved.  This
behavior was obtained by iterating only once on the * elliptic equa-
tion; additional iterations were noted to aggravate the convergence of
                             317

-------
¥ as well as the transported quantities n,  A,  and r.   It is entirely
possible that the system would have converged  equally well  if even
fewer ¥ iterations were performed,  that is, iteration on v  only once
for every two, three or perhaps five outer  iterations.   This facet
was not investigated in the present study,  but such experimentation
could yield fruitful results in terms of computer time.
                            318

-------
                             CHAPTER 8
           NUMERICAL EXPERIMENTS FOR SHALLOW WATER CASES
      Material presented in this chapter deals with application of the
numerical techniques discussed earlier in shallow water situations.
All computer runs presented here are for cases where the assumed water
depth is ten or less port diameters above the outfall discharge.  The
techniques used are identical to those applied for the verification
studies presented in the previous chapter.
      Unlike cases in Chapter 7, however, applicable data are not
available except for one case where surface temperature field data are
available.  Hence, we rely substantially on the verification study
as an indicator of the validity of the computational techniques.
Table 8.1 summarizes the cases to be discussed and illustrated in this
chapter; those listed are only a portion of the total shallow water
computer runs made during the course of the present research.  None-
theless, these cases are typical and space limitations preclude further
illustrations.
8.1  Modeling the Vertical  Eddy Diffusivity Multiplier, FZ
      In the region of the lateral surface spread of shallow water
plumes, modeling the vertical component of the pointwise eddy diffus-
ivity plays an important role in determining the flow behavior.  Con-
siderable effort was devoted to this subject in Section 7.2.2; the
computational methods used to obtain results presented in this chapter
will be briefly reviewed.
                           319

-------
CO
r^o
o
Case
No.
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
Grid
Size
26x25
26x25
26x25
26x25
26x25
26x25
26x25
26x25
31x34
31x34
31x34
31x34
30x26
30x26
30x26
30x26
30x26
30x26
29x20
A£
.14690
.14690
.14690
.14690
.12591
.12591
.12591
.12591
.14690
.14690
.14690
.14690
.14690
.14690
.14690
.14690
.14690
.14690
.12591
AZ
.5
.5
.5
.5
.25
.25
.25
.25
.20
.20
.175
.175
.40
.40
.40
.40
.40
.40
.15
TABLE 8.1
Depth R
2 oo
10
10
10
10
5
5
5
5
5.6
5.6
4.97
4.97
8.4
8.4
8.4
8.4
8.4
8.4
1.42
19.66
19.66
19.66
19.66
11.62
11.62
11.62
11.62
41.00
41.00
41.00
41.00
35.40
35.40
35.40
35.40
35.40
35.40
16.97
SUMMARY OF
F D
o ft
100
25
5
1
100
25
5
1
51
51
105
105
45
45
45
45
45
45
.111
10
10
10
10
10
10
10
10
14
14
16
16
10
10
10
10
10
10
21
SHALLOW WATER CASES
vo Boundary o
ft/ sec Type* °C
10.15
5.075
2.270
1.015
10.15
5.075
2.270
1.015
7.25
7.25
11.10
11.10
7.00
7.00
7.00
7.00
7.00
7.00
.574
2,N=10
2,N=10
2,N=10
2,N=10
2,N=10
2,N=10
2,N=10
2,N=10
2,N=7
2,N=7
2,N=7
2,N=7
4
4
4
4
4
4
2,N=10
18.12
18.12
18.12
18.12
18.12
18.12
18.12
18.12
8.33
8.33
8.33
8.33
11.10
11.10
11.10
11.10
11.10
11.10
13.80
Gtr
(amb)
27.98
27.98
27.98
27.98
27.98
27.98
27.98
27.98
18.01
18.01
18.01
18.01
27.78
24.78
24.78
24.78
24.78
24.78
25.57
Stratifi-
cation**
None
None
None
None
None
None
None
None
None
None
None
None
None
2 °C
3 oc
4 °C
5 °C
5 oc
None
      *  See  Table 7.1

    **  Stratification extends somewhat deeper than in Case 63.

-------
      The general form of the vertical diffusion coefficient is
            e= e,         + e                                (8.1)
                   ambient      plume
In the region of plume flow the ambient contribution will  be insigni-
ficant; hence,
            EZ^EZ      .                                      (8.2)
                   plume
The plume generated turbulence is a function of both mean  flow char-
acter and thermal character.
      Recall from Section 7.2.2,
e
                   o
                                                               (8.3)
where ez  is the vertical diffusion coefficient for neutrally buoyant
        o
conditions and f is a function of the point Richardson number, RI.
Likewise,  the vertical component multiplier, FZ, may be expressed as
            FZ • FZQ f(RI),                                    (8.4)
where FZ  is the neutral buoyancy multiplier.  The model for FZQ may
be expressed as (See Section 7.2.2)
                                   •                           (8-5)
where Z1/2 is the radial plume half-depth, and the radial velocity
difference,
             max" mi n "  max '
 since Um1n - 0 .  Then
            Fzo ' Zl/2 Umax'                                   (8'6>
                            321

-------
If we followed the same method used for computing the radial
multiplier,

            FR = Rl/2                                          (8'7)
Zi, /0 and V    would be computed iteratively and pointwise to establish
  i / c.      max
FZQ (Method 1, Section 7.2.2).   However, all  attempts to calculate
FZ  based on local  values of I-, /9 and IJ    led to numerical  instability.
  0                           B / u      ffloX
Exactly why this condition persisted, especially in view of excellent
success with Equation (8.7), was  never ascertained.  After several
numerical experiments and correctional efforts without success, it
was decided to stabilize the computation by restricting the computed
value of the plume half-depth,  Zy2» since this value seemed to ex-
hibit the most unstable character in previous experiments.  This
decision led to Methods 2S 3 and  4, described in Section 7.2.2.
      Method 2 used Z. .  based  on R-, .^ computed at the elevation of
lateral flow.  Z, ,„ was then held constant for that iteration but the
local value of Umax was used. This method led to eddy diffusion coef-
ficients which were quite large and a correspondingly unrealistic
flow field; hence, the method was quickly abandoned.
      Method 3 computed a constant value of FZQ to be applied every-
where in the lateral plume spread.  The value of FZQ in this method
was set by computing Z    at the radial position corresponding to the
maximum radial velocity; the value of Umax at this point along with
Zy2 was used 1>n Equation (8.6).   Results from this method are pre-
sented in Section 8.4.  Again, this method yielded diffusivities
of excessive magnitude.
                             322

-------
      However, experimentation with this method was carried out, in
every case, in conjunction with thermal flows (as opposed to neutrally
buoyant conditions).  It is possible that the Richardson number
modifier, f(RI), was inaccurate.
      Finally, the most realistic results were obtained by Method 4
which in principle uses the technique of computing Z,/2 of Method 3,
but bases U ,v on the local value.  This method was found to be always
           iTIaX
stable once the general, but approximate, flow patterns were
established.
      Table 7.5 summarizes several models for f(RI); however, the
Mamayev correlation has been employed exclusively in this work which
has the form,
            f(RI) = e"8RI,                                     (8.8)
where 6 is an empirical constant.
      The value of 3 to be used presents an additional uncertainty in
computing FZ.  Wada [108] used the value  @ = .8.  However, Stolzen-
bach [94], based on the data of Ellison and Turner [25], suggests the
value  6 y 5.0 when using the gross Richardson number.  We have used
values ranging from .4 to 2.0 in this work (Table 8.2).
      Table 8.2 below summarizes the computation of FZ for results
presented in this chapter.
                            323

-------
                 TABLE 8.2   SUMMARY  OF  FZ COMPUTATION

            Case                Method
            48-55                  4                  1.0
             56                    3                   .4
             57                    3                   .8
             58                    3                  1.0
             59                    4                  1.0
            60-65                  4                   .8
             66                    4                  2.0
      Actual  computation of FZ proceeds  as follows:   (sequence of
operations for one outer iteration).
      Based on the computed values of Us V and A-|  for the present
iteration:
      •  Compute the array of local Richardson numbers,
                                          ,„ 2

                                              J,K
      9  Scan the U array to establish the maximum value of U and the
         corresponding index, J.
      ®  Compute the plume half -depth, Z.  .   at index J.
      ®  Use this value of Z, ._ to compute
         where Umax takes on different values at each radial grid
         point.
         Compute
            FZ(jsK) = FZJJ.K) e' eRI(J'K)
         Use the above value of FZ in computing transported quantities
         for the next iteration.
                              324

-------
8.2  Results for Homogeneous Receiving Water 10 Port Diameters Deep
      Results for plumes issuing in homogeneous receiving water at a
depth of ten port diameters are reported as Cases 48 through 51.   For
these cases, the initial temperature excess is 18.12 °C and the value
of sigma-t for both the effluent and reference ambient is 27.98.
Each case represents a different densimetric Froude number as indi-
cated in Table 8.1.  Changes in the Froude number were effected by
varying the effluent velocity.  All initial velocity profiles are
assumed to be turbulent and follow a profile given by Equation (7.17)
with the exponent equal to 1/10.  The port  diameter is held constant
at 10 feet and the lateral spread is computed out to about 10 port
diameters.
      Centerline distributions of velocity and temperature excess are
illustrated in Figure 8.1 for all four cases.  Note that the plume
accelerates for low Froude numbers (FQ = 1,5), but for Froude numbers
of 25 and above, very little acceleration is noted  even though tur-
bulent mixing (as a function of distance from the port) is decreased
(temperature excess curves).  Velocity of the lateral surface spread
is illustrated in Figure 8.2 for these same cases.  Maximum velocity
in each case occurs at radial distance between 1.5 and 2.0 diameters.
In the highly buoyant Case 51, the maximum lateral velocity is nearly
as great as the initial velocity.  Note that these results are norm-
alized to the average effluent velocity; hence, for FQ = 1 the maximum
lateral velocity is about 1 fps, whereas for FQ = 1, the corresponding
velocity is about 3.5 fps.  Vertical profiles of lateral velocity for
                             325

-------
 3.0
 1 .0
 0.5  -
0. 15
                         AXIAL  DISTANCE,  Z
    Figure 8.1.   Computed Center!ine Velocity and Temperature
                 Excess for Intermediate Depth, Cases 48
                 Through 51 (10 diameters deep)
                          326

-------
co
ro
       o
<:
t—I
Q
<:
o:



GO
       o
       *—t
       CO
           1.0  U
             .8  \-
     .6  U
             .4 U
             .2
                                            3456

                                              RADIAL DISTANCE,  r/D
                            Figure 8.2.  Surface Distribution of Radial Velocity,

                                        Cases 48 Through  51 (see Table 8.1)

-------
Case 50 are illustrated in Figure 8.3.
      Comparison of the radial  velocity profiles are illustrated by
Figures 8.4 and 8.5.  Figure 8.4 is for a radial position of r/D=1.9,
which corresponds approximately to the position of maximum velocity
in all four cases.  This figure also illustrates that radial entrain-
ment occurs below the depth of about 10.5 (1.5 diameters from the
surface) for these cases.   At 7.32 diameters (Figure 8.5) the spread-
ing surface layer is slightly thinner.
      The small cross-hatched rectangle shown in Figure 8.5 illustrates
the variation of the spreading depth for these cases.  Greater penetra?
tion is noted at Froude number 100; FQ = 1  shows the least penetration.
      The distributions of temperature excess at the surface
(ATS/ATQ) are illustrated  by Figure 8.6.  Vertical profiles of excess
temperature (iTs/ATQ) for  Cases 48 and 50 are shown in Figure 8.7
(A and B).  Note that the  temperature profiles penetrate slightly
deeper than the velocity profiles and indicate some minor recircula-
tion of the heated water takes place.
      A complete set of contour plots and three-dimensional illustra-
tions for the stream lines, temperature and vorticity for Cases 48
through 51 are given in Figures 8.8 through 8.32.
                             328

-------
RS
DIA
i— «
rss
i— «
o
IA

oo
rw

i   '
§
o
                «—»
              SCALE: U =.5
                    SURFACE
           ''V  Y        V\
           ~~i—r	T—I
0
                        RADIAL
                        VELOCITY
                        PROFILES
                 REG I ON OF
                 NEGATIVEU
          •DISCHARGE PORT
             ro - D/2
            l


    468

RADIAL POSITION, r/D
FREE BOUNDARY
r
                                     10
       Figure 8.3. Distributions of Radial Velocity
               Case 50
                329

-------
 12
 11
 10
c/l
ee.
H-•

Q




rw 8
  7
o
a.
>*  c
<  6
       ELEVATION

       •   OF

         PORT
                                RADIAL POSITION rr/D'v,!. 9



                                    l    I    I    i    I    t
    -0.1               0                    0.5

              DIMENSIONLESS-RADIAL VELOCITY,  U
1.0
Figure 8.4.   Maximum Radial  Velocity Profiles, Cases 48 Through 51
                           330

-------
  12
  11  —
£10
UJ
H4
   9  -
   8  —
CO

o
Q_
   7
X
   6  -
   5  —
  4  -
  -0.3           0                    0.5

         DIMENSIONLESS  RADIAL  VELOCITY, U


     Figure 8.5.  Radial  Velocity Profiles at r/D=7.32,
                 Cases 48 Through 51
1.0
                      331

-------
CO
CO
ro
                                           3456

                                              RADIAL DISTANCE,  r/D
                            Figure 8.6.  Surface Temperature Excess Distribution
                                         Cases 48 Through 51 (See Table 8.1)

-------
CO
CO
CO
                                                 .1
                                               4T/4T
FIGURE  8.7 .   VERTICAL TEHPERATURE EXCESS DISTRIBUTIONS
              FOR  VARIOUS  RADIAL POSITIONS.  CASES 48 >  SO.
                                                                                                                                         RflOlRL OIRECTIBN. R/d

                                                                                                             FIGURE  8.8-   STRERHLINES FBR COSE 48 - 8U8TBNT DISCHARGE.  FO = 100
                                                 RBDIBL DIRECT1BH. R/D

                     FIOURE  8.9.   1S3THERHS  FBR CPSE 48 - BUBrnNT  DISCHARGE.   FO = 100
                                                                                                                                                         rte
                                                                                                                                        RBDIBL OIRECTIBN. R/l)
                                                                     FieORE 8.10.  VBRTIC1TT  LEVEL LIMES FBR CBSE  48  -  BUBYHNT DISCHflRGE. FO =

-------
OJ
GO
                  FIBURE8.il.  3D  ILLUSTRRTI8N  8F  STRERH  FUNCTiaK  —  PSl .    CBSE MB. 48



                               INTERtlEOIRTE HRTER  BUTFHLL.  3URFRCE 10 DIHHETERS  HBBVE  P8RT.  F0=100
FIGURE 8.12.  30 ILLUSTRHT18N BF rEHPERBTURE FIELD —4T.    CBSE NB. 48



             INTERHE01HTE HRTER 8UTFHLL. 8URFBCE 10 DIBHETERS BBBVE PBRT.  FD=100
                FIOURE 8.13.  30 1LLU3IRRTI8N ttf FLUID  VORTICITT  -  BflEOB.    CR8C NB.4*



                             INTCRHEOIRTE MBTE* BUTFRLL. 8URFBCE  10 D1R«TER8 M8VE r«KT. F0=|00
                                 te	rte	rte	rte~
                            RROIBL OIRECTI8N. R/0

  FIOURE 8.H.   STKBI1LIIIES FBR CBSE  49 - BUBTflNT OISCHRROE. FO = 25

-------
co
CO
171
                     FIGURE 8.15.   ISBTHERflS   FOR CRSE 49 - BUBYRNT OISCHRRGE. FO = 25
                                                                                                                                          RRD1RL DIRECTION. R/D

                                                                                                               FIGURE 8.16.  V6RTICITY  LEVEL LINES F6R CflSE 49  -  8UOf»NT  OISCHHRGE.  FO  =  25
                  FIGURE  8.17.   30  ILLUSTRRTIBN BF STREBM FUNCTIBN — PS1 .    CASE H8.  49


                                INTERHEDIRTE NRIER BUTFRLL.  SURFHCE 10 DIMETERS BBBVE PBRT. FO
FIOURE  8.11.  3D ILLUSTRBTIUM OF STRERd FUNCTIBN -- PSI.    CRSE NO. 49


             INTERdEOIflTE URTER OUTFRLL.  SUPFflCK 10 OIRnETERS RBBVE PORT. FO =25

-------
co
oo
en
                   FIGURE 8.19.  3D  ILLUSTRBT18N BF TEHPERHTURE FIELD —»T.    CRSE NB.49



                                INTERtlEOlfiTE  MRTER BUTFRLL. SURFACE 10   DIMETERS  RBBVE PORT. FO =Z5
FIGURE B.20.  3D ILLUSTRRT1BN BF FLUID  VORTICITY  - BHEGB.    CRSE MB. 49



             IHTERHED1BTE MRTER BUTFBLL.  SURFHCE 10 DIRflETERS HBBVE PBRT. FO =ZS


  g
                                                  RH01RL OIRECT1BN.  R/D


                       FIGURE  8.21. STRERHLINES  FOR  CRSE  SO- BU8VRNT D1SCHRRGE. FO - S
                               RRDIRL DIRECT IBM. R/D


    FIGURE B.22.  ISOTHERMS  FOR CRSE 50- BUBVRNT OI-SCHRRGE.  FO = 5

-------
CO
Co
                                                             RfiOIRL  DIRECTION.  R/0

                                                           LEVEL LINES FOR  CASE SO- BUflVflNT  OISCMHRGE. FO  =  5
FIGURE s.24.   30 ILLUSTRATION BF STRERH FUNCTION -- PSI-    CASE  NQ.SO


                 IffTEKDlATC HATCH OUTFAU. SUtFACE 10 DIJWTERS WOVE PORT. FO •  5
                                              FIGURE I.ZS. 30 ILUfSTWTIOd OF TDTCMTUK FIELD - AT.   CASC K.  50

                                                   INTEIKOIATC WTU OUTFAU. SUtfACE 10 OMC7EH MOV KIT. FO - S
_              _

lOUffC •.».   30 ILLUSTRATION  0F FLU 10 m


                        e HRTCR OUTFRLL.
                                                 - 0HEGfl.

                                                 t 10  oin
 CASE Hfl. 50


rCRS ABflve rasr, FO  =  s

-------
CO
co
oo
                     RfiDIflL DIRECTION. R/D

:e.27.   sTRERniiNES  FOR  CRSE  51  -  BUOYRNT OISCHHRGE.  FO = i
                                                                                                                                       RROIRL DIRECTION. R/D


                                                                                                             FIGURE 8.28.   ISOTHERfIS   FOR  CHSE  51  -  BUOYRNT  OISCHflRGE.  FO = 1
                                            TTfeo         j.tw
                                                 RflOIflL  DIRECTION,  R/D


                      FIGURE  8-M.   VORTICITY  LEVEL LINES FOR CRSE SI - BUeTRNT DISCHRRGE. FO = 1
                                                                              •"IOURE 8.30.  30 ILLUSTRRTIBN 8F STRERH FUNCTION -- P3I .    CRSE NO. 51



                                                                                           INTERHEDIRTE HBTER 8UTFSLL. SURFRCE 10 OIBHETERS RB8VE P8RT,  FO  =  1

-------
FIGURE 8.31.  30 ILLUSTRRTI0N 3F TEMPERRTURE FIELD  --4T.    CRSE  NO. 51




             1NTERMEDIRTE HRTER 0UTFRLL.  SURFRCE  10  OIRMETERS  RS0VE  PQRT.  FO = 1
FIGURE 8.32.   30  ILLUSTRRTI0N 0F FLUID VORTICITY    0MEGR.    CRSE NO. 51




              INTERMEOIRTE  WRTER 0UTFRLL.  SURFRCE 10 DIRMETERS RB0VE PORT, FO  -  1
                                 339

-------
8.3  Results for Homogeneous Receiving Water 5 Port Diameters Deep
      Results for outfalls issuing to receiving water 5 port diameters
deep are given as Cases 52 through 55 for Froude numbers of 100, 25,
5 and 1, respectively (See Table 8.1).  All  boundary conditions and
parameters for these cases correspond to those- of similar Froude
numbers for the 10 diameter deep cases given in Section 8.2.  Actual
water depth here is 6 diameters with the outfall port rising one diam-
eter above the bottom.
      Center!ine distributions of velocity and temperature excess are
shown in Figure 8.33 for Cases 52 through 55.   As was illustrated by
Case 50 and 51, the plume also accelerates for Cases 54 and 55 as a
result of dominant buoyant forces.   For Froude numbers of 25 and
above the centerline velocity remain essentially constant until sur-
face effects are encountered.  On comparing Figure 8.1 with 8.33, one
notes that at 5 diameters the temperature excess given in Figure 8.33
is slightly higher than for corresponding cases given in Figure 8.1.
The decreased dilution is a result of the surface proximity.
      The vertical distribution of radial velocity, U, is illustrated
by Figure 8.34 for Case 52.   The lateral spread is seen to be quite
thin (approximately .8 D) at least out to 4 diameters.  Figure 8.35
shows that temperature effects somewhat deeper (approximately 1.2 D)
and some recirculation of heated water is indicated.  At r/D = 1.0,
the temperature distribution lies within the rising portion of the
plume above 2 ^ 2.5 (1.5 above the port) and is not to be interpreted
as penetration of the lateral spread.

                              340

-------
co
-p.
             1 .0
               5   -
                                                     .5             1.0


                                               Axial Distance,  Z (Diameters)
           Figure 8.33.  Computed  Centerline Dimensionless Velocity and Temperature Excess for Shallow

                         Water Cases  52  Through 55 (5 Diameters Deep)

-------
CO
-p»
ro
                                                       r/0
                                                     o 1 .08

                                                     a 2.10

                                                     A 3.97
                                                     I
                                                           l
I
                 -0.4        0           0.4       0.8         1.2

                               Dimension!ess Radial Velocity,  U

             Figure 8.34.  Vertical Distribution of Radial  Velocity at Various
                          Radial Positions, Case 52
                                                                                     ^  4
                                                                                              AMBIENT,
                                                                      r/D
                                                                    o 1 .00

                                                                    D 2.24

                                                                    * 4.23
                                             TEMPERATURE,  T,  C
                                                                                                                                                   10
                    Figure 8.35.  Vertical Distribution of Temperature Excess  at Various
                                 Radial  Positions, Case 52

-------
      Contour plots and 3-dimensional  illustrations  of  the  stream
function, temperature and vorticity are given in  Figures  8.36
through 8.41.
                           343

-------
CO
-p.
                                            RBOlflL DIRECTION. R/0
                 FIGURE B.36.  STREflHLINES F9R CflSE 52 - BUBYflNT DISCHARGE. FO = 1
                              RBOIHL DIRECTION. R/0
          8.37.   ISOTHERMS  F8R CBSE 52 - BUBYRNT DISCHHRGE. FO  =
                                           RROIHL DIRECT IBM.  R/0
                FIGURE 8.38.   V8RTICITY   LEVEL LINES F8R CRSE 62 - BU6TRNT DISCHARGE. FO = 1
FIGURE B.JS.  30 ILLUSTRBT1BN OF STRERH FUNCTIBN — fSl.    CHSE HO. 55
             VERT SHHLLBH HBTER BUTFBLL. SURFBCE S OIBHETERS BBBVE PORT. F0=  1

-------
FIGURE 6.40.   30  ILLUSTRflTION OF TEMPERRTURE FIELD --4T.    CflSE NO.  55




              VERY  SHRLLOH HRTER OUTFflLL.  SURFRCE 5 DIRHETERS RBOVE  PORT.  F0=  1
FIGURES.Hi.  3D ILLUSTRHTION OF FLUID  VORTICITY    OflEGH.    CfiSE NQ. 55




             VERY SHHL1.3W WHTER 3UTFHI.L.  SURFRCE S OIRMETERS HBQVE  PORT. F0=  1
                               345

-------
8.4  Results for Two Different Methods of Computing FZ
      Cases 56 through 59 are results illustrating the effects of
using Methods 3 and 4, and different values of the constant 3, for
computing the vertical eddy diffusivity multiplier, FZ (refer to
Tables 8.1 and 8.2).  Cases 56 and 57 are for receiving water 5.6
diameters deep, using Method 3 to compute FZ with Froude number,
F  =51.  Case 58 has FQ = 105, with 4.97 diameter deep water using
Method 3.  Case 59 is the same as Case 58 except Method 4 is used to
compute  FZ.
      Cases 56 and 57 were run to observe the effect of changing
6 =  .4 to 3 = .8, respectively.  Comparative results are not shown,
but  this change of 3 did not alter the computed velocity and tempera-
ture profile a great deal.
      It was observed, however, that computation of FZ  by Method 3
resulted in excessive vertical diffusivities.  Case 58 also employed
Method 3 and exhibited excessive vertical diffusivities (in this
Case 6 = 1.0).  As pointed out in Section 8.2, Stolzenbach suggests
the  value of 3 = 5.0 based on the gross Richardson number; however,
values using 3 > 1.0 were not tried in these cases. Using the larger
value of 3 could have a major effect on the velocity and thermal dis-
tributions computed by the present techniques using Method 3.  The
use  of large 3 would significantly reduce vertical mixing in the
thermal  boundary region, but allow substantial vertical exchange
within the spreading plume where thermal gradients are expected to be
small.
                           346

-------
      Figure 8.42 shows the comparison of surface spread  velocity
between Cases 58 and 59.   The difference here is not of major impor-
tance, but Figures 8.43 and 8.44 illustrate a significant difference
in vertical  entrainment.   Significant differences between stream-
line patterns is revealed by comparing Figures 8.45 and 8.46.  The
contours shown in Figure 8.45 (Case 58) are more indicative of
creeping flow in the spreading portion of the plume than  a high
Reynolds number flow (Case 59, Figure 8.46).
      The distribution of surface temperature excess is shown in
Figure 8.47 for Cases 57, 58 and 59.  Case 57 shows lower temperature
at the centerline as a result of the port being in deeper water.
Case 58 may be compared to Case 59 and exhibits a lower surface temp-
erature (also, refer to Figures 8.48 and 8.49).  This result is due
to the larger values of vertical mixing employed in the computation
of Case 58.  Isotherms for Case 59 are illustrated by Figure 8.50.
                            347

-------
CO
-p>
CO
                                            5                        10

                                                 RADIAL POSITION, r /D
15
                       Figure  8.42.   Computed Radial Velocity at Surface, Cases  58  and 59

-------
CO
               ?.84
                    -0.04
                                                       CASE  58
                                           RADIAL POSITION:r/D=9.83


                                        I	1	I	I
 0.05       0.1        0.15

DIHENSIONLESS RADIAL VELOCITY,  U
                                                                        0.2
                    Figure 8.43.  Ve-tical Distribution of Radial Velocity. U.
                                 Cases 58 and 59.
                                                                                0.24
                                                                            RADIAL POSITION:r/D=13.19

                                                                                         I           I
          0         0.05       0.1       0.15

             DIMENSIONLESS RADIAL VELOCITY, U

Figure 8.44.  Vertical Distribution of Radial Velocity,
             Case 58 and 59
                                                                                                                                                   0.20
                                                                                                                                                  U.

-------
CO
en
O
                                                                                10.00
                                                              RROIRL DIRECTION. R/D


                   FIGURE 8.15. STRERMLINES FOR  RN  RXI SYMMETRIC.  VERTICPL PLUME. CONFINED BY R FREE  SURFRCc.

                                CASE 58.
                                                                                                              15.00

-------
oo
en
                                                      s-bo
                                                                                     IQ.OO
                                                                   RRD1RL DIRECTION.  R/0
                                                                                                                     1S.OD
                        FIGURE 8. 46.  STRERMLINES  FOR  RN  RXISYMMETRIC , VERTICRL  PLUME.  CONFINED BY R FREE SURFRCE .

                                     CASE 59.

-------
CO
U1
ro
                   5  -
                   4   -
                   3   -
                   2
                   *-   ••••
               o.
               LU  1
               (J  I
               «t
               u_
               o:
               3
               01
                         Figure  8.47.
                 RADIAL POSITION  r/D



Surface  Temperature  Excess, ATS> Cases 57, 58, and  59

-------
CO
en
CO
                                      RADIAL POSITION:r/D=9.13
                                                                             2.5
                                                                                                                RADIAL POSITION:r/DH4.2
                                                                                                                                                     0.25
                       Figure 8.48.   Vertical Temperature Excess Distribution.
                                     Cases 58 and 59
Figure 8.49.  Vertical Temperature Excess Distribution.
             Cases  58 and 59

-------
co
en
                                                                                  10-00

                                                           RRDIIU DIRECTION.  R/D


                FIGURE 8.50.  ISOTHERMS    F0R RN RX[SYMMETRIC. VERTICRL FLUME.  CONFINED  BY  R  FREE SURFRCE.

                              CASE 59.

-------
8.5  Numerical Experiments Involving Ambient Stratification
      Results involving the effects of stratification are given by
Cases 60 through 65.  Case 60 is a base case to be used for comparison
and is for a homogeneous ambient.  The remaining cases have different
degrees of ambient stratification.  In all cases the ambient (also,
effluent) salinity is constant at 35 ppts hence the ambient density
structure is a function of the temperature distribution alone.   In
this section, all results use Method 4 to compute F  and 6 = 1.0.
Unlike all previous cases presented in this chapter, the effluent
velocity profile is assumed flat.
      Figure 8.51 illustrates the assumed ambient density structure
for the six cases.
      Results for the base Case 60 are illustrated by Figures 8.52
through 8.59.  One significant feature of the Case 60 results concern
velocity distribution and may be noted in Figures 8.52 and 8.54.
Figure 8.52 illustrates that radial velocity profiles for the spreading
plume continue to penetrate deeper into the ambient with increasing
radial distance from the outfall.  For this case, temperature differ-
ences are small between the plume flow and ambient as illustrated  by
Figure 8.53.  The upward-distorted streamlines illustrated in Figure
8.54 indicate that there is significant upward entrainment into the
plume lateral spread.
      The influence of a 2 °C ambient thermocline situated as shown
by Figure 8.51 is illustrated by Figures 8.60 through 8.64.  Comparison
of Figures 8.62 and 8.54 shows that the presence of the thermocline
                              355

-------
             ELEVATION OF PORT
   20
21     22    23    24   25    26


     TEMPERATURE,  °C
Figure 8.51.   Ambient Temperature Profiles
              for Cases 60 Through 65
                   356

-------
CO
en
               10




                9.6  I—




                9.2   —




             £• 8.8   —
             CK

             &


             I ••«--
             O


             * 8.0  I—
             2  7.6
                 7.2




                 6.8




                 6.4




                 6.0




                 5.6




                 5.2
                 4.8
                    -0.1
                                   0             0.1           0.2


                                  MHEMSIONLESS RADIAL VELOCITY, U
                                                                            0.3
                         Figure 3.52.   Vertical Distribution Of. Radial Velocity. Case 60
Figure  8.53.  Vertical  Temperature Excess Distribution. Case 60

-------
CO
en
00
SH01RL DIRECTION.
                                              FIGURE  a.54.   STREAMLINES  FOR AN AXISTMHETRIC, VERTICAL PLUME. CONFINED BY  A  FREE
                                                            SURFACE  CASE  60 -  INTERMEDIATE DEPTH, HOMOGENEOUS AMBIENT,  HAMATEV
                                                                                                                                                                       22       tl.5       21    20.75       T - 20.5°C
                                                                                              RflDIRL DIRECTION.  R/O                        ""™

                                                                    FIGURE «,55.  ISOTHERMS FOR All AI1STMMETRIC,  VERTICAL  PLUME,  CORFINEO  BY  A  FREE
                                                                                 SURFACE CASE 60 - INTERMEDIATE  DEPTH,  HOMOGENEOUS  AMBIENT,  HAMATEV
                                                           VOKTICITY CONTOURS AlISYMNETRIC, VERTICAL FLUNC.  CONFINED IT  I FtEI
                                                           SURFACE CASE 60 - INTERMEDIATE DEPTH.  HOMKENCOUS AHIIENT. MMTEV
                                                                                                                                             FIMIRE 1.97.  IB ILLUSTRATION OF STKM FMCTION -  PSI.   USE NO.  CO

-------
CO
(Jl
VO
                                                      8.58.  3D  ILLUSTRflTIW Of TCnfeMTUftC FIELD  —JT.  CMC W. M
                                                                                                                                          riOUK 8.5*.  SO  ILLUSTRDTI^M OF FLUID TOTIC1TY - MlCOfl.   USE M. 60
                                                                                                                                                                                              THERHOCLIHE -  2°
                                                                                                                                                                              r/0
                                                                                                                                                                           o!J.26
                                                                                                                                                                           a  9.13
                                                                                                                                                                           •  (.80
                                                                                                                                                                           •  5.07
T... - 20°C
                                                                                                                                                                                   1.0

                                                                                                                                                                                     it. °C

-------
CO
CT>
O
                                                                                                                                                 r«R m miSTiwCTRic.  »EKTIC«. rune, cwfinto IT R mcc jwrncc
                                                                                                                                                 tEmCOIKU OtPTH. KITH I KOttl THCMKl.INC.  MMTC>
                       FtCURCLG* .  VQftllCi" (.OHTOURr- M1S rnntl RIC.  VCRT1CAL ft -rtC. CaNTINCO 6r ft rlfEC SURFRCE
                                     CBScVl-  illlCRBCOIRTE DEPTH. MITH 2 D€«t£E TME«nBtLl«E.  MH1BYE*

-------
causes significant flattening of the streamlines, or reduced vertical
entrainment by the spreading plume.  This reduction of vertical entrap-
ment is caused by suppression of vertical mixing by the presence of the
thermocline.  In this case  the plume flow spreads above the thermo-
cline,  Also, the plume destroys the thermocline in the discharge
locale but the "convecting in" of the ambient density structure has a
significant effect  beginning at distances approximately 5 diameters
out.  Note the diverging of isotherms in Figure 8.63 and the tendency
for the isotherms to attain the ambient condition.
      Increasing the magnitude of the thermocline results in further
reducing the vertical entrainment and stream line flattening as
illustrated by the results of Case 63 (Figures 8.65 through 8.71,
respectively).  In this case the vertical location of the thermocline
is the same as in Case 61, but the magnitude of the thermocline is
4 °C instead of 2 °C.
      The effects of a thermocline on the temperature structure are
most clearly revealed by Figures 8.66 and 8.67.  Also note that out
to about 5 diameters the ambient density structure is again completely
destroyed by the plume flow.  This feature coupled with the upwelling
of cooler water from beneath the thermocline results in a phenomenon
whereby there is a thermal peak above the outfall, but this peak
rapidly deteriorates radially to a temperature which is cooler than
the surface  (see Figure 8.67).   Unlike the base Case 60 where
vertical entrainment cools the plume, vertical entrainment warms the
                              361

-------
1



f 	 	
c2
z
Q

n
;y
0
1
p™
00
O
Q.

<
X







0
9.
9.
8.
8.

8
7

7


6

6

6
5
5
4

4
4
3
?

6
2
8
4



^p
	 1 	 1 	 p 	 71
£> J* ^^
Ai^^"^
sg/

- I//
••on
o- I//
*h "I'
1
on
.2- II


.8

.4

.0
.6
.2
.8

.4
.0
.6
?
CD
tl
— u
a
\
D
1
?
—
_

—
—
—
|
i


—

—




—

—
—
r/D ~
o 13.19
a 9.83
• 7.32 ~"
• 5.45 _
—
1 1 1
 -0. 1
 0          0.1        0.2

DIMENSIONLESS  RADIAL VELOCITY, U
0.3
0.4
Figure 8.65.  Vertical Distribution of  Radial Velocity.
             Case 63
                       362

-------
                 I    I    I    I    I    I    I    I    I    I
                                         2.0
                          AT,  °C
Figure 8.66.
Vertical Temperature Excess Distribution.
Case 63
                                           2.8
                     363

-------
to
                  o
                  o
                      5  -
                      4  -
                      3  -
                      2  -
                                                               T
                                   BASED ON 20 °C AMBIENT TEMPERATURE
                                   (SEE FIGURE  8.51 FOR VERTICAL STRUCTURE, CASE 63)
CASE 63
                                                              10
                                                     RADIAL POSITION, r/D
                                                                                      AMBIENT SURFACE
                                                                                      CASE 63  (ATC =4°C]
              15
20
                            Figure 8.67.  Surface Temperature  Excess, AT<- for Cases 60 and 63

-------
CO
CT>
                                                                       RADIAL DIRECT1BN. R/DIA
                                                                                                                                                                           lfH. OIRCCTIBN. R/OIA
                               riniRF  •«   3T«FBBLIHt3 FM (» BIISrnnET«IC. VCRTICHL FLUHE.  Off I"™ •I-iJ"" SU«FBCE           FIGimE  U.   ISOTncuns   rgi B« DKISTKnCTKIC. »E«TIC«. TLunt.  C»«fl«D er d r«EE SUKFHCE
                               FIGURE  !•»•  3crfJE"n^1_"YM;[';1,,E|,1BT£ OErTH. """ • OEOIEE THEBnOCLIHE. MmBYEV                                   Co5t B- l«Tt«BtolBI[ OErlH. «ITH « OECUtE  IHEHnKLIMt. mtMrcv
                                             voRTicnr  CONTOURS Hxisi'finETRic. VERTICBI PLUHE. cowfiNeo BT n FREE SURFHCE
                                              CBSL  O-  INTCRnEOIftTC DEPTH.  HITH  « DEGREE THEHMOCLLNE.  HRtlHYEV
                                                                                                                              FICURPS-71-   30 ILLUSTRflTI9N 6F STREflH FUNCTieN  -- PS1-



                                                                                                                                             CflSE 63 -  INTERPIEOfflTE DEPTH.  HITH  4 DEGREE THERH0CLINE.  nfitlHYEV

-------
lateral spreading flow since the cooler water is now on the surface
in the region of radial spread.  This is, of course, a thermally
unstable situation, but the configuration is maintained by the flow
dynamic forces.  This phenomenon is not uncommon and has been observed
on several occasions by Eliason [24] through areal infrared photog-
raphy.  We would expect, however, that once dynamic forces are
mitigated to the point where buoyant forces (if they still persist)
dominate, local upwelling within the lateral spread would occur.  Our
steady flow computer program cannot reveal these local time dependent
effects, but they are indicated by numerical cycling and reluctance to
converge.  Since the case converged without difficulty, we conclude
that the flow field is dynamically stable, at least for the parameters
used.
      Figures 8.69 and 8.72 again show the thermal effects of "con-
vecting in" or recirculating the ambient thermal structure and the
tendency of the thermal distribution to attain the ambient structure.
      Figures 8.74 through 8.81 show results for Case 64 where the
thermocline is 5 °C, although the thermal gradient is identical to
Case 63 (see Figure 8.51).  Comparison with appropriate results of
Case 63 shows little influence from this change.
      In Case 65 the shape of the thermocline was assumed to be the
same as in Case 64 except situated at a somewhat greater depth
(Figure 8.51).  Figures 8.82 through 8.90 illustrate results for this
case.  For the problem posed, computation could not be carried out
to achieve a steady flow converged solution.  Instead numerical
                              366

-------
FIGURE 8-72-  30  ILLUSTRRTI8N  3F  TEMPERflTURE FIELD --4T.




             CflSE 63-  INTERMEDIRTE  DEPTH.  H!TH 4 DEC-REE  THERM3CLINE. MRflRYCV
FIGURE 8.73.   30 ILLUSTRPiTI 3N 3F FLUID  VQRT1CTY    3HEGH.




              CHSE 63   INTERMt'DIfiTE DEPTH.  WITH  4  DEGREE  THERMOCLINE
                                 367

-------
co
o
00
          10.0
           9.2  -
           8.4
           7.6 -
           6.8 _
           6.0 -
            5.2 -
            4.4 -
                                                                          10.0
                   -0.05  0        O.I        0.2        0.3

                           DIMErtSIONLESS RADIAL  VELOCITY, U
            Figure 8.74.  Vertical Distribution of Radial Velocity.
                         Case 64
                                                                           4.4  -
                     1.0
                                          2.0
                                                            2.8
                          AT,  °C
Figure  8.75. Vertical,  Excess Temperature Distribution.
            Case 64.

-------
co
en
<£>
                                          RRDIRL DIRECTION. R/o

FIGURE 1.76.  STRERHUNES FBR AN RXISTfnlETRIC. VERTICAL PLUHE. CONFINED BT R FREE SURFACE
              CBSE M - IHTERnEOIRTE DEPTH. HITH 5 DECREE THERIWCLI HE. IUUIRTEV
                                          RRO"IBL DIRECTION. R/tr

FIGURE 1.77.  ISBTHERHS   FDR B« RXI3THIETR1C. VERTICRt. PtUHC. CONFIICO »T R FREE SUKFRCE
             CRSE  H - IHIERItEOIBIt DEPTH.  HITH 5 DEOHEE  THERMtl-IIC. MM)rE>
                                                                                                                                                FIGURE 1.19.   30 ILLUSTmilM or VI3CMJS SIKEBn FWCTIW

                                                                                                                                                             CRSE M - IHTERNCDIRTE DEPTH. HITH 5 DCCiCC Tl
                                                                                                                                                                                                         HCmmCLIIC. MMTEV

-------
FIGURE I.SO.  30 ILLUSTRATION BF TEMPERATURE FIELD—AT                     .
             CASE H - INTERMEDIATE DEPTH.  WITH S DECREE THERXOCLINE.  NfmRTEV
 FIGURE ».«!•
              3D ILLUSTRATION OF  VORTICITT  —  ONEGA
              CASE M - INTERNED IBTE DEPTH.  MITH 5  DEGREE  THERMOCLINE.  HAHAYEV
                             370

-------
IU.O
9.2
CO
UJ
Ejj 8.4
0
rw 7.6
ft
0
s 6-8
D.
_J
*~r*
< 6.0
5.2
4.4
3.6
-
—
- //
/*/*/y
• •00
I///

-1
• A
- \\
- \\
- \
: "
i
J^^^J "
.^D^^
^Jx

TAMB = 25


i
'
THERMOCLINE = 5
r/D
n 1 "\ 1 Q
w 1 o . I y ^~~~
i
a 9.83
• 7.32 T = 20
. 3.45 AMB
1 1 1 1




°C




°C

Op
-0.05 0 0.1 0.2 0.3 0.4
               DIMENSIONLESS RADIAL VELOCITY,  U
Figure 8.82.   Vertical  Distribution of Radial  Velocity.
              Case 65
                     371

-------
GO
~«4
IN5
                         10.0
                          9.2  -
                     to
                     ae.
                     at   Q A
                     i—*   O . •*
                     o
                     «t
                     »—I
                     X
                          7.6  -
6.8  -
                          6.0
                          5.2  -
                          4.4   ~
                                                 2.0
                                          3.0
                                                              AT,  °C
4.0
                                  Figure 8.83.
                       Vertical, Excess  Temperature Distribution.
                       Case 65

-------
CO
^J
CO
"RHDIRL DIRECTION. R/Q
                                                                                                    RHDIRL  0
                                                                                                            IRECTION.  R/D
                             FIGURE S-flU -  STRERI1LINES FOR RN RXISYHHETR1C. VERTICAL PLUME. CONFINED BY R FREE 9URFRCE
                                           CRSE  ffi •  IHTERflEDlflTE  DEPTH.  WITH 5 DEGREE THERHOCLINE. HRHRYEV
                                                         FIGURE 8,85 .   VORTICITY  CONTOURS  RXISYRRETRIC •  VERTICHL  PLUHE
                                                                       CR3ES -  INTERHEDIRTE DEPTH. HITIi 5 DECREE THE
                                                                                                                                                                                            .  CONFINED1 BY  R  FREE  SURFRCE
                                          ISOTHERMS    F3R  RN  RXiSYnflETRIC.  VERTICAL  PLUME.  CONFINED BY R FREE SURFRCE
                                          CRSE ffi - INTERMEOIRTE DEPTH. HITH 5 DEGREE THERHOCL1NE. MfmRYEV
                                                                                                   'RflOIflL DIRECTION.  R/D
                                                         FIGURE B.W .   STRERHLINES FOR  RN RxisrnMETRic.  VCRTICRL  PLUHC.  CONFIWEO er R FREE SURFRCE
                                                                      CHSE 65-    iNTERttEOIRTE DEPTH. CONTINUED  1TERRTION.

-------
                                          RBD°IAL DIRECTION.  R/D

FIGURE 8.88.  ISOTHERMS   FOR FIN AXI3YMNETR1C.  VERTICAL PLUME.  CONFINED BY ft FREE SURFACE
             CASE 65  -   INTERMEDIftTE  DEPTH. CONTINUED  ITERATION.
                                           RBDIAL  DIRECTION.  R/0

 FIGURE 8.89.   VORTICITY CONTOURS RX1SYMMETRIC.  VERTICAL  PLUME.  CONFINED  BT  A FREE  SURFACE
              CASE  65-   INTERMEDIATE DEPTH. CONTINUED  ITERATION.
                                   374

-------
cycling occurred.  Results after 1000 iteration cycles are shown by
Figures 8.84 through 8.86.  Figures 8.87 through 8.89 reveal  results
after 300 additional iterations.
      Although, the case as posed may not conform to a physically real
situation (in particular, the ambient density structure), a thermal
instability is suspected which may be either real or perhaps  incited
by numerical perturbations.  Inspecting Figure 8.86 illustrates a
large region of cooler water above the thermocline.  Continued itera-
tion showed that the ambient isotherms, within the circulating ambient,
begin to fluctuate vertically out to about 7 diameters.  Further itera-
tion resulted in the development of two recirculating regions: one
above the thermocline and the other below (see Figure 8.87).   That is,
some of the plume flow attempts to spread beneath the thermocline.  If
the iterative computation is continued, streamline patterns closely
resembling those shown in Figure 8.84 will redevelop (single recircu-
lating region).
      The investigation of Case 65 was carried out through approximately
three cycles of the flow changing from one recirculating region, to
two regions and back to one region again.  These computations showed
neither the tendency for the solution to converge or diverge numeric-
ally.  It is difficult to derive much incite from steady flow computa-
tions possessing such behavior except that a thermal instability is
either present or close at hand.  A transient computation of the same
flow conditions would doubtless reveal similar oscillations during the
initial transient, caused by the pulsed plume flow starting condition.
However, we would expect the oscillations to damp out with time except
                             375

-------
if a true thermal instability were present.  For our steady flow
computation, real conditions may be near to those for a real thermal
instability and the nature of the steady flow numerical techniques
may be perturbing the solution to a point which prevents a converging
result.
                                                                   
-------
computed velocity occurs about 0.7 diameter from the plume centerline,
which is about the edge of the "boil" for a real outfall of these pro-
portions.  Figure 8.91 indicates that the plume has undergone only
slight cooling on reaching the surface (^ 3/4 °C), but cools very
rapidly out to about 2 diameters and decreases to about 2 °C above
ambient at 8 diameters.
      The radial velocity profiles at selected locations are shown in
Figure 8.92 which shows that the plume along with entrained flow,
spreads in a fairly shallow sheet at the surface, penetration being
less than 0.4 diameter.  Temperature profiles (Figure 8.94) penetrate
slightly deeper.  In fact, the computation shows that plume thermal
effects penetrate into the negative flow region, hence there is some
indication of plume heat recirculation.
      Streamlines, isotherms and level lines of vorticity are illus-
trated in Figures 8.94, 8.95 and 8.96, respectively.  The inward
bending of the streamlines (Figure 8.94) above the discharge port
indicates considerable acceleration of the effluent.  Maximum vorticity
for this case occurs near the surface and near the point of maximum
lateral spread.  This region of high vorticity is also the region where
one would expect the edge of the surface boil to occur in a real flow.
Three-dimensional surfaces are plotted in Figures 8.97 through 8.101
for the stream function, temperature excess and vorticity.
                             377

-------
3.0
2.5  .
2.0  _
1.5  -
1.0
                    23456
                             RADIAL DISTANCE, r/D

                 Figure 8.90.  Surface Radial Velocity.
                               Case 66
                    23456
                              RADIAL DISTANCE, r/D

                  Figure  8.91.  Surface Temperature Excess.
                               Case 66.
                           378

-------
 1.425
  1.20  _
                                               RADIAL POSITION
                                                     r/D
                                                     2.10
                                                     3.97
                                                     4.81
                                                     7.48
   .30  _
                                    .5             1.0
                                    RADIAL  VELOCITY, U
                                           1.5
                                                          2.0
  1.425
      Figure 8.92.  Verttcal Distribution of Radial Velocity ik Various Radial Positions.
                   case 06
fc
o
Q.
    .90
    .60
    .30
                  SURFACE
\J
                      RADIAL POSITION
                             r/D
                          •  2.24
                          •  4.23
                          D  5.45
                          O  7.02
                                 TEMPERATURE  EXCESS. AT UC
       Figure 8.93.  Vertical Distribution of Temperature Excess at Various Radial Positions.
                   Case 66
                                   379

-------
co
8
FIGURE 3.34.  STREflnUNES  FOR  CflSE  66 I 1 .T ulP DLtf i  FO ; .III
                                    Tte	•	rte-
                           RBOIHL OIREETIBN. R/0

F1BURE 8.95.  IS6IHERHS  F8R CflSE 66 (I.0 OIR DEEP)  FO r  .111
                                             RflOlHL OIRECTI8N.  R/D


                  FIGURE 8.96.  YeRTIClTY   F8R  COSE 66 11 .0 DIB DEEP) FO =: .111
                                                                                                       FIGURE 8.97.  3D ILLUSIRfll 18N aF FLUID VBRTICTr  -  SIIEGfl.     CR3E  N8.  66

-------
OJ
00
                 FIGURE 8,38.  30 ILLUSTRRTI8N BF  STREfln  FUNCTION — PSI.     CR3E N8.
                                                                                                      FIGURE S.99.   30  ILLUSTRHTI8N 6F STRERn FUHCTI8N -- PS1.    CRSE N8. 66
                 FIGURE 8.100.  3D 1LLUSTRRT1BN BF TEBPERRTURE FIELD -- T.    CHSE HO. 66
                                                                                                             8.101.  3D  IU.U3TRRTIBN  BF  TEHrERRTURE  FIELD —»T.     CR3E  MB. 66

-------
8.7  Comparison with. Field Data
     At a late-date in this study* the author was able to obtain
reliable field data for one shallow water application.  This data,
obtained for a customer by Battelle-Northwest, is proprietary and
details cannot be disclosed.  However, the discharge depth is less
than one port diameter and the densimetric Froude number is on the
order of 2.5.
     Figure 8.102 shows a comparison between the computed results and
the field measurements.  As can be seen, there is reasonably good
agreement between data and computation.   The computer program predicts
surface temperatures which are about 50% high out to about 10 diameters.
Temperatures equal to the effluent temperature are predicted at the
surface directly over the outfall, whereas the field data indicates
an average of about 70% of this value.  This discrepancy illustrates
that an improved turbulence model is needed for the transition region
and perhaps a better representation of the cascading caused by the boil
formation.  Nonetheless, this result is  very encouraging because the
computation was performed before the infra-red field data were reduced
to temperature information, indicating that at least for very shallow
water cases the computer code is a useful predictive device which
requires little use of empirical constants.
     This result is only one check point and additional field or
laboratory data are certainly needed for further verification.  Such
information could also be used for improvement of the eddy diffusivity
model—which is sorely needed.
                               382

-------
oo
CO
    JU



   ,28



    26



    24



    22



    20
oo
oo
uj   1;
    16



    14



    12



    10


     8



     6



     4



     2
                                                                                   ,
                                                                          A FIELD  DATA

                                                                          • MODEL  DATA
                           MAXIMUM OBSERVED SURFACE TEMPERATURE
                         A
                          A
                      i    i    i
                                      i    i    i
                                                      i    i    r
                                                        1    i    !    i    1    i
                                 100             200              300             400


                                        RADIAL DISTANCE  FROM  DISCHARGE,  FEET


                     Figure 8.102.  Comparison  of Computed Surface Temperature with Field Data
                                                                                        500

-------
                              CHAPTER 9
                             CONCLUSIONS
      The work contained in this manuscript represents an extensive
numerical study of axisymmetric plume flow.  Various computational
details dealing with practical  applications have been investigated
along with an extensive verification study comparing numerical
results with available published data.
      The objective of developing a computer code for general  use for
vertical plume rise in shallow water and the ensuing lateral  spread
was riot entirely realized.   The code developed is more of a research
tool than a design tool.  The primary reason for this result was the
difficulty in modeling turbulent diffusivities.   Such models are well
established for the vertical  rise, but relatively little is known
about vertical diffusivities in the lateral spread.   Hence, for this
and other investigative reasons the computer code suffered through
various changes and adaptions during the study;  the  code listed in
Appendix E is one of these later versions.
      The more significant conclusions from this study are as
follows:
      ,  The steady flow vorticity-stream function technique
         along with the use of a coupled buoyancy transport equa-
         tion is an effective and accurate method for computing
         buoyant plume hydrodynamics up to our ability to model
         turbulent transport coefficients.
      *  The iterative use of Prandtl mixing length  theory
         (Prandtl's second hypothesis) is entirely satisfactory

                               384

-------
for computing radial eddy transport coefficients in
the plume-rise regime.   In addition
   -  the computations predicted  er to be essentially
      constant for a pure initial  inertia!  flow which is
      also demonstrated  by published experimental data).
   -  depending on the extent of  buoyancy, the computations
      predicted er to vary a great deal with axial positiop,
      and that using a constant value of e  in a buoyant
      flow can lead to large errors in the computed plume
      velocity and temperature distributions.
The iterative use of Prandtl mixing length theory for the
vertical eddy transport  coefficient was used in this work
but was found not  to  be   entirely  satisfactory for the plume
lateral surface spread.  That is, limitations had to be
imposed on the maximum size of the computed mixing length
to prohibit numerical instability resulting from an unstable
mixing length computation.  Vertical eddy diffusion was found
to have little effect on computed quantities within the plume
vertical rise.
Mixing length theory was found to be entirely unsatisfactory
                                                           /
for the circulating (ambient) flow field.
Solution convergence was slowed dramatically by:
   -  Iterative computation of eddy transport coefficient
      (as opposed to constant values),
   -  flow coupled with  buoyancy  transport (as opposed to a
                    385

-------
      pure inertial flow),
   -  multiple iteration on the stream function elliptic equa-
      tion between each iteration of the transport equations.
In addition to the third point mentioned immediately above,  in
every case tried one psi inner iteration (stream function) per out-
ter iteration (vorticity and buoyancy transport) was found to
be satisfactory for convergence.  It is strongly suspected that
once the approach to convergence for the stream function has
become smooth more than one outer iteration per inner itera-
tion would not significantly affect the convergence rate.
This action would, however, result in decreased computation  time.
The numerical techniques were found to be stable for every case
tried except for the following two instances:
   -  over relaxation of the transport equations,
   -  use of iteratively computed eddy transport coefficients
    ,  before reasonable.velocity profiles were obtained by
      using constant coefficients.
It was found that over-relaxation of the vorticity equation
always led to a numerical  instability for the cases tried.
This problem was rectified by using LT = .999.  In no case
using constant transport coefficients and LT <_  .999, was an
instability noted.
The stream function elliptic equation could be  over-relaxed
in some cases (deep water  cases) using LF = 1.6.
                        386

-------
         However,  in the shallow water cases  (Z$ <_ 5) numerical insta-
         bilities  were noted using  LE = 1.6.  Subsequently, L < 1 was
         used with general success.
      •  Based on  results shown in  Figure 8.102, it is concluded that
         the computational methods  presented  herein can be a very accur-
         ate mechanism for computing the surface temperature distribu-
         tion in the near field of  a large, vertical, shallow water
         coastal thermal outfall.   Hence, the primary objective of this
         study is  successfully accomplished.
         The result shown in Figure 8.102 is very encouraging since
         the computed surface temperature distribution was found to be
         in excellent agreement with field measurements and the fact
         that this agreement was obtained without prior knowledge of
         the field results.  However, this is the only case where
         computation was compared to field data and other situations
         may reveal discrepancy.  Obviously, complete validity of the
         model can only be ascertained by further comparison with field
         measurement.
      From the results of this study it is generally concluded that the
numerical techniques used are a viable and practical  method for comput-
ing thermal dispersion in confined steady-flow plumes up to our ability
to model the plume-generated turbulence.   The numerical  approach is
extremely attractive from the viewpoint that important complexities
can be incorporated in the analysis which cannot be accommodated with
                               387

-------
similarity techniques.Hence, the numerical model, which may be cali-
brated with field data, will yield reliable computed information and
permit a more competent thermal analysis.   However, this study has
shown that there is indeed a great need for research in turbulence
modeling and the application of these models in numerical  computation.
                               388

-------
                              BIBLIOGRAPHY

 1.   Abraham,  G.   Jet  diffusion  in  stagnant ambient fluid.  Delft,
     Delft  Hydraulics  Laboratory, 1963.   183 p.   (Publication No.
     29*  series  1, group  14,  section  14.42)


 2.   Abraham,  G.   Horizontal  jets in  stagnant fluid of others density.
     Proceedings  of the American Society  of Civil Engineers, Journal
     of the Hydraulics Division  86:1-13.  1960.


 3.   Abraham,  G.   Jets with negative  buoyancy in  homogenous fluid.
     Journal of  Hydraulic Research  4:235-248.  1967.

 4.   Albertson,  M. L., Y.  B.  Dai, R.  A. Jensen, and H. Rouse.
     Diffusion of submerged jets.   Transactions of the American
     Society of  Civil  Engineers  115:639-664.  1950.

 5.   Amsden, A.  A. and F.  H.  Harlow.  The SMAC method - a numerical
     technique for calculating incompressible fluid flows.
     Los Alamos,  Los Alamos Scientific Laboratory.  1970.  85 P-
     (Document No. LA-4370)

 6.   Anwar, H. 0.   Behavior of buoyant jet in calm fluid.  Proceedings
     of the American Society  of  Civil Engineers,  Journal of the
     Hydraulics  Division  97(4):1289-1303.  1969.

 7.   Aziz,  K.  A numerical solution of cellular convection.  Doctoral
     dissertation.  Houston,  Rice University, 1966,  116 numb.
     leaves.   (Microfilm)


 8.   Baines, Douglas W.   Discussion of submerged  jets by Albertson
     et al. Transactions of  the American Society of Civil Engineers
     115:677-684.   1950.

 9.   Barakat,  H.   Transient natural convection flows in closed
     containers.   Doctoral dissertation.  Ann Arbor, University  of
     Michigan, 1965.  220 numb,  leaves.

10.   Batchelor,  G. K.  An introduction to fluid dynamics.  London,
     Cambridge University Press, 1967.   615  p.

11.   Baumgartner, D. J.   Vertical jet diffusion in  non-linear density
     stratified  fluid. Doctor dissertation.  Corvallis, Oregon  State
     University,  1966. 166 numb, leaves.
                                   389

-------
12.  Baumgartner,  D.  J.  and D.  S.  Trent.   Ocean  outfall  design,
     part one,  literature review and  theoretical  development.
     Corvallis, 1970.   129 p.   (Federal Water  Quality Administra-
     tion, Environmental  Protection Agency)

13.  Bird, R.  G.,  W.  E.  Stewart, and  E. N. Lightfoot.  Transport
     phenomena.  New York, John Wiley and  Son, 1960. 780 p.


14.  Bosanqet,  C.  H.,  G.  Horn  and  M.  W. Thring.   The effect  of
     density differences  on the path  of jets.   Proceedings of the
     Royal Society of London,  Ser. A, 263:340-352.  1961.


15.  Bradshaw,  P.  and D.  H. Ferriss.   Application of a general
     method of calculating turbulent  shear layers.  London,  Imperial
     College,  1970.   17  p.  (Department of Aeronautics.   Report
     No. 70-06)

16.  Brooks, N. H.  Conceptual  design of  submarine outfalls  - I.
     Pasadena,  California Institute of Technology,  1969.   19 p.


17.  Cederwall, K.  Hydraulics  of marine water waste disposal.
     Goteborg Hydraulics  Division, Chalmers  Institute of Technology,
     1968.  273 p.  (Publication No.  42)


18.  Chou, P.  Y.  On velocity  correlations and solutions of  equations
     of turbulent fluctuations.  Quarterly Journal  of Applied
     Mathematics 3(l):38-54.  1945.


19.  Chou, P.  Y.  Pressure flow of turbulent fluid between two
     infinite parallel planes.   Quarterly  Journal of Applied
     Mathematics 3(3):185-197.   1945.


20.  Corrsin, S. and M.  S. Uberoi.   Further  experiments  on the  flow
     and heat transfer in a heated turbulent jet.  Washington,  D.  C.
     1947.  61  p.


21.  Crowley, W. P.  Numerical  advection  experiments.  Monthly
     Weather Review 96(1):1-11.  1968.


22.  Csanady, G. T.  The buoyant motion  within a hot  gas plume  in  a
     horizontal wind.  Journal  of Fluid  Mechanics 22(2):225-239.
     1965.
                                  390

-------
23.   Donovan, L.  F.   Numerical  solution of the  unsteady Navier-Stokes
     equations and application  to flow in a rectangular cavity with
     a moving wall.   Washington,  D. C. 1971.   56  p.   (NASA TN D-6312)


24.   Eliason, J.  R.   Battelle-Northwest.  Personal  communication.
     Richland, Washington.   1972.


25.   Ellison, T.  H.  and J.  S. Turner.   Turbulent  entrainment  in
     stratified flows.   Journal  of Fluid Mechanics  6(3):423-448.
     1959.


26.   Fan, Loh-Nien.   Turbulent  buoyant jets into  stratified or flowing
     ambient fluids.  Pasadena,  California Institute of Technology,
     1967.  196 p.  (W. M.  Keck Laboratory of Hydraulics  and  Water
     Resources.  Report No. KH-R-15)

27.   Fan, Loh-Nien and  Norman  H.  Brooks.  Numerical  solution  of
     turbulent buoyant  jet problems.  Pasadena, California  Institute
     of Technology,  1969.  94  p.   (W.  M. Keck Laboratory  Hydraulics
     and Water Resources.  Report No.  KH-R-18)

28.   Fay, J. A.,  M.  Escudier and D. P. Holt.   A correlation of field
     observations of plume rise.   Journal of the  Air Pollution
     Control Association 20(6):391-397.  1960.


29.   Fox, D. G.  Forced plumes  in a stratified fluid.  Journal of
     Geophysical  Research 75(33):6818-6835.  1970.


30.   Frankel, R.  J.  and J.  D.  Cummings.  Turbulent  mixing phenomena
     of ocean outfalls.  Proceedings of the American Society  of
     Civil Engineers.  Journal  of the Sanitary Engineering  Division
     91(2):33-59.  1965.

31.   Fromm, J. E.  Numerical methods for computing  non-linear,  time
     dependent, buoyant circulation of air in rooms.  IBM Journal
     of Research and Development 15(3):186-196.  1971.


32.   Gaunter, J.  W., J. N.  B.  Livingood and P.  Hrycak.  Survey  of
     Literature on flow characteristics of a single turbulent jet
     impinging on a flat plate.   Washington, D. C.   1970.  42 p.
     (NASA TN D-5652)
                                391

-------
33.  Golubeva,  V.  Nf   The  formation  of  the  temperature  field  in a
     stratified sea,  TR. by  F.  Goodspeed.   Bulletin of  Academy of
     Science of the USSR Geophysical  Series  5:4670-4671.   1964.


34.  Gortler, H.   Berechung  von aufgaben der.freinen  turbulenz auf
     grund eine$  neuen naherungsansatzes. zeitschrift  fur  Angewandte
     Mathematik Und Mechanik 22:257-267.  1936.


35.  Gosman, A. D.  et al.  Heat and  mass transfer  in  recirculating
     flows.  London,  Academic Press,  1969.   338  p.

36.  Harremoes, P.   Diffuser design  for discharge  to  a  stratified
     water.  Copenhagen, Danish Isotope Center,  1967.   18 p.


37.  Hart, W. E.   Jet discharge into a  fluid with  a density gradient.
     Proceedings  of the American Society of Civil  Engineers,  Journal
     of the Hydraulics Division 87:171-200.  1961.

38.  Hayahsi, T.  and N. Shuto.   Diffusion of warm  water jets
     discharged horizontally at the  water surface.  In:   Proceedings
     of the Twelfth Congress of the  International  Association of
     Hydraulic Research, Fort Collins,  Colorado, 1967.  p. 47-59.

39.  Hill, M. N.  (Editor).   The sea.  Volume 1.  New  York Interscience
     publishers,  1962.


40.  Hinze, J.  0.   Turbulence.   New  York, McGraw-Hill,  1959.  586 p.


41.  Hinze, J.  0.  and B. G.  Van Der  Hegge Zijnen.  Transfer of heat
     and matter in the turbulent mixing zone of  an axially symmet-
     rical jet.  Applied Scientific  Research (A) 1:435-461.   1949.


42.  Hirasaki, G.  J.  and J.  D.  Heliums.  Boundary  conditions  on the
     vector and scalar potentials in viscous three-dimensional
     hydrodynamics.  Quarterly of Applied Mathematics 28(2):293-296.
     1970.


43.  Hirst, E.  A.   Analysis  of buoyant  jets  within the  zone of flow
     establishment.  Oak Ridge, Oak  Ridge National Laboratory, 1971,
     41 p. (Report No. ORNL-TM-3470)
                                 392

-------
44.  Hirst,  E.  A.   Analysis of round turbulent,  buoyant  jets  dis-
     charged to flowing stratified ambients.   Oak Ridge, Oak  Ridge
     National  Laboratory,  1971.   36 p.   (Report  No.  ORNL-4685)


45.  Hoult,  D.  P.,  J.  A.  Fay and L. J.  Forney.   A theory of plume
     rise compared  with field observations.   Journal  of  the Air
     Pollution Control Association 19(8):585-590.  1969.


46.  Hwang,  J.  D.   On  numerical  solution of  the  general  Navier-Stokes
     equations for  two-layered stratified flows.   Doctoral disser-
     tation.  Corvallis,  Oregon State University, 1968.  134  numb.
     leaves.  (Microfilm)

47.  Isayeva,  T. and I. L. Isayev.  Determination of vertical eddy
     diffusion in the  upper layer of the black sea by a  direct
     method.  TR.  by Scripta Technica,  Inc.   Soviet Oceanography
     Trans,  of the  Marine Hydrophysical  Institute, Academy of Science
     of the USSR,  1963 series 2:22-24.   1963.

48.  Jen, Y.,  R. L. Wiegel and I. Mobarek.   Surface discharge of
     horizontal warm water jet.   Proceedings  of  the American  Society
     of Civil  Engineers,  Journal of the Power Division 92(2):i-3U.
     1966.


49.  Keagy,  W.  R. and  A.  E. Weller.  A  study  of  freely expanding
     inhomogeneous  jets.   In:  Proceedings of the Heat Transfer and
     Fluid Mechanics Institute,  University of California, Berkeley,
     June 22-24, 1949.  New York, American Society of Mechanical
     Engineers, p.  89-98.   1949.

50.  Keffer, J. F.  and W.  D. Baines.  The round  turbulent jet in a
     crosswind.  Journal  of Fluid Mechanics  15(4):481-497.  1963.

51.  Kent, R.  E. and D. W. Pritchard.  A test of mixing  length
     theories  in a  coastal plain estuary.  Journal of Marine
     Research  18(l):62-72.  1959.

52.  Koh, R. C. Y., and L. N. Fan.  Mathematical  models  for the pre-
     diction of temperature distributions resulting from the  dis-
     charge  of heated  water into large  bodies of heated  water.
     Washington, D. C.  1970.  219 p.  (FWQA  Publication No,  16130
     DW010/70,  Tetra-Tech Inc.).
                                393

-------
53.   Larkin,  B.  K.   Some  stable  explicit  difference approximations
     to the diffusion  equation.   Math. Computation 18(86):196-202.
     1964.

54.   Leenderste, J.  J.  Aspects  of  a  computational model  for  long
     period water-wave propagation.   Santa Monica, The  Rand Corpor-
     ation.  165 p.   1967.   (Document No. RM-5294-PR)


55.   Ma, A. S.  C.  and  K.  S.  Ong.  Impulsively  started incompressible
     jet.   London,  Imperial  College,  1970.   28 p.  (Department of
     Mechanical  Engineering.   Report  No.  EF/TN/A/28)


56.   Manabe,  K., Y.  Watanabe and  A. Wada.  Study on recirculation
     of cooling water  of  Tsuruga  nuclear  power station  sited  on
     Urazoko  Bay.   Civil  Engineering  in Japan  9:156-171.  1966.


57.   Masch, F.  D.  et al.  A  numerical  model  for the simulation of
     tidal  hydrodynamics  in  shallow irregular  estuaries.  Austin,
     The University of Texas, 1969.   123  p.  (Office of Water
     Resources  Research,  D.  I. Document No.  PB 184834)

58.   Morton,  B.  R.   Forced plumes.  Journal  of Fluid Mechanics
     5:151-197.   1959.


59.   Morton,  B.  R.   The ascent of turbulent  forced plumes in  a calm
     atmosphere.  International  Journal of Air Pollution  1:184-197.
     1959.


60.   Morton,  B.  R., Sir Geoffrey Taylor,  and J. S. Turner.  Turbulent
     gravitational  convection from  maintained  and instantaneous
     sources.  Proceedings of the Royal Society of London, Ser. A,
     234:1-23.   1956.


61.   Muragi,  M.  P.  and H. W.  Emmons.   Natural  convection  above  fires.
     Journal  of Fluid Mechanics  8(4):611-624.   1960.


62.   Murota,  A.  and K.  Muraoka.   Turbulent diffusion of the vertically
     upward jet.  In:   Proceedings  of the Twelfth Congress of the
     International  Association of Hydraulic  Research, Fort Collins,
     Colorado,  1967.  p.  60-70.


63.   Neuman,  G.  and W.  J. Pierson Jr.  Principles of physical ocean-
     ography.  Engelwood  Cliffs,  Printice Hall, 1966.   545 p.
                               394

-------
64.   Okubo, A.   Horizontal  diffusion from an instantaneous  point
     source due to oceanic  turbulence.   Baltimore,  Johns  Hopkins
     University, 1962.   22  p.   (Chesapeake Bay Institute, Tech.
     Report No. 32)


65.   Okubo, A.   Fourth report on the rising plume problem in  the  sea.
     Baltimore, Johns Hopkins University, 1968.   19 p.   (Chesapeake
     Bay Institute, Reference 68-3)


66.   Oregon State University.   Department of Civil  Engineering.
     Tidal  flats in esturine water quality analysis.  Corvallis,  1969.
     10 numb, leaves.  (Progress Report.   Grant  No. WP-01385-01 of the
     Federal Water Quality  Administration)


67.   Pagnani, B. R.  An explicit finite-difference  solution for
     natural convection in  air in rectangular enclosures.  Doctoral
     dissertation.  Corvallis, Oregon State University,  1968.
     176 numb,  leaves.   (Microfilm)

68.   Pai, S. I. and T.  Y.  Hsieh.  Numerical solution of  laminar jet
     mixing with and without free stream.  College  Park,  University
     of Maryland, 1969.  32 p.  (The Institute for  Fluid  Dynamics and
     Applied Mathematics.   Technical Note BN-627)
69.
70.
Patankar, S. V. and D. B. Spalding.  A finite-difference
cedure for solving the equations of the two-dimensional
boundary layer.  International  Journal of Heat and Mass
Transfer 10(10):1389-1412.  1967.
      pro-
Phillips, 0. M.  The dynamics of the upper ocean.
Cambridge University Press, 1966.  261  p.
London,
71.  Pond, S.  Oregon State University.  Personal  communication.
     Corvallis, Oregon.  1969.

72.  Prandtl, L.  Uber die ausgebildete turbulenz.  Zeitschrift Fur
     Angewandte Mathematik Und Mechanik 5:136-139.  1925.


73.  Pristly, C. H. B.  A working theory of the bent-over  plume of
     hot gas.  Quarterly Journal of the Royal  Meteorological
     Society 81:144-157.  1955.
                               395

-------
74.  Priestly, C.  H.  B.  and K.  F.  Ball.   Continuous  convection from
     an isolated source  of heat.   Quarterly Journal  of the Royal
     Meteorological  Society 81:144-157.   1955.


75.  Ramsey, J. W. and R.  J.  Goldstein.   Interaction of a  heated  jet
     with a deflecting stream.  Transactions of the  America!  Society
     of Mechanical Engineers, Journal  of Heat Transfer, Series C,
     93(4):365-372.   1971.

76.  Rawn, A. M.,  F.  R.  Bowerman  and N.  H.  Brooks.   Diffusers for
     disposal of sewage  in sea  water.   Proceedings of the  American
     Society of Civil  Engineers,  Journal  of the Hydraulic  Division
     86:65-105.  1960.

77.  Reichardt, H.  Impuls - und  warmeaustauch  in freier turbulenz.
     Zeitschrist Fur Angewandte Mathematik  Und  Mechanik 24:268-272.
     1944.

78.  Reynolds, 0.   On the  dynamical  theory  of incompressible  viscous
     fluids 'and on the determination of the criterion.   Philosophical
     Transactions  of the Royal  Society,  Series  A 186:123-164.  1894.


79.  Rotta, J. C.   Statistische theorie nichthomogener turbulenz.
     Z. Fur Physik 129:547-572.  1951.


80.  Rotta, J. C.   Statistische theorie nichthomogener turbulenz.
     Z. Fur Physik 131:51-77.  1951.


81.  Rouse, H., C. S.  Yih  and H.  W.  Humphreys.   Gravitational
     convection from a boundary source.   Tellus 4:201-210.  1952.

82.  Runchal, A. K.  and  M. Wolfshtein.   Numerical integration
     procedure for the steady state  Navier-Stokes equations.
     Journal of Mechanical Engineering Science  11(5):445-453.  1969.


83.  Sami, Sedat.   Space correlation in round turbulent jet.   Proceed-
     ings of the American  Society of Civil  Engineers, Journal of the
     Hydraulics Division 97(3):907-917.


84.  Schlichting,  H.  Boundary layer theory, TR. by  J. Kerstin.
     6th Ed.  New York,  McGraw-Hill, 647 p.  1968.
                                 396

-------
85.  Schmidt, Wilhelm.   Turbulente ausbreiting eines stromes
     erhitzter lift.  Zeitschrift Fur Angewandte Mathematik Und
     Mechanik 21:265-278, 351-363.  1941.


86.  Schmidt, F. H.  On the rise of hot plumes in the atmosphere.
     International Journal of Air and Water Pollution 9:175-198.
     1964.


87.  Scorer, R. S. The behavior of chimney plumes.   International
     Journal of Air Pollution 1:198-220.   1959.


88.  Sharp, J. E.  Spread of buoyant jets  at the free surface.
     Proceedings of the American Society  of Civil  Engineers,  Journal
     of the Hydraulics Division 95(3):811-825.  1967.

89.  Sharp, J. J.  Unsteady spread of buoyant surface discharge.
     Proceeding of the American Society of Civil Engineers, Journal
     of the Hydraulics Division 97(9):1471-1491.  1971.

90.  Spalding, D. B.  Concentration fluctuations in  a  round turbulent
     free jet.  Chemical Engineering Science 26:95-107.   1971.


91.  Spalding, D. B.  Numerical methods in fluid mechanics  and heat
     transfer.  London, Imperial College,  1970.   37  p.   (Department
     of Mechanical Engineering.  Report No. EF/TN/A/34)

92.  Spalding, D. B.  Heat and mass transfer in  recirculating flows.
     London, Imperial College, 1968.  28  p.  (Department of
     Mechanical Engineering.  Report No.  EF/TN/A/8)

93.  Spalding, D. B.  Mathematical models  of turbulent flames.
     London, Imperial College, 1969.  24  p.  (Department of Mechan-
     ical Engineering.   Report No. BL/TN/B/22)

94.  Stolzenbach, K. D. and D. F. Harleman.  An  analytical  and
     experimental investigation of surface discharges of heated
     water.  Cambridge, Massachusetts Institute  of Technology, 1971.
     212 p.  (Ralph M.  Parsons Laboratory for Water  Resources and
     Hydrodynamics.  Report No. 135)

95.  Stone and Webster Corp.  Circulating water  discharge system,
     Shoreham nuclear power station unit  1.  Hicksville, Long
     Island, Long Island Lighting Comp.  February 22,  1971.
                               397

-------
 96.   Tamai,  N.,  R.  L. Wiegel and G.  F- Tornberg.  Horizontal surface
      discharge of warm water jets.   Proceeding of the American
      Society of  Civil Engineers, Journal of the Hydraulic Division
      95(2)253-276.   1969.

 97.   Taylor, J.  F.,  H. L. Grimmett and E. wV Comings.  Isothermal
      free jets of air mixing with air.  Chemical Engineering
      Progress 47:175-180.   1951.

 98.   Tollmien, W.   Strahlverbreiterung.  Zeitschrift Fur Angewandte
      Mathematik  Und  Mechanik 6:468-478.  1926.


 99.   Tomich, J.  Heat and momentum transfer from compressible
      turbulent jets  of hot  air  impinging normally on a surface.
      Doctoral dissertation.  St. Louis, Washington University,
      1967.   216  numb, leaves.   (Microfilm)


100.   Torrance, K. E. and J. A.  Rockett.  Numerical study of natural
      convection  in  an enclosure with localized heating from below
      creeping flow  to the onset of laminar instability.  Journal
      of Fluid Mechanics 36(l):33-54.  1969.

101.   Tulin,  M. P. and J. Shwartz.  Hydrodynamic aspects of waste
      discharge.  Los Angeles, AIAA 3rd Fluid and Plasma Dynamics
      Conference, June, 1970.  (AIAA  Paper No. 70-755)


102.   Turner, J.  S.   Jets and plumes  of negative or reversing
      buoyancy.   Journal of  Fluid Mechanics 26(4):779-792.  1966.

103.   U. S.  Navy  Hydrographic Office.  Tables for sea water density.
      Washington, D.  C. 1952.  265 p.  (H. 0. Pub. No. 615)


104.   Van Sant, J. H.  Laminar free convection of heat generating
      fluids  in horizontal pipes.  Livermore, Lawrence Radiation
      Laboratory, 1969.  15  p.   (Report No. UCRL-71569)


105.   Van Sant, J. H.  Lawrence  Radiation Laboratory.  Personal
      communication.  Corvallis, Oregon.  1969.


106.   Wada, Akira.   Numerical analysis of distribution flow and
      thermal diffusion caused by outfall of cooling water.
      Tokyo,  Central  Research Institute of Electric Power  Industry,
      1969.  23 p.   (Technical Report:C:67004)
                               398

-------
107.   Wada, Akira.   Effects of winds on a two layered  bay.  Japan
      Society of Civil  Engineers 9:137-156.   1966.


108.   Wada, Akira.   Study on recirculation of cooling  water of  power
      stations sited on a bay.  Japan Society of Civil  Engineers
      10:143-170.  1967.


109.   Welch, J. E., F.  H. Harlow, J. P. Shannon and B.  J.  Daly.
      The MAC method -  a computing technique for solving  viscous,
      incompressible, transient fluid-flow problems involving free
      surfaces.  Los Alamos, Los Alamos Scientific  Laboratory,  1965.
      146 p.  (Document No. LA-3425)

110.   Welty, J. R., C.  E. Wicks and R. E. Wilson.  Fundamentals of
      momentum, heat and mass transfer.  New York,  John Wiley and
      Sons, 1969.  697  p.

111.   Wilkes, J. 0. and S. W. Churchill.  The finite-difference com-
      putation of natural convection in a rectangular  enclosure.
      American Institute of Chemical Engineers Journal  12(1):161-166.
      1966.

112.   Zeller, R. W.  Cooling water discharge to Lake Monona.
      Doctoral dissertation.  Madison, University of Wisconsin, 1967.
      287 numb, leaves.

113.   Zeller, R. W. and R. L. Rulifson.  A survey of coastal  power
      plants.  Portland, Oregon, 1969.  27 p.  (Federal Water Quality
      Administration Report)
                                399

-------
                              APPENDIX A
             CONVECTIVE TRANSPORT DIFFERENCE APPROXIMATION
      Differencing the convective terms is the most troublesome
aspect of solving transport equations numerically.   The mathematical
principles for treating these quantities are available, but one must
exercise extreme caution when applying these principles or grossly
inaccurate solutions will result if not numerical  instabilities.
When forming difference equations for convective transport, prime
consideration must be given to the directional nature of these terms.
      A number of papers have been written and studies made con-
cerning numerical convection experiments.   Perhaps  one of the best
studies on higher order methods has been carried out by Crowley [21 J.
Crowley carried out numerical experiments  using a  number of difference
techniques in solving the "color equation" due to  R. Lelevier,
Here r is a scalar quantity transported with the flow in a manner
such the total derivative is zero along an instantaneous streamline.
Crowley refers to Equation (A-l) as the advective form of the r trans
port equation.  An alternative way to write Equation (A-l) is
              at    3x     ay         sx    3y                       -
which Crowley refers to as the "conservative" form of the transport
equation. By continuity,
                             400

-------
                    . -
             3X    9y ~  U>
However,  in  numerical  approximation,
             3U    3V    .
             97 +  3j <<: ]
but never zero.   For this  reason,  the  right  hand  side of Equation
(A-2) is  sometimes  included with the analysis  in  an attempt to reduce
accumulating numerical error.
      As a point  of criticism,  in  view of transport physics, it is
correct to write
and
             3r + 3(ur) + 3(vr;
             3t   3x       3y
instead of Equations (A-l) and  (A-2), respectively.
      In the paper cited, Crowley carried out various numerical
experiments with first, second  and fourth order approximations for
Equations (A-l) and (A-2), and  the one-dimensional counterpart of
these equations.  For the one-dimensional tests, he concluded that a
second order process using the  "conservative" Equation (A-2) was the
most accurate.  In two dimensions he found that fourth order methods
were the most accurate but could not ascertain which equation gave
the best results.  However, he  does recommend that the conservative
equation be used.
      Reference [66] reports results of numerical experiments con-
cerning the one-dimensional transport equation,
                             401

-------
Unlike Crowley's work, this work was concerned with the directional
nature of u and the proper method for differencing 9F/3X (forward,
backward or central) to minimize numerical  error and achieve stable
computation.
      For these experiments u was assumed positive and steady, with
the corresponding explicit difference equation written as:
           n+1     n _ uAt           n    n      , n   n
         r     ' r   ~
          1       i  " IT  Lv'-»xM1i-rV "6x ui   i+l'J   (A-5)

where the superscript n refers to the nth time step.  The parameter
£  varies from 0 to 1.  The following difference techniques are
obtained from Equation (A-5) for the corresponding values of £ :
                                                              /\
            £  = 0   backwards or upstream method
            X  = .25 so-called "quarter point" method
            i  = .5  central method
             A
            f$x = 1. 0 forward or downstream method
      The results of these numerical experiments are compared with the
analytical results for various time steps and total elapsed time, and
found that the upstream difference (backward to the direction of flow)
gave the superior results.
      Note, that in all but the upstream method, downstream quantities,
to some extent, are used to establish upstream results.  In the case
of pure convection these formulations are physically incorrect.

                             402

-------
      Lelevier (cf. [21]) was evidently the first to introduce the
upstream differencing technique.  Crowley reports that a great deal
of numerical damping results with this method, applied to the
"advective" equation, over long integration periods.  Nevertheless,
the upstream method (also called, unidirectional or one-sided deriva-
tive), has been used extensively in solving transport equations.
For instance, Van Sant [104] used the "advective" form to solve the
vorticity transport equation.  Torrance and Rockett [100] solved  the
"conservative" form of the vorticity equation in this fashion, and
Runchal  and Wolfshtein [84] used upstream differencing to solve for
steady flow vorticity transport in "advective" form.  Van Sant [105]
stated that he was unable to obtain a solution to the steady flow
vorticity equation using central differences.
      One trouble with using any method except the upstream method
is that truncation and numerical round off can cause serious errors
and even destroy the solution through numerical instability.  Higher
order methods (central difference, for instance) in spite of their
purported higher degree of accuracy may be inferior if the direction
nature of the flow is not considered.  Runchal and Wolfshtein present
some clarification of this subject.  We will pursue the matter here
by formulating convective difference schemes using one-sided and
central  techniques.
      Consider the incompressible steady flow transport equations, with
constant eddy coefficients for a conservative scalar quantity r in
(x,y) coordinates:
                                  403

-------
            sr L „ sr     1
 2       2
) r  .   9   r
                                                              (A-6)
            dA     dy   IHrj II






where NR is the Reynolds number and



          .,   _ momentum diffusivity

           r        r diffusivity




The finite difference grid system (Figure A-l) has constant and  equal



 spacing  in  the  x  and  y  directions.
                   I
                   j-l
               k+l
                                             — k
         j+l
     Figure A-l:   Finite-Difference Grid System
                              404

-------
      Suppose we now apply a general difference scheme to the convec-



tive terms of Equation (A-6) which, for the time being, disregards  the



directional sense of the velocity components u and v.   Then,
            VP
            F
where the constant subscript has been suppressed and point (j,k)  is



replaced by p for convenience.  In the above equation, X  and X  are
                                                        x      y


factors corresponding to difference schemes in the x and y directions.



These quantities (X,  and X ) take values of 0, 1/2 and 1 for backward,
                   x      y


central, and forward differences, respectively.  The quantity F is



equal to NDN_h.  Solving for r  yields
          K 1                 p
   =  UP   1-rj-r*xrj+i  + VP
                                                             (A-8)
Case 1.  Central difference scheme, |JX and X  = 2




Equation (A-8) reduces to
                              405

-------
           V
                                                             (A-9)
      If F is very small,  implying  a  very  small  Reynolds  number
(creeping flow)  or a very  small  grid  spacing,  h,  Equation  (A-9) will
usually converge.   However,  for  large F,
            rp  a        'j-r      +    'k-rVi           (A-io)
Hence, small  errors in the  differences  are magnified by a  large coef
ficient, F, which will eventually  destroy the computation  through
instability.   For this reason  the  central difference scheme  is not
desirable for either transient or  steady state application for inter
mediate and large values  of F.
Case 2.  Forward difference scheme,  X   and $  =  1-
                                    x      y
Equation (A-8) reduces to
                                                        •     (A-ll)

Equation (A-ll) poses additional  complications  because  of  the presence
of the negative sign in the coefficient multiplying  r  .  For positive
u  and v  and
                              406

-------
Equation (A-ll) is unmanageable.  For large values of F, the differ-
ence scheme becomes
If either u  or v  is positive, this equation is physically incorrect
because we would be basing upstream computation on downstream informa
tion.  On the other hand, if both u  and v  are negative, then
                                   r      r
Equation (A-ll) becomes

         if +lJrp-f  <|up|rj+1*  |»p|rw)
which may be shown to be computationally stable for all  values of F
and is a preferred scheme.  This equation is also physically correct
since upstream quantities are used for downstream computation.
Case 3.  Backward difference scheme, X  and ((  = 0.
                                      x      y
      Equation (A-8)  reduces to
If velocities u  and v  are both positive we have a computationally
stable scheme which is posed physically correct.  However if either
velocity component is negative, we have the same type of situation
discussed in Case 2 where the scheme may be unstable and is not posed
                             407

-------
correctly with regard to transport physics.
      Clearly, it is necessary to have a computationally stable and
correctly posed difference scheme for all values of F.  It is impos-
sible to meet this criterion in a general flow system without cogniz-
ance of velocity directional sense and magnitude at each and every
boundary and computation point in the difference network.   A sound
scheme may be obtained by choosing <(  and <(  according to the sign of
                                    x      y
the velocity components.  We disregard /  and X  = 1/2 because of
                                        A      y
instability at large F.
                                          Vp
Figure A-2  Values of <(  and /  for a Preferred Difference Scheme
                       A      Jf
      Figure A-2 summarizes the upstream difference method.   Since the
velocity sign must be checked at each point in order to decide which
value of rfx and rf  is to be used, an alternate method is formed which
is well  adapted to computer application.   Consider Equation  (A-6),
specifically the term
            up[(l-iJx)(rp-r..1)-(5x(rp-r.+1)].
                           408

-------
Let


                i                  r UDF^ UD ^s  p°s''tive

      p   ^x    ?   '  pi      p'   I 0,if u  is  negative   '
                                  i.       p

                               ,if u  is positive

                                ,if u  is negative    '
hence,




        II
     UP 3X
                                            l»pl  ' uP>(VrM»
which always gives the correct difference regardless  of  the sign of u .



     The upstream difference technique applied  to  Equation (A-8) yields









            1                   l
          + 7TIU   ~U)r.   + 7T
            2  '  p'     p  J^l   2




          + ? d vpi  • Vvi+ r'
Solving for r  yields
         rP
              !{(lu
                           4+ F  (|u| +  Jv|
                                           p
                               409

-------
Upstream Differencing for Conservative Forms



      Previous discussion of upstream differencing has dealt entirely


with convective differences in the "advective"  form,  u.ar/3x..
                                                      J     J

However, this form is a result of mathematical  manipulation of


the correct "conservative" form, a(u.r)/ax..  The conservative  form
                                    J     J

is a direct result of a r balance in terms of infinitesimal quantities


and is the correct method for proper conservation of  a transported


quantity in numerical analysis.


      Consider the convective balance of r in r,z coordinates


(Figure A-3).
       r(v-n)dA
     '1
 fr(v-n


\»
                                  )dA
                                (r.z)
    I
                                                  7 4- AZ
                                                  z + 2-
                     •/
r(v-n)dA
                                                   z -
                                                       Az
            j:        Ar   I  r(\r-n)dA        Ar
                  r - o~~    «              r + -s—
       Figure A-3  Convective r Flux for an Infinitesimal
                   Axisymmetric Volume Element
                             410

-------
      The steady flow convective balance equation for volume element

p is given by



             I  r(v-n)dA = J   r(v-n)dA +  J   r(v-n)dA
             A              A               A
             MT            rt]              A2

                                                              (A-17)
                            J   r(v-n)dA +  I
                          +  /   r(vn)dA +  I   r(v-n)dA = 0.

                             A3              A4


      In Equation (A-17) and Figure A-3, A-j, Ap, etc., are element
                                            S*.
areas corresponding to side 1, 2, etc., and n is a unit normal  vector,

with outward, the positive sense and inward, negative.  Like direc-

tional sense is used for the boundary velocity vector v.

      Now refer to the grid system shown in Figure A-4.  This grid has

constant Ar and A.Z, and velocities u and v are specified at the cell

face, whereas r is cell centered at point p (also see Figure A-2).

In setting up the difference scheme based on Equation (A-17) we want

to:

      1)  convect into the cell, p, the value of r at the upstream

          neighbor, and

      2)  convect out of cell  p, the value of r at p.

Hence, the value of r to be used in Equation (A-17) is given by
            r =
                 r , for |v-n| = v-n
                .value at upstream neighbor for |v-n|  ^ v-n.  (A-18)
                              411

-------




j-i
* j ^ 1

'.-.-1

r .

"2
k + 1

r * *
P
O___ . .__ _

k-l

r . . , ».

Jtl I
r i+i- • A z
I




Figure A-4  Axisymmetric  Finite-Difference  Cell,
            p,  with the Four Immediate  Neighbor
            Cells
                    412

-------
      Unlike typical difference schemes, Equation (A-17) provides
flexibility of convecting into or out of any cell face.  For the ele-
ment, Equation (A-17) may be written as
                (v°n)
                   )AZF (v-n)
                Ar   r(v^n)
+i:r Ar  r(v-n|      = 0
                           z—
                             AZ
                AZ
 Dividing by volume (2-irrArAz) yields
                                         r(v-n)

                  rAr
      rAr
r(v
-i 	
/\
•n)
r
Z- —IF
	 =— 4- —
(v
A,
•n)
**!
                 AZ
    AZ
             = 0-
                                                              (A-19)
In accordance with Equation (A-18) and Figure A-4, Equation (A-19) may
be expressed as
                              r Ar
                               P
          2 rj

-------
4
                                   - r
                                      k+l
                                   AZ  ;
      <=?>
  1_  a(rur) +
  r    ar
                                                              (A-20)
The above form is used throughout in this thesis for convective dif-
ferences.  Vorticity transport has a slightly different form in the
convective terms,
           3r
which amounts to deletion of rj_-|/2» rj+i/2 and rn 1n the first two
terms of Equation (A-20).
                              414

-------
                              APPENDIX B



            FINITE-DIFFERENCES FOR IRREGULAR NODE SPACING



A.I  General


      Consider the irregular grid shown in Figure B-l  below.
                                       Bh
,1,
*
AX . . fc
•*- nA-,_l »
I
i
- AX- f
* 1 *]
,1
. AX n »
^ 1+1 m
\
                  Figure B-l. Irregular Spaced Grid
                                                       i



The width of node i is designated AXi and the nodal points are all


cell centered.  Finite-difference approximations for the first and


second derivatives at node  i are developed as follows.


      Let,
            h = \ (AX.^ + AX.)
and
            6h  = \ (AX. +



      Then  a Taylor  series  expansion  of a function f about point i is
given by  the  equations:
                              415

-------
fi*r  fi
                    i
                 ehf
     .3.3   „,    4h4
f-+tf f,  +6h
                                                           (B-l)
and,
                                               TW
                                                                (B-2)
Now, divide Equation (B-l)  by 6 and  add the result to Equation (B-2)

to obtain the difference approximation for the second derivative of f:
           *^ + _^
                                     2f.
                                       1  +  (3-1) Oh + Oh .       (B-3)
                          h'(3+l)


Fbr 3=1, Equation (B-3)  reduces  to  the familiar central difference

form:
             fi+l+fi-T2fi      2
              1+1  I  '	2.  + oh^
                                                           (B-4)
A finite-difference approximation  for the first derivative of f at

point i may be found by subtracting  Equation  (B-2) from  (B-l), up to

and including terms involving  f".  Hence,
         fi+rfi-i
      9f
      3X
 Again with  6 =  1 the familiar central difference form results:
                                                           (B-5)
9l
                    -1
                      '
                ~2T
                                                            (B-6)
                           416

-------
 Equation  (B-5)  is  a  first  order  approximation  of |y- .  A  second order
 method may  be developed  by reducing  the  coefficients of f." to 1  in
 Equations (B-l) and  (B-2).   Equation (B-2)  is  then  subtracted from
 (B-l) to obtain:
                   1    _.     9_    _    _ «_1_         9
                                                               (B-7)
Equation  (B-7) collapses to  (B-6) for  6 =  1.
A.2  Computer Application
      For computer application,  irregular  spaced first and second
derivatives difference forms are needed for both points (j,k) and (p,q)
in the vertical direction  (Figure B-2).
t
A;

i
AZ

i
AZ
I
1
i


k


k-1


X 	


X - — —


x 	

- J.K+I
	 P,q+l


	 P.q
_ i k-1

	 P,q-l

          Figure B-2.  Grid Layout for Vertical Differences
                               417

-------
The following forms  are  used  for  differencing a general quantity, F(the

subscripts p and j  have  been  suppressed).


Point (j.k)
     First derivative of F:
          3F
AZk Fkfl            AZk+1Fk-l
     Second derivative of F:
                       Fk+rFk
          3Z'
                                                     AZ, -/
                                                       !w4Zk
                                                                 (B-8)
                          Fk"Fk-l
                                                                 (B-9)
Point (p,q)
     First derivative of F:
                     q+l
If.
3Z
     Second derivative of F:
              q-l
                                       -  D
[k-l      AZk+l
                                           AZ|<+AZk_1
                                                                (B-10)

            q    7 AZk^AZk+l+AZk)     7
                                                                 (B-ll)
                             418

-------
                              APPENDIX C


                      COORDINATE TRANSFORMATION


      The required partial differential equations are given  in

Chapter 5 by Equations (5.10) through (5.14) and are restated here

for reference.


      Stream Function:



            4  - 1  U  +  4  - - Rn.                    (5.10)
            9R     R  9R     3Z



      Vorticity:
                                  w;
      Buoyancy Parameter:

         =  	._  . 	+ ^  	   + _!	J.  ,          (5.12)



along with

            U = - 113L   ,                                     (5-13)

and

                1  aw
            V"lf  W  '                                      (5-14)



                           419

-------
     These same expressions are given in transformed coordinates by
Equations (5.16) through (5.20), respectively.   The transformation
to £ coordinates by setting
          R = sinh £                                           (C-l)
has the desirable properties mentioned in Section 5.5.  Details of
the transformation are given in the following discussion.
     Consider a quantity F and first and second derivatives of this
quantity in R coordinates.  The general  transformation of  these deriva-
tives to 5 coordinates is derived as follows:
          dF _ dF    d?  _                                        -
Then
 or
          d2F    d£ _  d£   d£  _  d__ r df_   d£-,   d£
             ~ ~  dR "  d?  ' dR  "  dc L de  ' dRJ  ' dR
                                    d?  ,  d
                                    dR  ;  d?
Now,
          d_  /d^x   dH^   dH   d_R
          dc  ^dR; " dc ~ dR  * d?
                  = dj*

Hence,
                             420

-------
From Equation (C-l),


          41 =  	1                                            (C-4)
          dR    cosh £


and
Then,
                      f                                        (C-6)
and

           ?_              ?_
                                     e If.).                      (c-7)
Substitution of Equations  (C-l),(C-6) and (C-7) into Equations (5.10)

through (5.14) yields the  transformed set (5.16) through (5.20).


     One discomforting feature of non-linear transformations

is that small errors are introduced in calculating areas and dis-

tances in the transformed  coordinates.  For instance the distance AR

in real coordinates is given by



            ARA = sinh (5  + AC)  - sinh (5) .


In the difference computation,



            ARC = cosh (5  + T^- )  A5  .


Taking the ratio of these  two expression yields, after manipulation of


identities:
             A  .  Actual  spacing    = 2_  sinh  {  ^i }              (C-8)
                  Computed  spacing    AC        v  7- ; .


                            421

-------
As Figure C-l  indicates, AC should  be  kept as small as possible.
       1 .05
       1 .04  -
       1 .03  -
    o
       1 .02  -
       1 .01  -
                   I    I   I    I   I    I   I    I    I
                   .2     .4     .6      .8      1.0
           Figure  C-l.  Ratio of Actual  to Computed
                       Node Spacing
                      422

-------
                             APPENDIX D


                   SOME RELATIONSHIPS BETWEEN TIME
                 DEPENDENT AND STEADY STATE NUMERICAL
               METHODS IN HEAT TRANSFER AND FLUID FLOW


      The general transport equation for a conservative quantity,


T, is written in tensor form as:



           9U T
where the summation convention does not extend over the underscored


indices and source and sink terms are negligible.  The symbols in


the above equations are:


      t   =   time


      x.  =   jth spatial coordinate
       J

      U.  =   jth velocity component
       J

      a.  =   diffusion coefficient along the jth coordinate
       J


For simplicity in this discussion, we will ignore the convective


terms, consider a as a constant, and write Equation (D-l)  as
             \  J   J



For steady flow,
      ax. ax.
        J   J
                                                                (D-3)
The usual technique for solving the above equation is either by


Gauss-Seidel or Gauss iteration, where the former is much faster
                             423

-------
than the latter and, consequently, the most popular technique.   In
both cases successive over-relaxation (SOR, extrapolated Liebmann
method) is employed.
It is the task here to illustrate that certain methods for solving
Equations (D-2) and (D-3) above are identical  up to the Liebmann
extrapolation factor, L,  in the steady state technique ahd the time
scale factor, a, in certain time dependent methods.
D.I  Correspondence Between the Classical  Explicit and Gauss Methods
      The classical explicit and most common method for solving Equa-
tion (D-2) is given in difference form for an evenly spaced grid as
follows:
      T 3k1  -T Jk ' « .  (D-4)
where a =     .
          AX
The superscript n denotes the nth time step.   One may rearrange
Equation (D-4) to give

           • « 
-------
An algorithm for Gauss iteration of Equation (D-3) may be written as
       Tjk' '  l TJ*k  +   <'-L>  Tjk,                             (D-7)
where s denotes the sth  iteration  and L  is  again  the  Liebmann
extrapolation  (or SOR) factor.  We note  that Equations  (D-6) and
(D-7) are  identical insofar  as
       L = 4 a.                                                 (D_8)
In Equation (D-7), L  is  greater than 1,  but must  be less than 2
to prevent solution divergence; that is,  for over-relaxation

       1   <_ L 1 2.
Hence, as  a maximum value
       4 aAt   0      aAt
       —  ~- »
        AX            AX
which is exactly the explicit method stability criterion.
D.2  Correspondence Between ADEP Transient Methods and the Gauss-
     Seidel Technique
      Alternating direction explicit procedures (ADEP) are relative
newcomers to the field of applied numerical analysis.  The prototype
ADEP was conceived by the Russian mathematician, Saul 'ev, in 1957.
Since then other methods have been presented such as those proposed
by Larkin [53] and Barakat [9].  These methods, which have been
demonstrated to have good accuracy and incredible stability, have
basic algorithms identical to the Gauss-Seidel method with SOR.
                           425

-------
A.  Saul 'ev Method


      The Saul  'ev method consists of alternate directional sweep-

ing of the grid system.   A forward sweep  is written as


  n+1   Tn      /Tn+l   .  Tn+l        Tn       Tn      2 Tn  - 2 Tn+1 }.
Tjk  - Tjk = ° {Tj-U  + Tjk-i    +  Tj+l  k + Tjk+l   * Tjk     'jk  >•
                                                               (D-9)

Note that there is equal  weighting  on  the  n and n+1 time levels.

Rearranging Equation (D-9)  into the context of Gauss-Seidel iteration

with SOR yields
                             *

               (1  +  2 .)  1$  . 4 . T™ + (1-2 .) T^


               4T
               Tn+l _  4 a  1 T*n+l  |l-2 gj ,n
Hence,         Tjk  -  ^p^j Tjk  + \^-^J T..k
                                                                (D-10)


 Comparing Equation (D-10)  to  the Gauss-Seidel algorithm,



                                   (1-0  Tj^                    (D-ll)


 again  shows equivalence insofar as



               1=  -4a


or



               L =
                           426

-------
      Now           Lim  —r -> 2;
hence, the upper limit of the Liebmann extrapolation constant is

satisfied from the standpoint of stability irregardless of the

size of the time step, At.  Fora = .5,

                    Tn+l    T*n+1
                    'jk  "   jk

which is identical to the Gauss-Si edel method without SOR.

For the Saul 'ev method, the next time level  computation involves a

similar backward sweep.

B.  Larkin's ADEP

      Larkin's ADEP is actually one of several methods discussed by

Larkin in the cited reference.  The method here is very similar in

the mechanics to the prototype Saul 'ev ADEP, except that the for-

ward and backward sweeps are averaged to form a time level.

Larkin's methods yield the same relationship between L and a given

in Equation (D-12).

D.3  Further Comparisons Between Larkin's ADEP And The Gauss-Seidel
     Iterative Technique

      Consider the two-dimensional form of Equation (D-1),

                           / ?       9 \
      3T  . 3UT  ,  SVT    a/3T  +  lfl\ .                      (D-13)
                    --  a  — ~     —
Based on upstream differencing of the convective terms, the forward

sweep ADEP finite-difference equation would be,

                            427

-------
                                Tn+l
                                Tjk
                           -  U.
        Tn+l
         jk
                                                    jk
                               Tn+l
                                Jk
                                   Tn+l
                                   Tjk
                      - Vjk
jk+1 >
                                   jk
           {Tn+l     Tn      Tn    Tn+l   Tn+     Tn      Tn    Tn
           'j-1k + 'j+1k "  'jk  "  'jk  ^  'jk-1 + 'jk+1 " 'jk " 'j

                    ^7~~^                    1?

                                                             (D-14)

Figure D-l  illustrates a  finite-difference cell  and the relative

locations of  the quantities T,  U, and V.
                          -D
                            VJk
x      O

Tjk
                                    ujk
              Figure  D-l.  Finite-Difference Cell
                             428

-------
Note that in Equation (D-14)  if velocity  is negative T.. is evalu-
                                                      ' J

uated at n, whereas for positive velocity T.. is evaluated at n+1.
                                           ' J

The backward sweep would use  the opposite sense.  Also, this


convention is not a necessity and other time level evaluation schemes


may be used as long as they are computationally explicit.


             n+1
Solving for T'k  yields
     At
     At
          'V
4. u
                                  v      4.     4.
                                '  Vjk-l    «?"
                                aAtn+l
                                                               (D-15)
                             429

-------
 For a short hand  notation let:
          D..  = a<-V+-
           J      ' A     AY2
Then,
At T
                                                             n+1
                                                             jk-1
                                                        'jk+1



                                                           (D-16)
                         430

-------
The Gauss-Seidel  scheme yields
                                                 + -^'M-lk
                                                           '     (D-17)
Substituting Equation (D-17)  into  (D-16) yields,
(Cjk + Djk)  ^}  ^  -  (Cj
Yields
         (C.k + 2D  )At
     L =
                           431
       jk    jk               jk     j
                                                                (D-18)
or in terms of iterations s,
    s+,                         ,
Comparing to

          = L
                                                           (D-20)

-------
Thus,
     Lim   CJk + 2Djk     _ fo + 2Djk
     A t-**0  I    Z     -     C. i  + D. i
     uv    •	4. r   -u n       ilc     TK
           TT" ~ ".Six ~ u-ib    J1^    J *
           At    j f>    j i>
The condition
           2V
leads to some restrictions on the over-relaxation  factor t.
For the case where convection effects are  very  small,  characteristic of
a creeping flow,
     2D.,,
   -, - & - < 6                                                (D-22)
   k+DJk  "
The question is what values of 6 are possible in Equation (D-22).
     For  At -H», 6 = 2 and for At •*•• 0,6  =  0.
     "   0 l L i 2.
For very high Reynolds number  flow, viscous effects become relatively
small and

for At -*», 6 -»• 1  and for
    At = 0, 6 = 0; hence,

     0 <   L  < 1
                               432

-------
This preceding analysis indicates that it is impossible to accelerate
the Gauss-Seidel technique for flows where viscous effects are
negligible.  In the general case there will be regions in the flow
field where the local Reynolds number will be such that D..  ^ 0.
If the condition 0 < L < 1 is violated, then an instability will propa-
gate from this local point.
                                 433

-------
                                       APPENDIX E
                      LISTING C-F SYMJET COMPUTER CODE - UNlvAc  HO*  VERSION
              FOLLOWING ARE THF PROGRAM PARAMETERS AND DIMENSIONS.
CO
4s.
c*
c*
CGMLST*  FCOPY
      PARAMETER
      COMMON
     1
              5
      COMMON
      COMMON
               COMMON
              i
              2
               REAL
               INTEGER
               LOGICAL
               END
GAM(LJrLK),SC(LJf 15 > • R  'UGRAD(LjfLK) 'PlCH(LjtLK> »RB(LK) »
R051LK) »TEMdO) » XR (LJ) »ZC (LK ) »ISOLM(5»30) »N3DPT(5) »
OATE(2) »TjM(2) »TLABEL
                 DPSAX » DOMAX » DDMAX » DZc rEZ • S IGTS » VMB » DMB » S I GTB » ZRP »
                 G » OMB • SALR r SAuJ » DS/iLT»RMIN»tXR»EXS» EXT » BETA »AK1»
                 SIGTJ» TINT » START* RATIO rGAWEND»ERATTO»VMBl»PLX
                 MJ»HJJ»NH,NL»NOUT»NPT»IN»OHT»IPMAX»KASE»KT»INMODE»
                 ITMAX»ITf!0»NB»NTTY»NK.»NKK»NCR»NPT»MMAD»NX»DZT5»TO»TR»
                 I TEMP » NED » NEDDY » ^JOTE^•P t JpORT » KPORT r I TNOO »
                 ISOLN
                 CUT
                 CONTRL
         ApDIM*   FCOPY
               PARAMETER
               END
         C*
         C

-------
C     UN1VAC 1108-VERSION

C*    THIS VERSION oF THE SYMJET RRoGnAM HAS BEEN CHECKED-OUT FOR  THE
c*    FOLLOWING OPTIONS:
C*
C*        1)   AOUTr OPTION  (OUTPUT ARRAYS)
c*                  PSIP»'PSIV» »LELT».OMEG'
C*                  UFAC»'VFAC»»RICH»
C*        2)   PLOT* OPTION(CREATES PLOT ARRAYS FOR SUBSEOUENT
C*                            CONTOURING AND 3-D PLOTS)
C*                  ALL OPTIONS WORK
C*        3)   TERP» OPTION  (CALCULATrS UNORDERED CONTOUR  VALUES)
C*                  ALL OPTIONS WORKED CORRECTLY ON PREVIOUS VERSION'
C*                  BUT HAVE NOT BEEN CHECKED FOR THIS VERSION.
C*        **>   COMT» OPTION  IPROGRAM CONTROL)
C*                  BUOY*»TRAN»»TEMP».MONT»»TAPE»»SAVEr»INVS»»TUR8»r
C*                  CENO»'CENI»
C*
C*
C*        INMODE    = *

C*
C**** NOTE ****
C*
C*   THIS CODF VtPSIO!; HAS BEEi^ DEBUGGED FOR OPTION INMODE = * ONLY
C*   -HICH TREATS SHALLOW WATER PLUMES WITH PQ«FR LAW INFLOW VELOCITY
C*   PROFILE.  <-,QME CHANGES  HAVE BEEN MADE ON THL LATERAL  DIFFUSIVITY
C*   MODEL SINCE RUNNING OF  CASE 66. HENCF RESULTS WILL NOT CHECK
C*   PRECISELY.
C*
C*    SYSTEMS ROUTINES USED  bY  CODE.
C*
C*             TOY    (F)
C*             OOY    (F)
C*             ETIME  (F>

-------
CO
c*
L*
C*
C*
C*
C*
C*
c*
c*
c*
c*
c*
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c


WHEf-'
THEr
THE POLL
THE OPT






********

CELT
RSEA»
           Dl^LNSlONLEbS VELOCITY AT POINT (X»Z)»
           RADIAL EDDY MULTIPLICATION FACTORS-
              WHEN NEDOY=0 »FR(K)=1.178
              rtHEN NEDDYri .FR(K)=R.5*V^AX»WITH R
              WHEN NEDt)Y=2 r FR ( K ) =R ,5*VMAX » WITH R
                                             EXECUTION
           VERTICAL EDDY MULTIPLICATION FACTORS
           RADIAL DISTANCE TO OnTER CELL SIDE J'K
           RADIAL DISTANCE TO CENTER CELL 0»K
           DEfjsiTY STRATIFICATION OF AMRIENT(SIGMA UNITS)
           TRANSFORMED RADIAL DIMENSION TO riGDAL POTNT

           DENSITY DIFFERENCE BETWEEN PLUME AT PORT AND REF.AMBIENT
           CENTEPLINF VALUE OF CELTA AT Z=Z3
                                                                 X-COMPoNENT
                                                                 Z-COMPONENT
                                                                 ,5
                                                                 .5
SPECIFIED
CALC DURING
   OF PROGRAM

-------
CO
c
C
C
C
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
DSIGT

DZ
DX
DX2
DZ
DZ2
DIAC
EZ
FO

G

6MB
R.5
RER
REZ

RRP
RO
SALJ
SALR
DSALT

SIGTJ
SIGTR
SIGTB

TLABEL
TLIST(
                           VERTICAL
                           VERTICAL
                           WIDTH  OF
                           DX*OX
                           VID1H  OF
                           DZ*DZ
                           DIAMETER
                           VERTICAL
              DENSITY CHANGE OVFR DZ IF CONSTANT(SIGMA UNITS

              NODE THICKNESS IF CcNSTANT»DELTAZ/DTA
              NODEr X-DIRECTICN


              NODE* Z-DlRrCTION
              OF OUTFALL PORT
              EDDY TRANSPORT COEFFI IENT
     DENSIMETFIC FROUDE NyMBE" AT OUTFALL PORT
     LINEAR STRATIFICATION PARAMETER SIMILARITY SOLUTION

     CENTERLINE GAMMA (GA^M/GAMO) AT 2 = ZB
     RADIAL DISTANCE TO HALF VELOCITY (MIXING LENGTH APpRQX)
     RADIAL REFERENCE TL'RPULEMT REYNOLDS NUMBER
     VERTICAL PEFEKENCE TijRBULENT REYNOLDS NUMBER
     RADIAL REFERENCE pRApDTL MUMpEP
     RADIUS OF OUTFALL POPT
     SALINITY OF PLUME AT OUTFALL PORT
     SALINITY OF REFERENCE AMPlENT (ASSUMED CONTSTANT WITH Z
     SALR-SALJ

     DENSITY Op PLUME AT OUTFALL(SIGMA UNITS)
     DENSITY OF REFERENCE AMBIENT(SIGMA UNITS)
     DEUS. OF REF. AMBIENT AT Z - Zp (SIGMA UNITS)
(J)ALPHANUMERIC CASE HEADfR ARRAY
J) ALPHANUMERIC DATA INPUT FOP CERTAIN CONTROLS AS FOLLOWS:
                     SET UP ARRAY WRITER WITH OLIST(I)
                     OPTIONS.
                     INTEFPOLATE ARRAYS GIVEN BY ELIST(I)
                     FINDS ISOLINES OF VALUE ISOLN(K»N)  FOR
                     ARRAY MATCHING ELlST
-------
CO
oo
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
                           TLIST = CONTr
                SET ijP PPOGRAM LOGICAL CONTROL FROM
                 DIRECT(I) DATA.
               VO
               VMB
               ZR
               ZRP

               IPMAX
               INMODE
                ITMAX
                ITAPE
                ITtMP
               KASE
               KT
               NEDDY
INMODE=2
1NMODE=3
INMODE=i*
               NCR
               NJ
CENTERLI^E VELOCITY THERMAL pLUME AT SYSTEM IN-BOUNDARY
CEN'TERLINE VELOCITY (VM/VO) AT Z=ZB
ELEVATION JO GRID ROTTOM PHYSICAL BOUNDARY »Z/DI A
VERTICAL REFERENCE Pr-ANQTL NUMBER

MAXIMUM NUMBER OF ITERATIONS FOR psi ITERATION
INFLOW BOUNDARY INPUT DATA MnDE +
   INMODE=0   INPUT FROM DATA
              GAUSSlAN-FlOW ESTABLISHMENT ZONE
              GAUSSIAN-ESTABLISHMENT
              INPUT CALCULATED FROM SIMILARITY SOLUTION
              INFLOW DATA AT PORT ORlFICF
TOTAL NUM8ER OF ITERATIONS
SIGNAL FOR CONTINUED ITERATION OF OLD CASE*
   ITAPE=0 » NEW CASE
   ITAPE=1 » CONTINUE ITERATIONS OF OLD CASE
SIGNAL FOR DENSITY OR TEMPERATURE INPUT
   ITEMP r 0» SIGMA-T INPUT
   ITEMP = 1» TEMPERATURE IN^UT
CASE NUMBER
SIGNAL FOR TRANSFORM OF LINEAR RADIAL COORDINATES+
   KT    = 0» LINEAR RADIAL COORDINATES
   KT    r 1» TRANSFORMED ACCORDING TO R=SANH(X)
SIGNAL FOR TYPE OF RADIAL EDDY TRANSPORT COEFF CALCULAT
   NEDDY=Q • ER = CONSTANT
   NEDDY=1 ' ER = FO*R.5*VMAX'PRIOP SPECIFICATION OF R.
   NEDDY=2 » ER = EO*R.5*VMAX»RUNNING CALCULATION OF R.
NUNBER OF ITERATIONS PERFORMED AT ER=Eo*i.i78 BEFORE
   RUNNING MIXING LENGTH CALCULATIONS
   USED WHEN NEDDY = 2
RAnlAL CONVERGEN E RANGE
NUMBER OF MODES» RADIAL DIRECTION

-------
-p.
CO
c
C
C
C
C
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
NJJ
NK

NKK
NL

NH

NOX(J)
NOUT

NMAD
                           NJ-1

                           NUf-'bEP OF NODES'  VERTICAL DIRECTION
                           NK-i
                         NJ+1
                         NK-H
                           NUVBEPING

                           NUMBER OF
                     FOR OUTPUT HEADING* SET  IN MAIN PROGRAM
                     ITERATIONS FOR LINE PRINTER OUTPUT
                         SIGNAL TO CALL RICHARDSON VODIFIER ROUTINE
            NMAD  = Of DONOT CALL
            NMAD  = 1-b  S£E SUBROUTINE RCHMOD
NTTY       NUMBER OF ITERATIONS FOR CALCULATION MONITORING OUTPUT
              uPT =1 » CALL PLABAK
NPI        NUMbER OF ITERATIONS ON STREAM FUNCTION  IN MAIN COMP
WRITE(J)   SIGNAL TO CALL OUTPU- OF SPECIFIC DATA
NX       MAXIMUM VALUE OF INDFX j FOR PLOTTING
OLIST(J)   CHARACTER DATA INPUT SIGNAL OUTPUT ARRAYS DESIRED
           OPLIST(J) MATCHES IM«EDDFD DATA DLIST(J)
           TO SET VALUE OF NRlTf(J)
                                     POTENTIAL FLOW STREAM FUNCT
                                     VISCOUS FLOW STREM FUNCT,
                                     DENSITY DISPARITY
                                     VORTIClTY
                                     VERTICAL VELOCITY
                                     RADIAL VELOCITY
                                     GAMMA CONSTITUENT
                                     TEMPERATURES
                                     NORMALIZED DENS. DISP.
                                     NORMALIZED VERT. VELOCITY
                                     NORMALIZED TEMPERATURE
                                     RADIAL EODY FACTORS
                                     VERTICAL EDDY  FACTORS
                                     RICHARDSON NUMBERS
                                     ISENT
DIRECT(I)  LOGICAL  CHARACTER DATA FOR PROGRAM CONTROL
           REAL)  IN  UNDER TLIST OPTION CONT».
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
OLIST
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
1)
2)
3)
4)
5)
6)
7)
R)
9)
10
11
12
13
14
15
-
—
—
~
—
~
—
—
—
) =
) =
) =
) =
) =
) =
PSIP
PSIV
DELT
OMEG
VELV
VELR
GAVIA
TEMP
NOEL
NVEL
NTEM
PFAC
VFAC
RICH
f
t
t
f
f
»
t
>
t
i
r
»
r
r
BLANK
slRITF.
viRITF
'/RITE
-RITE
VRITE
WRITE
^RITE
WRITE
'.-,'RITF
WRITF
•JRIT^
'.--'RITE
AiRITE
WRITE
«T PR

-------
-PS-
CD
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
DIRECT (D = 6UOY» :

            BLANK

DIRECT(2> = UNCP»:



l,TPECT(3) = GRAD' :

            bLANK

uIREcTu) = TRAN» :


            bLANK

DlPECT(5) = TEMP»:
                                       BLANK
                                     = MOMT»:
                           DIRECT(7>  =
            BLANK
            NPCH»:

            BLANK
 Bt'OYANCY COUPLED FLOW.
 CONTRLU) = .TRUE.
 MOMENTUM FLOW ONLY* NO BUOYANCY
 CONTRLd) = .FALSE.
 NO BUOYANT INTERACTION1* BUT BOTH
 TEMPERATURE AND SALINITY OR
 CONCENTRATION APE COMPUTED.
 CONTKH2) = .TRUE.
 AMBIENT STRATIFICATION
 CONTRL<3> = .TRUE.
 IF HOMOGENEOUS AMBIENT
 CONTRL<3) = .FALSE.
 TRANSFORM RADIAL COORDINATE
 ACCORDING TO  R = SINH(XI>
 CONTRLU) = .TRUE.
 FOR LINEAR PADIAL
 CONTRL(U) = .FALSE-
 FLUID STATE INPUT DATA TO BE
 GIVEN IN TLRMS OF TEMPERATURE
 (DEG. C OR F)  AND SALINITY (PPT)
 IF TEMP* OPTION USED WITH INPUT
 IN DEGRFES C»  THEN CENI» OPTION
 MUST ALSO BE USED.
 CONTRL<5) = .TRUE.
 FLUID STATE GIVEN IN TERMS OF
 SIGMA-T AND SALINITY.
 CONTRL<5) = .FALSE*
MONITOR INFORMATION TO BE PRINTED
 AT EACH ITERATION.
 CONTRL<6) = .TRUE.
 DO NOT MONITOR.
 PUNCH RESTART DATA TO CARDS
 CONTRL<7) =• .TRUE.
 DO NOT PUNCH

-------
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
              DIPECT(8> = TAPF»:
                          BLANK
              DlRECT(g) = SAVF»:
                          BLANK
              GlRECT(10)=
                          BLANK
              DIPECT(11)= TURB»:
              DIRECT(12)= CENT' :


                          BLANK

              DIRECT(13)=
                          BLANK
                       INITIALIZE ARRAYS FROM RESTART
                       DATA FILL OR TAPE. MUST EQUIP OR
                       ASSIGN LUN 7.
                       CONTRL(P) = .TRUE.
                       00 NOT READ RESTART DATA FILE
                       SAVE ARRAYS FOR RESTART FILE' OR
                       PLOT FILE. MUST EQUIP OR ASSIGN
                       CONTRL19) = .TRUE.
                       00 NOT SAVE
                       PERFORM INVISCID FLOW COMPUTATIo
                       FOR CASF INITIALIZATION
                       CONTRL<10)= .TRUE.
                       NO INVISCID COMPUTATION
                       COMPUTE AMBIENT TURBULENCE AND/O
                       CONSIDER OERIVATIVES OF THE EDDY
                       TRANSPORT TFRMS.
                       CONTRL<11)= -TRUE.
                       TEMPERATURE INPUT DATA SPECIFIED
                       IM DEGREES CENTIGRADE.
                       CHNTRL<12)= .TRUE.
                       TEMPERATURE INPUT DATA SPECIFIED
                       IN PEGREES FAHRENHEIT.
                       TEMPERATURE OUTp-jT RESULTS
                       SPECIFIFO IM DLiritES CENTIGRADE
                       CONTRL(13)= .TRUE.
                       TEMPERATURE OUTPUT RESULTS
                       SPECIFIED IN DEGREES FAHRENHEIT.
IN
OUT
   UNUSEC CONTRL OPTIONS :
CARD READER  LOGICAL UNIT
LINE PRINTED LOGICAL UNIT
                                          r »CL15» »CL16» »CL17» ,

-------
  DIMENSION  OLIbT(l5> .DLIST(IR) .pLlSK*) .ELISTA) »RLIST<5)
  DIMENSION  OPTIONU) »DATA(l5) .CLISTU7) »DTRECT(17)
  INCLUDE    COMLST.LlST
  DATA < DL1ST(I),1 = 1.15)/bHPSIr..5nPSIV..bHDELT..5HOMEG»»5HVELV,,
 1           5HVE.LR . » 5HGAMA » » 5HTE"'P» . SHND^L • r 5HMVEL • ,
 1           5HNTEM..^HRFAC..^HVFAC..SHPICH.»5H       /
  DATA.1=1»5)/5HPbIV..SHnELTr.^HGAMA,.SHTE^P.»5HOMEG»/
  DATA
  IF(h'PI.EU.U) NPI =  1
  IF(IPN'AX.EO.C)  IPMAX  = 100   '
  IFlNCR.rQ.o) NCR =  ^J-l
  IF(NX.EO.O) NX  = NJ-1
  IF(KASE.EO.O)  STOP
  WRITE (OUT. 1001+) TLABEL'DATE.TIM
  READ
-------
-p*
.£»
OJ

7
6

9
10


11
12

13
14

15










18









GO TO 15
DO 81 =
OLIbT(l)
60 TO b
DO 10 I
RLISTd)
KPLQT
60 TO 6
DO 12 I
PLlST(l)
GO TO 6
DO 14 I
CLISTd)
GO TO 6
NJJ
KKK
NH
NL
JPOHT
KPOh'T
NB
ITERS
ITNOO
ITUO
DO 18 "-'
NOX(M)
DO 20 J
DO 20 K
PSl(J'K)
OELTt J'K
OVEG
-------
 20
 30
100
lOb
107
    RlCHlJri') =  0.
    UGR/\D(J,K) = 0.
    L,GKAD( j»K) = 0.
    CONTINUE
    DO  30 n
    K
    MOMK + 1)
    MOi\ilK+2)
    MONIK+3)
                1»3
                (N-1>*10
                2
                MK/5
    MCN(K+b) = ?
    MON-(K+7)
    MON(K+8)
    MON(K+9)
               t.'K
               I.JJ/2
               liK
               fJJ-1
CONTINUE
DO 100 T = 1§15
DO 100 J = 1 »15
IF(oLIST(T) .EG.OLIST(J) )  NRlTE(j)  = I
CONTINUE
DO 105 I = 1,5
DO 105 J =
IF(ELIST(T)
!F(tlLIST(I)
EQ
EO
RLIST(J) )
PLIST(J) )
                              NiDPT(j)
                              ISOPT(j)
                                       = I
                                       = I
                               CONT«L < I )  =
    CONTINUE
    DO 107 I = 1,17
    DO 107 J = 1 »17
    IF (QlRECT(I) .EQ.CLIST(J) )
    CONTINUE
    IF(KPLOT.fr&.l.ANC'..NCT.CoNTRL(9) )  GO TO 160
    KT       =0
    iitnP    = o
    IF(CONTPL(4) ) KT =  1
    IF(CONTRL(b) ) 1TEMP =  1

-------
cn
 110

 120
 150

 160

1000
1001
1002
1003
     CALL INPUT
     CALL READY
     CALL PL4BAK
     CALL ETIMEF (START)
     WR1TE
              FORMAT (////
              1  3bH      St4*
              2/26H      PROGRAM
                      SET-UP TIME = F5.2' 5H SFC /
                      tSsSSs^fsS'*;*
1001 FOHMAT(///12A6»5XHA6)
1005 FORHAT(//Q     YOU CAN NOT SAVE A PLOT plLE WITHOUT ASSIGNING
    *VE FILE TO LUN 8C/W     EITHER DELETE PLOT FILE CALL OR EQUIP
    *8ffl/Q     PUtJ ABORTED - - TRY AGAINQ)
     ENU
                                                                             A  SA
                                                                             LUN
     SUBROUTINE
     DIMENSION
     INCLUDE
                          INPUT
                          DATAdO)
                          COriLST»LlST

-------
            10
            20
CTl
            25

            30
IF( .NOT.CCUTRUS) )  GO ro 10
READ < 7)  ITNO»OMEG»D^LT»UX »UZ»PS T *
ITNOO    = ITNO
REWIND 7
N6       = 0
REIAQ(IN»1000) DATA»JI»KI»N1
      (100»20»30»*40'60»60»60»60»70»70»7P»RO)
         = DATA(l)
         = DATA (2)
         = DATA (3)
NI
GO TO
DIA
DX       =
DZC      =
ZO       =
VO       =
JPORT    =
KPORT    =
IF(DATA(8)
IF(DATA(9).F:(5.0.)
IF(DATA(10).EQ.O.)
                          DATA (5)
                          OATA(6)+.Q1
                          DATA(7)+.Q1
                                     = 1.
                                 DMB = i»
RO
IF
DO  25
DZ(K)
60 TO
TBOT
TO
TR
                      K  =
      10
           ,5*DIA
           )  GO TO
           1»NH
               DATA<2)
               DATA(3)
               CONTINUE
               DATA<1)
               DA1A(2)
               DATA<3)
                                  10
           QATA(3)
           DATA(D
           DATA(3>
          12)) GO TO 32
          3.0) GO TO 32
           5./9.*(DATAd)-32.)
           5«/9.*
-------
-pi
-vl
             60
SIGTJ    = PATA(l)
SIOTR    = DATA(2)
SIGTB    = PATA(3)
OSIGT    =: DAT A (1)
SALP     = DATA(5)
SALJ     = HATA(6)
LXb      = DATA<7)
EXK      = DATA(S<
IFlEXS.iy.o.) tXS =  .999
IFltXR.EQ.O.) tlXR =  .999
D5ALT    = SALK-SALJ
bELTJ    = SlGTR-SlC-TJ
R5tA ERATIO =- .01
iFtf.Z.Er .0. ) EZ = .1
IF(RRP.!=:Q.U. ) RRP =  I./.714
IF(ZRP.rQ.O.) ZRP =  l./.71<*
CONTINUE
GO TO lo
DO  65 -I = JIfKl
KAT      = N-Jl+1

-------
                IF(NI*EQ*5>
uz m»i
DtLT(M»l
RSLA(N)
00

65


70




75

60



85

1000
100

!F(Nl.Eu
CONTINUE
iFChl.EO
GC TO 10
JI
KI
DO 75 r;
KAT
MON(N)
CONTINUE
GO TO 10
NN
DO 85 M
NA
ISOLN(KI
CONTINUE
GO TO 10
FORMAT ( 1
RETURN
ENu
.8)

.5)

—
—
—
TI
—


—
—
—
DZ(N) = D

MB = KI

(fJ!-9)*l041
JI+9
JI»KI
N-JI+1
DATA(KAT)+.0001


JI-1
1»10
NN+N
»MA) = DATA(N)


OF5




.0*315)


DATA(KAT)
DATA(KAT)
DATA(KAT)
DATA(KAT)*?.
                SUtJROUTTNE RFADY
                INCLUDE    COMLST»LIST
                NOTE^P   = 1
                1F(KPORT.E(J.O) KPORTz 1
                IFdNMOOE.EQ.M-.AND.CONTRLCf) )
                RER      = 39,
                REZ      = RO*VO/EZ
                2<1)     = ZB
                ZC(1)    = 2(1)-.25*OZU)
        = .8pl37^59/(JPORT-i)

-------
   ZPORT    r o.
   DO   5 K = 2rNH
   IF(K.LE.KPOPT) ZPORT = ZPQRT-«-.5*DZ (v )
   Z    = l.b+ZC(K)-ZPOKT
 5 CONTINUE
   DZTOT    = ?(NK)-ZB
   IF( iNMOoE.LQ.it) DZTOT = Z(NK)-ZrORT
   DZT5     = DZTOT*. 5
   IF(DSlGT.Eti.O) GO TO 15
   RSEAd)  = SIGTR
   DO  10 K = 2rNH
   RSEA(K)  = R5EA(K-D+DSIbT*DZ(Kl/(Z(NK)-ZR)*.5
10 CONTINUE
15 DO  20 K = 1»NH
   IF(.NOT.CONTRL<12).ANO.lTEMP.r E.O) PSEA(K) = 5./9.*(RSEA (K)-32« )
   IF(.NOT.CONTRL(3) ) RSEA(K) = Slf;TR
   IF(CONTRL(3» RSEA(K) = blGMAT < f,ALR,RSLA (K ) , JTEMP)
   RStA'K)  = RSEA<«)+IOOO.
20 CONTINUE
   DXi:      = DX*DX
   FO       = VO*VO/(DELTJ/(SIGTJ+100C.)*2.*PO*32.2)
   IF(PELTJ.EQ.O.) FO =0.
   DO  50 K = 2rMH
   DELT/DELTj
   6AM(NL»K)= 0.
   iF(.NOT.CClMTRLdl) > GO TO 50
   EDI      = (RSEA(K)-R5EA(K-1))/;RSEA(1)*RO*DZ(K)*.3048)
   IFCED1.EQ.O.) EDl=-l.E-t*
   IFC.NOT.COi^TRLl^J) EDI = -1.E-*
   ED       = -l.E-7/tDl
   FZ(1»K)  = EO/tZ
   FZ(UL»K) = FZ(1»K)

-------
C
C
C
4-0
5(J
55
60
      AK1      -
      DO  40 J =
      FZ(J'K)  =
      CONTINUE
      CONTINUE
      IF(KPORT.LE
      DO  55 J =
      DO  55 K =
      FZ(J'K) =  .
      CONTINUF
              .5*uZTOT*SQKT(.689)/(Z(NK>-Zp
              prMJ
              EXp(-(AKl*Zl>**2)+FZ(l»K)
              .1.0R.CONTRL<11> >  60 TO 60
              If ML
              1'NH
              0001
              ABS( (SI6TR-SI6TR)/?B)/DELTJ
               G = 0.
              UZ(NK)
   SET-UP FOR Z-DIRECTlON CONSTANTS
70
DO  70 K
SZ(K'D
SZ(K'2)
SZ(K'3)
SZ(K'4)
SZ(K'5)
SZlK '6)
SZ(K»7)  =
SZ(K'B) = 2
SZ(K'9) = i
SZ(K'IO) =
SZ(K'll) =
SZ(K»12) =
SZ(K'13) =
SZ(K'l^) =
SZ(K'15) =
SZ(K'l6) =
CONTINUF
                 2»NK
                  2./REZ*
                                                             )/DZ(K)
                                            
-------
c
C     SET-UP FOR  R-DIRECTION
C
R(
XX
X(
XR
uo
x(
IF
XX
XR
RC
M
1)

i)
(1

J)
(J 	



}
00 j

• EG.
«
—
z
—
~
—
2)
0.
n.
^ *
0.
?»
X(
x(


5*DX

ML
J-D
1) =





+QX
0«



































= xx+ux
(J
(J
J)
)
)

_
~
n
XX
5 A
SA

NH 'KT
X»KT)

)










(K
X
{


(
»KT)



()(X»KT)*CASH(XX»KT)/K
*DV)**P)/RER
J) »KT
CASH(


/(CASH t
*RRP
( J»f
*sci
)/


• *DX)


J]
(J)


>KT)*DX
+CONC)
-CONC >
*
*


)**2)
•C!
. c.
                                                            J)>
                         J)*CASh(X(J)
      lF(CONTf?L(ii) )  SC(J»5)  =  0.
      IF(.NOT.CONTRL(1»  5c(j»b)= 0.
   80 COIMTlNUF
      RC(1>    = -RC(2)

-------
    DO  90 J = ] »riL
    DO  90 K - 1»NH
    IF(RB(K).LT.KCIJ).O^.K.LT.KPOPT) FR(J'K) = EkATlO
 90 CONTINUE
    iF(UJMOrE.NE.O) GO TO 15U
    DO 100 J = 2rNJ
    PSi(J'l) = PSI(J-lrl) + uZ(j»l>*RC(j)*CASH(X(vJ) »KT)*DX
100 CONTINUE
150 lF(lNMODE.Ln.3) CALL sIMjET  (fCn»ZB»DZ (1 > »G»FO r V«B» VMB1 »r,MB»DMB)
              LG.l) CALL GAUSS (1)
              E-Q.S.OR.INi^OUE.EQ.S). CALL GAUSS (2)
    UZ(1»D  = UZ(2»D
    IF IINMOOE.NE.^) GO TO 20U
    DO 160 J = \fJpOPT
    DELT(J»KPOKT)= 1.0
    GAM(J'KPOPT) = 1.0
    IF(J»EQ,1) GO TO IbO
    UZU'KPORT)   =  (PLX + 1)*(2*PLX + 1)/(2*PLX*PLX)*(1»-RC(J) )**(1./PLX)
    PSI(J'KPORT)  = PSI
160 CONTINUE
    PSIR     = PSKJPORTrKPOKT)
    UZH'KPORT)   = UZ(2,KPORT)
    LO 170 K = 1»KPOPT
17U PSI(JPORT»K)= PSIB
    DO 180 J r JPORTiNJ
    PSIU'I) = PSIB
180 CONTINUE
    NB       = JPORT
200 CONTINUE
    RETURN
    END
    SUBROUTINE  PLABAK

-------
    INCLUDE    COMLST.LTST
    DATA/DF/lHP/CF/lHC/
    TU       r pF
    IF(CONTPL(1?) )  TU  =  CF
    EPR      = l./KRP
    EPZ      = l
    toRITE(OUT»lPOl)
    M H 1 TE ( Oi iT » 1 0 02 ) N J » NK » OX » RO . VO » FO » TU » TU r TU » TK r SAL J » SALR » S I GT J
   i                SIGTR»KEK»KEZrPL.XrUZ(2'KPORT)
    V.R I TF ( OUT • i (} 07 ) NMAD » KPOKT » JPOR? » NEDDY r EXS ? BETA » AK 1
  5 ftRlTE(0'jT»10in) DAT?:»T1M, (NOX(K) »K=lffl) '
    DO 1^0 j r l,rjj
    v\R!TE(OUT»10l2) J» X < J) »RC ( J) »R ( J) » KC ( J,L) »L=1 »fl)
    CONTINUE
    WRITE(OUT»101Q) DATF-»TIM» (NOX(K) »K=Q»15)
    DO 145 J = ItHj
    WRITE (0( IT. 10 12) J»XL=9»15)
    COI4TINUE
    •A'RITE(OUT»1014) OArErTIM» (NOX(K) »K=1»P)
    DO 150 y = 1»MH
150 D»RITE(0«.iTilOl2) K »DZ (K ) »ZC (K ) »Z (K ) » (SZ (K »D »L=1 '8)
    WRITE(OUTrl01*i) DATE»TIM» (NOX(K) »K=q»16)
    DO 155 K = 1»NH
    WRITE (OUT ,1012) K»D2(K) 'ZC(K) »Z(K) • (SZ(K»L) »L=9»16)
155 CONTINUF
    WRITE(OUT»lQle») CATE»TIM» (NOX(K) »K=  17»20)
    DO 165 K = IrNH
    WR ITE( OUT , 1006) K »U? (K) '1C (K ) »Z (K ) » (S7(K»L ) »L=1?»20 )
   1           RSEA(K)
165 CONTINUE
    WRITECOL.'T»1004)
190 DO 200 J = I»NJ
    L        = KPORT

-------
               VvRITE<0'iT.1005)  PSI < J»D »U^ (J»L) *UX (J»L) *DE.LT(J'L) *GAM(J*L)
           200 CONTINUF
cn
1001 FOKf>.AT(/// <+OH PARAMETERS FOR THERMAL PLUME CASE 13)
1002 FORMAT (///
1/55H NUMBER OF RADIAL NODFS ( X-DIPLCTION) - - - - -
3/55H NUMRFH OF VERTICAL NODFS (?-DIRECTIOM) - - - - -
U/55H RADIAL NODE THICKNESS (X-DIRFCTl^N) * DX 	
b/5b(i PLUME OUTFALL PORT RADIUS (X-CoORD)* RO- - - - -
7/55H PLUME OUTFALL PORT VELOCITY *FT/SEC )» - - - - -
£/5bH DENSlMETRiC FROUDE NO* AT OUTFALL PGRT(VO**2)
A/5XQTEMPERATURL OF REFERENCE AMpIF.NT
-------
                      CLNTE.KLINE  VALUt.  OF  GAyMA-CONSTlULNT     ' GMB = F9«3
                      CLNTFRLINE  VALUE  OF  BUCYANCY  PARAMETER   »OMB = F9.3)
        1004 FOKMAT(//Q     RADIAL DlbTRlHUTlONS  O/
            1/OX65HPSI       VERT vtLO     RAn  VEI 0       OE'LT          GAMMA
            2       //)
        lOOb FORMATltP6(tTl3.3»lX»
        1006 FOR!iAT( 13 » F7.2» 2F10 • 3 t iPbEll. * »Fl3. 5 >
        1010 FORMAT (1H1»K    DATE Q2Ab»ioi  Tl^r  Q2A6/
            i bun     COMPUTED CONSTANTS  FOR RADIAL DIFFERENCED - -SC(J»D   //
            S 30h  J   X(J)       R/RO      R(J)      »8(I6»5X>/)
        1012 FORMAT(i3fF7.2r2F10-2rlPoFilO)
        1013 FORMAT(/J
        1014 FORMAT ( 1H1 tQ    DATE Q2Ab»r»i  TTMf  02A6/
            1 60H     COMPUTED CONSTANTS  FOP VERTICAL  DIFFERENCES - -SZ  //
in           c. 3UH  K   LiZ(K)      Z/UO      7//
            2 30H  K   UZ(K)     Z/DO      Z      »i*(I6»5Y)  '
            3 30H DELT(lMLfK)
             RETURN
             END
             SUBROUTINE  STREAM  (iT
             SUBROUTINE CALCULATES THL  TWO   OlMENSlO^AL   STREAM  FUNCTION'  PSKj
             INCLUDE    CO^LST^LIST
             OM£GA3= 0.
          10 DO 120 I = 1,JT
             QPMAX    = 0.
             SET OUT-BOUNDARY STREAM  FUNCTION  FOP  NEXT  ITERATION CYCLE
             DO  20 K = 2»NKK
             PS I ( N J» K ) = ?,. *RS I ( N J-1»K ) -PS I ( N J-2 »K >
          20 CONTlNUr
             DO 100 J = 2» i\|jJ

-------
    Al       = SC(J,11)
    A2       = SC(J»12)
    A3       = 5C(J.13)
    00 100 K = ?»MK'
    IF(J.LE,JPGRT.Ar'C'.K»LE«KPOrtT)
                                               TO
                         AH-SZ(K»B)
                         PSKJ'K)
                         .O) GO TO 50
                       = ,5*(OMEG

                       = OMEGA1+SZ(K»12)*S I1J t K-l) +SZ (K»10) *PSI (J, K+l > -»-OMEGA3
                         *R(J))/CON
                                           )/PSI(J»K))
tn
    CON
    PS 10
    IFtNSKlp
    OML6A1
    OMLGA2
 **5 CONTINUE:
    OM£GA3
 50 PSKJ'K)
   1
   2
    DEL      = ABS(
    IF(J.GT.NCR)  GO TO 95
    QPMAX = AMAXKDPA'AX'DED
    iFtDPf^Ay.Gl .DEL) GO TO 9b
    NODE(5)  = J
    NODE(6)  = K
 9b PSKJ'K) = PSIO+EXT * (PSI (JrK )-pSIO)
100 CONTINUE
    IF(DPMAX-LE..0005) GO TO 130
120 CONTINUE
130 iF(fjSKlp.EG.l) GC TO IbO
    ITNC     = I
    CALCULATE VELOCITY FIELD
15U DO 250 J r 2»NJ
    Al       = SC(J»14)
    DO 250 v= 2»NK
    IF(J.LT.JPORT.ANC.K.LT.KPORT) GO TO 250
    UX(J'K)  = -(PSKJ»K)-PSKJ»K-D )/(R(J)*D7(K)
    UZ(J'K)  = (PSlf J'K>-PbKJ-l»K)
    IFlj-Eu.2) UZd'Kl = UZ(2»K)

-------
                 UX(J»NH> = UX
     IF
-------
en
   UMAX    = ITNO+ITMAX
   N2       = IT^AX
   NED      = NED + IT^O
   IF(NED.GT.ITNO.OR.NEDDY»tO.O) GO TO 15
   CALL EUpY(NEOnY)
15 CALL ETIMEF(START)
   DO «00 L - M»N2
   ITNO     = L
   DDMAX    - 0.
   IF( .NOT.CONTRLU) .AND. .N Go TO 200
   MASK     = 0
   IF(J.EGI.JPORT.AND.K.EQ.KPOHT+1) MASK = 1
20 Al       = SC(J»7)
   A2       = SC(J»8)
   A3       = SC(J»9)*FR(J»K-1)
   AU       = 'SC(0»10)*FR(J»K-1)
   A5       = SC(J»6)*Fp(o»K-l)
   UPOSl    = ABS(UX(J-1»K» * UY(J-1»K)
   UPOS2    = ABS(UX(J»K))   * UX(j»K)
   UNLGl    = ABS(UX(J-1»K)) - UX(j-l»K>
   UNE62    = ABStUXU'K) )   - UX
-------
             DGAM     = GZCON*DFZ*(DPi*GAM(J,K + l)-nMl*GArv-(JfK-l)-DCO* -AM(J
             DGKAD(J»K)    = DPl*DELT(JfK+D-DMl*DELT(J»K-l)-DCO*bELT(J»K)
             CDELT    = GZCON*DFZ*DGRAO(J»K)
          5C COUTlNur
             BO       = FZ(J'K)*SZ(K»b)
             AJ1      = (Al*UPOSl+A3)
             AJ2      =
             AK1      =
             AK2      = (SZ(K»4)*VNEG2+SZ(K'7)*FZ(J'K»
             DJi      = AJ1*OELT*OELT
             GJt      = AK2*GAM(JfK+l)
             GAMO     = GAi-I(J»K)
             DELTO    - DELT
             GAM (J»K)=
             DEL      - ABS((DELT(jrK)-DELTO)/DELT(J,K)
             IF(J.GT.NCK) GO TO 195
             DDMAX    r AMAXl (DL)MAX»DEL)
             iF(DDMAy.GT.DEL) GO TO 195
             NOUEtl)  = J
             NOUE<2)  = K
         195 CELT(J»K>= CELTO+EXR*(DELT(J»K>-DELTO)
             GAM(J»K) = eANO+EXH*(GAM(J»K)-GAMO)
         200 CONTINUE
             SET BOUMDAKY VALUES FOR UELTA(J»K)
             DO 230 J = 2fNJ
             GAM(J»NU)= GAM(J»NK>
                  Jr'iH)= DELT(J»NK>

-------
    IF(J*LT.NP) (30 TO  230
    DELT= DELTAS'
    GAM(J'l) = GA,y(Jr2)
230 CONTINUE
    IF(lNMOHE.NE.t) GO TQ  25Q
    DO 2<+0 K. = 2»KPOPT
    DELT
    GAi*i( JPORTfK)  = GAM
    UNEti2    = ABS(UX(J'K»   -  UX(j»K)
    VPOSl    = ARS(UZ(J'K-D ) *  UZ(J»K-1)
    VPOS2    - ABS(UZ(J'K))   +  U7

-------
               UP1      =  .5*(UX(
               UCO      =  ,5*(UX(J'K   )HIX(J-1.K
               DFZ      =  SZ(K»15)*FZ(J»K-H)-S? ) )*UGRAD(J'K)
           293 CONTINUE
               BO       =  FZ(J»K)*Sz(K»l)

               01       =  1»/-OM£GO)
           300 CONTINUE
               CALL STREAM (ITL'PS»1)
         C     SET CENTEPLINE AND CUT BOUNDARY
               DO 310 K = 2»MK
               OMEG
-------
                VORT2
                00
                DUZ
                DVR
en
ro
      SET BOTTOM BOUNDARY VORTICITY
      IF(INMOOE.EG.'4) GO TO 35u
      SET SLIP BOUNDARY
      DO 330 J = UBrNJ
      OMLG(J»t>= -OMEG(Jr2)
  330 CONTINUE
      SET INFLOW BOUNDARY VORTICITY
               = l./DZ+UX(J-:
      DVR      = .5*(UZ(J+1»2)-UZ(J-1»2))/(RC(J+l)-PC= DUZ-DVR
  360 CONTINUE
C     SET PORT SIDE NO-SLlP BOUNDARY
                                                  DOES NOT ENTER IN CALCULATIONS
                DO 370
                UKP1
                UK Ml
                UKC
                DVR
             K = 2»KPOPT
               = .5*(UX(JPORT»K+1)+HX(JPORT+1»K+1))
               = .5*1UX(JPORT»K-1)+UX(JPOPT+1»K-1))
               = ,5*(UX(jPORT»K   )+uX(JPORT+1tK   ))
               =  SZ(K»15 >*UKPl-bZ(K'14)*UKM1-SZ(K »16> *UKC
               = . 5* (UZ (JPORT+2»K) +UZ (JPORT+2, K-l) -HjZ (JPORT+11 K)
                 UZ(JPORT + 1»K-l))/(RC(JPORT+2)-RC(JPORT))
            370 CONTINUE
                SET INFLOW BOUNDARY VORTICITY

-------
              LO  380  j  =  ?>jpORT
              DVR       =  (UZ(J+1»KPQRT)-UZ(J-1»KPORT) )/(RC(J+D-RC
-------
    IF(KOD(L»NTTY) .EG.O) CALL.  OUTPUT  (£)
800 CONTINUE
    IF(lNMonE.LO.U) TEMPER(JPORT»KPORT)=TFMPEP(JPORT-1»KPORT)
    1F(.NOT.CONTRL(9)) GO  TO 880
    iivR ITE (8) I Tf ;0»OMEG»DEL.T»UX r(JZ» PC I' GA M
    N3UPTS   =  0
    DO 810 j =  i»5
    IFtf--}3DPT(J) .EO.O) GO TO 810
    N3UF-TS   =  N3DPTS+1
810 CONTINUC
    UZ       =  NK
    IFCr,l3DPTS.£.e.C) GO TO  830
    WRITE(8) KASE»DATE»TlM»TLA6EL»N3DPTS'JPoRTrKPORT»NX»NZ
    DO 820 J =  1,5
    L        =  N3DPT»(Z  (N)»N=1»NZ)»
   1                    ((PSi  (NrM)»N=1»NX)»M=1»NZ)
    IF
    IF(L«EQ.5)  URITE(8) L»(RC(N)»N=1»NX)»(ZC(N)»u=2»NH)
   1                    ((OMtG(N»M)»N=lrNX)»M=2»NH>
    LL       =  L
    IFtL'EQ.l)  WRITE(OUTrlOOH) LL
    IF(L-E0.2)  WRITE(OUT»1005) LL
    IF

-------
      IF(TEMPrR(2,NK>-TEMPER(NL»NK).LT..1)  GO TO BfaO
      IF(CONTf>L(l3» WRITE(OUT»1001>  OC
      IF(.NOT.CONTRLll3)) WRITE(OUT.lOOl)  DF
      DO 850 j r 1 ,p,o
      TDLl_T(J) = TE"!pFp(uL,NK)+J
      IF(TOELT(J).LT.TEMPER(2»NK))  GO TO  850
      LAbT     - J-l
      GO TO 855
  3bU CONTINUE
  855 DO 870 L = l»LAf,T
      DO 860 J = ?»-\>j
      IF(TEMPER(J»NK).GT.TOELT(L))  GO TO  860
      RAD      = (RC(J-D + (TDELT(L)-TEMPER(J-1»NK)
     1           -TEMPER(J-1»NK))**DZ(K)+GAM(NCR»K)*i jFACE
      lF(uFACt.LT.O,)GAMCON=GAMCON+2.*R(NCR)*DZ(K)*GAM(NCp+l»K)*UFACE
      GAMP IF   ~ GAMC*FR (NCR' K ) *DZ < K > * (GA^ (NCR»K ) -G <\M < NCR-H t K ) ) +GAMDIF
  895 CONTINUE
      GAMCON   r 100.*GAtfCON/GAMlN

-------
      6AMDIF   = 100.*GAMDlF/GAMIN
      GAMSUM   = GAMCON+GAMDIF
      GAMERR   = GAMSUM-lGO.
      WR I TE ( OUT » 1 0 1 1 ) J » GASCON » 6 AMD IF r GAMFRR
  900 CONTINUE
      RETURN-
 1000 FORMAT(I6»**Xl5(F6.tt»2X))
 1001 FORMAT (//JSOH     SURFACE ISOTHERM DATA      //
     1/15H       DEGREES  'Al»3bH      AREA  IN           RADIUS  OF
     2/60H      APOVE AM6      SQ. FF-^T      ISOTHERM »  FEET
     3/>
 1002 FORMAT (lln»?(luXFlO*l) )
 1003 FORMAT (///bXl5f ® THREE-D PLOT RECORDS WRTTTEM ON TAPEQ/
     1        fa)         SET  PLOT PARAMETERS NJ =  Ql3»0  NK  =  QI3)
 1004 FORMAT (.-D STREAM FUNCTION PECOPD Vi/RlTTEN TO  TAPE - RECORD NO    SI3)
 lOOb FORMATtQ BUOYANCY PARAMETER RECORD WRITTEN  TO TAPE - RECORD  N00I3)
 1006 FORMAT (E GAMMA-COMSTlTUENT RECORD WRITTEN TO TAPE -  RECORD NO  013)
 j.007 FORMAT (Q TEMPERATURE RECORD WplTTEN TO TAPE - RECORD NO        013)
 1008 FORMAT (Q VURTICITY RECORD WRITTEN TO  TAPE - RFCORD NO          013)
 1010 FORMAT (1H1»//Q     GAMMA-CONSTITUENT  BALANCE ERROR    Q  ///   •
     1   loi     NET COMVECTIVE      h'ET DIFFUSIVE           GAMMA BALANCE
     2I3/ 0     OUTFLOW » PERCENT    OUTFLOW » PrRCENT        ERROR »  PERCEN
 1011
      END
      SUBROUTINE
      INCLUDE    COMLST»LlST
      DIMENSION  FCORE(LK>fRb(LK)»KR
      RATIO    = REZ/RER
      VEDC     = .015
      GO TO ( I0»i:0»120»20'20»5u0r50p) rM
C*    CALCULATE RADIAL EDDY FACTORS USING  PRESCRIBED  MIXING  LENGTH

-------
   10 DO   15  K  =  2,NK
      VMAX      =  .5*IUZ(2'K)-UZ(2»K-1) )
      FR(J'K-D=  . 180*(Z(K)-.25*DZ(K) )*VMAX
   15 CONTINUE
   20 C OUT I NUT
C*    CALCULATE HADIAI.  EQPY  FACTORS BASED ON /i RULING CALC. OF MIXING L
C*    BASL LENGTH  OF POTENTIAL CORE Or PERCENT GAMvA DECREASE AT CENTERL
      iFUNMonL.LT.u) GO  TO  40
      DO  25  K  =  K PORT • Nil
      IF(6AM(2fK ) .LT.GAMtNQ)  GO TO '0
   25 CONTINUE
      IF    =  1.
      IFlK-LT.KCORE) GO TO 60
      R5(K>    = RCCNSO + ^iJZtNbOfKJ-VsO/dJZtNSO^KJ-UZCNISn + lfK) )*
   60 ROb(K)   - Fi'C(N05) + (UZ(Nu5fK)-\/n5)/(UZ(N05fK}-UZ(MOtS+lfK) )*

-------
     1           (PC(N05+D-KC(N05)
   70 CONTINUE
      IFU"»EQ.l) GO TO 100
      DO  90 J = 2'NJ
      IFU"«EQ.2) GO TO 75
      IF(HB(K) .LT.RCU-D } GO TO 100
   75 CONTINUE
      FR(J'K)  r (P5(K)-KCORE(FR( J'K-i>+FR(J»K) )
   90 CONTINUE
  100 CONTINUE
      IFIK.EQ.1 .OR.M.EQ.2) GO TO 400
C*    CALCULATE VERTICAL EDDY FACTORS FZ(J'K) IN SURFACE SPREAD
  120 CONTINUE
      00 1^0 K = KPOKTfNK
      DO 130 j = g,NJ
      IF(RC(J-1).GT.RR(K» GO TO 140
      FZ(J'K)  r RATIO*FH
-------
             152 CONTINUE:
             155 CONTINUF
                 FZC      = VFOC*ZLE.N*UWAX*REZ*2.
                 DO 200 J z 2»NJ
                 DO 150 K = X'NH
                 UAVE:     = .5*(ux(j-irK>+ux(j»K) )
                 IF(UAV£.LT.O. )  KB(J) = K
             150 CONTINUE:
                 DO 160 K = 2»NK
                 IF(RC(J-1) .LE.RB(K) .AND.K.GT.KPoRT)  GO  JO  1S5
                 FZ(J»K)  = FZ(NLrK)
                 IF(K+2.LT.KtMj) .OR.K.LE.NK/2) Go TO  160
                 IF(J.NE.NJ)  Ff;(J»K-D = FR(2rK-l)
                 FZ(J'K) = FZ(J»K) + FZC*UX(J»NK)/UM4X
             15b lF(f,MAD.E(?.0) GO TO 160
           C*    MODIFY RY KICHARDSON NUMBEH MODrL  (IN LATFRAL  PLUME  SPRF.AD ONLY)
^                RICHNO   z +.b/FO*DGRAD(J»K)/(U<$RAD(J»K)**2)
10                !F(HlCh,jO.LT.O. )  RICHNO = 0
                 RICH = FZ(JrNK)
             200 CONTINUE
             UOO IF(?"OD( TTNO»NOuT) ,NE.O>
                 ftRITE(0!.!T»1001) ITNO
                 DO 450 K z 2»NK
                 KN       = KK+2-K
                 Dl       = .5*KCCR£.
-------
 500 RETURN
inuu f-OKf;AT( IHlX/fo)      PLUME COKE F.XTENDS  TO  3UKFACE K
    I/            fc      CORE AbSUwEH TO END AT K  = MH FOR PURPOSES  OF  Er,
    ^DY CALCI'LATJOM,  FR(JtK-l) 0 >
3001 POKVAT(//    I?      pLiiME LATERAL SECTION  GFO^hTRY» ITERATION N0»  0
    lIb//R      K         Z/D
1002 F-CRi AT ( Tlf:»Fo.ii»2X) '
    1            HO' AX»NO^E(i) »NOPF('U »n{jvAX»NOnt(I) f NODE (2)
     DO    5 J -  lf';J
     DO    5 K = IrNK
     IF(i'Z(J»K) .LE.5. )GO TO 5
     iB/RUE (OUT r 1003)
     ISTCP    =  °9999
     GO TO IP
   5 CONTIMUF
     *FSTART)/NOUT
     DO 100 J -  1»15
     L        =  Ni'ITEU)
     IF (L .fu.6)  NOTL'MP =ii
     IF(L»F-:U.O)  GO  TO 10°
     IF (MODE.. EO. 1 )  GO TO Q0
     IF(L.EG.I)  CALL  APoUT(L»Zr R»rSl»
    1                 <+2r|STRLAivi FUNCTION  -  IKROTAT10NAL FLOW

-------
    IF(,'10DL.EQ.O) tO TO  10U
 90 IF(L»EQ.2) CALL AROUT(L»L»R.PSl»
   j                U2HSTREAM  FUNCTION  -  VISCOUS FLOW                )
    IF(L«E^.3) CALL AROUT(L»ZC»RC»D£LT»
   1                42KBUOYANCY  PARAMETER " DFLT                     )
    IF(L'EG.U) CALL AROUT(Lr^C»RC»0"EG»
   1                42HVORTICITY  -  CME6                              )
    iF(L-EQ.b) CALL ARO^T(L»2»RC»UZ.
   1                42HVERTICAL  VFL-OCITY  COMPONENT - U^              )
    IF(L«EQ.6) CALL AF^QUT (L»^C»  R»Ux>
   i                42HRADIAL  VELOCITY  COMPONENT - ux                )
    IFCL-EQ.7) CALL AROUr(L»ZC»RCf6AM»
   i                42HGAMMA-CONSTITUENT                              )
    IF(.NOT.CO(JTRL(13) )  GO Tu  92
    IF(L»E0.8) CALL AROUT(L»ZC»RC»DPLTr
   1                42HTEMPERATURE' DEGREES CENTIGRADE                )
    tO TO 94
 92 IF(L'EQ.ti) CALL AROUT (L»^.C »RC »Dr"LT»
   1                U2HTEMPEKATURF' DEGREFS
 94 CONTINUE
    IF(L»EO,9) CALL APO^T (L »ZC » RC »Dr.LT»
   1                U2HNORMALIZED BUOYANCY
    lF(L«EQ.lnjCALL AROUT(L»^»RCrUZ,
   i                <42HNORMALIZED VFRTICAL  VELOCITY  COMPONENT         >
    IF(L«EQ.11)CALL AROUT(Lt2lC»RC»D|-:LT»
   1                U2HNORMALIZED TEMPERATURE  DISTRIBUTION            )
    IF(L*Eu.i?)CALL AROUT(U»ZC»RC»FP»
   i                12HRADIAL  EDDY  •. IXIMG FACTORS                     )
    IF(L»EQ.1?)CALL APOUj (L »ZC f RC »F;?»
   i                12HVERTICAL  EDDY MIXING FACTORS                   >
    IF(L»EG.14)CALL ARuUT(L»ZC»RC»RfCHr
   1                U2HPICHARDSON N
    IF(L»EQ.lf))no TO loO
100 CONTINUE
    lF(lSTOp.E(i.0999q> STOP

-------
     CALL ETIMEF(START)
     RETURN
1000 FORMAT(]H1
              NO.
              MAX
    2/
    *»7H
    3/  55H
    *»7H
    <+/  35H
    *»7H       (
 1001 FORMAT(iHl
    I/
)    ^
    35H SEC
 1002 FORMATI/
   26H
OF PSI
CHANGE
    KESULTS
ITERATIONS-
IN Pbl  	
                                     FOR ITER.
              MAX CHANGE IN OMEG
                  CHANGE IN DELT
NO.    T5/
  I7»10XfHNODF
  1PE10.3

  ElO»3

  ElO«3
                U5H     STREAM FljNCTjON RESULTS FOR ITERATION
                30H     MAXIMUM RELATIVE ERROR IS   1PF12.3/
                23H     TIME REQUIRED FO" I3»14H ITERATIONS = F6«2»
                   ///)
                1CH ITERATION
                                                                   I5/
    1 10H 	 VERTICAL VELOCITY 	
    2 40H ********* RADIAL VELOCITY *************
    3 ^Oh 	 BUOYANCY PARAMETpR ..........
    U/10H  NUMBER    15(2n (12,lH'12,1H))//)
1003 FORMAT(  ///a     THIS CASE IS APPARENTLY UNSTABLE* RUN ABORTED 0)
     END
  10
     INCLUDE
     DIMENSION
     DIMENSION
     REAL
     DO  10 J =
     HCOC-RD(J) =
     COWTlNUf
     M2       = 0
     Nl
     N2
                COM'LSTrLlST
                ARrJAME(LJ»LK>
                                                 HCOORD(LJ) »VCOORD(LK)
                LABEL
                1 r/NL
                  5*RCOORD(J>

-------
     IF(N2.GT«NL)N2=NL
     fe/RITE(OUT»1000)  DATE*T!MrLABEL»lTNO»TINT
     WRITE (OUT* 10-0*»)
     WRITE(OUTfiOOl)  (NOX(K)»K=N1»N2)
  65 WRITE (OUT »1002)  (HCOOROdO »K. = Nl»N2>
  70 DO  200  K  =  1»NH
     KN        =  NH-K*1
     IF(N•EQ.fa.OR.N.EQ.ll)  60 TO 160
     IF(N«NE.9.AND.N»NE.10) GO TO I5n
     AMAX     =  ARNAME(2»KM)
     lF(ARNA;-*E(d»KN> .GT.AMAX) AMAX=ApNAME<3
     DO  100  J  =  N1»N2
 loo ANORM(J)  =  ARNAME:(J'KN)/AMAX
     WR1TE(OUT»1003)  KN»VcOORD(KM) »   = TEMP(SALrSIGT)
     IF(.NOT.CONTRL(13) ) TEMPER(J'KN)  = 1.8*TE!*PER (JiKN)+32.
 165 CONTINUE
     1MAX     = TEMPER(2»KN)
     IF(K:.EQ.ID GO TO  i?o
 167 WRITE»J=NlfN2»
 200 CONTINUE
     IFCN2.NE«NL) GO TO 60
1000 FORMAT! 1H1»D    DATE  Q2Ab»U  Tlf/E C
-------
                    7      ITERATION'  DUMBER    0 »I5»
    1 25H     COMPUTATION  SPEED = F6,3»1SH  SEC/ITERATION       )
1001 FORMAT(/17X3HJ =  lC(I8'3X))
1002 FORMAT (l2Xh;RCOORD =I3» 10 (F9»2»2X)/12X(3ZCoOPDQ)
1003 FORMTC bH K - 12, 5H  Z = F6.2 »2XlPlOEl 1.3)
1004 FOkr.;AT(;.!     COOFDlNATtS GIVEN  IN  PnPT DlAMETFRSf  Z/D OR R/D (3 >
FUUCTIO':
IF(N»EO.
SIGU
b
A
SUMT
     KETURM
  10 SIGVAT
     RETURN
     END
                GO TO 10
              = LE-6*T*(
              = .Of'l*T*((
         = T
                                             .814^) *SAL-. 093
                                                   -SUMT
     FUNCTIOii
     ERROR
     T
     SI60
     SI60
     DO 100
     TSQD
     TQBC
     F
     DSUI'.T
            TEMP(SALT»SIGMA)
          •uEWTON RA^HSON METHOD  FOR  CALCULATING TEMP.
               SALINITY AND REFERENCE  DENSITY
        I =
.01
20.
-.093+.ei/+9*SALT-.On0482*SALT*bALT
SIGO+6.8EI
] »bO
T*T
          = SIGMAT(SALT»Trl)-SlGMA
          = (215.74* (T-3. Q8 )**2)/ (503 .570* (T + 67 .26 )**'2>

-------
             100
             150
DSUMT
DA
DB
DF
Tl
ER
ER
T
IF(£R.LT»ERROR)GO
CONTINUE
TEMP      = T
RETURN
                           = DSUMT-2*(T-3.9e)*(Tf293.)/(503.57n*(T-»-b7.26)
                           = .001*U'7867-.196?7*T+.P03252C>*TSQD>
                           = l.E.-6*(lR.03-1.6328*T+.05*TsQn)
                           = (SIGO+.1324)*(-DA+DR*(SIGO-.132'4) ) +DSUMT
                           = T-F/DF
                           = Tl-T
                          = ABS(ER)
                           = Tl
                                   TO
en
                 FUNCTION   RCHM~!D(M»RlCHrBETA)
           I      CHOOSE r-ETA CONSTANT FOR APPROPRIATE MOpEL AT INPUT
                 GO TO (I0»20»30»^0»50fb0)»M
           C*  10 ROSSBY /\ND MONTGOMERY  (1935)
              10 RChMOD   = l./(l.+BETA*RICH>
                 RETURN
           C*  20 ROSSBY ANP MONTGOMERY (1935)
              20 RCHXOD   = l./(l.+BETA*RICH)**2
                 RETURN
           C*  30 hOLZMAN (1935)
              30 RCHMOD   = AMAXl(0.»l.-BtTA*RlCn)
                 RETURN
           C*  ^0 YAMAMOTr> (1959)
              i^O RCHMOD   = 5Qt-'T(ANiAXl(0.rl.-RFTA*RICH) )
                 RETURN
           C*  50 MAMAYEV (19^8)
              50 RCHMOD   - EXP(-RETA*RICH)
                 RETURN
           C*  60 MUNK ANC: ANDERSON

-------
60 RCril^OD   =  (l.+RETA*PICH)**l,^
   RETURN
   END
   SUdROUTINF  ISOGEN(L'R »PSI»ISOLN,L »NJ» NK »LABEL)
   INCLUDE     ARQlMrLlST
   DIMENSION   Z(LK>'R(LJ)rPSllLJ»LK)»ISOLN(5»3n)tLABEL(6)
   DIMENSION   XP(iiOO) »ZP(200) »ROOT(3)
   REAL        LABEL»IS°LN
   INTEGER     OUT
   OUT      =  6
   V»RITE
-------
C     ********* INTERPOLATION *********
   40 M=K-1
C     S,$$$S$$$$ QUADRATIC  INTERPOLATION  ,„,,,„„	
C     EQUATION FOR INTERPOLATION  IS Op FORM  Y =  AX**2+BX+f>
      IF(43,45,45
C     ML CORRESPONDS TO I~l
C     MM CORRESPONDS TO I
C     MM CORRESPONDS TO I+l
C     BRANCH TO 43—USE pOlNTS K-2'K-l»AND K F0» THE
C     QUADRATIC INTERPOLATION
   43 ML=K-2
      MM=K-1
      MH=K
      GO TO 44
C     BRANCH TO 45—USE POINTS K-1»K» AND K+l FOR THE
C     QUADRATIC INTERPOLATION
   45 ML=K-1
      MM=K
      MH=K+1
   44 DENOM=(Z)
      BNUM=(PSI(J» MH)-PSI< J,MM))*(Z(MM)**?-Z(ML)**£)-(PSI(J»MM)
     1-PSIU»ML))*(Z(MH)**2-Z
      AA = ANL'M/QENOM
      BB = BNUM/DENOM
      D=PSl (J»MM)-AA*Z(MMH*2-BB*Z(r/M)
      TERM=SQRT(dP**2-4.*AA*CD-PSIC))
      ROOT (i > = t -BB+TL.RM f /(2. *A A >
      ROOT<2)=(-fB-TERM)/^2.*AA)
      DO 57 1=1,2
      IF(MM.EQ.K>60 TO 61

-------
                IF(ROOTd) .LT.Z(MH) .AND.ROOT .GT.Z(MM) )GO TO 60
                IF I ROOT d) .LT,Z .GT.ZGO T^ 60
                IF(ROOT(I).LT.Z(ML)»AND«KOOT(I>.GT.Z= XCOORD>
4=.               GO TO 5
»            85 CONTINUE
                DO 185 K=KN»NK
                j        = i
             90 IFlpSI (jfK)-PSlC    )100»200»300
            100 J=J+1
                IF(J.GT.NJ)GO TO 185
                IF(PSI (J»K)-PSK
            300 J=J+1
                IF(J«GT.NJ)GO TO 185
                IF(PSI (J»K J-PSIC
          c     ********* INTERPOLATION *********
            400 M=J-1
                         QUADRATIC IMTLRPOLATIOM $$$$$$$$
            410 IF( (J-1)-1
           <+20  IF((PSIC    -PSI<^»K)
          C     ML CORRESPONDS TO  1-1
          C     MM CORRESPONDS TO  I
          C     MH CORRESPONDS TO  I+l

-------
C     BRAIiCH JO 130--USE  POINTS  J-2,J-1»AND J FOR THE
C     QUADRATIC INTERPOLATION
  130 ML=J-2
      MM=J-1
      MH=J
      GO TO 1-40
C     BRANCH TO 450—USE  POINTS  J-liJr  AND J+l  FOft THE
C     QUADRATIC INTERPOLATION
  150 ML=J-1
  11 u DENOM=(o(MM)* t^-R(ML)**2)*(R(MH)-RI MM))-(R(MH)**2-R(MM)
     1*(K(MM)-R(ML))
      ANUM=
-------
            615  XPUOUNT+1)  = XCCORD(ROOT
                60  TO 90
            185  CONTINUE
                      ouTrj.001) NI»PSIC»KOUNT
                      01JT f 1002)
                DO  500 KK= l»KOUNTfln
                KT        = ABSlKK-l)
                KS        = KT+10
                IF(KS.GE*KOUNT) KS =  KOUNT
                'HRITE(OUT»1003) KTr ( XP (Khi) »KR  = KK»KS)
                wRITE(OUTrlOOU) (ZP
           1004  FORMAT(5X10(F3.2»2X> )
           1005  FORMAT(//& *****SET PLOT  PARAMETERS* NJ=0»I3rQ NK=BI3»0 *****Q)
                END
                SUBROUTINE
                INCLUDE    COVLST»LIST
                DO  100  J = 1»5
                L         = ISOPT(J)

-------
              IF(L-EG.O) GO  TO  100
              IF(L-EQ.l) CALL ISO<5FNlZ»XR»Pc;If ISOLN»L.NJ»NK»
             i           36HVISCOUS  STREAMLINES
              IF(L»EG.«n CALL ISOGEN (ZC » X»DrLj» !SnLN»L »MJ»fjK »
             1           36HbUOYANcY PARAMETER  ISOLlNES*
              IF(L.E0.3) CALL ISOGEN(ZC» A»GAM. ISOLN»L,NJ»riK »
             1           36HSALIUITY iSOLINESfPARTS PER THOUSAND
              IF(L.EQ,<*> CALL isoGEN(zc»XrTEMPER*isoLN'LrNj»NK»
             1           36MTEMPERATURE  CONTOURS»  DF6 CENTIGRADE
          100 CONTINUF
              RETURN
              END
oo
                         GAUSS(N)
                         COMLST»LlST
                         PS8*.5
                         2.*(ZB-DZ(2>)
SUBROUTINE
INCLUDE
DIMENSION

DELS     =
ZR       =
ZP       =
ZP1      =
PSB(D   =
IF(N.NE.I) GO TO 100
RMIND    = 1.-ZP/8.
           1.-ZR/9.
           1.-ZR1/9*
           1./9.
           l./tt.
           2»NJ
           1.0
           1.
IF(RC(J).GE.RK'IN) UZ(J»1) = EXP (-**0 . 5* (C* (RC < J) "1 • > /ZR) **2)
IFCRCCJV.GL.RMINI) Up, = EXP (-»0 .5* (C+(RC (J)-l . ) /ZR1) **2)
              RMIfjl
              C
              Cl
              DO  10 j
              UZ(J'l)
              UB

-------
    iFUiZ(Jrl) .LE..01) UZ(JH)
    IF(UB.LE..01) UB = 0.
    PSKJ'H = PSI (J-1»1)+UZ(J» 1)*RC+UB*Rt(J)*CASH(X(J)»KT)*DX
 10 CONTINUE
    DO  50 j = 2»NJ
    DEL"T.LE«.01> DtLT(Jrl) = 0.
    GAM(J'l) = DELT(J»1>
 bO CONTINUE
    GO TO 120
100 DO 110 J = 2tNJ
    UZ(J'D  = EXP(-92.*(RC(J)/ZR)**2)*VMB
     UB       = EXP<-92»*(RC(J)/ZRl)**2)*VMBl
    lF((UZ(jil)/VMB>.LE.-.Ol> UZ(J»1) = 0.
     IK (UB/VMB1.LE«»01) UB = 0.
    PSI(J'l) = PSI(J-l»D+UZ(J»l)*Rr(J)*CASH(X(j) »KT)*OX
     Pb[3(J)   = PSB(J-1>*UB*RC(J)*CASH(X(J)»KT)*DX
    EXPART   = FXP(-68.*(RC(J)/ZP)**2)
    IF(EXPAr?T.LE..OD EXPART = 0.
    DELT= EXPART*QMR+DELP
    GAM(J»1) = EXPART+GMB
11U CONTINUE
120 DO 150 J = 2»NJ
    IF-
-------
00
GO
  SUBROUTINE
  SUBROUTINE
  DIMENSION
  AK       =
  A(l)     =
  A<2)     =
  A(3)     =
  AU)     =
  FRACT    r
  EX1      =
  DELZ     =
  ZB3      =
  ZP2      -
  zai      =
  FlUD  LE^GTf
  ZO       =
  DO   10 K =
  ZE       =
  DEL      =
   iFtoEL.LE.
10 CONTINUE
Ib I
   L-B1
   EB2
   EB3
   VMB1
   SB
   VMB
   DMti
                         SI ff JET ( NCO» ZB > DZr/ » T » ^0 » VP'B » WB1 »GMR » DMB)
                         OBTAINS SIMILARITY  SOLUTION  FOP VERTICAL  PLUME
                         A (4) r AF.C*>»AR(**) »D7lC*>»Ml^)
                         84.
                         0.
                         .5
                         .5
                         1.
                         l./o.
                         1./3.
                         .5*UZM
                         2B
ZR-DELZ
i FOK  FLOW  ESTABLISHMENT
                          5.57/( (1
                          ABSUZE-ZO)
                          0001)  GO  TO
                         ))  GO TO  30
                          (i*./ZE)**3+3./32./FO*(ZBl**2-ZF**2>
                          (4./ZE)**3+3./32./FO*(ZB2**2-ZE**2)
                          (4./ZE)**3+3./3ii./FO*(ZB3**3-ZE**2)
                          EB1**EX1*AK**.5/ZB1               '
                          l./SB
                          DMti
               RETURN
            30 DZl(l)
            -  INTIZE+1O-ZE

-------
00











50



100




150


160
200


DZK2)
DZK3)
DZ1C+)
Nl ( 1 >
Nl(2>
Nl(3>
Niu)
£
R
Z
DO 200 L
DZ
NSTEPS
DO 100 J
DO 50 K
AE(K>
AR
CONTINUE
Z
E
R
CONTINUE
IF(L*NE.3
0Mb
6MB
60 TO 160
IF(L»EQ.l
VMB
IFlL.EQ.2
CONTINUE
CONTINUE
RETURN
END
=
~
—
-
—
—
—
—
—
-
.1
DEi_Z/10»
DZK3)
1
10.*(ZB1-ZE>
5
5
(4./ZE>**3
.25
ZE
li<4
DZKL)
NKD
l»NSTtPS
1»M
DZ*FE ( A ( K ) *L)Z+Z » A < K > * AE ( K-l ) +E » A ( K ) *AR ( K-l ) *R ,FO )
=QZ*FR(A(K)*DZ+Z»A(K)*AE(K-i)+E»A(K)*AR(K-l)+R»T)

—
—
~

)
—
—

)
—
)





Z+DZ
E+FK ACT * < AE < 1 > +2 . * ( Ar < 2 ) + AE ( 3 » ^AE ( 1 > >
R+FRACT*URd)+2.*(AP(2)-»-AP(3) )+AR(<*) >

GO TO 150
1«S.*R/(E**FX1*Z)
l./(.245*Z*£**EXD

GO TO 160
F**£X1*AK**.5/Z
VM81 = V^B





-------
                FUNCTION
                RETURNS ZERO WHENEVER N IS EVpNtY DIVISIBLE BY M
                END
                FUNCTION  SANH(XrN)
                SANh     = .5*(EXP(X)-EXP(-X))
                IF(N.EQ.O) SANH=X
                END
                FUNCTION  CftSH(X»N)
ro               CASH     = .5*(EXP(X)-»-EXP(-X))
-                IF(N.EQ.O) CASH=1.
                END
                FUNCTION XCOORU(X)
                XCOORD   = .5*SINH(X)
                RETURN
                END
                FUNCTION   FR(Z»E»R»T)
                FR        = -.109*E**(1./3.)*T*Z
                RETURN
                END

-------
FUNCTION   FE
-------
  SELECTED WATER
  RESOURCES ABSTRACTS
  INPUT TRANSACTION FORM
                       Rti-
                                        w
    Numerical Thermal Plume Model for Vertical
      Outfalls in Shallow Water
                                         5, R

                                         6.
                                         H.
                                                                     Vp.'totmiD-- Oiga;' f.tioa
    Donald S.  Trent and James R. Welty
     Department of Mechanical Engineering
     Oregon State University
     Corvallis, Oregon 97331
                                             16130 DGM
  12, Span: if lag O
Environmental Protection Agency
      Environmental Protection Agency
      Report Number,, EPA-R2-73-162, March 1973.
             A theoretical study of the heat and momentum transfer resulting from a flow
)f power plant condenser effluent discharged vertically to shallow, quiescent coastal re-
ceiving water is presented. The complete partial differential equations governing steady,
Incompressible,  turbulent flow driven by both initial momentum and buoyancy are solved
ising finite-difference techniques to obtain temperature and velocity distributions in the
lear field of the thermal discharge.
             Turbulent quantities were treated through the use of Reynolds stresses with
further simplificati"on"uti±izing the concept of eddy diffusivities computed by Prandtl's
aixing length theory. A Richardson number correlation was used to account for the effects
   density gradients on the computed diffusivities.
             Results were obtained for over 100 cases, 66 of which are reported, using the
computer program presented in this manuscript. These results ranged from cases of pure
moyancy to pure momentum and for receiving water depths from 1 to 80 discharge diameters
ieep.  Various computed gross aspects of the flow were compared to published data and
round to be in excellent agreement.  Data for shallow water plumes and the ensuing lateral
spread are not readily available; however, one computed surface temperature distribution
?as compared to proprietary data and found also to be in reasonable agreement.
  17s..
      Waste heat disposal, heated shallow discharge, turbulent buoyant jets,
       temperature prediction, thermal pollution.
                       05G
                         15,  S- urity C *ss.
                             (Kepon)
                                           21.
                                                 of
                             Send To:
                                                      WATER RESOURCES SCIENTIFIC INFORMATION CENTER
                                                      U.S. DEPARTMENT OF THE INTERIOR
                                                      WASHINGTON, D. C. 2O24O
            Author
                         Oregon  State Un-fvers-n-y

-------