UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Flocculation of Fraser River sediments due to pulp mill effluents Evans, Wayne John 1996

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1996-0361.pdf [ 7.72MB ]
Metadata
JSON: 831-1.0050356.json
JSON-LD: 831-1.0050356-ld.json
RDF/XML (Pretty): 831-1.0050356-rdf.xml
RDF/JSON: 831-1.0050356-rdf.json
Turtle: 831-1.0050356-turtle.txt
N-Triples: 831-1.0050356-rdf-ntriples.txt
Original Record: 831-1.0050356-source.json
Full Text
831-1.0050356-fulltext.txt
Citation
831-1.0050356.ris

Full Text

FLOCCULATION OF FRASER RIVER SEDIMENTS DUE TO PULP M I L L EFFLUENTS  by  Wayne John Evans  B. E. (Hons), University of Queensland, 1992  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in  THE FACULTY OF GRADUATE STUDIES Department of Civil Engineering  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA June 1996 © Wayne John Evans, 1996  In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It  is understood that copying or  publication of this thesis for financial gain shall not be allowed without my written permission.  Department of C)\ll'  / SlA<^ (ft e.-erf  The University of British Columbia Vancouver, Canada Date  DE-6 (2/88)  3 1 j'  b  ( Q/n  ABSTRACT Recently, researchers noted an aggregation phenomenon downstream of a pulp mill effluent discharge. A large quantity of the sediment in the Athabasca River deposited downstream of the discharge of the pulp mill effluent (Krishnappan et al., 1994). This effect may be significant on the Fraser River, where a number of pulp mills along its length discharge effluent.  This aggregation phenomenon could not be explained by existing coagulation / flocculation theories, and an experimental investigation was undertaken. Experiments involved simple jar testing and settling tests in settling columns. The jar tests were completed in Prince George using pulp mill effluent, and Fraser River water.  Throughout these tests, turbidity measurements were made which resulted in some ambiguity  surrounding interpretation of the results. Nonetheless, on addition of pulp mill effluent to river water, it was noted that a reduction in turbidity occurred with respect to expected values immediately after mixing. The magnitude of this turbidity reduction was also shown to be greatest when effluent and river water were mixed in equal proportions. The turbidity reduction was thought to indicate aggregation of solids in the effluent river water mixture.  Settling experiments were completed in a settling column, and measurements of total suspended solids were made. This enabled information regarding settling velocities to be obtained, which are related to aggregate size. It was found that on mixing effluent with a suspension of illite, the settleability of the suspended solids was greatly improved. The same results were not achieved when a suspension of Fraser River sediment and effluent were mixed, and it appeared that the settleability of suspended solids was actually impeded due to the addition of effluent in this instance.  An analysis of the field data collected by field researchers (Krishnappan, 1994; Droppo, 1994) was also completed. The analysis completed was more detailed than any previously completed, and indicated that there may be aggregation of particles within the effluent plume of the Northwood mill at Prince George. The degree of aggregation appeared to be slight, and hence not likely to affect the overall transport of suspended solids within the Fraser River.  ii  TABLE OF CONTENTS ABSTRACT  "  TABLE OF CONTENTS  Hi  LIST OF TABLES  vm  LIST OF FIGURES  xi  ACKNOWLEDGEMENT  1.0 INTRODUCTION.  xiv  '.  '.  /  1.1 RELEVANT FIELD OBSERVATIONS  4  1.2 IMPACTS OF FLOCCULATION  5  1.3 STUDY SCOPE  6  2.0 LITERATURE  REVIEW AND BACKGROUND  INFORMATION  2.1 AGGREGATES AND THEIR COMPONENTS  7 7  2.1.1 Aggregate Characteristics  7  2.1.2 Theory of Colloids  9  2.2 AGGREGATION / DISAGGREGATION THEORY 2.2.1 Collision Theory  11 11  2.2.1.1 Integral Forms  11  2.2.1.2 Numerical Form  12  2.2.2 Disaggregation Theory  13  2.3 AGGREGATE FORMATION MECHANISMS  15  2.3.1 Differential Settling  15  2.3.2 Turbulent Shear  16  2.3.3 The Role of Organic Material  17  2.4 IMPLICATIONS OF FLOCCULATION FOR TRANSPORT MODELS  iii  18  2.4.1 Sediment Transport  18  2.4.2 Contaminant Transport  ••••• 20  2.5 AGGREGATE MEASUREMENT TECHNIQUES  21  2.5.1 Electrical Resistance  22  2.5.2 Image Analysis  22  2.5.3 Laser BackScatter / Diffraction  23  2.5.4 Turbidity Measurements  J  24  3.0 APPLICATION OF EXISTING THEORIES  25  3.1 CALCULATION OF VELOCITY GRADIENTS IN TURBULENT FLOWS  25  3.2 SECOND ORDER COAGULATION AND SETTLING THEORY  26  3.2.1 Comparison of Results to Theory  26  3.2.2 Calculation of Removal Rate Parameters  29  3.3 COAGULATION OF PARTICLES IN A PLANE PLUME  31  3.4 POSSIBLE COAGULATION / FLOCCULATION MECHANISMS PRESENT  35  3.5 RESEARCH OBJECTIVES  37  4.0 EXPERIMENTS  38  4.1 BASELINE DATA COLLECTION  38  4.2 SAMPLING PROTOCOLS  38  4.2.1 Effluent  38  4.2.2 River Water  39  4.2.3 Transportation and Storage  39  4.3 CRITICAL COAGULATION CONCENTRATION  40  4.4 SETTLING TESTS MEASURING TURBIDITY  40  4.4.1 Experimental Procedures and Apparatus  41  4.4.2 Conversion of Settling Test Data to Settling Velocity Distributions  42  4.4.3 Mixtures of Effluent and River Water Considered  43  iv  4.5 INSTANTANEOUS TURBIDITY MEASUREMENTS  44  4.6 SETTLING TESTS MEASURING TSS  45  4.6.1 Two Litre Jars  45  4.6.2 Settling Column  46  4.6.2.1 Description of Apparatus  '.  4.6.2.2 Experimental Procedure  46 48  4.6.2.3 Experiments with lllite Suspensions  50  4.6.2.4 Experiments with Fraser River Sediment Suspensions  50  52  5.0 RESULTS OF EXPERIMENTATION 5.1 CRITICAL COAGULATION CONCENTRATION  52  5.1.1 Ionic Species in the Effluent  52  5.1.2 Experiments  53  5.2 SETTLING TESTS MEASURING TURBIDITY  54  5.2.1 Turbidity Variation as a Function of Time  55  5.2.2 Settling Velocity Distributions  57  5.3 INSTANTANEOUS TURBIDITY MEASUREMENTS  58  5.3.1 Analysis Techniques and Sample Calculations  58  5.3.2 Dependence of Turbidity Reduction on Mixing Proportions  61  5.3.3 Dependence of Turbidity Reduction on TSSftver  62  5.3.4 Theoretical Discussion of Turbidity Reduction  63  5.4 SETTLING TESTS MEASURING TSS  64  5.4.1 Jar Scale  65  5.4.2 Settling Column Scale  65  5.4.2.1 Experiment 1  67  5.4.2.2 Experiment 2  72  5.4.2.3 Experiments 3 and 4  77  5.5 PRIMARY PARTICLE SIZE DISTRIBUTIONS  84  v  6.0 RESULTS  OF ANALYSIS  EXPERIMENTAL  OF PREVIOUSLY  COLLECTED  FIELD  AND  DATA  87  6.1 ANNULAR FLUME EXPERIMENTS (KRISHNAPPAN AND ENGEL, 1994) 6.1.1 Solids Concentration Measurements  87 •  6.1.2 Aggregate Size Distribution Measurements  88 89  6.2 IN SITU AGGREGATE SIZE DISTRIBUTIONS (IAN DROPPO, 1995, UNPUBLISHED DATA)  93  6.2.1 Median Diameters  93  6.2.2 Aggregate Size Distributions  94  6.3 FIELD TRANSECTS - KRISHNAPPAN (1996, UNPUBLISHED DATA)  95  6.3.1 Total Suspended Solids Concentrations  96  6.3.2 TSS Predictions  97  6.3.2 Median Diameters  99  6.3.3 Aggregate Size Distributions  102  6.3.4 Explanations for Aggregate Size Distribution Changes  105  7.0 CONCLUSIONS AND RECOMMENDATIONS  109  7.1 GENERAL REMARKS  109  7.2 SUMMARY OF RESULTS  109  7.2.1 Experimental Work  109  7.2.2 Analysis of Other Researchers' Data  110  7.3 CONCLUSIONS  112  7.4 RECOMMENDATIONS FOR FUTURE RESEARCH  113  8.0 REFERENCES  115  APPENDIX  A: CALCULATION  KOLMOGOROFF  OF TURBULENT  MICROSCALES  VELOCITY GRADIENTS  IN THE FRASER RIVER vi  AND 120  APPENDIX  B:  CALCULATION  OF VELOCITY GRADIENTS  IN SETTLING  COLUMN  '. APPENDIX COMPLETED  122 C: RESULTS FROM SETTLING  TESTS MEASURING  TURBIDITY  AT THE UNBC  125  APPENDIX  D: RESULTS  OF MIXING EXPERIMENTS  APPENDIX  E: RESULTS OF SETTLING  MEASURING  COLUMN EXPERIMENTS  TURBIDITY.... MEASURING  130 TSS 134  APPENDIX  F: CROSS-SECTIONS  IN THE VICINITY OF THE  DIFFUSER  NORTHWOOD 139  vii  LIST OF TABLES Table 3.1. Summary of the data of Krishnappan and Engel (1994)  27  Table 3.2. Calculated Coagulation Removal Rate Parameters from experiments completed by Krishnappan and Engel (1994)  30  Table 3.3. Second order coagulation removal rate parameters as determined by Hunt (1980) for montmorillonite.3 Table 4.1. Summary of settling test information  43  Table 4.2. Volume combinations considered for instantaneous turbidity reduction experiments  45  Table 4.3. Combinations of effluent, distilled water and illite used in Experiment 2  50  Table 4.4. Combinations of effluent, distilled water, and Fraser River sediments used in Experiments 3 and 4  51  Table 5.1. Average concentration of some predominant ions in Northwood effluent (Evans & Hall, 1996) and calculated effective Na concentrations  52  +  Table 5.2. Average concentration of some predominant ions in sea water ((Wilson, 1975) as cited by Chester (1990)) and calculated effective Na concentrations  53  +  Table 5.3. Summary of the settling tests completed, initial sample turbidities (x ), TSS (TSS ) and the sampling 0  0  point depth for each settling test  55  Table 5.4. Results of turbidity measurements, predicted turbidities and calculated percent reductions in turbidity for river water with Turbidity - 69 NTU, and effluent with Turbidity = 5.0 NTU.  59  Table 5.5. Results of TSS measurements, predicted TSS values and calculated percent reductions in TSS for river water with TSS = 153 mg/L, and effluent with TSS - 37 mg/L  60  Table 5.6. Average percent turbidity reduction for effluent and river water mixtures and all data sets  62  Table 5.7. Average percent turbidity reduction for river water and distilled water data sets  63  Table 5.8. Summary of settling tests completed in settling column at UBC  67  Table 5.9. Median settling velocities as determined from the plots of settling velocity distribution for Experiment 1. Table 5.10. Summary of median settling velocity information  :  77  Table 5.11. Summary of median settling velocities (v )for Experiments 3 and 4  80  so  Table 6.1. Results for volume median spherical diameter (d$ )from Droppo (1995, unpublished data). Numbers 1 0  and 2 indicate replicate samples  94  viii  Table 6.2. Transect averaged TSS data and relevant flows at Shelley (Krishnappan, 1996, unpublished data)  96  Table 6.3. Summary of Fraser River flows (Shelley gauging station), predicted TSS using Equation 6.1 and Equation 6.2, and measured TSS.  ,  98  Table 6.4. Average median diameters noted at the various transects-(Krishnappan, 1996, unpublished data)  10  Table 6.5. Calculated velocity gradients for plume and control locations at the transects near Northwood (Krishnappan, 1996, unpublished data)  104  Table 6.6. Summary of TSS and velocity gradient information, their product, and measured median diameters... 1 Table 6.6. Data used in calculations for Ad$o and summary of some important results  108  Table 7.1. Summary of experimental results obtained by the author.  709  Table 7.2. Summary of analysis offielddata, and results of analysis of experimental data collected by other researchers  Ill  Table D-l. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  39.7 mg/L, and Turb  = 58 mg/L, Turb  River  = 31 NTU, and TSS  = 6.0 NTU.  Efflumt  Effluent  =  130  Table D-2. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  = 58 mg/L, Turb  River  = 31 NTU.  130  Table D-3. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  = 40 mg/L, and Turb  = 141 mg/L, Turb  River  = 52 NTU, and TSS  , = 5.5 NTU.  Effluent  131  Effluen  Table D-4. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  = 141 mg/L, and Turb  River  = 52 NTU.  131  Table D-5. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  - 47.8 mg/L, and Turb  Efflunt  = 286 mg/L, Turb  River  = 95 NTU, and TSS  = 5.2 NTU.  Effluent  132  Table D-6. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  ix  = 286 mg/L, and Turb  River  = 95 NTU.  132  Table D-7. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  = 40.2 mg/L, and Turb  Effluem  = 156 mg/L, Turb  = 5.0 NTU.  River  = 62 NTU, and TSS  Effluent  133  Table E-l. Results for TSS and settling height collected during Experiment 1  135  Table E-2. Results for TSS and settling height collected during Experiment 2  136  Table E-3. Results for TSS, VSS and settling height collected during Experiment 3  137  Table E-4. Results for TSS, VSS and settling height collected during Experiment 4  138  x  LIST OF  FIGURES  Fisure 1.1 Map of the Fraser River Basin (Zrymiak & Tassone, 1986)  2  Figure 1.2 (a) Summer photograph of the Fraser River just down stream of the Northwood Pulp Mill effluent outfall, (b) Photograph showing the final effluent lagoon, and the effluent offtake structure  3  Fisure 1.3 Diagram showing the hypothetical situation downstream of a pulp mill effluent discharge  5  Figure 2.1. Photograph offine-grainedsediments. Disaggregated Lake Erie bottom sediments. The solids concentration is about 100 mg/L, while the average diameter of the particles is about 6 \im (Lick and Huang, 1993)  8  Figure 2.2. Typical Floe formed with Lake Erie sediments having a diameter of around 100 \im (Lick and Huang, 1993)  8  Figure 2.3. Diagram showing the formation of a diffuse electrical double layer (Adapted from Everett, 1994)  10  Fisure 2.5. Photograph of a floe formed as a result of the action of differential settling (Lick and Huang, 1993).. 16 Figure 3.1. Comparison of results from deposition test to lines of bestfit(linear): (a) for shear stresses 0.0563  N/m and 0.121 N/m , and (b)for shear stresses 0.169 N/m and 0.213 N/m . Raw data used with permission of 2  2  2  2  Krishnappan (1995)  28  Figure 3.2. Evolution of a Coagulating Plume (adapted from Holman, 1986)  33  Figure 3.3. Contour plot of the product normalised concentrations of the plume and ambient i.e. C g * C e  amb  .... 34  Figure 3.4. Diagram showing the possible effects of discharging pulp mill effluent into the Fraser River.  36  Fisure 4.1. Dimensions ofjar and paddle apparatus as per standards (after Hudson, 1981)  42  Figure 4.2. Diagram of settling column set up and dimensions  47  Fisure 5.1. Removal of turbidity comparison between a control case and an effluent and river water mixture for a river water sample with TSS - 286 mg/L  56  Figure 5.2. Settling velocity distribution comparison for control and river water and pulp mill effluent mixture  57  Fisure 5.3. Plot of mean turbidity reduction as a function of dilution for data sets generated with TSS = 141, 153, 156 and 286 mg/L  61  Figure 5.4. (a) Plot of the variation in TSS throughout the settling test, (b) Normalised plot of the variation in TS throughout the settling test.  66  xi  Fieure 5.5. (a) Plot of the smoothed variation in TSS as a function of time for Experiment 1. (b) Rate of solids removal for Experiment 1  69  Figure 5.6. (a) Normalised plot of the variation of TSS in the settling tests of Experiment 1. (b) Normalised plot o the variation in TSS removal rates in the settling tests of Experiment 1  70 72  Fisure 5.7. Settling velocity distributions for Experiment 1  Fieure 5.8. (a) Variation of TSS throughout the settling tests in Experiment 2. (b) Variation of rate of removal of TSS throughout the settling tests in Experiment 2  73  Figure 5.9. (a) Normalised plot of the variation of TSS in the settling tests of Experiment 1. (b) Normalised plot o 75  the variation in TSS removal rates in the settling tests of Experiment 1 Figure 5.10. Settling velocity distributions for Experiment 2  76  Fisure 5.11. (a) Smoothed variation in TSS for Experiment 3. (b) TSS removal rates in Experiment 3  78  Fisure 5.12. Variation in VSS in Experiment 3  79  Figure 5.13. (a) Normalised variation in TSS for Exp. 3. (b) Normalised TSS removal rate in Experiment 3 Figure 5.14. Settling velocity distributions in Experiment 3  81  82  Figure 5.15. Comparison of primary particle size distributions for miscellaneous samples  86  Figure 6.1. Effects of pulp mill effluent addition during deposition tests (Krishnappan, 1994). (a) Shear stress of 0.121 N/m . (b) Shear stress of 0.213 N/m 2  89  2  Figure 6.2. Variation in d (\i.m) as a function of time in deposition test - shear stress = 0.121 N/m . Raw data 2  50  used with permission of Krishnappan (1995)  90  Figure 6.3. Comparison of aggregate size distributions as a function of time during a deposition test with a shear stress of 0.121 N/m . Raw data used with permission of Krishnappan (1995)  92  2  Figure 6.4. Aggregate size distribution comparisons for upstream and downstream samples. Raw data used with permission ofDroppo (1995)  95  Figure 6.5. Location offield transects and sampling locations downstream of the diffuser superimposed on the predicted plume location for Q , = 525 m /s. Plume location used with permission of Vine (1996) 3  r ven  97  Figure 6.6. Plots of depth averaged median diameters for the five transects in the vicinity of Northwood for (a) x 30 and 100 m, and (b) x = 300 and 1000 m. Raw data used with permission of Krishnappan (1995)  xn  101  Figure 6.7. Comparison of aggregate size distributions for plume and control locations from the transects at Northwood. Raw data used with permission of Krishnappan (1995)  103  Figure 6.8. Correlations between CG and d . (a) Using information transects except x = 1000 m, and(b) using m  information from all transects. Raw data used with permission of Krishnappan (1995) Figure C-l. Plot of turbidity variation during settling tests A and B  706 125  Figure C-2. Plot of turbidity variation during settling tests C and D  125  Figure C-3. Plot of turbidity variation during settling tests E and F.  126  Fisure C-4. Plot of turbidity variation during settling tests G, H, IandJ.  126  Figure C-5. Plot of turbidity variation during settling tests K, L, M and N.  727  Figure C-6. Settling velocity distributions obtained from tests A and B  727  Figure C-7. Settling velocity distributions obtained from tests C and D  728  Fisure C-8. Settling velocity distributions obtained from tests E and F.  128  Fisure C-9. Settling velocity distributions obtained from tests G, H, I andJ.  729  Fisure C-10. Settling velocity distributions obtained from tests K, L, M andN.  729  Fisure F-l. Cross-section and vertically averaged velocities 2000 m upstream of the diffuser.  139  Figure F-2. Cross-section and vertically averaged velocities 30 m downstream of the diffuser.  140  Figure F-3. Cross-section and vertically averaged velocities 100 m downstream of the diffuser.  140  Figure F-4. Cross-section and vertically averaged velocities 300 m downstream of the diffuser.  141  Figure F-5. Cross-section and vertically averaged velocities 1000 m downstream of the diffuser.  141  xiii  ACKNOWLEDGEMENT Special thanks must be extended to a number of people, without whom this thesis would not have been possible. •  To my supervisors Eric Hall and Greg Lawrence. Without the guidance and encouragement from them, I would probably still be reading articles about the fractal characteristics of turbulent jets and wondering why I was getting nowhere with my research. Our weekly meetings were never dull as we tried to come up with new ideas when the last one had proved fruitless.  Their financial support for my field work at Prince George and  laboratory equipment needs must also be acknowledged and their critical review of my thesis will hopefully result in a readable thesis and improve my writing skills in general. •  To Krish Krishnappan and Ian Droppo at CCIW. Their field and experimental data provided me with something vital I wanted all along, but couldn't get....aggregate size distribution measurements! Also, their helpful advice on many topics was greatly appreciated.  •  To my co-workers in the FRAP PACK Jason Vine and Bonnie Marks. Their collaboration and help on a variety of issues helped to enhance the team feeling that we had when we thought we would come up with one product at the end of our work.  •  To Ellen Petticrew at the University of Northern British Columbia (UNBC).  The use of her brand new  laboratory at UNBC was a great advantage when I was faced with conducting experimental work in the rain. Also, her general help in making sure I had everything I needed in Prince George was invaluable. •  To Ron Fujino at Northwood Pulp and Timber Ltd. The freedom that I was allowed on the Northwood site and the samples that were always sent promptly and free of charge were greatly appreciated.  •  To the staff in the Environmental Engineering laboratory at UBC. The help and tolerance in the laboratory that I received was invaluable considering my inexperience and general mess creation ability.  •  To Michael Church. His final reading of my thesis provided some very useful comments which I have been able to incorporate.  •  To all the people I befriended in Canada. It's been great fun outside school as well, but it's just a pity that I had such a bad run with injuries while I was here as I could have done more with you all. As it was, I've worked more than I planned, so it's going to be a pretty close race JP!  •  To my family. As always....right behind me! Hang in there Old Feller!  xiv  LO INTRODUCTION The importance of the Fraser River in British Columbia as a resource that must be preserved is well known. In recent times, the River and its drainage basin have come under increasing pressures due to anthropogenic activities. These pressures have resulted in the river being forced to carry rapidly increasing loadings of a wide variety of pollutants. At the same time, it is expected that it will be capable of maintaining the same biological productivity for which it is famous (EQTWG, 1995).  The most significant industrial source of pollutants is the pulp and paper industry. In all, there are seven mills (EQTWG, 1995) which discharge their effluents into the river, contributing 39 % by volume of all the effluents discharged into the river. Three of these mills are located near Prince George in central BC (Figure 1.1). The largest mill in Prince George is Northwood Pulp and Timber which has been the subject of a number of studies (including this one) as it discharges to the river before its waters have been seriously influenced by anthropogenic activities. Figure 1.2(a) shows the reach of the River into which the Northwood effluent is discharged during the summer, and Figure 1.2 (b) shows the final effluent lagoon and offtake structure. The study of the transport of contaminants associated with pulp mill effluents is of importance because of the potentially detrimental impact that these mills may have on water quality.  Contaminant transport modelling for pulp mill effluents is particularly complicated because of the organic nature of the contaminants. Many of these contaminants have a high affinity for solid or organic material. Hence, these contaminants tend not to exist in dissolved form in the water column, but rather they attach to the suspended sediment in the river, or the biosolids in the effluent. As a result, contaminant transport modelling must address both the issues of transport of a solute, and particle transport as well.  The dynamic processes which result in  contaminants changing from the dissolved to the attached state must also be considered, resulting in an extremely complicated problem (Ongley et al, 1992). One of the key parameters in modelling the transport of any solids in a fluvial system is the size of the particles. A particle's size is important because it largely dictates its settling velocity which in turn influences the way the particle will be transported along the river.  1  Figure 1.1 Map of the Fraser River Basin (Zrymiak & Tassone, 1986). 2  (a)  Figure 1.2 (a) Summer photograph of the Fraser River just down stream of the Northwood Pulp Mill effluent outfall, (b) Photograph showing thefinaleffluent lagoon, and the effluent offtake structure.  3  1.1 RELEVANT FIELD OBSERVATIONS Field studies have shown in the Athabasca River, (Krishnappan et al,  1994), that significant deposition of the  suspended sediment load occurred downstream of the discharge of a pulp mill effluent into the river. Two different sediment surveys were carried out, the first in the fall of 1993, and the second in the winter of that year. In the fall survey, it was found that approximately 50 % of the total sediment load of the river settled out after travelling 175 km downstream of the effluent discharge. In the winter survey, the results were even more impressive, with 70 % of the solids load of the river settling out after travelling just 20 km downstream of the effluent discharge. It was thought that this could only occur due to the aggregation of suspended sediment and effluent biosolids resulting in increases in particle sizes, associated settling velocities, and the observed deposition (Krishnappan et ai,  1994).  This mechanism is summarised in Figure 1.3. Other mechanisms often responsible for sediment deposition were ruled out as contributing factors. More detailed measurements were also made on the Fraser River in the fall of 1994, and during the spring freshet of 1995, with the majority of measurements being made in the immediate vicinity of the pulp mill. These data have not previously been analysed in any detail or any conclusions made regarding the observations, and an analysis of available data has formed part of this study.  In other field studies, Sekela et al. (1995) recorded a general increase in contaminant concentration during the onset of freshet. This was thought to be due to the resuspension of fine bottom sediments, and particulate organic matter during this period of increasing flow. It was suggested that resin and fatty acids were associated with the organic matter, and that dioxins, furans and PAHs were found to partition almost exclusively in the suspended sediments. The publication of Hall and Liver (1996) suggests that the majority of resin acids discharged in a pulp mill effluent would indeed be partitioned on the biosolids of the effluent so the explanation of Sekela et al. (1995) of high contaminant levels during spring sounds reasonable.  4  Suspended sediment as primary particles  biosolid deposition as aggregates Figure 1.3 Diagram showing the hypothetical situation downstream of a pulp mill effluent discharge.  1.2 I M P A C T S O F F L O C C U L A T I O N Flocculation of river sediments due to the discharge of pulp mill effluents could have an impact on sediment transport.  If rapid aggregation occurs and large quantities of solids are deposited downstream of an effluent  discharge, traditional sediment transport theory and modelling would be unable to predict this. Thus, investigation of the phenomenon is required to determine if sediment transport will be affected. Flocculation of effluent biosolids and river sediment could result in the deposition of large quantities of contaminated solids during periods of low flow, such as during the fall and winter. At the onset of the spring freshet, these solids could be resuspended due to the increase in flows, resulting in an increase in contaminant concentrations in suspended sediments during this period. This mechanism may be responsible for the effects noted by Sekela et al. (1995) despite time dependent degradation effects.  Contaminant transport could also be greatly affected by flocculation of biosolids and river sediments.  If, for  example, the majority of contaminants in pulp mills are associated with effluent biosolids which, when discharged  5  into a river, aggregate with the sediments, then modelling contaminants as tracers would be invalid. Contaminant transport would instead be more closely linked to that of sediment particles. Significant changes would have to be made to existing modelling approaches before accurate results could be obtained. For this reason, the phenomenon of aggregation is an important issue if the transport and fate of contaminants associated with pulp mill effluents are to be accurately modelled. Even the most recent contaminant transport models fail to consider flocculation of finegrained sediments themselves, stating that "flocculation, consolidation and erosion/deposition are not accounted for owing to their poor definition" (Ng et al, 1996) without even contemplating interaction effects between solids in the effluent and receiving water.  1.3 S T U D Y  SCOPE  Interest has been generated in the transport of contaminants associated with pulp mill effluents as a result of the Fraser River Action Plan (FRAP). Gobas (1996) produced a contaminant fate model of the River which is capable of predicting with reasonable accuracy the concentrations of selected contaminants in both the sediments, and the tissues of fish. The model presently assumes that the Fraser River can be modelled as being comprised of 23 completely mixed reactors. It was thought that significant improvements could be made to this model by the inclusion of more detailed mixing information and by taking account of flocculation effects.  The scope of this study, was initially to model the processes of flocculation and aggregation in the river, as a result of the discharge of the pulp mill effluent. It was intended that the Northwood mill at Prince George be focused on initially, as data from remote sensing operations (Borstad, 1995, pers. comm.) identifying the plume were available. These data were to be used for verification of any modelling attempts. The discussions in Chapters 2 and 3 show that there are still many poorly understood aspects of particle aggregation and disaggregation for simple sediments only, without considering the effects due to the addition of pulp mill effluents. Some experimental data were available and were used to test an existing "flocculation model theory" and even speculate a new model. Largely, it was found that these models were not useful, and this study instead focused on the reproduction of observed effects under laboratory conditions. Once the phenomenon had been confirmed, an investigation of the effect on particle size distribution or settling velocity distribution under a variety of conditions was completed.  6  2.0 LITERATURE  REVIEW  AND BACKGROUND  INFORMATION  2.1 AGGREGATES AND THEIR COMPONENTS 2.1.1 Aggregate Characteristics Before investigation of flocculation issues, it is important to understand something about aggregates themselves. A definition for the aggregates present in natural systems is offered by Droppo and Ongley (1994). Floes are defined as sediment transported in rivers which take one of the following three forms: (1) water-stable soil aggregates washed into the system via overland flow (Dorich et al, 1984); (2) sediment flocculated within the river system through physical, chemical and biological means (van Olphen, 1977; Muschenheim et al, 1989); and (3) a combination of both.  The existence of aggregates or floes in natural aquatic systems has been recognised for some time (Sherman, 1953; Kranck, 1979).  The majority of early in situ sediment sizing research had focused on the marine/estuarine  environment. It has been only in recent times that aggregates were recognised as being of importance in freshwater fluvial systems, and recent studies revealed that the majority of the volume of sediment particles in a river are transported in the form of aggregates (Droppo and Ongley, 1994). They observed that while floes may not always comprise the majority of particles by-number in transport, in Ontario, they were found to comprise more than 90 % of the total volume of sediment transported.  Lick and Huang (1993) demonstrated photographically the difference between flocculated sediments and disaggregated sediments. In Figure 2.1 disaggregated particles with a mean diameter of 6 um are visible. In Figure 2.2 a floe with a diameter of around 100 um is shown. This floe is comprised of the same particles shown in Figure 2.1. The open structure of the floe is obvious, as is its potential fragility. The fragile nature of floes has been noted by many authors (Bale and Morris, 1991; Gibbs and Konwar, 1982; Woodward and Walling, 1992).  The particles from which aggregates are composed are known as discrete or primary particles. Primary particles can be either organic or inorganic in nature, and may also be colloidal depending on their size. Colloidal particles are defined as having a diameter less than 1 um. A wealth of literature exists regarding the behaviour of colloidal  7  Fisure 2.1. Photograph of fine-grained sediments. Disaggregated Lake Erie bottom sediments. The solids  concentration is about 100 mg/L, while the average diameter of the particles is about 6 \im (Lick and Huang, 1993  Figure 2.2. Typical Floe formed with Lake Erie sediments having a diameter of around 100 \ym (Lick and Huang, 1993).  8  particles in relation to aggregation (Everett, 1994; van Olphen, 1977), and we can be sure that particles of this size are involved in natural flocculation processes. The degree to which larger particles present in natural systems participate in these flocculation processes is not clear. A description of the concepts responsible for the physicochemical behaviour of colloids is useful in the understanding of many flocculation phenomena. In particular, we are interested in the behaviour of clay colloid systems.  2.1.2 Theory of Colloids Colloids are an important group of primary particles and an understanding of their associated electrical double layers is important. If a clay is added to water, it appears to dissolve, however the solution is changed as a result and will appear turbid if a light is shone through the solution (van Olphen, 1977). Light is scattered by what is actually a dispersion of very small clay particles which will not settle from the suspension. Most colloidal particles carry a charge, and it is this charge that holds particles apart in the suspension (Benefield et al, 1982) through simple electrostatic repulsive forces.  In response to particles having this charge, an electrical double layer is formed (Benefield et al, 1982). Like an ionic solution, a colloidal system does not have a net charge, and hence particle charges must be compensated for internally. The concept of the electrical double layer has arisen from a need to explain internal charge neutralisation (van Olphen, 1977). The double layer encloses the colloidal particle and then extends into the solution and contains sufficient oppositely charged ions to neutralise the charge on the colloid as shown in Figure 2.3. The tendency for oppositely charged particles to cluster on the surface of the particle is balanced by the effects of thermal diffusion which results in the formation of a diffuse double layer with a finite thickness.  The thickness of the electrical double layer dictates the ease with which particles can be brought into contact with one another. When particles are brought close enough together, attractive Van der Waal's forces will overcome the repulsive electrostatic forces. To get this requisite distance apart, a certain amount of energy must be supplied to overcome the repulsive forces. The magnitude of the energy barrier is a function of the thickness of the double-layer (Holman, 1986). The following formulation (Everett, 1994) has been used to describe the thickness of the double  9  Diffuse e l e c t r i c a l d o u b l e layer  Bulk solution  0  0  0 0  0 0 0  0  0  0  0 0  0  0 0  0  © 0  Particle S u r f a c e  0 0 0  0  0  0  Figure 2.3. Diagram showing the formation of a diffuse electrical double layer (Adapted from Everett, 1994).  layer (—) K xl/2  EkT K  2v  2  (2.1)  where e represents the permittivity of a vacuum, k is the Boltzman constant, T is the absolute temperature, c, represents the concentration of the ith ion, and z, represents its charge. The thickness of the electrical double layer and associated potential hill can be reduced by an increase in the ionic concentration of the suspension. Higher valence ions will have an increased effect in producing compression of the electrical double layer. Similarly, 10  temperature can also be seen to have an effect on the thickness of the electrical double layer such that as temperature is reduced, double layer thickness is reduced. Thus, any of these factors can cause reduction in the thickness of the electrical double layer, meaning that the potential hill is reduced and particles can be brought into contact more easily.  2.2 AGGREGATION / DISAGGREGATION THEORY 2.2.1 Collision Theory Aggregates can form due to biological, physical and chemical factors. Of these, physico-chemical mechanisms have been most closely studied (Pearson et ai, 1984; Lick and Lick, 1988; Tsai et al., 1987; Hunt, 1980; Burban et al., 1989) and theory has been developed for the explanation of these mechanisms. In order to understand the formation of aggregates, it is first necessary to understand collision theory because collision of particles is essential to bring particles close enough together that binding forces can form aggregates.  2.2.1.1 Integral Forms Collision theory in its most basic form for continuous particle size distributions is outlined by Pearson et al. (1984). Firstly, the particle size distribution can be defined as  dN = n(y)dv  (2.2)  so that dN is the number of particles per fluid volume whose sizes (volumes) lie in the range v to v + dv. The collision rate per unit volume of fluid, of particles of volumes v; and vj is given by the product of their respective concentrations n(v,) and n(y,-) and a collision frequency function (3(v- ,vj) representing the geometry and dynamics of z  the collision mechanism, so that:  collision rate = (3(v,., v^niv^niv^dv^Vj  (2.3)  Then the change with time of the concentration of particles of size v is given by the general dynamic equation:  11  dn(v) I(v) +  -Jj3(v', v -  v')n(v')n(v - v')dv'  - fp(v,v')n(v)n(v')dv' +  (2.4)  5(v)^ r)  JO  where I(v) represents a source of particles (resulting from condensation for example) and S(v)dn/dz is a particle 7  sink resulting from particles sedimenting in the z-direction at their Stokes settling velocity S(v). This form of the dynamic equation is rarely used as the terms for the source and the sink are typically neglected. The sink term is neglected because spatial derivatives are taken as zero for a homogeneous situation. This leaves what is known as the coagulation equation, i.e.  (2.5)  The first term on the right hand side of the equation represents the rate of gain of particles of volume v by coagulation of smaller particles, and the second term represents the loss of particles of size v due to their coagulation with particles of all sizes. The function (5 will vary depending on the mechanism of aggregation being considered. Pearson et al. (1984) tabulates this function for the mechanisms of Brownian motion, laminar shear, pure strain, isotropic turbulent shear, turbulent inertia and differential sedimentation, although in most modelling studies, all mechanisms other than turbulent shear and differential sedimentation are neglected.  One basic assumption made in this formulation is that the volume of particles is conserved during the collision process and that only binary collisions are important. The assumption of conservation of volume has been disputed by some authors who maintain that larger floes have lower densities because they incorporate more water (Lick and Lick, 1988). These authors recommended that the assumption of conservation of mass is more reasonable.  2.2.1.2 Numerical Form The integral form of the coagulation equation can only be solved analytically. This may not be possible, particularly when the variation of particle volume cannot be easily characterised. As a result, in practical modelling situations, numerical forms of the coagulation equation are used which allow for discretisation of the particle volumes into a  12  number of size classes. A complete form of such an equation was presented by Lick and Lick (1988) and is reproduced below:  dt  1 j i+  =k  / = 1  (2.6) j=k+l  where,  /=!  j=k+l  i'=l  = the concentration of particles within size range k, Ay = an empirically determined coefficient representing the probability of cohesion of particles of size i and j after collision, Bfc = an empirically determined coefficient dependent on shear, floe diameter and effective density, C = the probability of disaggregation of a particle of size k after collision with a particle of size i, and ik  y,jt = the probability that a particle of size k will be formed after the disaggregation of a particle of size j, and is described by the following. 7y*=-^T  7-1  (2-7)  The existence of this theory demonstrates that flocculation modelling is achievable with a theoretical basis. At the same time however, there is a heavy dependence on the use of poorly understood coefficients which must be determined experimentally. The theory also indicates the dynamic nature of the process so that at equilibrium, aggregation and disaggregation must be evenly matched.  2.2.2 Disaggregation Theory Consideration of aggregate break-up mechanisms and floe strength allows prediction of maximum floe diameters. The theory developed for the processes controlling the disaggregation of floes is not as well developed as for the physical processes governing aggregation.  There are two main mechanisms of floe break-up, both of which are involved in turbulent systems (Miihle, 1993), and the floe strength varies according to the mechanism by which it is stressed. Floe splitting refers to situations in  13  which the floe is stressed as a whole and large scale splitting of the floe occurs. Floe erosion refers to the detachment of single particles from the surface of the aggregate. The size of resistive force and the force required to detach the particle depend on the size of the particle being removed. A number of models of floe strength in turbulent flows have been developed, and all are highly empirical (Miihle, 1993; Partheniades, 1993), yet in many models the Kolmogoroff microscale is treated as a very significant quantity for turbulent flows. The Kolmogoroff microscale in a flow is defined as the scale of the smallest eddies and is an important parameter because these eddies are primarily responsible for the dissipation of energy into heat (Landahl and Mollo-Christensen, 1987).  The reason that the Kolmogoroff microscale is such an important factor in the disaggregation process was discussed by Miihle (1993). The flow within the smallest turbulent eddies can be considered to be laminar, and it is only at a scale larger than this that the flow can be considered turbulent at all. If a floe is larger than these eddies, then it will tend to disaggregate due to floe splitting as the difference in shear between one eddy and the next results in the tearing apart of the floe. This tearing may fragment the floe into particles of similar size or particles very different in size, and depends entirely on the arrangement of the floe and the relevant turbulent eddies. If, on the other hand the floe is smaller than the smallest eddies in the flow, it is likely that a given floe will be completely engulfed in the eddies, in which case it is actually subjected to a laminar flow field, and hence only laminar shears are present. In this case then, floe erosion is considered the likely disaggregation mechanism.  Based on consideration of these mechanisms, Partheniades (1993) presents two formulae (one for floes larger than the turbulent microscale, and one for floes smaller) for the maximum diameter (d ) of floes in turbulent flow. a max  They are reproduced below: -1/(1+Xp)  -i/a+v = K,  ford  a  P  » ^ ,and 0  (2.8)  -3/2(3+Kp )  0'  where:  14  (2.9)  V  K =k x  x  2(l+if ) p  2(3+AT ) p  ; and K = k 7  (2.10)  7  VP k and k are constants depending on the properties of the sediment-water system. The other variables are defined as {  2  follows: X = Kolmogoroff microscale of turbulence, 0  e  = energy dissipation rate or per unit mass ( £ ' ), and  K  = reported to be in the range from 1 to 1.5 (Tambo and Watanabe, 1979).  p  The values of the constants can be assumed to vary depending on the material from which the primary particles are composed. Similar predictive models based on floe strength considerations have also been developed by other researchers, and the model described above should be considered only an example. The strongly empirical nature of Equations 2.8 to 2.10 is also obvious. This makes application of the theory to individual situations difficult without extensive experimental work.  2.3 A G G R E G A T E FORMATION MECHANISMS A number of aggregation mechanisms in natural systems have been identified by various researchers, and an understanding of important mechanisms is necessary. The formation of aggregates involves two distinct processes. First, particles must be transported close to one another ensuring a collision. Secondly, the particles must bind together resulting in the formation of an aggregate. The two most important transport mechanisms for sediment sized particles in natural environments are differential settling and turbulent shear.  2.3.1 Differential Settling Differential settling occurs when a particle settles faster than those around it, and collides with particles beneath it. This mechanism is envisaged as an important mechanism for the behaviour of particles larger than 2 um in diameter when conditions are reasonably quiescent. Such conditions will occur far from flow boundaries or where the depth is quite large. The formula for the collision frequency function derived by Findheisen (1939) was presented by Pearson et al. (1984) as:  15  2ng 9u  P,  -Pw  (r,+r,) r, -V 2  V  Pw  y  2  where g represents the acceleration due to gravity, v represents the kinematic viscosity of the fluid and  (2.11)  and p are w  the floe and fluid densities respectively. The structure of a floe formed due to differential settling is shown in Figure 2.5. The floe appears to be very fluffy and has been shaped as a result of its downward passage through the water column. The generally fragile nature of these floes is very similar to that of the floes that are formed as a result of Brownian motion.  Figure 2.5. Photograph of a floe formed as a result of the action of differential settling (Lick and Huang, 1993).  2.3.2 Turbulent Shear Turbulent shear flows are also responsible for bringing particles together due to the random nature of turbulence. The collision function for this mechanism in isotropic turbulence was determined by Saffman and Turner (1956), as cited by Pearson et al. (1984). The function is given as:  16  l3, =2.3(^.+r.) (e/v) 3  2  7  (2.12)  where e is the rate of turbulent energy dissipation. This formula is also often written substituting the mean shear rate of the fluid (G) for (e/v) . m  The parameter G is used commonly in flocculation work in the environmental  engineering literature. In this field, G is defined as (Metcalf and Eddy, 1993):  G=  (2.13)  where P represents the power applied to the fluid by the mixing device that is in use, u represents the viscosity of the fluid, and V represents the volume of the mixing reactor.  In fluvial systems, other methods of calculating the velocity gradient have been proposed due to the high degree of non-uniformity in the flow field. Open channel flows are characterised by highly non-uniform turbulence with a relatively narrow zone of high shear rates and non-uniform turbulence near the flow boundary, and a main zone of low shear rates and near homogeneous turbulence far from the boundary (Partheniades, 1993). The details of this theory for open channel flow have been presented in Appendix A.  The mechanism of turbulent shear is probably the most important flocculation mechanism at work in a fluvial system such as the Fraser River. The other mechanism of differential settling is important under conditions of quiescent flow, and has been utilised in previous modelling work (Krishnappan, 1990; Lick et al, 1993 (b)). Aggregates formed by collisions due to turbulent shear have been subjected to considerable stresses, they reach their steady-state size distribution after a period of breaking and recombining, and have higher density and strength than aggregates formed by differential settling (Partheniades, 1993).  2.3.3 The Role of Organic Material It has generally been accepted that physico-chemical forces are responsible for the binding of primary particles into aggregates. More recently, the role of biological activity and material in the formation of aggregates from inorganic and organic material has been recognised. Ongley et al. (1992) suggested that chemical flocculation in fresh-water is probably subordinate to other physical and biological mechanisms.  17  Muschenheim et al. (1989) identified two distinct modes of flocculation in experiments involving Spartina detritus and granitic till. Some aggregation was observed to occur over almost instantaneous time scales, and it was thought that this could not be a microbially mediated process due to the rapidity of the effect. Possible reasons were thought to be physico-chemical effects or some binding action due to the organic material present. It was also suggested that elevated dissolved organic carbon levels would increase the negative surface charge on particles and thus make them less prone to flocculation. The second mode of aggregation identified was biologically mediated, took place over longer time scales and produced larger aggregates. This type of aggregation did not occur until between 24 and 48 hours after the start of the experimentation.  According to Muschenheim et al. (1989) a clear picture is still yet to emerge from the above-mentioned work to explain the degree of interdependence between microbial growth and detrital aggregate formation. The high rates of microbial production and aggregate formation associated with turbid environments indicate that the two processes may be linked. The actual role played by the bacteria is also under some debate, as various authors have arrived at different conclusions. Riley (1963) proposed that bubbles and detrital particles provided nucleation sites for the formation of aggregates, suggesting that only subsequently, did the aggregates provide surfaces for microbial growth. Work by Paerl (1974) and Biddanda (1985) has demonstrated the potential for microbial mediation of aggregate formation. There is little doubt that bacteria do attach to particles in aquatic environments, and it was found by Bell and Albright (1981) that in the Fraser River Estuary, it was only in saline waters that bacterial attachment was not as apparent. Within the fresh water portion of the River itself, it was not uncommon that as much as 90 % of the bacteria in the water column was attached to particulate material. In conclusion then, it is likely that microbial material will play some role in flocculation of fine-grained sediments in fluvial systems. The nature of this role is yet to be defined.  2.4 IMPLICATIONS OF FLOCCULATION FOR TRANSPORT MODELS 2.4.1 Sediment Transport Flocculation of fine-grained particles will change the fluvial transport behaviour of these particles. Traditionally, fine particles are treated as discrete, however some recent work has attempted to incorporate aggregation and disaggregation into sediment transport models (Gailani et al, 1991; Lick et al, 1993 (a); Krishnappan, 1990; 18  Lavelle, 1993; Hunt, 1990). Invariably, gross simplifications of the coagulation theory discussed previously are made to allow computationally reasonable modelling.  In the work of Lick et al. (1993 (a)) simple empirically derived relationships were incorporated into a cohesive transport model to account for flocculation. When fluid shear is the dominant aggregation mechanism, they showed that solids concentration and fluid shear have a similar effect on the steady-state median floe size (d ), and on the m  time to steady state (T ). s  From their experimentation with Detroit River sediments, they developed the  approximation given as, d  m  =9.0(CG)-°  (2.14)  56  where C represents the solids concentration (g/cm^), and G is the fluid shear (s"l). Similarly, a relationship for the time to steady state conditions was derived, T =12.2(CG)-°  (2.15)  36  s  where the units for 7^ are minutes. Different coefficients in both Equation 2.14 and 2.15 would be obtained for different sediments in either salt or fresh water, so these results cannot be applied universally.  The reasons for the dependence of d and T on the product CG are different for each quantity (Lick et al., 1993 (a)). m  s  The rate determining step in achieving steady state conditions is the aggregation of the smallest particles, so T is a s  function of the rate of aggregation for the smallest particles. In this case, binary collisions are most important, and it can be shown that collision rate is proportional to the product CG. Median diameter however, is an equilibrium property, and hence independent of the initial rate of aggregation. At equilibrium, d is determined by a balance m  between aggregation and disaggregation. While the aggregation rate is primarily determined by binary collisions, the disaggregation rate is primarily dependent on three body collisions (Lick and Lick, 1988; Burban et ai, 1989). The ratio between these two rates is a function of CG, and hence d should also be a function of CG. m  Quantification of settling velocities was also necessary for the model of Lick et al. (1993 (a)), and requires consideration of the fact that aggregate densities change with the size of the floe. The following experimentally derived relationship for floe settling speed (w ) was used, s  19  w = ad  (2.16)  s  where both a and m are parameters dependent on fluid shear, sediment concentration and salinity.  An example of an even more simple model incorporating flocculation was constructed by Hunt (1990). This model was constructed for the purpose of modelling effluent particle transport after discharge of a municipal waste plume into oceanic waters. Coagulation of effluent solids was thought to be enhanced by the discharge to saline water, resulting in double layer compression.  A simple modelling approach based on dimensional analysis and  observations (Hunt, 1982) was used. According to these observations, the overall kinetics for removal of a suspended-particle mass from a fluid element is second order in particle concentration, i.e. —  = -bC  2  dt  (2.17)  where C represents the particle concentration in mg/L, and b is a rate parameter dependent on particle aggregate properties and the fluid shear rate. This equation can be solved analytically to give, 1 , — = bt + a  C  (2.18)  where a is a constant of integration. The constant can be evaluated by consideration of the initial condition that C = 1 C at t = 0, giving a = 0  . The experimental determination of b was necessary for application to the actual  modelling situation.  In conclusion then, flocculation offine-grainedsediments does have an effect on sediment transport. Some attempts have been made to quantify the effect, although more work is necessary before sediment transport calculations can be made realistically while taking account of flocculation effects.  2.4.2 Contaminant Transport There is no evidence in the literature of the inclusion of flocculation effects in contaminant transport models. Despite this, the importance of flocculation to contaminant transport has been recognised by Irvine et al. (1995). Merv Palmer (1995, pers. comm.) also suggested that biological floes are responsible for transport of contaminants long distances from their source. 20  Flocculation of fine grained (< 63 urn) sediments can have a profound effect on the transport of contaminants in fluvial systems (Droppo and Ongley, 1989). Several studies have shown that it is the smaller size classes of sediment particles that are associated with the transport of contaminants. Horowitz and Elrick (1987) found the < 63 um and < 125 um size fractions to be most closely correlated with the bulk metal chemistry for a wide variety of stream sediments from 17 geographically and hydrologically diverse areas of the United States. Stone and Mudroch (1989) found that mineralogical and chemical composition of the finest fraction (< 13 um) produced the largest measured absorption. Karickhoff (1981) showed that the particulate organic carbon content of the suspended solids plays a central role in organic contaminant transport.  In conclusion then, the flocculation of fine-grained sediments in fluvial systems is indeed important and will affect sediment transport as well as the transport of hydrophobic contaminants associated with the sediments. Flocculation is an issue on which some researchers have made some progress on, but in applications to contaminant transport, there has been very little success achieved due to the incredible complexity of the problem. According to Ongley et al. (1992) knowledge gaps exist in the areas of: (a)  the physical behaviour of the cohesive fraction of suspended sediment vis-a-vis freshwater flocculation,  including microbiological controls of this process; (b) the uncertainty in predicting sediment transport for the < 63 u.m fraction; (c) the role of suspended solids in toxicity assessment of river water; (d) the chemical partitioning on suspended sediment. All of these areas will have to be addressed if accurate answers are to be obtained regarding the fate and pathways of contaminants in aquatic systems.  2.5 A G G R E G A T E MEASUREMENT TECHNIQUES The fragility of aggregates formed from organic and inorganic material has made their measurement particularly difficult. Despite this, a number of measurement techniques have been developed, all of which have their own weaknesses under certain conditions. A detailed account of all known aggregate measurement methods is provided by Farrow and Warren (1993). The methods used to obtain aggregate or particle size data in this study will be discussed here.  21  2.5.1 Electrical Resistance The use of the property of electrical resistance has been applied in many machines such as the Coulter Counter. With this technique, particles in an electrolyte pass one by one through an aperture across which there is an electric field (Farrow and Warren, 1993). The change in the electrical resistance across the aperture is proportional to the volume of the particle. Particle sizes determined by this technique are reported in terms of the diameter of the sphere having the same volume as the particle. A number of researchers have used these devices for determination of particle size distributions (Hunt, 1980; Petticrew, in press; Kranck, 1979). A slightly more modern device working on the same basic principle is the ELZONE as used by Teeter (1993).  There are two well recognised limitations of this technique (Farrow and Warren, 1993). The first lies in the problem of break-up of aggregates as they pass through the aperture. Hunt (1980) in particular noted some limitations of the device in that larger aggregates would break up before they entered the sensing zone of the orifice and thus, could not be sized. Secondly, dilution of the sample with an electrolytic solution may effect the aggregate size distribution. A conducting suspension medium is necessary to ensure that an electrical field can be maintained across the aperture. Other criticisms of the technique were made by Treweek and Morgan (1977). For loosely bound porous aggregates, a correction for porosity of the aggregate must also be made to ensure that realistic results are obtained. They suggested that aggregate sizes were underestimated due to the fact that the counter senses only the presence of particulate matter within the aggregate. Also, on passing through the aperture, an aggregate may be elongated in response to the flow field resulting in an underestimation of the particle's true size. For this reason, such devices are best suited to quantitation of primary particle size distributions.  2.5.2 Image Analysis A recently developed method of aggregate size distribution measurement is the use of image analysis. This method has one significant advantage over other techniques in that visualisation of the actual floe particles is possible. Visualisation allows a feel for the nature of the floes to be obtained, and makes detection of disturbance a simple matter. The primary criticism of this technique is that measurements cannot be made in situ.  22  Various methods of image analysis have been developed in recent times, with one of the most promising having been developed by Droppo and Ongley (1992). In this publication they document a method of sample collection and analysis that they developed. A sample is collected from a field site in a special plankton settling chamber. The size of the chamber must be chosen based on the sediment concentration in the water course at the point of interest. The particulate matter is allowed to settle out to the bottom of the plankton chamber, and then remains there. After settling of the solids, the plankton chamber is then placed on an inverted microscope and the image of the natural flocculent particles observed. Analysis of the image is then required. In their publication, Droppo and Ongley (1992) describe how the image was projected, and the individual particles were then digitised. This procedure was particularly time consuming and since this publication (Droppo, 1995, pers. comm.) significant advances have been made in the areas of image capture and processing. Image capture has been improved by the connection of a Charge Coupled Device (CCD) to the microscope, allowing for the direct digitisation of the image. The image is then analysed by a software package known as OPTOMAX. This software both defines and sizes the aggregates in the image it receives, resulting in a much quicker analysis.  2.5.3 Laser BackScatter / Diffraction The use of laser devices for the measurement of particle size distributions in aquatic systems has also been documented (Krishnappan et al., 1994, Phillips and Walling, 1995). With each device, it is possible to simply immerse the probe in the water for which aggregate size distribution measurements are desired.  The laser diffraction device used by Krishnappan et al. (1994) consists of a 2 mW laser, a receiving lens, a detector plate, an electronic interface and a microcomputer. The laser beam passes through the sample volume and diffracted light is collected and focused onto the detector plate. At this point the measured distribution of diffracted light is used to calculate the size distribution of the particles in the sensing volume. The laser backscatter equipment described by Phillips and Walling (1995) operates on a slightly different principle but produces similar measurements.  These laser devices are perfect for field applications as they can be easily transported and a short period is required for an accurate measurement to be made in situ. Measurements can also be made non-intrusively, minimising the  23  disturbance to aggregates. Despite this, they allow no visualisation of aggregates, which means there is no way of telling what is being measured in a water course.  2.5.4 Turbidity Measurements Turbidity measurements have been used in a number of situations for the measurement of particle sizes or for the determination of the presence of coagulation or flocculation (Holman, 1986; Brown et al., 1985). The location of a turbidity maximum in an estuary (a zone of higher suspended matter concentrations) is also determined with the use of turbidity measurements (Grabemann and Krause, 1989). Turbidity is a measure of the quantity of light scattered as it passes through a given suspension, yet this information has often been used to infer information about the particle size distribution of the sample. A very good account of the relationship between suspension turbidities and the aggregate size distribution is given by Treweek and Morgan (1980). Here, turbidity is defined in the following equation: T = ^- = e- "' =e' NC  (2.19)  xl  where T represents the transmission, 7, represents the light intensity transmitted through a suspension of length /, 7 the incident light intensity, N the number of particles in suspension, C  ext  0  the extinction cross-section, and x the  extinction coefficient, or suspension turbidity. After studying the turbidities and particle size distributions of a number of suspensions of E. coli bacteria, Treweek and Morgan (1980) showed that measurement of turbidity alone was not sufficient to ascertain whether or not a system was undergoing coagulation or flocculation. In a flocculating system, it was possible that the turbidity could increase, decrease, or remain the same. Factors affecting this included the refractive index of the particle material, the particle size distribution, and the particle concentration. Still, despite this, turbidity has been used in many applications for the identification of the onset of flocculation or coagulation.  24  3.0 APPLICATION OF EXISTING THEORIES The flocculation of fluvial sediments in freshwater systems due to the discharge of pulp mill effluents has not been studied in any depth. It is a phenomenon that has been observed by only one research group (Krishnappan et al., 1994; Krishnappan et ai, 1996 unpublished data). As such, there is no real theory on which to base modelling attempts at this time. Even the development of theory is impossible before the mechanisms responsible for the flocculation and the magnitude of the resulting effect have been established. In examining some potential methods for modelling the situation, comparisons were made between a simple theory and available relevant data and a simple model was proposed.  3.1 CALCULATION OF VELOCITY GRADIENTS IN TURBULENT FLOWS Velocity gradient is an important parameter with respect to floes because it provides a convenient description of the break-up forces to which a floe is subject in a particular flow regime. A calculation of this parameter is useful to gain a feel for the conditions present in the River for comparison with the literature. The theory presented by Partheniades (1993) allows for calculation of the magnitude of the turbulent shear in the Fraser River in regions close to the bed, as well as far from the bed. Assuming a channel depth of 2.15m (in a rectangular channel (Vine, 1996)), and a slope of 425 x 10" m/m (Northwest Hydraulic Consultants Ltd., 1993) it is possible to determine that 6  the turbulent shear in the main body of the flow is 85 /s. Within the viscous sub-layer, the magnitude of the turbulent velocity gradient is estimated as being 22,500 /s. The Kolmogoroff microscale can also be calculated from this theory and was calculated as being of the order of 10 |im for the viscous sub layer, and of the order of 110 um in the near homogeneous zone far from the bed. The theory and assumptions used to obtain these results are included in Appendix A.  The actual calculation of flocculated particle sizes from this information is not possible without the determination of some constants through experimentation. If this was done, the methods discussed in §2.2.2 and §2.4.1 could be used to evaluate the in situ maximum and median aggregate diameters respectively.  25  3.2 SECOND ORDER COAGULATION AND SETTLING THEORY One of the potentially most useful modelling strategies for flocculation studies is that of second order coagulation and settling as devised by Hunt (1980) and described in §2.4.1. The experimental work from which the 2nd order coagulation and settling theory was derived is similar to the experimental work performed by Krishnappan and Engel (1994). In the experiments of Hunt (1980), clays were dispersed in a suspension subject to a constant velocity gradient while the concentration of solids and aggregate size distribution were measured as a function of time. The beauty of the theory developed lies in its simplicity, and Hunt (1990) applied it to the modelling of the discharge of a sewage effluent into oceanic waters.  It was thought that second order coagulation and removal may be applicable to the work of Krishnappan and Engel (1994) for flocculation and settling of cohesive Fraser River sediments in an annular flume. In the experiments of Krishnappan and Engel (1994), the deposition of Fraser River sediment was observed under conditions of constant shear and a brief description of the method utilised follows. Adequate sediment was added to the flume to yield a fully suspended concentration of 250 mg/L. Mechanical mixing of the suspension was performed to ensure that all floes were broken up. Rotation of the flume was then commenced, at a speed that was higher than would be utilised for the actual deposition test. This speed was high enough that floes would not be formed, and no deposition of sediment would occur. Once this had been maintained for 20 minutes, deposition was allowed to proceed at a lower rotational speed, and hence lower bed shear stress. As in Hunt's experiments, measurements of total suspended sediment concentration (TSS) and the aggregate size distribution were made throughout the tests.  3.2.1 Comparison of Results to Theory The observations of concentration as a function of time from Krishnappan and Engel (1994) can be analysed to see if they produce results consistent with second order coagulation theory. The data supplied by Krishnappan and Engel (1994) are tabulated in Table 3.1.  If second order coagulation theory is followed, then a plot of inverse  concentration as a function of time should yield a straight line.  It was decided that only a portion of the data set should be analysed to ensure that the greatest similarity to Hunt's experiments was achieved. Since during the first twenty minutes of the test no deposition occurs due to the high 26  Table 3.1. Summary of the data of Krishnappan and Engel (1994). TSS (mg/L) Time (min)  Shear Stress (N/m ) 2  0.0563  0.121  0.169  0.213  0  186.8  186.2  197.8  191.8  5  181  -  195.9  195.9  10  180.5  181.5  184.5  179.2  15  174.4  -  187.9  200.3  20  167.3  178  175.8  184.7  25  137.9  -  148.3  180.8  30  67.5  94.7  115.3  151  35  35.5  -  114.3  140.7  40  20.5  60.5  100.5  137.2  45  22.5  -  106.1  -  50  19.7  48.4  106.3  140.8  60  9.6  46.7  102.1  137.5  70  13.4  37.4  102  127.7  80  10.4  41.3  96.9  133.2  90  16  41.5  100.6  136.1  100  10.5  33.4  95.4  143.1  110  -  -  98.1  -  120  -  32.8  -  134.1  -  97  -  130 140  -  29.6  -  130  150  -  -  98  -  160  -  30.5  -  131.6  170  -  -  93  -  180  -  30.8  -  124.2  190  -  -  92.8  -  200  -  29.2  -  125.8  210  -  -  93  -  220  -  31.8  -  123.9  230  -  -  92.4  -  27  0.120 i•  0.100  /  0.080  Sh«sar = 0.0563N/m2  E ^. 0.060 o  <  She ar = 0.121 N  >/ •  /•  0.040  n L,l •  0.020 0.000  20  ri  —H  E  60  40  h—n  80  100  120  140  Time (min) (a)  0.012 She;ir = 0.16$ N/m2  !  ^ 4  >  0.010 •  c 0.008  o S  •  <  •  a  •  | 0.006 o  [  c o  Shear = 0.213 N/Tl2  o 0.004  w  > - 0.002 0.000 0  20  40  60  80  100  120  140  160  180  Time (min) (b) Fisure 3.1. Comparison of results from deposition test to lines of bestfit(linear): (a) for shear stresses 0.0563  N/m and 0.121 N/m , and (b) for shear stresses 0.169 N/m and 0.213 N/m . Raw data used with permission of 2  2  2  Krishnappan (1995). 28  2  shear in the flume, this portion of the data can be eliminated. It also seems reasonable that some time period would be required for the turbulent velocity gradients in the flume to reduce from the value present during the initial 20 minute mixing period to those sustained at the constant shear relevant to that particular test. This dissipation has been assumed to take 10 minutes. Hence, in accordance with this reasoning the data from the first 30 minutes of the test can be ignored. Some data from the end of the test can also be eliminated from consideration. After some time in each test, an equilibrium flume sediment concentration was reached for each constant shear. Since Hunt's theory describes a non-equilibrium situation, the theory cannot be used to describe conditions under steady state when  dC dt  approaches zero, and C * 0. The times to steady state flume sediment concentration were different for each shear, and were assumed to be 60 min for a shear of 0.0563 N/m , 140 min for a shear of 0.121 N/m , 170 min for a shear 2  2  of 0.121 N/m and 180 min for a shear of 0.213 N/m . Data collected at times greater than these selected times were 2  2  then eliminated from the analysis.  A comparison of these selected data to second order coagulation theory is presented in Figure 3.1. It would appear in plotting these data, and their associated lines of best fit, that the second order theory is a good fit for all cases except for a shear of 0.0563 N/m . For this case, second order theory still provides a reasonable fit to the data. The 2  results presented by Hunt (1980) showed a similarly good agreement between experimental data and second order removal theory.  3.2.2 Calculation of Removal Rate Parameters The parameter b in Equation 2.17 can now be evaluated. This was done by simply calculating the slope of the lines of best fit plotted in Figures 3.1 (a) and (b), which provides b for the conditions in the annular flume and for Fraser River sediments during the deposition phase of the test at constant shear. These results have been tabulated in Table 3.2.  29  Table 3.2. Calculated Coagulation Removal Rate Parameters from experiments completed by Krishnappan and Engel (1994). Turbulent Velocity Gradient (G)  Hunt's Removal Rate Parameter  (/s)  (*) (L/mg.s)  0.0563  25.8  2.7 x 10"  3  0.121  55.7  1.8 x 10"  4  0.169  77.8  1.1 x 10"  0.213  98.0  5.6 x 10"  Bed Shear Stress (N/m ) 2  5  6  Calculations of the values of velocity gradient in the annular flume were made based on a calibrated k - £ model fitted to the flow characteristics measured in the flume (Krishnappan, 1996, pers. comm.). At a location just outside the laminar sub-layer, and inside the turbulent boundary layer near the bed, the calibration ^ 0.46T G=  (3.1)  lOOOv can be used to describe the turbulent velocity gradient. The bed shear stress is represented by T, and v represents the kinematic viscosity of water. It could be expected that the shear far from the bed would be somewhat less than this, but of the same order of magnitude.  It is difficult to compare these calculated removal rate parameters to those obtained by Hunt (1980). Typical values obtained by Hunt (1980) are tabulated in Table 3.3 along with the shear for which they were derived. It is only possible to realistically compare the values for b obtained from the data of Krishnappan and Engel (1994) for G = 25.8 /s to the values obtained by Hunt (1980) for G = 16 /s. The result from Hunt (1980) at 3.2 x 10" L/mg.s is two 5  orders of magnitude lower than the result for G = 25.8 /s from Table 3.2 of 2.7 x 10" L/mg.s. This indicates that 3  flocculation and settling of the Fraser River sediments in the annular flume occurs much more quickly than for Hunt's montmorillonite clay particles under similar velocity gradients. This is a sensible result considering the difference in primary particle sizes between the Fraser River sediment and the clay particles.  30  Table 3.3. Second order coagulation removal rate parameters as determined by Hunt (1980) for montmorillonite. Velocity gradient (G)  Hunt's Removal rate  Us)  parameter (b) (L/mg.s)  0.5  6.1 x 10"  1  6.1 x 10"  2  9.0 x 10  -5  4  5.5 x 10  -5  8  3.1 x 10"  16  3.2 x 10"  32  0.0  5  5  5  5  In conclusion then, Hunt's theory can be used to describe the flocculation and deposition of Fraser River sediments in a turbulent shear flow.  Despite this, further investigation of the suitability of second order theory to the  flocculation phenomenon in the Fraser River was not considered prudent until more was known about the nature of the phenomenon.  3.3 COAGULATION OF PARTICLES IN A PLANE PLUME Holman (1986) developed theory for the coagulation of particles in a plane plume in a destabilising ambient using dimensional analysis. Expressions were obtained for a coagulation initiation length scale, and a coagulation quenching length scale. The zone between these two length scales represents the area where most coagulation occurs within the plume. Figure 3.2 shows the arrangement of length scales and the region of rapid particle coagulation. The onset of rapid coagulation is governed by the time required for sufficient mixing to occur between the solids discharged in the plume and the destabilising ambient fluid, and the maintenance of solids concentrations at high enough levels that high collision rates will be present. At the point at which rapid coagulation begins, the concentration of particles is high, and sufficient time has passed for adequate mixing to occur. Holman (1986) described the coagulation initiation length scale as  1  -  ^1  ~  V  D  l/3  f  Ji Aon 31  (3.2)  where B is the buoyancy flux at the plume orifice, / is a function and D „ is the diffusivity of the destabilising solute 0  io  ions. Rapid coagulation ceases when the concentration of solids decreases sufficiently, that collision rates are reduced. The dilution of the effluent is responsible for the reduction in solids concentration as one moves away from the plume outlet which occurs beyond the coagulation quenching length scale (^ ). This length scale was defined as q  K =  —f Jq  v  (3.3)  1/3  0  where D is the diffusivity of suspended particles in the plume. p  This theory suggests that if coagulation is occurring in the buoyant jets of the Northwood diffuser, then it is likely to occur in the near field mixing zone where concentrations of the effluent are at their highest. However, it is difficult to apply the results to the Northwood situation, even neglecting the vast geometrical differences between Holman's experiments and the Northwood effluent diffuser.  The reason for this is that the coagulation/flocculation  mechanisms may not be the same. At Northwood, the effluent may be the destabilising fluid, and interaction of biosolids and entrained sediment may also be responsible for aggregation. Hence, it is unlikely that Holman's theory could be realistically applied.  It was thought possible that a simple relationship may exist between the concentrations of the effluent and the ambient, and the zone of rapid coagulation in the plume evidenced in Holman's (1986) work. If so, this basic relationship could possibly be extended to the Northwood scenario. The form taken by such a relationship would be,  coagulation rate °c C* * C* ff  where C  (3.4)  amb  is the effluent concentration normalised by the initial effluent concentration, and C*  eff  amb  is the ambient  concentration normalised by the ambient concentration far from the plume. For a plane plume, expressions for C and C eff  amb  can be determined based on the theory in Fischer et al. (1979).  32  Figure 3.2. Evolution of a Coagulating Plume (adapted from Holman, 1986).  Calculations were made for the plume case considered by Holman (1986) with a plume Reynold's number of 51, an initial buoyancy flux of 28 cm /s and a plume slot width of 0.16 cm. The spatial variation of the product of L g and e  C  amb  is shown in Figure 3.4. Superimposed on this is the measured location of the zone of rapid coagulation which,  according to Holman (1986), lies between 3.4 and 6.7 cm from the plume outlet. The location where C g * C e  anib  peaks is at around 1 cm from the plume outlet as evidenced from the contours in Figure 3.3. This does not correspond to the location determined from measurements.  To accurately predict the location of this zone,  experimental work as well as consideration of the time scales involved in mixing and diffusion may be necessary. Consideration of these time scales may result in translation of the predicted zone of rapid coagulation upwards from the plume orifice and closer to the location of the zone as measured. Such a detailed study cannot be justified at this time when there are so many uncertainties regarding the nature of flocculation between pulp mill effluents and 33  -  2  -  1  0  1  2  x (cm)  Fisure 3.3. Contour plot of the product normalised concentrations of the plume and ambient i.e. C*  eff  * C*  sediments. The simple approach investigated was inadequate for identification of the zone of rapid coagulation and was not pursued in relation to the Northwood diffusers.  34  amb  .  3.4 POSSIBLE COAGULATION / FLOCCULATION MECHANISMS PRESENT Consideration of all the issues discussed in Chapter 2 suggests that a suite of mechanisms may be responsible for flocculation of pulp mill effluents with Fraser River sediments. The immediate potential for modelling the situation, then, is very low. The sketch in Figure 3.4 provides an overall description of the situation, and the processes that may be involved.  The discharge of the effluent into the river through the Northwood diffuser as a buoyant jet (Marks, 1996) results in increased turbulence in the near field region. The near field is defined as occupying 20 m downstream of the diffuser (Marks, 1996).  The buoyant jets result in higher localised turbulent velocity gradients, one of the factors  traditionally considered to have an impact on flocculation. The gradual mixing of the effluent with the waters of the Fraser River results in the entrainment of sediment into the plume, and thus a localised increase in the concentrations of suspended solids in this mixing zone. The effect of the increased solids concentration and the higher turbulent velocity gradients leads to a higher collision rate between particles in this near field area. Once outside this region, where the influence of jet momentum is small, the shear is reduced to a much gentler level. This variation in shear is similar to that used in traditional coagulation for water treatment where a coagulant is mixed for a short period of time at high shear to encourage coagulation, and then a lower shear is applied for some time after to allow aggregation and deposition. If the effluent behaves as a coagulant of some kind for the river sediments, then it would seem that mixing conditions may be ideal for aflocculationeffect to occur. The variation in shear is also similar to that utilised in Krishnappan's flume experiments. Since Hunt's 2nd order coagulation and removal theory was shown to fit the relevant portion of the flume data, there exists the possibility that this modelling strategy could be extended to the situation present on the Fraser River.  The probability that any collision will result in aggregation may also be increased. Measurements of the temperature of the river water downstream of the diffuser (Evans, 1995, unpublished data) showed that the temperature of the plume was barely elevated in comparison to the river despite the fact that before addition to the river, the effluent was around 20 °C warmer than the river water. The temperature reduction experienced by the effluent results in the reduction of electrical double layer thicknesses of the colloidal material present in the effluent, thus destabilising it.  35  The double layer thickness of the entrained fine sediment particles may also be reduced due to the high ionic concentration of the effluent having a destabilising effect on colloids there also. Sediment entrainment also provides Natural River  Effect on Flocculation  Impacts of Mixing  Effluent  § 'IIP  Potential Higher Velocity Gradient  Increased Turbulence  w  Sediment Entrainment  w Solids W  w  Increased  Concentration Biosolid & Sediment Interaction  ° More frequent collisions • Higher probability of aggregating collisions  Fisure 3.4. Diagram showing the possible effects of discharging pulp mill effluent into the Fraser River.  the opportunity for contact between effluent biosolids and the sediments in the river and, considering the fact that some affinity seems to exist between sediment and bacteria as discussed in Sherman (1953) and Muschenheim et al. (1989), then the probability that any collision will result in aggregation may be increased. The combination of the increase in the number of collisions and the increase in the probability that they will result in aggregation could be responsible for the field observations indicating flocculation (Krishnappan, 1994; Krishnappan, 1996, unpublished data).  36  3.5 R E S E A R C H O B J E C T I V E S In view of the preceding literature review and analysis, it is possible to set out the objectives of the present study. •  To reproduce, by some experimental means, the flocculation and deposition phenomenon noted in the Athabasca River (Krishnappan et al,  1994) and Fraser River (Krishnappan, 1996, unpublished data) since theory and  previous work can only be used to postulate the possible existence of this behaviour. •  To determine the magnitude of any reproducible flocculation effect. By determining the magnitude of the effect, it will be possible to ascertain the relative importance of this phenomenon to sediment and contaminant transport modelling.  •  To investigate the effect that dilution of the effluent has on any observed flocculation effects and hence the region of the effluent plume that is important from the point of view of modelling.  •  To determine the effect that sediment type has on flocculation, and the likelihood that biosolids in the pulp mill effluent are an essential part of the phenomenon.  •  To investigate the effect that temperature has on the phenomenon, and hence the time of year, if any, when flocculation will be most important in the Fraser River. This will also provide insight to the role of temperature in the flocculation process.  •  To personally analyse any relevant field data. A n analysis of this data will seek to identify the magnitude of flocculation effects observed in the Fraser River.  As can be seen from this list, a wide range of objectives was pursued. It was hoped that conclusions could be reached which advance the understanding of the nature of the flocculation phenomenon. There is certainly little prospect for modelling this phenomenon before more is understood about its basic nature.  37  4.0 EXPERIMENTS As part of this study, a number of different standard flocculation and coagulation experiments were completed in the hope that the flocculation phenomenon could be reproduced experimentally. Some variations of standard tests were also completed. The experimental work was undertaken in two phases. The first phase included experiments completed in the laboratories at the University of Northern British Columbia (UNBC), and in the field near the Northwood Pulp and Timber Mill. This work was completed during July of 1995. The bulk of the experiments, which comprised the second phase of experiments, were conducted in the environmental engineering laboratories at the University of British Columbia (UBC).  4.1 BASELINE DATA COLLECTION While the author was in Prince George, temperature information was collected to provide general background for the condition of the effluent and the river. A thermistor was placed in the final effluent lagoon at a level 0.6 m below the surface. A thermistor was also placed in the river. It was located approximately 3 m from the water's edge at the time of deployment. Owing to a significant increase in the flow in the river, as the data collection period progressed, the thermistor came to be in deeper water, and a lot further from the bank. Previous measurements of temperature in the river and the effluent lagoon indicated that there was no temperature stratification in either, (Evans, 1995, unpublished data) so the temperatures at these locations could be considered to be representative of the entire water column in the effluent lagoon and the river respectively. Temperature data were collected for 8 days in total.  4.2 SAMPLING PROTOCOLS 4.2.1 Effluent The sampling protocol adopted for the work at Prince George was reasonably simple. Effluent was collected from the surface of the final lagoon beside the intake for the outfall pipe. This location was accessed using the cat-walk shown in Figure 1.2 (b). At some times, particularly after a rainy night, foam developed on the surface of the lagoon. This was always avoided in taking an effluent sample from the surface. A bucket was used to collect the effluent from the surface of the pond, and then was quickly emptied into a 20 L plastic container for ease of transport. The pouring was performed rapidly to ensure minimal settling of the solids in the effluent.  38  Samples of effluent for experiments at UBC were obtained from either Western Pulp Ltd at Squamish or Northwood Pulp and Timber at Prince George. Differences must be present between the effluents as although both mills are kraft mills, the mill at Squamish utilises a pure oxygen activated sludge process for effluent treatment (Werker, 1996, pers. comm.) and the Northwood mill uses aerated stabilisation basins (Derksen, 1995, pers. comm.). Wood furnish at each mill would also be different. The exact sampling protocols followed at each mill were not clear, although it was requested that samples be collected in the usual way as per normal procedures at that mill. At Northwood, this entailed the use of a pump which sampled at depth in the final lagoon. Samples collected this way showed a higher TSS than those collected by the author. Precautions were taken to ensure that samples shipped from Prince George in the winter did not freeze during transport.  4.2.2 River Water The sampling of river water was also a simple procedure. A bucket was used to scoop river water from the surface and slightly below, at a point as far from the bank as possible. This was then emptied into a 20 L plastic container. Solids concentrations were higher in the river than in the effluent pond and so performing this at speed, to ensure no settling of the solids occurred, was even more important.  4.2.3 Transportation and Storage At Prince George, except when experiments were completed on site, the containers were transported back to UNBC. The samples were not refrigerated during their transportation, or once in the laboratory at UNBC. It was thought that this was unnecessary since new effluent samples were typically collected daily. At the time of the experiments, the temperature of the river was approximately 13 °C, and the temperature of the effluent was 33 °C according to analysis of the measurements described in §4.1. One of the aims of these experiments was to maintain conditions as close to those in situ as possible, so it was not thought to be unreasonable that the two fluids be stored at room temperature for the short periods of time involved.  When effluent was shipped to UBC from either the Squamish or Northwood pulp mills a different procedure was adopted. It was not uncommon for effluent samples to require storage for a couple of weeks before they were used in experiments, so samples were stored at 4 °C. 39  4.3  CRITICAL COAGULATION CONCENTRATION  The first experiment completed was intended to obtain a measure of the critical coagulation concentration. The method used in performing these experiments is described in van Olphen (1977). This is an experiment commonly completed to determine the concentration of electrolyte that is capable of causing coagulation of colloidal sols. It is assumed that reduction in the double layer thickness through increased electrolyte concentration is responsible for coagulation. The time of standing and the solids concentration of the sol are arbitrary. A number of standing times were investigated, however the experiment was not completed for different sediment concentrations.  This experiment was conducted in the field at the actual site of the effluent outfall. This allowed for numerous attempts to be made, and easy access to river water and effluent.  The tubes used were obtained from the  environmental engineering laboratory, and had a marked volume of 100 mL. River water was first added to the tubes in volumes of between 10 and 90 mL in increments of 10 mL. Since this was a completely qualitative test, and results were to be based solely on observation, the tube rack was placed in a location where the best visual contrast could be gained between the background and the fluid in the tube. After the specified volume of river water was added to the tubes, they were filled to the 100 mL mark with effluent. The temperature difference between the effluent and the river water resulted in poor mixing between the two fluids and the effluent would simply sit on top of the river water in the tube. To facilitate good mixing between the two fluids, each tube was inverted a couple of times. Once placed in the rack, the tubes were then observed for a period of thirty minutes. Photographs were taken at various times during the test, and any differences in appearance noted.  4.4 S E T T L I N G T E S T S M E A S U R I N G T U R B I D I T Y Settling tests measuring turbidity were completed to investigate the effect of the addition of pulp mill effluent on the settling behaviour of suspended river sediments. For a given effluent and river water sample, this was achieved by comparing the settling from one litre of river water mixed with one litre of distilled water, to the settling when one litre of effluent was added to one litre of river water.  40  4.4.1 Experimental Procedures and Apparatus The procedure followed in the completion of one of these settling tests began with the measurement of the total suspended solids (TSS) and the turbidity of both the effluent and the river water. The TSS measurements were made via the method described in Eaton et al. (1995), and turbidity measurements were made using a HACH Model 2100A Turbidimeter. This was considered to be an adequate characterisation of the two fluids for the purposes of this research.  The next step was to measure the volumes of sample or distilled water applicable to that settling test. In doing this, and sampling from the 20 L containers, care was always necessary to ensure that solids were always suspended. The measured volume of river water was then added to a 2 L beaker, and a standard jar test mixer was used to stir the beaker. Figure 4.1 shows the approximate dimensions of the jar and mixer arrangement. The jar test mixer was operated at a rotational speed of 100 rpm, and this was sufficient to ensure the suspension of all solids, even when the river water was the most sediment laden. As soon as possible, the effluent or distilled water was added to the river water, and mixed for 1 minute at 100 rpm. This level of agitation is equivalent to a mean velocity gradient of approximately 75 /s, according to Hudson (1981). After this one minute mixing period, the mixers were turned off, and the settling of the suspended material was allowed to commence.  While settling proceeded, samples were withdrawn at intervals specified in the method of Hudson (1981), as well as at some later times since settling was allowed to continue for longer periods. Turbidity was measured and recorded. Sampling from the beaker was facilitated with the use of a small flexible tube fixed to the beaker. A bend was placed in the end of the tube to ensure that the sample was removed from a location away from the wall of the beaker. The depth of the sample port opening below the fluid surface was also recorded for calculation of settling velocities. Flow through the tube was ensured by the action of gravity alone thus minimising potential for aggregate break-up which could occur if sampling was more vigorous. The initial distance from the water surface to the sampling port was also measured, although it was typically more than the 10 cm recommended by Hudson (1981).  41  c  L.  20.1 cm 16.3 cm  V  V  6.5 cm  Fisure 4.1. Dimensions ofjar and paddle apparatus as per standards (after Hudson, 1981).  4.4.2 Conversion of Settling Test Data to Settling Velocity Distributions The turbidity data collected in these experiments could be converted into settling velocity distributions using the method recommended by Hudson (1981). The method described is as follows: The settled water turbidity values are measured on samples drawn from a fixed depth of not less than 10 cm below the liquid surface in 2 L beakers. If the fixed depth is equal to 10 cm, then the times of sampling after halting stirring should be 1,2, 4, 8, and 16 min. Samples withdrawn at these times represent settling velocities of 10, 5, 2.5, 1.25, and 0.625 cm/min, respectively. The turbidity of the settled water samples withdrawn at these stated times, divided by the raw-water turbidity represents the ratio of raw water turbidity remaining, which is then expressed as a percentage. These percentages may then be plotted directly against the corresponding settling velocities to produce a settling velocity distribution curve for each jar of the jar-test series. 42  The settling of particles encountered in the Fraser River was slow in comparison to Hudson's tests which could be completed in less than half an hour. Many of the settling tests completed in this research were continued for periods greater than 24 hours.  4.4.3 Mixtures of Effluent and River Water Considered Settling velocity data of this kind were obtained using river water and effluents collected on different days. The  Table 4.1. Summary of settling test information.  Test  Date  Vol. River water  Vol. Effluent  Vol. Distilled  Commenced  (mL)  (mL)  water (mL)  A  26-M-95  1000  B  26-M-95  1000  C  28-M-95  1000  D  28-Jul-95  1000  E  29-Jul-95  2000  F  29-Jul-95  G  30-Jul-95  1000  H  30-M-95  1000  1000  I  30-Jul-95  1200  800  J  30-Jul-95  800  1200  K  31-Jul-95  1000  L  31-Jul-95  1000  1000  M  31-M-95  1200  800  N  31-M-95  800  1200  1000 1000 1000 1000  2000 1000  1000  turbidity and total suspended solids of the river water and effluent differed from day to day. Table 4.1 contains the information regarding the details of the settling tests completed in this format. 43  The same batches of effluent and river water were used for the settling experiments of tests A and B; C, D, E and F; G, H, I, and J; and K, L, M and N. This allowed for direct comparison of the results within these groups of tests. In one case, settling tests were completed using two litres of river water only, and also for two litres of effluent only. Tests were also completed for some mixtures of effluent and river water other than 1 to 1.  4.5 I N S T A N T A N E O U S T U R B I D I T Y  MEASUREMENTS  After completion of some of the settling tests, it seemed that much of the turbidity changes took place immediately after mixing. Thus, it was decided that shorter duration tests could be completed to focus on the initial (i.e. before any settling occurred) turbidity behaviour of a mixture. As described in §2.5.4, the measurement of turbidity as a method for inferring the presence of flocculation is not foolproof, however in this instance, it was thought to be suitable for the indication of changes in aggregate size distribution.  In these simple experiments, quantities of river water and effluent were mixed with one another for one minute, and then the turbidity of the mixture measured directly afterwards. Again, samples for measurement were withdrawn by siphoning through a small tube, although in this instance, the location of the sampling port was not important as long as it was representative of the sample in the beaker. Mixing was again carried out using the jar test stirrer, and stirring was maintained for a period of 1 minute at 100 rpm. In these experiments, a 1 L volume was used in a one litre beaker. The velocity gradient due to the mixing in this situation was then 105 /s. The volume combinations for which these experiments were completed are presented in Table 4.2. For one of the series of mixing tests, a sample was also withdrawn after mixing for measurement of TSS.  The turbidity of each of the components was measured prior to mixing. Using this information, it was possible to compare the measured turbidity after mixing to the predicted turbidity of the mixture. The predicted turbidity was estimated by consideration of the initial turbidities of the mixture components and the volumes in which they are combined, assuming additivity.  44  Table 4.2. Volume combinations considered for instantaneous turbidity reduction experiments. Volume of river water (mL)  Volume of Effluent (mL)  1000  0  900  100  800  200  750  300  666  333  600  400  500  500  400  600  333  666  300  700  200  800  100  900  0  1000  4.6 SETTLING TESTS MEASURING TSS Settling tests were completed using TSS as an indication of material remaining in suspension. Measurements of TSS were considered preferable to turbidity because of the difficulties involved in interpreting turbidity measurements with respect to aggregate size distributions. Ultimately, TSS measurements would also be more useful for comparison with field observations where TSS measurements had been made.  4.6.1 Two Litre Jars Some jar settling tests were conducted at UBC, using effluent and river water collected towards the end of the experimental period in Prince George. These tests were completed in almost identical fashion to the settling tests measuring turbidity, except that TSS measurements were made on the samples withdrawn from the jar. The same procedure could be used to determine the settling velocity distributions produced in these experiments. In the  45  settling velocity distributions produced, calculated values of "percent turbidity remaining" in the experiments where turbidity was measured must be substituted with "percent solids remaining".  4.6.2 Settling Column Tests similar in principle were also carried out in a conventional settling column. The intent of completing tests in this apparatus as opposed to a 2 L beaker was that it allows for larger volumes of reactants to be utilised, and more sensitive results to be obtained due to larger settling distances. TSS measurements can also be made more accurately because larger volumes can be withdrawn without greatly affecting the volume in the settling column.  4.6.2.1 Description of Apparatus Some description of the test apparatus is necessary before discussion of the experimental procedure utilised. The settling column used was comprised of three sections. The top two sections were 960 mm long, and the bottom section was 660 mm long. This gave the column an overall height of approximately 2.6 m when assembled. The internal diameter of the column was 139 mm, giving a total volume of 39.4 L. Since effluent was to be used at very low dilutions in these experiments, it would have been possible that 20 L of effluent would be used in a single settling test. Effluent used had to be shipped from pulp mills within British Columbia, and hence it was thought that use of volumes this large would be difficult. It was decided to use only the bottom two sections of the column, and that the total fluid volume at the start of a settling test would be approximately 20 L. This resulted in an overall water column depth of 1.3 m in the column, a vast improvement over the 12 cm available with the 2 L jars. This was considered to be adequate for the purposes of this experiment. The settling column also had ports evenly spaced along its length which could be used for sampling and the connection to pumps and air supply. The overall column set up as used in experiments is depicted in Figure 4.2.  The lowest port was used for connection to a compressed air supply. An air supply was necessary for the provision of vigorous mixing in the column prior to the settling period. The air bubbles provided excellent mixing and also created adequate turbulence ensuring that all of the solid material in the column was suspended. Suspension of the solids in the fluid samples was necessary to simulate the conditions in the River, where turbulent eddies are responsible for the suspension of fine sediments. If the solid material was not kept in suspension, there would be no 46  13.9 cm I.D. I<  -^1  20 cm  20 cm  20 cm  20 cm  24 cm Sampling Port 20 cm  20 cm 8.5 cm  figure 4.2. Diagram of settling column set up and dimensions.  Effluent or .Distilled Water Air supply for mixing  opportunity for flocculation to occur. Compressed air from a laboratory supply was regulated to a pressure of 138 kPa, and was maintained at a constant flow of approximately 835 mL/min while used for mixing throughout the experiments. The average velocity gradient that this imparts on the flow is difficult to calculate exactly (see Appendix B), yet may be of the order of 1000 /s. Although bubbles could be thought to add another complication to the flocculation problem, and hence could not be considered the most ideal mixing tool, the ease with which they could be introduced to the column led them to be a favoured tool over any type of mechanical mixer.  For different experiments, the air flow outlet needed some variation. With very fine grained sediments it was sufficient to allow the bubbles to simply enter at the side of the column where the bottom sampling port ended. This resulted in a very turbulent flow in the column. On the side of the column opposite the air inlet port, the flow direction tended to be down in response to the rapid updrafts caused by the large bubbles which would travel up the wall before breaking up. When larger grained sediments were introduced to the column, this flow characteristic made it difficult to keep sediments in suspension. An alteration was made to the air outlet to ensure a more uniform upward bubble flow through the column. This alteration took the form of a thin tube bent towards the bottom with three holes cut in it. Simple experiments indicated that this mechanism was indeed successful in keeping the sediments with higher settling velocities suspended.  The second port in the column was used to simulate the effluent diffuser in a very rudimentary fashion. It was connected to a peristaltic pump which was used to pump effluent or distilled water into the column. The third port was then used for drawing sample from the column. In early settling tests, a port even higher up was also used for drawing samples, but the use of this was abandoned in later tests.  4.6.2.2 Experimental Procedure The procedure by which a typical experiment was run is set out below. •  Compressed air flow turned on.  •  Column filled with distilled water to about 7 L, and left to bubble.  •  Desired quantity of illite was weighed out.  •  Illite was added to distilled water in a 2 L beaker, and agitated by a magnetic stir bar.  48  •  Mite was allowed to soak, while some of the particle agglomerations dispersed.  •  The 2 L beaker was immersed in a sonicating bath for at least 5 min, while all sediment was suspended by stirring with a spatula. This ensured all of the agglomerations were dispersed so that in each settling test, the same particle size distribution was present initially and comparisons between tests would be valid.  •  The illite suspension was poured into the top of the column, ensuring maximum amount of sediment entered the column. A wash bottle was used to remove any sediment that had settled in the beaker as it was transferred from the sonicator to the column.  •  Distilled water was added to the column so that the volume of the sediment suspension reached 10 L. The true volume of the suspension was actually somewhat less since the flow of bubbles raised the water level. The disturbance of the water level due to the bubbles also complicated accurate volume measurements.  •  A sample was collected in a 250 mL plastic container from the lower sampling port, and the time clock for the overall experiment was started.  •  At t = 1.5 min, pumping of the effluent into the column was commenced.  •  Effluent was then pumped into the column until the aerated volume of the mixture reached 20 L. This took between 6 and 7 minutes to complete, and the time when pumping ceased was recorded.  •  The column was left to stand with the air flow on until t = 11 min, when another sample was collected from the lower port.  •  At t = 13 min, the air flow to the column was stopped, and settling was allowed to commence.  •  Samples were then taken at various time intervals, and the distance of the water level above the sampling port measured in each instance. A volume of between 200 and 250 mL was taken for each sample, and the sampling tube was flushed prior to each sample being taken.  Usually, the last sample would be taken the next day, and after this, the samples would be analysed for TSS. Where a control experiment was being completed, distilled water was treated in the same way as the effluent had been. Similarly in experiments completed with Fraser River sediments in place of illite, this was handled the same way as the handling of illite is outlined above.  49  4.6.2.3 Experiments with Illite Suspensions Settling tests were carried out with a variety of different controls and fluids. The first series of experiments was run using an effluent from Western Pulp and illite, a primary component of the fine fraction of the sediments in the Fraser River (Prahacs, 1994). The illite was obtained from the Soils Laboratory in the Civil Engineering Department at UBC. Experiment 2 was completed with the combinations of effluent, distilled water and illite shown in Table 4.3. The same combinations were used in Experiment 1 except that the final test, E, with a nominal initial illite concentration of 300 mg/L, was not completed. All other tests involving illite were completed with a nominal initial illite concentration of 200 mg/L. Experiment 1 was completed using an effluent collected in December 1995, and Experiment 2 was completed using effluent collected in January 1996. The repetition was considered important to ensure the results of Experiment 1 were repeatable.  Table 4.3. Combinations of effluent, distilled water and illite used in Experiment 2. Settling Test  Effluent Used  Distilled Water  Illite  A  -  half  half @ 200 mg/L  B  half - with all biosolids  -  half @ 200 mg/L  C  half - with all biosolids  half  -  D  half with reduced biosolids  -  half @ 200 mg/L  E  -  half  half @ 300 mg/L  For each case in which a test with distilled water was performed, it was treated as the control for the purposes of analysis. By subjecting it to exactly the same treatment (experimentally) as that to which the effluent was subjected when it was added to the column, it is possible to account for all of the shear and flocculation effects that may be present as a result of the incoming fluid jet. The timing of each experiment was also kept the same to ensure that direct comparison between results would be as realistic as possible.  4.6.2.4 Experiments with Fraser River Sediment Suspensions Effluent used in experiments with Fraser River sediment was obtained from the Northwood mill in Prince George. Sediment used was originally collected for a series of contaminant adsorption experiments completed at UBC 50  (Gomm, 1994). The sediment was collected using a sediment trap consisting of four plexiglass tubes, 30 cm long, and 4 cm in diameter (L. Gomm, 1996, pers. comm.). The tubes were fixed in a plastic holder, and placed in the river near Shelley (upstream of Prince George) at approximately mid-depth enabling collection of sediment over a 5 day period from July 1 to 6, 1992. Sediment collected had been frozen before use in this work. Primary particle size analysis completed on suspended sediments collected by the author in July 1995 had shown that the majority of particles were of diameters smaller than 60 um. It seemed reasonable then, that the sediment collected by Gomm should be sieved with a 63 um sieve to give the most realistic representation of the in situ suspended sediments. This was completed using both wet and dry sieving where appropriate.  Table 4.4 summarises the combinations of effluent, distilled water and sediment used in these experiments. The first set of three tests (Experiment 3) were completed at room temperature (23 °C), and the second set of tests (Experiment 4) were completed in one of the cold rooms (4 °C) in the environmental engineering laboratory at UBC. In addition to measurements of TSS, measurements of volatile suspended solids (VSS) were also made in these experiments.  Table 4.4. Combinations of effluent, distilled water, and Fraser River sediments used in Experiments 3 and 4. Settling Test  Effluent used  Distilled Water  River sediments  A  half  -  half @ 200 mg/L  B  -  half  half @ 200 mg/L  C  half  half  -  51  5.0 RESULTS  OF  EXPERIMENTATION  5.1 CRITICAL COAGULATION CONCENTRATION 5.1.1 Ionic Species in the Effluent The concentration of cationic species in the effluent could be significant. If the colloids are negatively charged as is expected for hydrophobic clay sols (van Olphen, 1977), then positively charged particles in the effluent could be responsible for the compression of the electrical double layer for suspended clay particles. Average concentrations for some of the more predominant ionic species in the effluent are presented in Table 5.1. The valency of ions is also important in relation to their relative effect in compressing the electrical double layers of suspended colloids as was evident in Equation 2.1. The higher the valency of an ion, then the larger the effect it can have on compressing the double layer. In view of this, Table 5.1 also contains indications of the relative importance (valence factor) of each of the ionic species as described by Benefield et al (1982). It is then possible to determine the effective concentration of the solution in terms of the concentration of Na ions (or monovalent ions). To do this, the factors +  for relative importance due to valency as well as a correction factor for the molecular weight of each ion must be considered. The effective concentrations tabulated in Table 5.1 have been obtained by taking the product of concentration, valence factor and molecular weight correction factor.  Simple summation reveals that the  concentration of ions in the effluent is similar to a solution of 12,950 mg/L of Na . +  Table 5.1. Average concentration of some predominant ions in Northwood effluent (Evans & Hall, 1996) and calculated effective Na concentrations. +  Ionic  Mean Concentration  Valence  Species  (mg/L)  Factor  Na  +  490  1  2+  195  Ca Al  3 +  Fe  3+  M.W. ratio  Effective Na cone. +  (mg/L) 23/23  490  64 (2 )  23/40  7176  6.79  729 (3 )  23/27  3077  7.37  729 (3 )  23/56  2207  6  6  6  The study by Burban et al. (1989) showed that particles flocculate faster in sea water than fresh water, though once steady state is reached, floes formed in sea water are actually smaller. By comparing the ionic strengths of the 52  effluent and sea water, an indication can be obtained of the likelihood that effluent could have a similar flocculating effect. A calculation of the effective ionic strength is presented for sea water in Table 5.2 using the compositional data of Wilson (1975) as cited by Chester (1990). Addition of the effective Na concentrations yields an effective +  Na concentration of around 108,000 mg/L. This is an order of magnitude higher than the effective Na +  +  concentration in the effluent. Hence, the effluent may cause a slight electrolytic flocculation effect, but it would not expected to be as strong as that caused by sea water.  Table 5.2. Average concentration of some predominant ions in sea water ((Wilson, 1975) as cited by Chester (1990)) and calculated effective Na concentrations. +  Ionic  Mean Concentration  Valence  Species  (mg/L)  Factor  Na  +  11,000  1  Mg  2+  1,320  Ca  2+  K Sr  +  2+  M.W. ratio  Effective Na cone. +  (mg/L) 23/23  11,000  64 (2 )  23/24  81,000  422  64 (2 )  23/40  15,500  409  1  23/39  240  8.1  64 (2 )  23/88  140  6  s  6  5.1.2 Experiments The experiments completed for assessing critical coagulation concentration involved the addition of effluent at various dilutions to the river water. This was intended to indicate whether electrolytic flocculation/coagulation of the sol would occur.  There was no evidence of coagulation at all in these experiments. Settling of the solids in the river water was the only observable effect regardless of the time frame. Better results may have been achieved if the river water had contained a higher suspended solids concentration, but it is likely that even then, the relative proportion of colloids in the suspension may not have been high enough to notice a difference. Benefield et al. (1982) noted that the concentration of colloids would not affect the critical coagulation concentration, so this may not have been the  53  explanation for the lack of an observable effect.  From these experiments then, we can conclude that  flocculation/coagulation of the sediments was not enhanced by the addition of effluent.  Despite this negative resuk, further experimentation was justified for two reasons.  Firstly, the result is only  indicative that coagulation due to increased electrolytic concentrations has not been evident. Other flocculation mechanisms requiring different sample treatment may be responsible for the field observations discussed in §1.1. Secondly, the test is only qualitative, and if effects were only marginal, they would not be observed without measurement of some quantitative parameters.  5.2 SETTLING TESTS MEASURING TURBIDITY Settling tests were completed in 2 L jars where turbidity was measured as a function of time. River water and effluent were mixed, and settling was observed. The same was completed for mixtures of river water and distilled water to provide control cases. The results from all the settling tests completed at UNBC for which turbidity was measured are included in Appendix C, as Figures C - 1 to C - 5. A summary of the tests completed is provided in Table 5.3. Using the measured depth of sampling location contained in Table 5.3, this information has also been converted into plots of "percent turbidity remaining" as a function of settling velocity distribution, which are also contained in Appendix C as Figures C - 6 to C - 10.  54  Table 5.3. Summary of the settling tests completed, initial sample turbidities (To), TSS (TSS ) and the sampling 0  point depth for each settling test.  river water Test  effluent  distilled  Date started  volume (mL)  to  TSS  0  (NTU) (mg/L)  volume  To  TSS  (mL)  (NTU)  (mg/L)  0  water  Sampling  volume  depth  (mL)  (cm)  1000  12.0  A  26-Jul-95  1000  27  58  B  26-Jul-95  1000  27  58  C  28-Jul-95  1000  52  141  D  28-Jul-95  1000  52  141  E  29-Jul-95  2000  52  141  F  29-Jul-95  G  30-Jul-95  1000  95  286  H  30-Jul-95  1000  95  286  1000  5.2  47.8  12.5  I  30-Jul-95  1200  95  286  800  5.2  47.8  13.0  J  30-Jul-95  800  95  286  1200  5.2  47.8  13.5  K  31-Jul-95  1000  62  156  L  31-Jul-95  1000  62  156  1000  5.0  40.2  12.5  M  31-Jul-95  1200  62  156  800  5.0  40.2  13.0  N  31-Jul-95  800  62  156  1200  5.0  40.2  12.5  . 1000  6.5  39.7  11.0 1000  1000  5.5  40  13.5 14.0 13.0  2000  5.5  40  13.5 1000  1000  12.5  12.5  5.2.1 Turbidity Variation as a Function of Time Since turbidity reduction was most noticeable for high TSS, it suggests some dependence of the turbidity reduction effects on the concentration of suspended solids in the river water. The effect of pulp mill effluent addition can best be seen by considering the results of tests G and H which used river water with the highest TSS of 286 mg/L. Figure  55  5.1 then, is a plot of turbidity as a function of time for tests G and H, and shows the effect that addition of pulp mill effluent had on turbidity.  50.0  40.0  fc; 30.0  '•5  '•S 20.0 * *  •o  10.0  0.0 100  10  1000  10000  Time (min) -•—Water dilution - control (G) - - o - - Effluent dilution (H)  Fieure 5.1. Removal of turbidity comparison between a control case and an effluent and river water mixture for a river water sample with TSS = 286 mg/L.  There are a number of observations that can be made from this result. It appears that there was an instantaneous reduction in the turbidity of the suspension so that at t = 1 min (immediately after mixing), although the turbidity in the control was 47.5 NTU, the turbidity in the river water and effluent mixture was 38 NTU. Since the effluent added to the river water had a small amount of measured turbidity, it was expected that a turbidity higher than that for the control would have been recorded. After this initial reduction in turbidity, it does not appear that there was any significant difference in the rate of removal in turbidity as the slopes of both plots are identical during the portion of the test where the bulk of turbidity removal occurred. The turbidity in the river water and effluent mixture thus remained less than that of the control throughout the test.  56  The reduction in turbidity was thought to be indicative of a change in aggregate size distributions due to aggregation. An analysis of the derived settling velocity distributions was then completed to determine if this aggregation had resulted in changes in settling velocity distributions. It was expected that an increase in the proportion of turbidity with a larger settling velocity would result.  5.2.2 Settling Velocity Distributions On the whole, it would appear that there was little difference between the settling velocity distributions for controls, and the cases where effluent was added. Figure C - 6 has been reproduced as Figure 5.2. The effect of the addition of effluent in this instance was an increase in the proportion of turbidity with a lower settling velocity. This effect is  itftfre-  .o  o  c  - '  ra  E  1  10.00  100.00  oU.U on n  ±s fin n OU.U  <u  !5  p  ^^^$f Cr  An n HU.U  0)  u  °n n  0.02 cm/minO.O cm  tU.U  7  \/ 0.01  /min  \/  Or00.10  1.00  Settling velocity (cm/min) -•—Water Dilution - control (A) - -o - Effluent Dilution (B)  Figure 5.2. Settling velocity distribution comparison for control and river water and pulp mill effluent mixture.  most noticeable for settling velocities between 0.1 cm/min and 1.0 cm/min where the difference between the percent turbidity remaining for a given settling velocity was the largest. At lower settling velocities, it appears as though the difference was reduced. From Figure 5.2, it is possible to make an estimate of the median settling velocity of the turbidity, although since only one data point was recorded beyond the 50th percentile in each settling test, the accuracy of this approximation may not be satisfactory. The median settling velocity of the turbidity in the river water only, was around 0.07 cm/min, and the median settling velocity of the turbidity in the river water and effluent 57  mixture was approximately 0.02 cm/min. These settling velocities are indeed low, as under quiescent conditions at a settling velocity of 0.07 cm/min, it would take a particle almost 1 day to settle through a i m water column.  In the other cases, the differences between the distributions were insignificant. It would seem that on the whole, in the river water and effluent suspensions, there is a larger proportion of turbidity with a lower settling velocity than for the controls. This trend is consistent with the results shown in Figure 5.2, and suggests that there is little evidence that the settleability of the turbidity in the river water was affected by the addition of a pulp mill effluent. Considering this result and that of §5.2.1, it seems reasonable that pulp mill effluent addition may have resulted in aggregation of particles but has no impact on the settling dynamics of the turbidity.  5.3 I N S T A N T A N E O U S T U R B I D I T Y M E A S U R E M E N T S The settling test results discussed in §5.2 suggested that for river water with a high enough TSS, the addition of a pulp mill effluent will result in an instantaneous and longer term reduction in turbidity. Instantaneous was taken to mean the length of time required for mixing and collection of the sample, and was slightly more than one minute. The instantaneous reduction in turbidity has been investigated in more detail in the experiments described in §4.5 to identify the roles that dilution and TSS of the river water played in turbidity reduction. The full results of these experiments are presented in Appendix D.  5.3.1 Analysis Techniques and Sample Calculations The measured turbidities obtained immediately after mixing of the various volumes of river water and effluent were compared to predicted turbidities. Predicted turbidities were calculated using a mass balance technique based on the initial turbidities of the river water and the effluent, and the volumes in which the two samples were combined, i.e.  Predicted Turbidity =  T  ^  X  v  ^  +  T  ^  X  l  W  ( 5 1 )  1000  where x  River  VEffluent  and x fflUem represent the respective turbidities of the river water and effluent, and similarly v E  River  and  represent the volume of river water and effluent being mixed. Additivity of turbidity has been assumed in  making this calculation. A relative percentage reduction in turbidity was then calculated for each result using the following relationship. 58  . . .... , . Predicted turbidity - Measured turbidity „ Relative % turbidity reduction = - x 100% (5.2) Predicted turbidity These turbidity reduction results are also presented in Appendix D. Similar experiments were also completed where distilled water was added to the river water to provide control cases, and these results are also tabulated in Appendix D. Table 5.4 presents the results for the most complete set of experiments undertaken with one set of effluent and  Table 5.4. Results of turbidity measurements, predicted turbidities and calculated percent reductions in turbidity for river water with Turbidity = 69 NTU, and effluent with Turbidity = 5.0 NTU. Effluent  River water  Measured  Predicted  Percent  Volume  Volume  Turbidity  Turbidity  turbidity  (mL)  (mL)  (NTU)  (NTU)  reduction  1000  0  5.0  5.0  0.0  900  100  12.0  11.4  -5.3  800  200  15.5  17.8  12.9  750  250  18.0  21.0  14.3  666  334  22.0  26.4  16.7  600  400  24.5  30.6  19.9  500  500  30.0  37.0  18.9  400  600  35.0  43.4  19.4  333  667  38.5  47.7  19.3  250  750  45.0  53.0  15.1  200  800  50.0  56.2  11.0  100  900  59.0  62.6  5.8  0  1000  69.0  69.0  0.0  river water samples. The average percentage reduction in turbidity due to the addition of all volumes of effluent to river water for experiments with these samples was 12.3 %. It also appears that the reduction was larger for the mixtures in which the two samples were combined in similar proportions.  59  For the set of experiments presented in Table 5.4, measurements of TSS were also made. This allowed for determination of whether a change in TSS accompanied the change in turbidity. The TSS measurements and predictions are presented in Table 5.5. Predicted values of TSS and calculated percent reductions were obtained in the same way as turbidity parameters were. If a reduction in TSS was noted, this would indicate that settling had occurred in the time required to sample after mixing was completed.  Table 5.5. Results of TSS measurements, predicted TSS values and calculated percent reductions in TSS for river water with TSS = 153 mg/L, and effluent with TSS = 37 mg/L. Effluent  River water  Measured  Predicted  Percent  Volume  Volume  TSS  TSS  TSS  (mL)  (mL)  (mg/L)  (mg/L)  reduction  1000  0  37.9  37.0  -2.4  900  100  51.5  48.6  -6.0  800  200  64.5  60.2  -7.1  750  250  64.9  66.0  1.7  666  334  78.3  75.6  -3.6  600  400  86.2  83.4  -3.4  500  500  95.9  95.0  -0.9  400  600  106.4  106.6  0.2  333  667  112.4  114.2  1.6  250  750  118.3  124.0  4.6  200  800  133.3  129.8  -2.7  100  900  144.3  141.4  -2.1  0  1000  151.4  153.0  1.0  The average reduction in TSS for the same mixing experiments was -1.6 %. This actually represents some apparent increase in the TSS of the mixtures, which was most likely due to experimental error. For TSS measurements made at 60 mg/L, the coefficient of variation for analytical precision was found to be 3.5 %, so it is likely that the apparent 1.6 % average TSS increase would be due to analytical variance. We can then conclude safely that no reduction in TSS occurred as would be expected if mass was conserved and no settling occurred in the time required for sampling. Despite this result, a consistent reduction in turbidity was apparent. 60  5.3.2 Dependence of Turbidity Reduction on Mixing Proportions A similar analysis of turbidity results for the majority of the data was performed to determine the dependence of turbidity reduction on mixing proportions or dilution. This has been evaluated by plotting the mean percentage reductions (calculated using Equation 5.2) as a function of the volume of effluent added to make 1 L of mixture. Figure 5.3 summarises this result for the data sets generated with TSS ver = 141, 153, 156 and 286 mg/L. The data Ri  from the fifth data set generated with TSS  River  = 58 mg/L have not been included as the results from this test were  30  0  200  400  600  800  1000  Volume of effluent in 1L mixture  Fisure 5.3. Plot of mean turbidity reduction as a function of dilution for data sets generated with TSS153, = 141, 156 and 286 mg/L. 61  thought to be less reliable than those obtained with a higher TSS  River  because of the smaller absolute changes in  turbidity that resulted. Inclusion of these results would produce quite a different relationship, with much larger error bars. Each point shown on this graph represents the average of up to four results available for that quantity of effluent added. The error bars indicate the standard deviation of these results.  A quadratic curve was fitted to these data. The fit appears to be quite reasonable giving an R of 0.85. This good 2  correlation has been achieved despite the fact that different concentrations of solids were present in each of the river water samples tested. The peak in the parabola for an effluent addition of around 500 mL (1:1 mixture) indicates that the largest percent reduction in turbidity occurred when equal volumes of effluent and river water were mixed. This result suggests that in the actual river, it will be the near field mixing zone that is most important in determining what flocculation occurs, since dilutions of the effluent will be lowest in this region. The result is also useful because it suggested that a 1 to 1 mixture should be utilised in future experimentation to ensure easiest possible identification of flocculation effects.  5.3.3 Dependence of Turbidity Reduction on TSSRiver The effect that TSS in the river water has on the reduction in turbidity was determined by investigation of the average turbidity reductions for each data set. Table 5.6 tabulates the average percentage turbidity reduction and TSSRiver-  Table 5.6. Average percent turbidity reduction for effluent and river water mixtures and all data sets. TSS  Mwr  (mg/L)  Percent turbidity reduction (%) Range  Mean for data set  58  -1.0 to 47.6  20.9  141  -8.1 to 20.0  10.1  153  -5.3 to 19.9  12.3  156  2.4 to 23.3  13.8  286  11.8 to 28.3  18.2  62  Apart from the result for TSS i er = 58 mg/L (considered to be least accurate) which is dominated by some large R V  percentage reductions, a trend is indicated in the data. As the TSS for the river water increases, the average percent turbidity reduction appears to increase. The TSS of the effluent collected on different days was reasonably constant, varying only marginally about 40 mg/L. For this reason, it can be assumed that the majority of variation in mean percent turbidity reduction shown in Table 5.6 would be due to the changes in TSSRi e . It is also worth looking at V  r  the same information obtained for the river water and distilled water mixtures (control cases) and these data are presented in Table 5.7.  Table 5.7. Average percent turbidity reduction for river water and distilled water data sets. TSSRi (mg/L) ver  Percent turbidity reduction (%) Range  Mean for data set  58  -8.8-8.4  -1.6  141  -13.3-4.0  -4.2  153  -13.8-2.2  -5.2  156  no results  no results  286  -9.5 - 1.4  -3.1  It would appear that on the whole there was a slight increase in the turbidity of the controls, as indicated by the negative calculated reductions in Table 5.7. This may be a result of the shears applied during mixing and dilution changing aggregate size distribution, or could simply be a result of the sampling technique. There does not appear to be any dependence of the magnitude of the reduction on the concentration of suspended solids in the river water in this instance. Regardless of the reason for the trend, it suggests that the control-corrected changes calculated for the river water and effluent mixtures would be slightly greater than the reductions reported in Table 5.6.  5.3.4 Theoretical Discussion of Turbidity Reduction The reduction in turbidity that occurred without any associated fall in the level of TSS suggests that some change has occurred in the particle size distributions. It could be assumed that a reduction in turbidity will result as smaller particles agglomerate, and it becomes easier for light to pass through the suspension. At the same time however, as 63  mentioned in §2.6.4, agglomeration may not be the only factor which affects the resulting turbidity. The turbidity of a monodispersed suspension containing spherical particles whose diameters are small compared to the wavelength of the incident light can be described as ((Kerker, 1969) as cited by Holman (1986)), 24KV (n -n ) ^ = ^ 7 ^ - 1 2  2  2  0  T r  (5-3)  N  where particle volume is represented by v , X represents the wavelength of incident light, n and n are the relative p  n  0  x  refractive indices of the medium and the particle respectively, and N is the total number of particles per unit volume T  of suspension. From this formula then, turbidity must vary with the square of aggregate volume, and with the number concentration of particles.  Thus, as flocculation proceeds, although the number concentration of the  particles in suspension decreases, the volume of the aggregates resulting must be increased by at least the same factor. If the aggregates formed are at all porous, then the volume of aggregates would increase by a factor greater than the factor by which number concentration is reduced. The effect of increasing particle volume will be dominant since turbidity is related to the square of this parameter. Hence, aggregation should result in increased turbidity.  Two important assumptions implicit in this formulation are not applicable to our situation. First, this formulation was derived for a monodispersed, suspension, i.e. a suspension where all particles have the same size. In the river water and effluent, this assumption would not be reasonable as discussed in §5.5. Furthermore, the particles in the effluent and river water were not spherical as evidenced in microscopic visualisation of the solids in both samples. The predictions of Equation 5.1 may not be directly applicable to our situation. In conclusion then, the turbidity reduction can be thought to represent some change in aggregate size distribution, but without actual distribution measurements, we cannot be sure if an increase in aggregate size is indicated or what the magnitude of this change might be.  5.4 SETTLING TESTS MEASURING TSS As discussed in §4.6, similar settling tests were completed for which TSS measurements were made instead of turbidity. This was intended to provide a more accurate measure of the solids in suspension.  64  5.4.1 Jar Scale Directly after field work and experiments were completed at Prince George, settling tests were performed in 2 L jars at UBC using effluent and river water collected at Prince George. Figure 5.4 (a) shows the comparison between the settling observed for a 1:1 mix of effluent and river water (TSSRj = 180 mg/L), and that observed for a 1:1 mix of ver  river water and distilled water. Due to the presence of suspended solids in the effluent, the concentration of the river water/effluent mix was higher initially, and remained that way for the 24 hour period of the test. Settling does not appear to have proceeded more rapidly in the presence of effluent. Replotting the data by normalising with respect to the initial concentration, (Figure 5.4 (b)) it can be seen that there was no discernible effect on the settling of the solids due to the addition of a pulp mill effluent. Increased settleability of solids was not indicated in the results from these settling tests which is a similar result to that presented in §5.2.2.  5.4.2 Settling Column Scale Settling tests in which TSS was measured were also completed in a settling column as described in §4.6.2.3. It was hoped that the use of the settling column instead of jars would provide a greater sensitivity to differences in settling behaviour. Table 5.8 summarises the experiments that were completed, and Appendix E contains all the results collected in these experiments. The difference between Experiments 3 and 4, was that Experiment 4 was conducted at a temperature of 4 °C, whereas Experiment 3 was conducted at a temperature of 23 °C. In Experiments 3 and 4, measurements were made of the volatile suspended solids (VSS) as well as TSS.  65  120.0 100.0 80.0 *  £ W  60.0  *  C  (J) H  a.. 40.0 20.0 0.0 0.00  * - « _  * . „  0.50  1.00  1.50  2.00  2.50  3.00  3.50  Log(time(min)) - • — Effluent dilution - - o - - Water dilution  (a)  100.0 * *  *n l_l  80.0  \ *  c  Ira  \  >  60.0  E  0)  W  40.0  20.0  • • . _ —, '»-,  0.0 0.00  •a 0.50  1.00  1.50  2.00  2.50  3.00  •  -a  3.50  Log(time(min)) - • — Effluent dilution - - o - - Water dilution (b)  Fisure 5.4. (a) Plot of the variation in TSS throughout the settling test, (b) Normalised plot of the variation in TS throughout the settling test.  66  Table 5.8. Summary of settling tests completed in settling column at UBC. Expt  1  Effluent A  B  C  D  2  3  Western Pulp - Squamish  4  Northwood - Prince George  • Illite (200 mg/L)  • Illite (200 mg/L)  • Sed. (200 mg/L)  • Sed. (200 mg/L)  • Effluent  • Effluent  • Effluent  • Effluent  • Illite (200 mg/L)  • Illite (200 mg/L)  • Sed. (200 mg/L)  • Sed. (200 mg/L)  • Distilled water  • Distilled water  • Distilled water  • Distilled water  • Effluent  • Effluent  • Effluent  • Effluent  • Distilled water  • Distilled water  • Distilled water  • Distilled water  • Illite (200 mg/L)  • Illite (300 mg/L)  • Clarified effluent  • Distilled water  E  • Illite (200 mg/L) • Clarified effluent  Note: Sed. - Fraser River sediments collected in a sediment trap by Leslie Gomm.  5.4.2.1 Experiment 1 The results of Experiment 1 indicated the occurrence of flocculation in a mixture of effluent and illite. Figure 5.5 (a) is a plot of the smoothed results for TSS in this experiment. The TSS data were smoothed to remove some of the variability using a moving average with a window width of three data points. The smoothed distributions can be easily differentiated using a three point Lagrangian interpolation (Hildebrand, 1987), and the results of these rate calculations are presented in Figure 5.5 (b). In examining these distributions, it must be remembered that the settling commenced at t = 13 min, and that the first point plotted at t = 11 min represents the TSS while all the solids were kept suspended artificially.  The most striking observation that can be made from these plots is that the addition of a pulp mill effluent enhanced the settling of the illite. At the beginning of the settling tests, the suspended solids concentration in the illite and  67  effluent mixture was 25 mg/L higher than that in the illite and distilled water control. Considering that the concentration of the control was 80 mg/L, this represents a percentage difference of 31 %. Before an hour of settling had elapsed, the TSS of the illite and effluent mixture decreased to the TSS present in the illite control settling test. After this time, it continued to decrease further with the difference between the two TSS values increasing till t = 120 min. Towards the end of the settling test (t = 6 hr), the TSS for the illite and effluent mixture was less than that of the illite control by approximately 30 %.  A similar experiment was completed using clarified effluent. Clarification was achieved by allowing the effluent to settle for at least 24 hours. The purpose of this was to investigate the effect that biosolids concentration had on enhancement of settleability. The addition of clarified effluent (TSS =15 mg/L) to the illite seems to have had a similar impact on the settling of the illite, but the magnitude of the effect was greatly reduced. The TSS of the clarified effluent and illite mixture was initially higher than that of the illite control, but at the end of the settling test, the TSS of the mixture with effluent was lower than that of the control by 20 %. The reduction of solids in comparison to the control was much lower than was evident when raw effluent was used. In light of this, it would seem reasonable that the enhanced settleability noted for the illite and effluent mixtures required that biosolids be present in the effluent.  The plot of TSS removal rates allows observations to be made regarding the rate of solids removal. From Figure 5.5 (b) it can be seen that the rate of settling was significantly faster for the illite and effluent mixture than for the control, or any other test completed in Experiment 1. This was especially noticeable in the first 100 minutes of the test. After this time, the rate of solids removal was similar to that displayed in the other settling tests. During the first 100 minutes, the second fastest solids removal rate was indicated for the mixture of illite and clarified effluent. It would appear that the addition of this effluent resulted in a slight increase in the suspended solids removal rate relative to the illite control, again reflecting the importance of biosolids in enhancing the settleability of solids.  Different initial sediment concentrations were present in most of the tests. Normalisation of the settling test results with respect to their respective initial concentration allows examination of these data in a different light. For the  68  ~I 1  120  —\ 1 1 1—I—I  r~  (a) J  20  _J I L_ 100  10  1000  I  10000  Time (min)  0.0 \'-^ J  -0.2 —  ;  n  6 \  \  \  (b)_  -  -0.4 —  \  oo  -  Illite (200 m g / L ) & Effluent  -a -0.6  -A  TO TO ET3  Illite (200 m g / L )  - \  Effluent  -0.8  Illite & Clarified Effluent  1.0 10  100  1000 Time (min)  10000  Fisure 5.5. (a) Plot of the smoothed variation in TSS as a function of time for Experiment 1. (b) Rate of solids removal for Experiment 1.  69  1.0  (a)_  0.8  0.6 h  0.4  0.2  0.0  10  100  —I 1 1 1 L_1000  10000  Time (min)  0.000  (b)  -0.002 h  Illite (200 m g / L ) & Effluent" Illite (200 m g / L ) Effluent Illite & Clarified Effluent -0.010 10  100  1000  10000  Time (min) Figure 5.6. (a) Normalised plot of the variation of TSS in the settling tests of Experiment 1. (b) Normalised plot the variation in TSS removal rates in the settling tests of Experiment 1. 70  calculations of removal rate in particular, we can expect that normalisation will enable better comparisons to be made. The reason for this is that we would expect absolute removal rates to be higher for a suspension of the same material with a higher initial concentration. Hence, to compare different tests with different initial TSS, examination of normalised data may be more reasonable. Figures 5.6 (a) and (b) then present the same data of Figure 5.5 in normalised form. These plots confirm that it is not the differences in initial TSS which were responsible for the observed increases in solids settleability in the effluent and illite mixtures. In Figure 5.6 (a), it is obvious that the fastest normalised settling was evident for the illite and effluent mixture. The increased rate of removal is also indicated in the first portion of the settling test, such that at t = 20 min, the normalised TSS removal rate for the illite and effluent mixture is approximately four times that for the illite control.  The measurements of settling distance in the settling column can be used to produce a settling velocity distribution as described in §4.4.2 except that TSS measurements are used instead of turbidity. This has been done in Figure 5.7. These settling velocity distributions can be easily compared through evaluation of the median settling velocity of the distribution. In Figure 5.7, the median settling velocities are indicated as the settling velocities that correspond to 50 % of the solids remaining.  The settling velocities evaluated from Figure 5.7 have been tabulated in Table 5.9. The increase in the median settling velocity (v ) due to the addition of the whole pulp mill effluent is apparent, in that v increased fivefold 50  50  from 0.13 cm/min for the illite control to 0.67 cm/min for the case with effluent added. Nonetheless these settling velocities are still quite low. Under quiescent conditions, it would take a particle approximately 2.5 h to travel 1 m at a velocity of 0.67 cm/min. A smaller increase in v appears to be apparent due to the addition of a clarified 50  effluent to the illite suspension.  Table 5.9. Median settling velocities as determinedfromthe plots of settling velocity distribution for Experiment 1. Mixture  v (cm/min)  Illite (200 mg/L) and effluent  0.67  Illite (200 mg/L) and distilled water  0.13  Effluent and distilled water  0.29  Illite (200 mg/L) and clarified effluent  0.23  so  71  1.0  "1  1—I  I I II I  1  1  1—I  1  M i l l  1  1—I  1—I  TT  I I I I  0.8  a • i—i c '0) 3 0-6 u m m C  O 1.— 1  o 0.4 £0  Illite ( 2 0 0 m g / L ) & Eff. Illite(200mg/L) Effluent  0.2  Illite &c Clarified Eff.  0.0 0.01  J  1  1  ft 1 1 111  .ftft..  0.10  ft  J  1  1  1.00  J  10.00  1  1 1 1111  100.00  Settling Velocity ( c m / m i n ) Fieure 5.7. Settling velocity distributions for Experiment 1.  5.4.2.2 Experiment 2 Experiment 2 was completed to replicate Experiment 1. A similar analysis was completed for the TSS information collected in Experiment 2. The variations of TSS with time, and the TSS removal rates are plotted in Figures 5.8 (a) and (b) respectively. Again the data have been smoothed to reduce some of the noise. These data have also been normalised, and are plotted in Figures 5.9 (a) and (b). The settling velocity distributions associated with these data have also been evaluated, and are plotted in Figure 5.10. 72  150  100  h-  ra  100  1000 Time  10000  (min)  0.0 (b) -0.2  j  -0.4 Illite (200  e  Illite  mg/L)  &  Effluent  (200mg/L)  -0.6 Effluent  ET3 1  Illite  (300mg/L)  -0.8 Illite (200  mg/L)  &c C l a r . E f f .  1.0 10  100  1000 Time  10000  (min)  Fisure 5.8. (a) Variation of TSS throughout the settling tests in Experiment 2. (b) Variation of rate of removal of TSS throughout the settling tests in Experiment 2.  73  In Experiment 2, as is evident in Figure 5.8 (a), the enhancement of removal of TSS by the addition of an effluent was not as significant as that observed Experiment 1. In Experiment 2, the initial TSS in the illite and effluent mixture was significantly higher due to the use of a different batch of effluent. The initial TSS in the illite control was around 85 mg/L, and the initial TSS in the illite and effluent mixture was approximately 130 mg/L, giving a percentage difference at the start of the test of 53 %. By the end of the settling tests, there was no longer any difference in TSS for the two tests. This again suggests that the addition of a pulp mill effluent improved the removal of solids from the suspension.  Further conclusions can also be drawn from Figure 5.8. In the settling test with the clarified effluent, there was improved removal of suspended solids noted with respect to the illite control. At the start of the test, the TSS in the illite and clarified effluent mixture was 15 mg/L higher than in the illite control. Towards the end of each test, there was no difference between the TSS in the control and the illite and clarified effluent mixture. This is consistent with the results from Experiment 1. Another useful comparison can be achieved by comparing the results for the illite at a 300 mg/L nominal initial concentration to those for the illite and effluent mixture. Since these cases have essentially the same starting concentration they can be easily compared. It can be seen that the removal of solids was more rapid initially for the illite and effluent mixture than for the illite at a 300 mg/L initial concentration although a similar final concentration results. This is a further confirmation that the improved solids removal is not simply due to differences in initial concentration.  Consideration of the plot of rate of TSS removal (Figure 5.8 (b)) is not as productive as for Experiment 1. There appears to be no real peak rate of solids removal rate in any of the distributions, but solids removal rates do appear to be consistently higher for the illite and effluent mixture than for any other case.  The normalised variation of TSS for Experiment 2 is presented in Figure 6.10 (a). As observed in Experiment 1, even on a normalised basis, the removal of solids from the illite and effluent mixture was more rapid than in the other tests. The normalised TSS removal for the two different illite settling tests are almost identical. The normalised TSS removal rate information (Figure 5.9 (b)) suggests that although the illite and effluent mixture displays the fastest rate it is not by as greater margin as was evident in Experiment 1.  74  1.0  (a)  0.8  J  0.6  o o 0.4  0.2  0.0 10  100  1000  _J l l I 10000 _ l  Time (min)  0.002  (b)  o.oooh-  -0.002 Illite (200 m g / L ) & Effluent " ^  -0.004  Illite (200mg/L) Effluent  -0.006 h  Illite (300mg/L) Illite (200 m g / L ) & Clar. Eff:  -0.008 10  100  1000  10000  Time (min) Fisure 5.9. (a) Normalised plot of the variation of TSS in the settling tests of Experiment 1. (b) Normalised plot of the variation in TSS removal rates in the settling tests of Experiment 1.  75  0.10  1.00 10.00 Settling Velocity ( c m / m i n )  100.00  Fisure 5.10. Settling velocity distributions for Experiment 2.  Determinations of v were made using Figure 5.10, and are tabulated in Table 5.10 along with the corresponding 50  results from Experiment 1. The values of v for illite from Experiment 2 compare well with the values obtained in 50  Experiment 1 suggesting that the overall technique is reasonably precise. There is a large difference in v o for the 5  effluent alone, i.e. 0.07 cm/min in Experiment 2 compared to 0.29 cm/min in Experiment 1. This is indicative of differences in the solids present for the two effluents and suggests that the solids in the effluent used in Experiment 2 were slower to settle. The addition of effluent to illite suspensions has resulted in an increase in v  50  in both  Experiments 1 and 2, although the magnitude of the increase has been different in each instance. In Experiment 1, 76  the effect was most dramatic with a fivefold increase in v occurring, whereas in Experiment 2, v increased by a 50  50  factor of 2.5. This disparity must have been due to the different effluents used in Experiments 1 and 2.. Once clarified however, both effluents have resulted in an almost identical increase in v  when added to the illite  50  suspension. Thus, in conclusion, the overall results of Experiment 1 were successfully replicated however the different effluents used in each experiment resulted in differences in the magnitude of increases in v . 50  Table 5.10. Summary of median settling velocity information. Mixture  v (cm/min) 50  Experiment 1  Experiment 2  Illite (200 mg/L) and effluent  0.67  0.38  Illite (200 mg/L) and distilled water  0.13  0.15  Effluent and distilled water  0.29  0.07  Illite (200 mg/L) and clarified effluent  0.23  0.21  Illite (300 mg/L) and distilled water  -  0.15  5.4.2.3 Experiments 3 and 4 Experiments 3 and 4 were completed with Fraser River sediments collected by Gomm (1994) and effluent from the Northwood mill to determine if the improved settleability effects observed with the illite and effluent were relevant to the "real" materials. Experiment 4 was completed at a lower temperature (4 °C) to determine if the cold temperatures present in the river for much of the year were important for flocculation. The results of each experiment were slightly different, reflecting the difference in the viscosity of water, and the associated reduction in settling velocity as a result of changing from room temperature to 4 °C. Only the full results for Experiment 3 will be presented due to the similarities between results from the two experiments. The results of Experiment 3 are summarised in Figures 5.11 to 5.14.  In Figure 5.11 (a) and (b), it is shown that there was no real difference between the settling for the sediment plus effluent mixture and the sediment only suspension. There was a slightly higher rate of TSS removal at the start of the experiment for the sediment and effluent case than for the control. The effluent appears not to have settled at all 77  150  100 bo Ul Ul  100 Time (min)  1000  (b)  Sediment (200 m g / L ) & Eff. Sediment (200 m g / L ) Effluent  10  Figure  100 Time (min)  1000  5.11. (a) Smoothed variation in TSS for Experiment 3. (b) TSS removal rates in Experiment 3.  78  40  i  i  i  i  i  i  i r  ~i  i  1  1  1—i—i—r  Sediment (200 m g / L ) & Eff. Sediment (200 m g / L ) 30 h  20  h  10  h  ui m >  10  100 Time (min)  J  i  i  i  i  i_  1000  Fisure 5.12. Variation in VSS in Experiment 3.  until after 100 min. Even then, the settling of solids from the Northwood effluent is slow compared to the settling observed for the effluent from the Squamish mill used in Experiments 1 and 2. This could be a reason why settleability of river sediments was not enhanced by the addition of this pulp mill effluent.  The VSS information was collected for the purpose of indicating whether the biosolids of the effluent were critical to the process of flocculation. Figure 5.12 shows the variation of VSS as a function of time in Experiment 3. Consideration of the effluent and sediment control cases show that the majority of VSS were present in the effluent 79  i.e. 25 mg/L compared to 4 mg/L in the sediment. This being so, changes in the behaviour of VSS and hence effluent biosolids due to mixing with sediments are best examined by comparing the results for the effluent only and sediment and effluent settling tests. The VSS in the sediment and effluent test was initially higher than in the effluent control by 20 %, yet between t — 200 min and the end of the test, was 30 - 40 % lower. This indicates that there was some preferential removal of biosolids from the suspension when Fraser River sediments were present. Despite the fact that flocculation and enhanced TSS deposition was not observed in Experiment 3, the preferential removal of VSS in the sediment and effluent mixture provides evidence that some sediment - biosolid interaction occurred.  The normalised information presented in Figures 5.13 (a) and (b) actually suggests that when the effects of increased concentration are taken into account, the settling for the sediment on its own was more rapid than for the sediment and effluent mixture.  The median settling velocities for the different tests can be evaluated from the distributions in Figure 5.14. These settling velocities have been tabulated in Table 5.11, along with the settling velocities determined in the same manner for Experiment 4.  Table 5.11. Summary of median settling velocities (v )for Experiments 3 and 4. 50  Expt 3 (23 °C)  Expt 4 (4 °C)  Settling case  v (cm/min)  vso (cm/min)  Sediment (200 mg/L) and effluent  1.22  1.24  Sediment (200 mg/L) and distilled water  3.55  1.80  Effluent and distilled water  0.057  0.049  50  These results indicate that the settleability of river sediments was not enhanced by the addition of pulp mill effluents. In Experiment 3, v for the effluent and sediment suspension was one third of that for the sediment only suspension. 50  In the experiment at 4 °C, the difference was not as great, but v for the sediment and effluent mixture was again 50  lower than for the sediment only.  80  1.01  —I  1 1 1—— I (a)  0.8  0.6  0.4  0.2  0.0 10  100 Time (min)  1000  0.00  (b)  ^ U \  -0.02  Sediment (200 m g / L ) & Eff.  o  Sediment (200 m g / L )  T3  Effluent  - 0 . 0 4 L_ 10  100 Time (min)  1000  Fisure 5.13. (a) Normalised variation in TSS for Exp. 3. (b) Normalised TSS removal rate in Experiment 3. 81  0.01  0.10  1.00  10.00  100.00  Settling Velocity (cm/min) Fisure 5.14. Settling velocity distributions in Experiment 3.  The effect of lower temperature can be seen by comparing results from Experiment 3 to 4. In decreasing the temperature, v for the effluent decreased from 0.057 cm/min to 0.049 cm/min although examination of the settling 50  velocity distributions suggests that this change may be insignificant. temperature seems to have been much more significant, as v  50  For the sediment, the impact of reduced  was reduced from 3.55 cm/min to 1.8 cm/min.  Finally for the sediment and effluent mixtures there was actually a slight increase in v at lower temperatures. 50  82  An explanation of the temperature effect observed for the sediment control can be obtained by consideration of Stokes' Law. In Stokes' formula for the terminal settling velocity of spherical particles (Streeter & Wylie, 1985), viscosity is an influencing factor, i.e.  (5.4) where u is the terminal velocity of the particle, d is the diameter of the particle, and p and p . are the densities of the p  s  fluid and the particle respectively. This indicates that the settling velocity is inversely proportional to the viscosity of the fluid. The viscosity of water, at 23 °C is 0.935 kg/ms, and at 4 °C, it is 1.57 kg/ms (Fischer et al, 1979) thus resulting in an increase in viscosity by a factor of 1.7 in decreasing temperature from 23 °C to 4 °C. In Experiment 4, v  50  for illite was reduced to a half of its velocity in experiment 3 which is in reasonable agreement with  expectations from Stokes' Law. This suggests that discrete settling was the predominant mechanism involved in the settling of the river sediments.  For cohesive fine-grained sediments, Lau (1994) found that decreased temperature enhanced settleability. Lower temperatures resulted in the compression of electrical double layers which enhanced flocculation, and despite reduced ambient viscosities, resulted in increased deposition at lower temperatures. Thus, simplistically, there are two competing effects present as temperature is reduced. Fluid viscosity is reduced at the same time that double layer thicknesses are reduced, enhancing the probability of flocculation. For discrete sediment behaviour, viscosity effects can be expected to dominate, resulting in lower settling velocities, whereas for cohesive sediment behaviour, double layer effects will dominate resulting in increased settling velocities. This explains the fact that essentially no decrease in v  50  has occurred for the effluent and sediment mixtures. In this instance, sediment behaviour was  cohesive to some degree resulting in the two competing effects cancelling out. For the effluent on its own, there was a 15 % decrease in v due to the reduction in temperature. 50  Given this, and the 50 % reduction in v  50  for the  sediment, we can again conclude that some interaction has occurred between the sediment and effluent biosolids to result in more cohesive behaviour of the suspended solids.  A further observation can be made regarding the median settling velocities in Experiments 3 and 4. It appears that v  50  produced for sediment and effluent mixtures lie between those for the sediment only and effluent only  83  suspensions. This suggests that there is no interaction whatsoever between the sediment and effluent biosolids, and that v for the mixture is simply the proportionate combination of the two. Closer examination reveals this is not so. 50  Assuming that in each mixture, two thirds of the solids mass were river sediment, and the remainder effluent biosolids (as at the start of the test), v for the sediment and effluent mixture was predicted to be 2.4 cm/min for 50  Experiment 3 and 1.2 cm/min for Experiment 4. Although the prediction for Experiment 4 is very similar to the observed value of 1.24 cm/min, there is little agreement between the prediction and the observed result (1.22 cm/min) for Experiment 3. The result from Experiment 3 suggests that some interaction may have occurred that actually slowed settling of the combined solids, whereas the result from Experiment 4 suggested that there was no interaction, and the solids must settle separately. In light of other observations, it is difficult to conclude that there is no interaction using this information, as many complicating factors may be involved.  In conclusion then, Experiments 3 and 4 provided some useful insights. There was no evidence to suggest that temperature reduction enhanced flocculation effects, but for river sediments and effluent from Northwood, no flocculation was observed under any conditions. One possible reason for this was that the settling of the sediment was primarily discrete. Despite no observable flocculation, the measurements of VSS suggested that effluent biosolids did associate with sediment particles of all sizes.  5.5 PRIMARY PARTICLE SIZE DISTRIBUTIONS Measurements of primary particle size distributions were completed to provide basic particle size distribution information for the materials used in experimentation. This information was useful in determining how well illite and trapped river sediments approximate the in situ primary particle size distribution displayed by suspended sediments.  An impression could also be gained of the general characteristics of the sediment from these  measurements.  Measurements were completed using an ELZONE, the operation of which was described in §2.5.1. This device was located in the Mining and Mineral Process Engineering Department at UBC, and measurements were carried out by a member of that department. The results from these measurements are summarised in Figure 5.15. Some of the  84  distributions have been extrapolated below the detection limit, and hence have very smooth curves for particle diameters less than 1.0 urn. The samples for which these measurements were completed included: •  a sample of Northwood effluent collected in July '95 as used in experiments at UNBC,  •  a sample of suspended Fraser River sediment also collected in July '95 as used in experiments at UNBC,  •  a sample of illite as used in the settling column experiments, and  •  a sample of Fraser River sediments as collected in a sediment trap (Gomm, 1994) and used in settling column experiments.  The difference in the primary particle size distributions for the sediment collected in a sediment trap, and that collected as a suspension from the River by the author is quite significant. The sizes of particles collected in the sediment trap were far greater than those collected as a suspension. The cut-off in the distribution for the trapped sediments at a diameter of just larger than 60 urn due to the sieving of this sediment is also obvious. According to Hawley (1988) and Baker et al. (1988), depending on the magnitude of the trap Reynolds Number and the aspect ratio of the trap, it is likely that there will be a bias towards the collection of larger particles in traps, and particles with lower settling velocities may not be included in the sample. Considering the existence of this bias, it is likely that the sediment collected using the traps was not truly representative of the suspended sediments in the Fraser River.  The primary particle size distribution of the illite that was used in experimentation appears similar to the size distribution obtained for the suspended Fraser River sediments.  The distribution shown for the illite is not  necessarily complete. The existence of a peak in the distribution near 1.5 urn could be indicative of the existence of a bi-modal distribution (S. Finora, 1996, pers.comm.). This was the detection limit of the ELZONE, and there is the possibility that the majority (by volume) of particles had diameters smaller than 1.5 u,m and that the peak at 1.5 pm is just part of the larger peak that would be present for diameters less than 1.5 um. Further measurements on the illite would be required with different equipment to determine this. However, if it is assumed that the distribution is not bi-modal, then the illite appears to be a reasonable approximation to the distribution obtained for the suspended sediments. Hence, we could assume that the settling test results achieved with the illite are more realistic than those  85  obtained using trapped Fraser River sediments. In saying this, it is assumed that particle size is more important than the nature of the material with regard to flocculation.  42-  0.1  1.0  10.0 Diameter (um)  ••— Fraser - Suspended - - o - - Effluent — A — Illite — X — Fraser - Trapped  Fisure 5.15. Comparison of primary particle size distributions for miscellaneous samples.  86  100.0  6.0 RESULTS OF ANALYSIS OF PREVIOUSLY COLLECTED FIELD AND EXPERIMENTAL DATA A number of researchers have previously collected data related to the flocculation of pulp mill effluents with Fraser River sediments. The first experimental work was completed by Krishnappan and Engel (1994). Some of the results from this work were analysed in §3.2 for their agreement with existing theories. Subsequently, in August 1994, additional field samples were collected from the Fraser River at locations just upstream of the diffuser, just downstream, and in the final effluent lagoon by Ian Droppo (1995, unpublished data). They were analysed for their in situ aggregate size distributions, although no primary particle size distribution information was obtained. Soon after that, in October 1994, Krishnappan (1996, unpublished data) collected data from a number of transects in the vicinity of the Northwood effluent diffuser. Information regarding the in situ aggregate size distribution, the velocity and depth of flow, and the total suspended solids concentration was collected. None of the field data have been completely analysed, and the experimental work was only considered briefly with respect to the presence of flocculation (Krishnappan and Engel, 1994). Since the results from the author's laboratory experiments were not completely conclusive, analysis of these field data was undertaken to determine precisely what flocculation effects may have been observed in the field.  6.1 ANNULAR F L U M E EXPERIMENTS (KRISHNAPPAN AND ENGEL, 1994) Two of the experiments completed by Krishnappan and Engel (1994) involved an investigation of the effect of the addition of pulp mill effluents on deposition of Fraser River sediments. The details of the experimental procedure and materials preparation are outlined in Krishnappan and Engel (1994) although a simple description of their experimental procedure is provided in §3.2. For the comparative tests completed with pulp mill effluents, the shear stress maintained during deposition was 0.121 N/m in one instance and 0.213 N/m in the other case. In the tests 2  2  with effluent, whole effluent previously collected from the Northwood mill was added to the flume in a volume such that the dilution of effluent was 3 %. Analysis of these data sets can be used to determine the impact of pulp mill effluent addition on the deposition and aggregation of Fraser River sediments.  87  6.1.1 Solids Concentration Measurements As previously mentioned in §3.2, measurements of aggregate size distribution and suspended solids concentration were made as a function of time as deposition progressed. Figures 6.1 (a) and (b) show the variation in suspended solids concentration as a function of time for the two cases. In the case of the experiment completed at a shear stress of 0.121 N/m , the steady state suspended solids concentration was not affected by the addition of the pulp mill 2  effluent. Immediately, and while deposition was occurring, solids concentrations were reduced due to the addition of the effluent, but not dramatically so. In the test with a deposition shear stress of 0.213 N/m , solids concentrations 2  were reduced throughout the whole test due to the addition of the pulp mill effluent but again, only marginally. There is thus a small indication that the addition of Northwood pulp mill effluent increased the settleability of solids in the suspension of river sediments.  i The addition of 3% of the flume volume of effluent should have resulted in a corresponding initial dilution of the sediment concentration. If it is assumed that the TSS of the effluent was around 80 mg/L (Evans and Hall, 1996), then the expected initial TSS in the flume would be 245 mg/L. If the TSS of the effluent was taken instead as 40 mg/L as found in the author's work, the expected initial TSS in the flume would be much the same at 244 mg/L. This expectation still does not account for the difference evident in the experiments which resulted in TSS immediately after effluent addition of 230 mg/L. In conclusion, some solids reduction of the order of 6 % was evident, although it may not be truly significant.  88  0  50  100 150 Time in minutes  200  250  (b) Figure 6.1. Effects of pulp mill effluent addition during deposition tests (Krishnappan, 1994). (a) Shear stress of 0.121 N/m . (b) Shear stress of 0.213 N/m . 2  2  6.1.2 Aggregate Size Distribution Measurements The effect of the addition of pulp mill effluents to the river water suspension can also be investigated by consideration of the aggregate size distribution measurements associated with the flume experiments described above. The aggregate size distribution measurements were made by Krishnappan and Engel (1994) using the laser diffraction device described in §2.6.3. A simple analysis has been completed for this thesis by considering the variation in median aggregate size (d ) as a function of time. This is a useful tool for assessing the behaviour of 50  aggregate sizes because it allows for a huge amount of information to be compressed into a very concise form. Figure 6.2 shows the variation in d for the deposition test with a shear stress of 0.121 N/m . 2  50  89  90.0 85.0  v  80.0 75.0 o m •o  70.0  /*'  i  •  p  •  h  /a  65.0 i  i  L  1  A  '  60.0 55.0 50.0 0  20  40  60  80  100  120  140  160  180  Time (min) o - - River Water only  Fisure 6.2. Variation in d  •  River water + effluent  (um) as a function of time in deposition test - shear stress = 0.121 N/m . Raw data 2  50  used with permission of Krishnappan (1995).  Figure 6.2 shows that there was some increase in d as a result of the addition of the pulp mill effluent. For the first 50  20 minutes of the test, the shear in the flume was kept high enough that flocculation was not possible, and during this period median particle diameter remained fairly constant. There does appear to be a slight reduction in d during 50  this period, suggesting that some of the larger discrete particles settled from suspension. During this period, d for 50  the river water and effluent mixture was higher by around 3 um, which is consistent with the lower TSS present from t = 0 to 20 min in Figure 6.1 (a).  As the test proceeded, larger median aggregate diameters were observed for the river water and effluent case. After the shear in the flume was reduced after the initial 20 minute portion of the test, there was a rapid increase in d for 50  both the river water, and the river water and effluent mixture. Strangely enough, after reaching a peak at the 30 and 40 minute marks, d for each suspension was reduced slightly before it increased rapidly again after 80 minutes. For 50  90  the river water only suspension, this pattern of gradual reduction followed by a rapid increase was particularly dramatic. At the 80 minute mark in each of the tests d seemed to remain reasonably constant, as some steady state 50  median diameter appeared to have been reached. The steady state diameter for the mixture with pulp mill effluent was approximately 10 urn larger than for the river water only suspension. This difference represents a 13 % increase in aggregate size diameter. The fact that larger aggregates were evident in suspension at steady state also suggests that the settling characteristics of the aggregates were not changed. Although aggregates formed with sediments and effluent biosolids may be larger, they may also be less dense meaning they will be retained in suspension anyway. This possibility is consistent with the observations made in §5.2.2.  Some investigation of the change in aggregate size distribution is also worthwhile to determine what has been responsible for the increase in median diameter. Figure 6.3 presents a comparison between the aggregate size distributions for river sediments only, and a river water and effluent mix, at t = 0, 18, 40, 60 and 80 min throughout the deposition test. With respect to both distributions, there was a significant change with time. At the start of the tests, there were two peaks in the distribution, one at a diameter of 30 pm, and the other at a diameter of 75 um. As the test progressed, these peaks were lost as the distribution built towards a peak at the upper end of the measurement range. The maximum diameter for which measurements were made was 188 pm, and it is possible that aggregates larger than this were produced, but were not measured. The distributions at t = 80 min are fairly representative of the distribution attained at steady state. From Figure 6.3 it can be seen that at all of these times but most noticeably at t = 60 and 80 min, there was a tendency for larger aggregates to be more prominent in the suspension which includes 3 % effluent. The fact that this was evidenced even at the start of the test suggests that any flocculation effect between pulp mill effluent and river sediments occurred almost instantaneously.  In conclusion then, it appears that the addition of a small quantity of pulp mill effluent to the flume resulted in some increase in the size of aggregates. Although the difference in aggregate sizes was of the order of 10 %, there was no corresponding effect on the deposition of the solids, particularly under steady state conditions. More noticeable effects may have been apparent if a greater proportion of effluent had been used in the test. Although a dilution of 3 % is low for the mid field zone (Vine, 1996), dilutions in the near field, can be expected to be around 10 % (Marks, 1996).  91  t = 0 min  "river s e d i m e n t only effluent  added  Aggregate Diameter (/im)  t  = 18 min  Aggregate Diameter (fJ.m)  t = 40 min  Aggregate Diameter (/im)  t = 60 min  Aggregate Diameter (/im)  t = 80 min  Aggregate Diameter (/im)  Figure 6.3. Comparison of aggregate size distributions as a function of time during a deposition test with a she stress of 0.121 N/m . Raw data used with permission of Krishnappan (1995). 2  92  6.2 IN SITU A G G R E G A T E SIZE DISTRIBUTIONS (IAN DROPPO, 1995, UNPUBLISHED DATA) Samples collected from upstream and downstream of the effluent diffuser were subjected to an aggregate size distribution analysis. These measurements were made using an image analysis technique, as outlined in §2.5.2. The sample taken upstream of the diffuser was collected 30 m upstream, and the sample below the diffuser was taken 50 m downstream of the diffuser. Both samples were collected between 20 and 30 m from the bank, and at each location two samples were collected. Samples were also taken from the final effluent lagoon.  6.2.1 Median Diameters As part of the measurement process employed by Droppo (1995, unpublished data) volume median spherical diameters were also obtained. Table 6.1 summarises the results for this parameter for all of the samples. From this information, it appears that in one of the samples taken below the outfall (Downstream - 2), there was quite an increase in d , going from around 70 pm upstream of the diffuser, to 102 pm just downstream, suggesting that 50  significant aggregation was occurring in the River whereas a decrease in d^o has been shown for the sample Downstream - 1. The large median diameters reported for the effluent would not be maintained on entry to the river as the shear conditions in the outfall pipe and diffuser jets would be most likely to destroy the floes rapidly. According to Droppo (1996, pers. comm.), the distribution yielding the larger volume median diameter (Effluent - 1 at 252 pm) should be considered to be the most accurate, as it included some of the larger floes present in the effluent.  The vast differences in d for replicate samples downstream and in the effluent lagoon suggest that this aggregate 50  measurement technique may not provide accurate results for this application. Hence, it is difficult to make any statistically significant conclusions based on these data. There are two possible reasons for the poor reproducibility of these results. Much of the problem could be due to the fact that grab samples are required for measurement. The problem with this for sampling at the location downstream of the diffuser is that at a distance of 50 m from the diffuser, the plume is visibly temporally and spatially unstable. Thus, it is possible that one sample was taken within the plume, and the other one was not. This could account for the differences in d  50  for the samples Downstream - 1  and 2. The other problem with this technique is a result of the comparatively small number of aggregates counted. 93  Table 6.1. Results for volume median spherical diameter (d^from Droppo (1995, unpublished data). Numbers 1 and 2 indicate replicate samples. Sample  d o (um)  Upstream - 1  70.3  Upstream - 2  73.6  Downstream - 1  65.4  Downstream - 2  102.4  Effluent - 1  251.8  Effluent - 2  69.7  5  This results in the tendency for a few larger particles to dominate the distribution. This factor was responsible for the difference between d  50  for the samples Effluent 1 and 2 (Droppo, 1995, pers. comm.). For these reasons, a more  thorough analysis was desired for the field data collected by Krishnappan using the laser diffraction particle size analyser This device counts more particles in determining aggregate size distributions, and hence is regarded as giving more accurate results (Droppo, 1995, pers. comm.).  6.2.2 Aggregate Size Distributions A brief analysis of the aggregate size distributions obtained by Droppo (1995, unpublished data) was conducted for completeness. The raw aggregate size distributions obtained for the river water samples are presented graphically in Figure 6.4. Sharp jumps in the distribution begin at aggregate sizes greater than 50 um. The jumps became larger at higher aggregate sizes until the distributions are almost indistinguishable. Hence it is difficult to draw conclusions about the distributions at aggregate sizes larger than 50 um.  One conclusion can be made from the distributions shown for d < 50 um. For both samples taken downstream of the diffuser, there are lower proportions of particles with diameters less than 40 um. This must be balanced by an increase in the proportion of larger particles, but it is difficult to determine at what diameters this may have occurred. This suggests that a small amount of aggregation of finer particles has occurred downstream of the effluent discharge. 94  Fisure 6.4. Aggregate size distribution comparisons for upstream and downstream samples. Raw data used with permission of Droppo (1995).  6.3 FIELD TRANSECTS - KRISHNAPPAN (1996, UNPUBLISHED DATA) Results from the most comprehensive field data collection program became available in the year of writing this thesis.  During October 1994, data were collected for flow depths and velocities, the in situ aggregate size  distribution, and the concentration of suspended solids in the flow. A l l of this information was collected for a number of transects, one of which was 2 km upstream of the diffuser, and the others which were all downstream of the diffuser at distances of 30, 100, 300 and 1000 m. Each transect was completed on consecutive days, which resulted in some complication in the analysis of the results. 95  6.3.1 Total Suspended Solids Concentrations Information regarding the average suspended solids concentration across each transect was available. These results are presented in Table 6.2. Considering the TSS information alone, it would appear that there was a significant drop in the suspended solids concentration of the river as the sampling proceeded further downstream from the diffuser. These results are quite dramatic, and are even more dramatic than the findings presented resulting from the work in the Athabasca River (Krishnappan et al, 1994).  Table 6.2. Transect averaged TSS data and relevant flows at Shelley (Krishnappan, 1996, unpublished data). Transect location  Date  TSS (mg/L)  Flow (m /s)  2000 m u/s  5 Oct 1994  255.6  597  30md/s  6 Oct 1994  129.6  584  100 m d/s  7 Oct 1994  67.5  523  300 m d/s  8 Oct 1994  49.1  480  1000 m d/s  9 Oct 1994  33.1  453  3  Consideration of the situation under which the data were collected is also necessary. The fact that each transect was completed on a different day becomes important because the flow in the river was dropping quite significantly over this period as shown in the flow data in Table 6.2. Also, it must be realised that these concentrations are average concentrations based on the entire transect. The locations of transects and sampling locations (x) in relation to the predicted location of the plume for a river flow of 525 m /s (Vine, 1996) is shown in Figure 6.5 and indicates that for 3  some transects, the majority of measurements were made outside the plume. In Vine's work, the effluent plume from the Northwood diffuser was modelled using a particle tracking model. In this diagram, the River has actually been straightened for ease of plotting, and the width of the River was obtained from topographic maps. Although this is only the predicted location of the plume, simulated results at a flow of 805 m /s agree well with the remote sensing 3  data of Borstad (1995, pers. comm.) collected at that flow. A flow of 525 m /s was chosen for this simulation since 3  it is the median of the range of flows present during the period of Krishnappan's field work, and the change in plume width over this range of flows would be negligible (Vine, 1996). It is impossible that the effluent could be 96  responsible for changes in TSS outside the plume, and since often only one TSS measurement was made within the plume, the majority of these changes must be related to TSS reductions in the river as a whole.  200  -2001 0  [  I  I  100  200  300  I I I I I I I I I I 400 500 600 700 800 900 1000 1100 1200 1300 Distance Downstream - x (m)  Fisure 6.5. Location of field transects and sampling locations downstream of the diffuser superimposed on the predicted plume location for Q  = 525 m /s. Plume location used with permission of Vine (1996). 3  river  6.3.2 TSS Predictions It is possible to make predictions of the TSS that could be expected in the River based on historical information, and the flow ( 0 in the River, upstream at Shelley. Church (1996, pers. comm.) determined calibrations for TSS and Q 97  at Hansard, upstream of Prince George. This correlation was produced by considering sediment concentration and flow information for 1984 - 1986 for the month of September at the Hansard station. The resulting correlation with an R of 0.811 was 2  In C = -5.272 + 1.576 In Q  (6.1)  where C represents TSS and Q represents the flow at Hansard on that day. An additional correlation was also computed by the present author based on the Hansard data reported by Zrymiak and Tassone (1986) for the years 1974, and 1976 - 1980. Only October data were considered, and the equation resulting from this correlation with an R of 0.699 was as follows. 2  C = 0.0002f2  (6.2)  20844  Suspended sediment concentrations obtained using both of these formulae have been tabulated in Table 6.3 along with the necessary flow information at Shelley. The correlation at Hansard has been assumed to be valid for Shelley where the flow data were available, and the conditions at Shelley were also considered to be equivalent to those present in the vicinity of Northwood.  Table 6.3. Summary of Fraser River flows (Shelley gauging station), predicted TSS using Equation 6.1 and Equation 6.2, and measured TSS. Transect  Date  Q (m /s) 3  Location 2000 m wis  Measured  Predicted TSS (mg/L)  TSS (mg/L)  Equation 6.1  Equation 6.2  5 Oct 1994  597  255.6  122  122  30 m d/s  6 Oct 1994  584  129.6  118  117  100 m d/s  7 Oct 1994  523  67.5  99  93  300 m d/s  8 Oct 1994  480  49.1  86  78  1000 m d/s  9 Oct 1994  453  33.1  79  69  The fall in TSS from day to day seems to be reasonably consistent with the trend predicted from knowledge of the River flows alone. However, the decrease in TSS at the Northwood site was much greater than the predictions for Hansard suggest. For October 5, 1994, the predicted sediment concentration is significantly lower than the measured concentration. Towards the end of the sampling period, this relationship is reversed as predicted TSS results are 98  greater than the measured TSS. Unfortunately, no TSS measurements were made at Hansard or Marguerite stations over the period that Krishnappan conducted his field study for comparison to the data collected at Northwood. Although the drop in TSS was significant, there is no evidence to suggest that it occurred because of interactions between the pulp mill effluent and river sediments.  Drops in TSS in the River of over 100 mg/L in a 24 hour period due to a small reduction in flow are not uncommon. During the summer of 1995, the author collected samples at the Northwood effluent diffuser which were analysed for TSS. Comparison of these results with the flow in the River at Shelley showed that between the July 29 and 30, as the flow dropped from 1630 m /s to 1520 m /s, TSS dropped from 286 mg/L to 156 mg/L. Hence, it is likely that the 3  3  TSS reduction observed by Krishnappan was due to natural day to day changes in the behaviour of the Fraser River.  6.3.2 Median Diameters As with the suspended solids information, median aggregate diameter data for each transect were collected on different days. A summary of the data, and the median diameters averaged across each transect is presented in Table 6.4. Median diameters of between 10 and 20 pm obtained in this work are much smaller than those reported by Droppo (1995, unpublished data) of around 70 pm. It would appear that there was an increase in d  50  as  measurements were taken further downstream, until the 300 m transect. The field data suggest that, between this location and the 1000 m transect, rapid settling of the larger aggregates resulted in a removal of the larger particles from the flow, and hence a reduction in d$o in the flow.  However, much of the variation in the observed median diameters, is thought to be due primarily to the day to day variation in conditions in the river. Since each transect was completed on a different day, comparison between these results is not realistic. Also by consideration of Figure 6.5 it would appear that many sampling locations could have been outside the plume where it is not possible that flocculation effects could be felt.  99  Table 6.4. Average median diameters noted at the various transects (Krishnappan, 1996, unpublished data).  Transect Location  dso (um)  2000 m u/s  8.77  30 m d/s  12.09  100 m d/s  14.96  300 m d/s  19.51  1000 m d/s  13.90  Knowing the location of sampling points with respect to the plume allows a determination of the effect of the plume on median diameters within 1000 m downstream of the diffuser. Figures 6.6 (a) and (b) show the median aggregate sizes plotted with position across the river for the five separate transects. These values for d o have been obtained by 5  averaging the d results obtained at the surface and at a depth of 1 m for each sampling location. The transect 50  plotted for x = -2000 m can be considered as the control for all other transects. Since we know that the points farthest from the Northwood bank of all transects were probably outside the plume (Figure 6.5), they represent the background conditions in the river, regardless of the presence of the plume. The increase in d  50  that occurred for  each transect can then be attributed to a natural increase due to changes in river conditions (Ad . 50  Natural)'  and an  increase due to the plume (Adso). These divisions have been marked on Figures 6.6 (a) and (b).  It does appear that there was a slight increase in d due to the plume at all transects. However, compared to the day 50  to day variation of the median diameters in the river, the increase is not large. The difference between the diurnal variation in aggregate size diameter and the spatial variation in aggregate size diameter on a particular day is also obvious. The transect at which the most significant plume flocculation effect appeared to be present is 100 m downstream of the diffuser. For this transect, the plot of median diameter is quite similar to the concentration profile that would be recorded due to the presence of an inert tracer in the plume, and Ad in the middle of the plume is 50  around 2 um, and around 1.5 um nearer the bank. At other locations, Ad  50  is typically of the order of 0.5 um.  Assuming that the aggregate density is maintained, increases in aggregate diameter of 0.5 um will result in a 7 % increase in settling velocity. Of course, it is most likely that aggregate density will be reduced as more particles agglomerate, and this relative increase in settling velocity must be considered as an upper bound. This analysis has 100  Fisure 6.6. Plots of depth averaged median diameters for the five transects in the vicinity of Northwood for (a 30 and 100 m, and (b) x = 300 and 1000 m. Raw data used with permission of Krishnappan (1995). 101  provided some insight to the magnitude of the flocculation effect actually occurring in the Fraser River. It does not appear to be as large as previously thought.  6.3.3 Aggregate Size Distributions An analysis was also made of the aggregate size distribution information collected during the field study of Krishnappan. From Figure 6.5, it would appear that for each transect, at least one measurement was made within the plume, and one in the body of the river flow outside the plume. The measurements taken outside the plume can then be considered as a control for the measurements within the plume. Figure 6.7 shows the comparisons between aggregate size distributions for "plume" and "control" locations. Although there is no plume present at the x = -2000 m transect, the location closest to the Northwood bank has been taken to be the "plume" and the sample location nearest the right bank has been taken as the "control" for simplicity. The sampling locations from which these distributions were obtained are also marked on Figure 6.7. The distributions shown were constructed by averaging the distributions measured at the surface with those 1 m below the surface.  There is little evidence of a significant flocculation effect in the comparisons of Figure 6.7. At most locations, the aggregate size distribution within the plume is almost identical to the control distribution. At locations 30 m and 100 m downstream, it appears as though there was a slight increase in the proportion of larger aggregates in the distributions. It is only for a distance of 100 m from the diffuser, that the plume distribution shows a genuine difference from the distribution at the control. For this location there is evidence of a greater predominance of larger particles, with a spike in the distribution being present for d = 75 um. The increased proportion of particles with d = 75 um has been produced by decreased proportions of particles with diameters of 20 and 40 um. There appears to be no difference in the proportion of fine particles.  There is some possibility, that because of the cross-sectional shape of the river, there are significant differences in shear between the plume and control locations which would affect aggregate size distributions. There is indeed some difference between the turbulent velocity gradients as calculated for the plume and control locations.  Using  measured velocities presented with cross-sectional information in Appendix F (Krishnappan, 1996, unpublished data), and based on the calculation method described in Appendix A, it is possible to calculate the value of G in the  102  zz  ^  §-  ^control  \-  y  .  1  p  1  1  1  1 — r  180 m  /  -2000 m  •  ,  •  r——i  1  I  \  = 40 m plume  P 1  1  1  r-  E  control  -f  plume  -i  1  I Aggregate Diameter (/im)  30 m  X =  y  = 60 m control  y J  = 40 m plume  Aggregate Diameter (/im)  100 m  X =  3 ^  y J  = 90 m control  y J  = 30 m plume  Aggregate Diameter (j±m)  x = 300 m  y  = 150 ml control  y  = 30 m plume  Aggregate Diameter (^m)  1000 m  zz  §§-  E  y  =  -E  * control  = 30 m  y J  \_y  ^<^--\  plume  E  -|  E  Aggregate Diameter  (fim)  Figure 6.7. Comparison of aggregate size distributions for plume and control locations from the transects at Northwood. Raw data used with permission of Krishnappan (1995). 103  flow far from the bed at the sample locations. This should be reasonably representative of the conditions where the measurements were made in the top 1 m of the flow. Table 6.5 summarises this information, and the percentage difference in G between the control and plume locations.  There are differences in calculated G from control to plume locations at each transect. The differences between G at plume and control locations are greater than 10 % for the transects upstream of the diffuser, and 300 m and 1000 m downstream. At the upstream location, there is no confusion of the two distributions due to the presence of an effluent. At this location, despite the difference in G of 15 % from one side of the river to the other, the aggregate size distributions are almost identical. It would appear in this situation that the difference in shear has little effect on the aggregate size distribution.  Table 6.5.  Calculated velocity gradients for plume and control locations at the transects near Northwood  (Krishnappan, 1996, unpublished data). Transect  Location  location 2000 m u/s  30 m d/s  100 m d/s  300 m d/s  1000 m d/s  Measured  e (m /s ) 2  Percentage  3  velocity(m/s)  difference (%)  Plume  0.7  0.00292  54.0  Control  0.97  0.00404  63.6  Plume  1.06  0.00442  66.5  Control  1.31  0.00546  73.9  Plume  1.16  0.00484  69.5  Control  1.4  0.00584  76.4  Plume  0.65  0.00271  52.1  Control  0.99  0.00413  64.2  Plume  0.46  0.00192  43.8  Control  0.26  0.00108  32.9  15.1  10.0  9.0  19.0  -33.0  The response time of aggregate size distribution to some change in the shear may also be an important factor in justifying these comparisons. As the water flows through different channel shapes, it is not likely that the aggregate  104  size distribution would be affected unless a particular cross-sectional shape and hence distribution of G was maintained for a long period of time. Since little is known about the effect of cross-sectional shape on aggregate size distributions, it is thought that the plume - control comparisons made in Figure 6.7 are indeed reasonable.  6.3.4  Explanations for Aggregate Size Distribution Changes  It is possible to describe the observed day to day changes in aggregate size distribution in the river using the theory of Lick et al. (1993 (a)). This theory was reviewed in §2.4.1 and in essence suggests that d o should be proportional 5  to the inverse square root of the product of shear and TSS, i.e.  <*50 ~  (6-3)  j =  This relationship is applicable to a given sediment, and the constants of proportionality would be different for different sediments. Table 6.6 summarises the measured transect average value of C, the calculated value for G at the "control" locations (Table 6.5), the product CG, and the measured transect average median aggregate diameters.  Table 6.6.  Summary of TSS and velocity gradient information, their product, and measured median diameters.  Transect  C(mg/L)  G (s-)  CG (mg/L/s)  d (urn)  2000 m u/s  255.6  67.2  17176  8.77  30 m d/s  129.6  78.1  10122  12.09  100m d/s  67.5  80.8  5454  14.96  300 m d/s  49.1  67.9  3339  19.51  1000 m d/s  33.1  34.8  1152  13.90  1  so  Plotting the first four data points yields a significant correlation with an R of 0.99 for the relationship 2  d  (6.4)  = 36.0(CG)" 0  50  50  where C is in g/L, G is in /s, and d is in um. Inclusion of the fifth data point changes the relationship of best fit m  significantly giving a power index very different to -0.5, and reduces the R to 0.42. 2  If we assume that the  relationship of equation 6.3 is valid for the sediment with a particular primary particle size distribution, then the most likely reason that the results at the 1000 m transect do not fit the relationship is some significant change in the 105  100  d = 36(CG)° R = 0.99  5  2  —  o  10  LO  10  100  CG (mg/L/s) (a)  100  d = 18.1(CG) R = 0.42  018  2  _  •  E 3o  10  •  LO  T3  1  10  100  CG (mg/L/s) (b) Fisure 6.8. Correlations between CG and d . (a) Using information transects except x = 1000 m, and (b) usi m  information from all transects. Raw data used with permission of Krishnappan (1995). 106  primary particle size distribution of the sediments which may occur due to dramatic erosional events in the river rapidly changing the nature of the sediment. Figure 6.8 (a) shows the correlation for the first four transects, and Figure 6.8 (b) shows the change in the relationship due to the inclusion of the result at the 1000 m d/s transect.  The similarity between the relationship of Lick et al. (1993 (a)), and the correlation shown in Figure 6.8 (a) is strong indeed. This suggests that the noted increase in particle size from day to day and transect to transect may be primarily due to changes in the sediment concentration, and turbulent velocity gradients in the river.  Despite this explanation for the day to day changes in d , there is still some increase in d in the plume (Ad ) that 50  50  50  remains to be explained. A simple calculation was completed which utilised data from Krishnappan's field work and the primary particle size analysis results presented in §5.5. If the assumption is made that all biosolids in the effluent bind to river sediments, then it is possible to calculate an expected median aggregate diameter the plume using the measured TSS  River  (d .noc) 50  at a point in  and natural median aggregate diameter (d o-Ri er) from Krishnappan's data. 5  V  The following assumptions were also made: •  sediment particle density  •  biosolid particle density (pBiosoiid) =1.1 g/mL,  •  suspended solids concentration of the effluent (TSS  •  median particle diameter for effluent particles (d .Effluent) = 4.1 pm from §5.5, and  •  dilution (S) can be assumed from the work of Vine (1996).  (psedimem)  = illite density = 2.6 g/mL (Terzaghi and Peck, 1960),  Effluent  ) = 80 mg/L (Evans and Hall, 1996),  50  The calculation of d o.fioc then proceeded as follows: 5  •  first, the mass of the median sediment particle was calculated.  •  The number of these particles per litre of sample was then calculated by dividing TSS  River  by the mass  of the median sediment particle. •  The same procedure was followed for the effluent to obtain the number of biosolid particles per litre of effluent.  •  For the relevant location in the plume, using a dilution factor from Vine (1996), the ratio of biosolid particles to effluent particles (N g/N ) at a particular transect can be determined. e  sed  107  •  Then, assuming that this number of biosolid particles stick to each sediment particle,  d -fioc  c a n  50  be  determined assuming conservation of volume. The results of these calculations for the four transects (x = 30, 100, 300 and 1000 m) are tabulated in Table 6.6 along with predicted and measured results for Ad . 50  Table 6.6. Data used in calculations for Adso and summary of some important results. Transect  TSSRi  dsO-River  (um)  (mg/L)  30 m  11.8  129.6  50  100 m  13.6  67.5  300 m  19.1  1000 m  13.8  The predicted values of Ad  50  N /N  S  ver  ef  sed  Ad  dso-doc  s o  (rim)  (um)  Predicted  Measured  0.70  11.91  0.11  0.4  70  1.5  13.78  0.18  2.1  49.1  90  4.3  19.37  0..27  0.6  33.1  100  2.2  14.06  0.26  0.5  can be seen to be lower than the measured results evaluated from the analysis of  Krishnappan's field work. This is reasonable since the assumption made in the calculation of dso-noc requires that the volume of each biosolid particle attached to a sediment particle is evenly distributed around the outside of the sediment particle like a coating. In reality, a biosolid will stick to a sediment particle without such a gross distortion. This will result in an aggregate with a larger effective diameter and lower overall density. This fact combined with the coarseness of the calculations completed suggests that it is not surprising that little agreement is present between the predicted and measured values for Ad . It is not considered unreasonable then, to conclude that biosolids were 50  indeed binding to sediment particles to give the increase in aggregate diameters noted in the plume in the Fraser River. It can also be concluded that this binding must occur quickly, since an increase in d was noted just 30 m 50  downstream of the diffuser. In travelling 30 m from the diffuser, less than a minute would have elapsed, indicating that the process is indeed rapid.  108  7.0 CONCLUSIONS  AND  RECOMMENDATIONS  7.1 GENERAL REMARKS It has been shown in Chapters 2 and 3 that the use of theory to describe the processes that may be occurring in the Fraser River near the Northwood diffuser is very difficult. Therefore, experimental data and field data were used to examine the phenomenon of flocculation of river sediments with pulp mill effluents. Consideration of the sum of these results altogether allows some useful conclusions to be drawn regarding the observed flocculation. In the remainder of this Chapter, results will be summarised, and the important conclusions that have been drawn from these results re-stated.  7.2 SUMMARY OF RESULTS 7.2.1 Experimental Work The results from the experimental work have indicated varying degrees of flocculation. The results and important conclusions based on these results are summarised in Table 7.1.  Table 7.1. Summary of experimental results obtained by the author. Experiment  Results  Conclusion  Critical coagulation  No indication that a critical  Electrolytic flocculation of the river  concentration  coagulation concentration was  sediments was not apparent.  present. Settling tests measuring  Apart from the initial turbidity  An instantaneous aggregation effect  turbidity  reduction, turbidity removal was  appeared to be present due to effluent  not enhanced by the addition of  addition, but settling velocity  effluent.  distributions were unchanged.  109  Table 7.1 Cont'd Experiment  Conclusion  Results  Instantaneous turbidity change  Turbidity reduction occurred for  Turbidity changes with no associated  measurements  additions of effluent to river water  reduction in TSS are thought to be  with maximum reductions  associated with aggregate size  occurring for a 1:1 mixture.  distribution changes. Flocculation will be controlled by conditions in the near field.  Settling tests measuring TSS  Settleability of illite was enhanced  Flocculation has occurred resulting in  (illite)  by the addition of effluent. It was  increased settling velocities. The  most profound when effluent  biosolids of the effluent may be  contained its naturally occurring  necessary for most significant effects.  biosolids. Settling Tests Measuring TSS  Addition of effluent hindered  Flocculation did not occur between the  (river sediments)  settleability of solids in sediment /  river sediments and the effluent since  effluent mixture. VSS  particles behave discretely. Biosolids do  preferentially removed from  aggregate with sediment even if  sediment / effluent mixture.  settleability is not enhanced.  Primary particle size  There were large differences  The trapped sediments were a poor  distribution measurements  between the distributions present  approximation to suspended sediments,  for the trapped sediment, and the  however illite provided a reasonable  sediment collected as a  approximation with respect to primary  suspension.  particle size distribution.  7.2.2 Analysis of Other Researchers' Data A large quantity of useful information has been collected by other researchers (Krishnappan and Engel, 1994; Krishnappan, 1996, unpublished data; Droppo, 1995, unpublished data), yet it has not been analysed in great depth or considered in the overall context of information available. These data were analysed in detail in Chapter 6, and 110  have proved extremely useful in completion of this work because they included some in situ aggregate size distribution measurements as well as distribution measurements made during experiments. It has been measurement of aggregate size distribution that has proved to be the stumbling block of the author's laboratory experiments. The results and major conclusions drawn from the analysis of the sum of these data are contained in Table 7.2.  Table 7.2. Summary of analysis of field data, and results of analysis of experimental data collected by other researchers. Analysis  Results  Conclusions  Annular flume experiments  Addition of effluent resulted in a  A slight increase in settleability of  (Krishnappan and Engel, 1994) -  slight reduction in suspended solids  suspended solids was observed for  TSS measurements.  concentration. The effect depended  one comparison but not for another  on the shear applied during the test.  at steady state.  Annular flume experiments  Addition of effluent resulted in an  Aggregation of suspended solids  (Krishnappan and Engel, 1994) -  increase d , immediately and under  appears to have resulted due to the  aggregate size measurements.  steady state conditions.  addition of pulp mill effluent.  In situ aggregate size distribution  Fewer fine particles and more large  Data quality seemed questionable  measurements (Droppo, 1995,  aggregates were present downstream  although significant aggregation of  unpublished data).  of the diffuser than upstream.  pulp mill effluents and river  50  sediments was observed 50 m downstream of the diffuser. In situ TSS measurements  A drop in suspended sediment  The drop TSS could not be  (Krishnappan, 1996, unpublished  concentrations was noted, although  attributed to the flocculation of  data).  much of this was due to falling flow  effluent with river sediments.  in the River. In situ aggregate size distribution  Median diameters of aggregates  Some flocculation was apparent  measurements (Krishnappan, 1996,  increased marginally within the  although the majority of changes in  unpublished data).  effluent plume.  aggregate sizes are due to natural changes in the River.  Ill  7.3 C O N C L U S I O N S A number of conclusions can be made based on the study completed. •  The most conclusive evidence of the flocculation phenomenon investigated was provided by the settling column experiments. The addition of pulp mill effluent to an illite suspension resulted in more rapid deposition of TSS than for control cases. This enhanced deposition is thought to be a result of flocculation of the biosolids in the effluent and illite particles. At most, the addition of a pulp mill effluent to an illite suspension resulted in median settling velocities being increased fivefold.  •  Evidence of aggregation was also provided in experiments with river sediments and effluent through observed turbidity reductions. The fact that turbidity changes did not appear to be reflected in changes in settling velocity distributions suggests that sediment dynamics may not be affected by agglomeration with biosolids.  •  The presence of biosolids was shown to be necessary for the most dramatic flocculation effects in the experiments with illite suspensions.  •  Flocculation was not enhanced by the addition of pulp mill effluent to a suspension of trapped Fraser River sediment. This suggests that the flocculation mechanism is only relevant to smaller discrete particles with diameters less than 10 pm. Biosolids were still shown to associate with sediment.  •  The effect that temperature has on flocculation was not ascertained, since for the sediments tested while temperature effects were investigated, no flocculation was evident.  •  The field data collected by Ian Droppo provides little evidence of flocculation effects immediately downstream of the Northwood diffuser.  •  The field data collected by Krish Krishnappan indicated that slight flocculation occurred within the effluent plume. Flocculation typically resulted in increases in d of approximately 0.5 pm. The majority of the 50  observed changes in aggregate diameters at locations further downstream were due to variations in the turbulent velocity gradient and sediment concentration in the River. •  The flocculation phenomenon appears to occur on time scales of less than a minute, and is most pronounced when dilution of the effluent is low. For these reasons conditions in the near field of the Northwood diffuser are most likely to dictate how flocculation occurs in the River.  112  •  It does not appear from the field data available that flocculation in the plume of the Northwood diffuser is an important consideration for sediment transport.  Seasonal factors may complicate this conclusion. The  reasonable likelihood that effluent biosolids do associate with sediment particles suggests that this flocculation must be regarded as being extremely important from the perspective of contaminant transport. Hence, the very coarse approach to mixing and transport of contaminants taken by Gobas (1996) could be dramatically improved by consideration of these effects if they can be quantified.  7.4 RECOMMENDATIONS FOR FUTURE RESEARCH There are a number of possibilities for future research related to this study. The justification for these should be that the flocculation is important from the point of view of predicting contaminant transport. Prior to further study, it could be assumed for modelling purposes that biosolids do become associated with the fine particles of a sediment, and their fate will be determined by the fate of these fine sediments.  Most importantly, more work needs to be done on the definition of the phenomenon. It is suggested that fine grained sediments are concentrated on in further experimentation, and that tight control be kept on the particle sizes that are dealt with in the sediments. Experiments should be completed with the measurement of aggregate size distributions only, and no measurements from which particle/aggregate sizes are inferred should be used. The magnitude of the flocculation effect does not appear to be large enough that experimentation can be completed without the use of this information as the basis of the study. Tight control of the shear to which aggregates are subject also needs to be ensured.  It is also recommended that further field work be under taken in the Fraser River. This should be completed in a more tightly controlled fashion, so that variations in aggregate size distributions in the river from day to day can be eliminated from analyses. Determination of the effect of cross-sectional shape, if any, on in situ aggregate size distribution will also be necessary for best analysis of field data. Finally, the collection of data regarding primary particle size distributions throughout the measurement period is also recommended. This will then assist in accurate evaluation of the in situ magnitude of the phenomenon.  113  The mechanism responsible for flocculation must also be determined. If this can be achieved, then it will be possible to develop some theory regarding the phenomenon. If this can be achieved, then the goal of modelling the process using simple theory such as Hunt's (1980) second order coagulation and settling theory will become more realistic.  114  8.0  REFERENCES  Baker, E.T., Milburn, H.B., and Tenant, D.A. (1988). Field assessment of sediment trap efficiency under varying flow conditions. Journal of Marine Research. 46. 573 - 592. Bale, A.J., and Morris, A.W. (1991). In situ size measurements of suspended particles in estuarine and coastal waters using laser diffraction. In Syvitski, J.P.M. (Ed.) Principles. Methods, and Application of Particle Size Analysis (pp. 197-208). Cambridge, Cambridge University Press. Bell, C.R., and Albright, L.J. (1981). Attached and free-floating bacteria in the Fraser River estuary, British Columbia, Canada. Marine Ecology - Progress Series. 6, 317-327. Benefield, L.D., Judkins, J.F., and Weand, B.L. (1982). Process Chemistry for Water and Wastewater Treatment. New Jersey, Prentice Hall. Biddanda, B.A. (1985). Microbial synthesis of macroparticulate matter. Marine Ecology - Progress Series. 20. 241251. Brown, G.M., Gregory, J., Jackson, P.J., Nelson, D.W., and Tomlinson, E.J. (1985). An on-line monitor for flocculation control. In Drake, R.A.R. (Ed.), Instrumentation and Control of Water and Wastewater Treatment and Transport Systems: Proceedings of the 4th IAWPRC Workshop held in Houston and Denver (pp. 239-245). New York, Pergamon Press. Burban, P., Lick, W., and Lick, J. (1989). The flocculation of fine-grained sediments in estuarine waters. Journal of Geophysical Research. 94 no. C6, 8323-8330. Chester, R. (1990). Marine Geochemistry. London: Unwin Hyman Ltd. Dorich, R.A., Nelson, D.W., and Sommers, L.E. (1984). Algal availability of phosphorous in suspended stream sediments of varying particle size. Journal of Environmental Quality. 13. 82-86. Droppo, I.G., and Ongley, E.D. (1989). Flocculation of suspended solids in southern Ontario rivers. In Sediment and the Environment. Proceedings of the Baltimore Symposium (IAHS Publication no. 184, 1989), Baltimore. Droppo, I.G., and Ongley, E.D. (1992). The state of suspended sediment in the freshwater fluvial environment: a method of analysis. Water Research. 26 (11. 65-72. Droppo, I.G., and Ongley, E.D. (1994). Flocculation of suspended sediment in rivers of southeastern Canada. Water Research. 28 (81. 1799-1809. Eaton, A.D., Clesceri, L.S., and Greenberg, A.E. (Eds.). (1995). Standard Methods for the Examination of Water and Wastewater. Washington D.C., American public Health Association. Environmental Quality Technical Working Group (EQTWG) of the Fraser River Action Plan. (1995). Measuring the health of the Fraser River. Environmental Quality Program. 1995 Status report. Vancouver, Nautilus Publications. Evans, W. and Hall, E.R. (1996). Fraser River pulp mill effluents: interpretation of Northwood effluent characterisation data. Vancouver, Department of Civil Engineering, University of B.C. Everett, D.H. (1994). Basic Principles of Colloid Science. Cambridge, The Royal Society of Chemistry. Farrow, J., and Warren, L. (1993). Measurement of the size of aggregates in suspension. In Dobias (Ed.) Coagulation and Flocculation: Theory and Applications (pp. 391-426). New York, Marcel Dekker Inc. 115  Fischer, H.B., List, E.J., Koh, R.C.J., Imberger, J., and Brooks, N.H. (1979). Mixing in Inland and Coastal Waters. London, Academic Press. Gailani, C , Ziegler, C.K., and Lick, W. (1991). Transport of suspended solids in the lower Fox River. Journal of Great Lakes Research. 17 (4), 479-494. Gibbs, R.J. and Konwar, L. (1982). Effect of pipetting on mineral floes. Environmental Science and Technology. 16,119. Gobas, F.A.P.C. (1996). Modelling the fate of contaminant discharges in the Fraser River basin. In Proceedings of the 3rd DOE research workshop. Vancouver. Grabemann, I. and Krause, G. (1989). Transport processed of suspended matter derived from time series in a tidal estuary. Journal of Geophysical Research. 94 (CIO). 14373 - 14379. Hall E.R., and Liver, S.F. (1996). Interactions of resin acids with aerobic and anaerobic biomass - II. Partitioning on biosolids. Water Research. 30 (3), 672 - 678. Hawley, N. (1988). Flow in cylindrical sediment traps. Journal of Great Lakes Research. 14 (1). 76 - 88. Hildebrand, F.B. (1987). Introduction to Numerical Analysis. New York, Dover Publishing. Holman, H.N. (1986). Particle Coagulation in a Turbulent Plume (PhD thesis). University of California, Berkeley. Horowitz, A.J., and Elrick, K.A. (1987). The relation of stream sediment surface area, grain size and composition to trace element chemistry. Applied Geochemistry. 2. 437-451. Hudson, H.E. Jr. (1981). Water Clarification Processes - Practical Design and Evaluation. New York, Van Nostrand Reinhold Company. Hunt, J.R. (1980). Coagulation in Continuous Particle Size Distributions; Theory and Experimental Verification (PhD thesis). California Institute of Technology. Hunt, J.R. (1982). Particle dynamics in sea water: implications for predicting the fate of discharged particles. Environmental Science and Technology. 16. 303-309. Hunt, J.R. (1990). Particle removal by coagulation and settling from a waste plume. In Baumgartner, D.J., and Duedall, I.W. (Eds.), Oceanic Processes in Marine Pollution (pp 109-118). Malabar, Florida; Robert E. Krieger Publishing Company. Irvine, K.N., Pettibone, G.W., Droppo, I.G., and Atkinson, J.F. (1995). Comment on "Linked sediment/contaminant transport model for rivers with application to the Buffalo River, New York" (J. Great Lakes Res. 20: 671 - 682). Journal of Great Lakes Research. 21 (3), 402 - 404. Karickhoff, S.W., (1981). Semi-empirical estimation of sorption of hydrophobic pollutants on natural sediments and soils. Chemosphere. 10. 833-846. Kim, S.D., Baker, C.G.J., and Bergougnou, M.A. (1977). Bubble characteristics in three phase fluidized beds. Chemical Engineering Science. 32. 1299 - 1306. Kranck, K. (1979). Dynamics and distribution of suspended particulate matter in the St Lawrence estuary. Naturaliste Canadien. 106. 163-173. Krishnappan, B.G. (1990). Modelling of settling and flocculation of fine sediments in still water. Canadian Journal of Civil Engineering, 17 (5), 763-770. 116  Krishnappan, B.G., and Engel, P. (1994). Critical Shear Stresses for Erosion and Deposition of Fine Suspended Sediments in the Fraser River. Burlington, National Water Research Institute. Krishnappan, B.G., Stephens, R., Kraft, J.A., and Moore, B.H. (1994). Size Distribution of Suspended Particles in the Athabasca River near Hinton. National Water Research Institute, Canadian Centre for Inland Waters. Landahl, M.T. and Mollo-Christensen, E. (1987). Cambridge: Cambridge University Press.  Turbulence and Random Processes in Fluid Mechanics.  Lau, Y.L. (1994). Temperature effect on settling velocity and deposition of cohesive sediments. Hydraulic Research. 32, (1), 41 - 51.  Journal of  Lau, Y.L., and Krishnappan, B.G. (1994). Does reentrainment occur during cohesive sediment settling? Journal of Hydraulic Engineering. 120 (2), 236 - 244. Lavelle, J.W. (1993). A model for estuarine sedimentation involving marine snow. In Mehta, A.J. (Ed.) Nearshore and Estuarine Cohesive Sediment Transport (pp. 148-166). Washington DC, American Geophysical Union. Lick, W. (1988). Modelling the transport of fine-grained sediments in aquatic systems. The Science of the Total Environment. 55. 219 - 228. Lick, W., and Huang, H. (1993). Flocculation and the physical properties of floes. In Mehta, A.J. (Ed.) Nearshore and Estuarine Cohesive Sediment Transport (pp. 21-39). Washington DC, American Geophysical Union. Lick, W., Huang, H. and Jepsen, R. (1993 (b)). Flocculation of fine-grained sediments due to differential settling. Journal or Geophysical Research, 98. C6, 10,279 - 10,288. Lick, W., and Lick, J. (1988). Aggregation and disaggregation of fine-grained lake sediments. Journal of Great Lakes Research. 14 (4), 514-523. Lick, W., Ziegler, C.K., Lick, J., and Joshi, A (1993 (a)). Effects of flocculation on particle transport. In Spaulding, M.L., Bedford, K., Blumberg, A., Cheng, R., Swanson, C. (Eds) (1994) Coastal and Estuarine Modelling. 3. Proceedings from the third International Conference. New York: American Society of Civil Engineers. Marks, B. (1996). Initial Dilution of a Horizontal Jet in a Strong Current. Draft M.A.Sc. thesis. University of British Columbia, Vancouver. Massey, B.S. (1983). Mechanics of Fluids: 5th Edition. London, Van Nostrand Reinhold. Metcalf and Eddy, Inc. (1993). Wastewater Engineering: Treatment. Disposal and Reuse. 3rd edition, McGraw-Hill Inc. Miihle, K. (1993). Floe stability in laminar and turbulent flow. In Dobias (Ed.) Coagulation and Flocculation: Theory and Applications (pp. 355-390). New York, Marcel Dekker Inc. Muschenheim, D.K., Kepay, P.E., and Kranck, K. (1989). Microbial growth in turbulent suspension and its relation to marine aggregate formation. Netherlands Journal of Sea Research. 23 (3), 283-292. Ng, B., Turner, A., Tyler, A.O., Falconer, R.A., and Millward, G.E. (1996). Modelling contaminant geochemistry in estuaries. Water Research. 30 (1), 63 - 74. Northwest Hydraulic Consultants Ltd. (1993). Determination of Sediment Deposition Zones Fraser River Basin. Vancouver. Ongley, E.D., Krishnappan, B.G., Droppo, I.G., Rao, S.S., and Maguire, R.J. (1992). Cohesive sediment transport: emerging issues for toxic chemical management. Hydrobiologia. 235/236. 177-187. 117  Paerl, H.W. (1974). Bacterial uptake of dissolved organic matter in relation to detrital aggregation in marine and freshwater systems. Limnology and Oceanography. 19. 966-972. Partheniades, E. (1993). Turbulence, flocculation and cohesive sediment dynamics. In Mehta, A.J. (Ed.) Nearshore and Estuarine Cohesive Sediment Transport (pp. 40-59). Washington DC, American Geophysical Union. Pearson, H.J., Valioulis, I.A., and List, E.J. (1984). Monte-carlo simulation of coagulation in discrete particle-size distributions. Parti. Brownian motion and fluid shearing. Journal of Fluid Mechanics. 143 367-385. Petticrew, E.L. (in press). Sediment aggregation and transport in northern interior British Columbian streams, in Erosion and Sediment Yield: Global and Regional Perspectives. Walling, D.E., and Webb, B.W. (Eds). International Association of Hydrological Sciences Publishing. Phillips, J.M., and Walling, D.E. (1995). An assessment of the effects of sample collection, storage and resuspension on the representativeness of measurements of the effective particle size distribution of fluvial suspended sediment. Water Research. 29 (11), 2498-2508. Prahacs, S.M. (1994). An Evaluation of Leeches as In situ Biomonitors of Chlorinated Phenolic Compounds Discharged from Bleached Kraft Pulp Mills. M.Sc. thesis, University of British Columbia, Vancouver. Riley, G.A. (1963). Organic aggregates in seawater, and the dynamics of their formation and utilisation. Limnology and Oceanography. 8. 372-381. Sekela, M., Brewer, R. Baldazzi, C , and Moyle, G. (1995). Change in contaminant concentration in Fraser River suspended sediments and water during the onset of freshet (Marguerite - 1993) (DRAFT REPORT). North Vancouver: Aquatic Section, Science Division, Environmental Conservation Branch, Pacific and Yukon Region, Environment Canada. Sherman, I. (1953). Flocculent structure of sediment suspended in Lake Mead. Transactions of the American Geophysical Union. 34. 394-406. Stone, M . and Mudrefoch, A. (1989). The effect of particle size, chemistry and mineralogy of river sediments on phosphate adsorption. Envir. Technol. Lett.. 10. 501-510. Streeter, V.L., and Wylie, E.B. (1985). Fluid Mechanics. Mc-Graw Hill Inc, USA. Tambo, N., and Watanabe, Y. (1979). Physical characteristics of floes. The floe density function and aluminium floe. Water Research. 13. 409-419. Teeter, A.M. (1993). Suspended transport and sediment-size transport effects in a well-mixed, meso-tidal estuary. In Mehta, A.J. (Ed.) Nearshore and Estuarine Cohesive Sediment Transport (pp. 148-166). Washington, D.C., American Geophysical Union. Terzaghi, K. and Peck, R.B. (1960). Soil Mechanics in Engineering Practice. New York: John Wiley and Sons, Inc. Thomas, D.G. (1964). Turbulent disruption of floes in small particle size suspensions. Journal of the American Institute of Chemical Engineers. 10. (4), 517-523. Treweek, G.P., and Morgan, J.J. (1977). Size distribution of flocculated particles: application of electronic particle counters. Environmental Science and Technology. 11. (7), 707-714. Treweek, G.P., and Morgan, J.J. (1980). Prediction of suspension turbidities from aggregate size distribution. In Kavanaugh, M.C., and Leckie, J.O. (Eds.) Particulates in Water: Characterisation. Fate. Effects, and Removal (pp. 329-352). Washington, D.C., American Chemical Society. Tsai, C.H., Iacobellis, S., and Lick, W. (1987). Flocculation offine-grainedlake sediments due to a uniform shear stress. Journal of Great Lakes Research. 13 (2). 135-146. 118  van Olphen, H. (1977). An Introduction to Clay Colloid Chemistry. New York, John Wiley and Sons. Vine, J. (1996). Random Walk Models Applicable to Rivers. Draft M.A.Sc. thesis, University of British Columbia, Vancouver. Woodward, J.C., and Walling, D.E. (1992). A field sampling method to obtain representative samples of composite fluvial suspended sediment particles for SEM analysis. Journal of Sediment Petrology. 64 742-744. Zrymiak, P., and Tassone, B. (1986). Sediment station analysis: Fraser River at Hansard. Environment Canada, Water Resources Branch, Inland Waters Directorate.  119  APPENDIX A: CALCULATION OF TURBULENT VELOCITY GRADIENTS AND KOLMOGOROFF MICROSCALES IN THE FRASER RIVER There are two important zones in an open channel flow where it is possible to calculate the turbulent shear velocity and the Kolmogoroff microscale. The first zone comprises the bulk of the flow but excludes the viscous sub-layer near the bed. The second zone is comprised of only the viscous sub-layer, where turbulent energy dissipation reaches its peak, and velocity gradients are much higher (Partheniades, 1993).  In the portion of the flow far from the bed, the average rate of turbulent energy dissipation (e ) in an open channel av  can be evaluated using the relationship,  where g represents gravity, V represents the velocity in the channel, and S represents the slope of the energy grade e  line (Partheniades, 1993). It is then a simple matter to evaluate turbulent velocity gradient (G) using the equation  as initially introduced by Camp and Stein (1943), and cited by Partheniades (1993). The symbol v represents the kinematic viscosity of water, which in calculations was assumed to be 1 x 10" m /s. Finally, it is possible to 6  2  calculate the Kolmogoroff microscale (KQ) using the following formulation (Partheniades, 1993).  K=  VV V  / 4  (A-3)  J  In calculations made using this theory (see §3.1 and §6.3.3), the following assumptions were made. •  The energy grade line can be easily determined by assuming uniform flow, and hence is equal to the bed slope of the river. The bed slope of the Fraser River at Northwood was determined from profiles (Northwest Hydraulic Consultants, 1993) to be 425 x IO" m/m. 6  •  Velocities were either obtained from the vertically averaged velocities measured in each transect by Krishnappan (1996, unpublished data), or calculated assuming a Manning's n of 0.02 and flow depth of 2.15 m as per Vine (1996).  120  In the zone comprising the viscous sub-layer the turbulent energy dissipation (e ) is calculated differently v  (Partheniades, 1993). The shear velocity (w*) of the flow must first be calculated using the following relationship, "* = ylgy Se 0  (A-4)  where g is gravity, and y is the depth of flow. This can then be substituted into an expression for turbulent energy 0  dissipation (ev) (Partheniades, 1993).  (A-5) V  v  J  Once Ev has been obtained, then Equations A-2 and A-3 can be used to obtain G and Xo respectively for this zone of the flow.  121  APPENDIX B: CALCULATION OF VELOCITY GRADIENTS IN SETTLING COLUMN The measurement of velocity gradients in the column while mixing with air bubbles is under way would be a major undertaking. Instead, if an estimate was obtained of the velocity, size, and occurrence of bubbles in the column then velocity gradients can be calculated. This is not a simple matter, and is presently an area of research in civil, and chemical engineering, and applied mathematics. A reasonable approximation to these parameters can be gained with the use of theory developed for fluidized bed reactors. There are however, a number of short comings of the theory with regard to the application required.  From the work of Kim et al. (1977) the following correlation was derived for bubble diameters (d ): B  d = 10.2U; U° y om  B  -° a  354  027  g  (B-i)  0163  where the symbols are defined as follows: Ui  —> the superficial velocity of the fluid, with units mm/s,  U  the superficial velocity of the gas, with units mm/s,  a  — ¥ the surface tension of the fluid, with units dyne/cm, and  y  -» generalised liquid viscosity constant, with units mNs/m .  g  2  The generalised liquid viscosity constant must also be defined in terms of other parameters: Y = rC8  (B-2)  n _ 1  where k represents the fluid consistency or fluid viscosity (mNs7m ), and n is the fluid behaviour index. 2  From Kim et al. (1977) then, the surface tension of water at 20 °C is given as being 72.8 dynes/cm. For the purposes of calculation, it has been assumed to be 70 dynes/cm, because without measurement, it is impossible to know the effect that the addition of the pulp mill effluent and the solids have on the surface tension of the mixture. The superficial gas velocity can be evaluated based on the gas flow in the column (Q ) of 835 mL/min, and the internal g  cross-sectional area of the column (A ). c  122  _  835 cm / min  ~n X13.9 /4 cm 2  2  = 5.50 cm / min This then equates to a superficial gas velocity of 0.9 mm/s. From Table 1 in Kim et al. (1977) k and n are given as being 1.00 for water. Substitution into equation B-2 then gives y = 1 mNs/m . We can then write equation B - l in 2  terms of the superficial liquid velocity which is zero, but if treated as such will result in undefined values from the correlation equation.  d = B  io.2u? u y-° <3 l38  021  0163  g  = 10.2 x £/,-° = 19.81/,-°  0 354  138  x 0.92  0354  x LO-  x 70°  0027  163  138  After trying some values for superficial liquid velocity, it seems reasonable to settle on a value of 0.05 mm/min, which yields a bubble diameter of 30 mm. It is then necessary to determine the rise velocity of the bubbles (U ), and bT  this can be done with the use of a second correlation determined by Kim et al. (1977). U  br  = 253L/^  1 3 3  C/ °-  3 4 ,  S  Y -°-  0  2  6  G  0  1  5  7  (B-3)  Using the same assumed superficial fluid velocity, we obtain that the rise velocity must be 713 mm/s. A value of 0.7 m/s will be assumed for the purpose of further calculations.  Using these assumed values for bubble diameter, and rise velocity, we can make a calculation for the velocity gradient in the column using equation 2.11. One of the terms in this expression is the power imparted to the fluid. For the situation where bubbles are doing the mixing of the fluid in the column, power (P) can be expressed using the form:  P = \c p,A U  3  d  B  br  (B-4)  where Q represents the drag coefficient of the bubble, and A represents the cross-sectional area of the bubble. This B  assumes that the bubbles formed are spherical in shape, which from observation is not completely accurate, but for the sake of this calculation is a reasonable assumption. By evaluating the bubble Reynolds number, it is possible to  123  obtain the drag coefficient using Massey (1983) to be 0.45. Substituting this information into equation 2.11, and assuming that the volume of the fluid is 20 L, we obtain:  C p,A U^  r  G=  D  B  2\iV 0.5 0.45 x 1000 x (rt x 0.03 / 4) x 0.7 3 A 2  2xlxl0"  6  x20xl0"  3  J  = 1650*-' so that it would be reasonable to assume that the shear in the column under these conditions must be of the order of 1000 s". 1  124  APPENDIX TURBIDITY  C: RESULTS COMPLETED  FROM  SETTLING  AT THE  TESTS  MEASURING  UNBC  16.0  •a  n  0.00  0.50  1.00  1.50 Log(time(min)) - • — A  2.00  2.50  3.00  2.50  3.00  - - 0 - - B  Figure C-l. Plot of turbidity variation during settling tests A and B. 30.0  3 20.0  0.00  0.50  1.00  1.50 Log(time(min))  2.00  —C --O--D Figure C-2. Plot of turbidity variation during settling tests C and D.  125  0.00  0.50  1.00  1.50  2.00  2.50  3.00  3.50  Log(time(min)) -»—E - - 0 - - F  Fisure C-3. Plot of turbidity variation during settling tests E and F. 50.0 lt=-"  « ft^fc-—  40.0  >• 30.0 •• ..  ^  I— 20.0 * >.  (k*°  10.0  0.0 0.00  ~ A - -  0.50  1.00  1.50  2.00  2.50  Log(time(min)) •  G•  •O  ••H -  -A-  •I  Figure C-4. Plot of turbidity variation during settling tests G, H, I and J.  126  3.00  3.50  Figure C-6. Settling velocity distributions obtained from tests A and B.  Ill  on n  o  yu.u^  at  o  (0  E 9>  '  'S  >• /U.U  n  ou.u  4) O 0)  on n ou.u  O  Q.  4^e-  0.01  0.10  1.00  10.00  100.00  Settling Velocity (cm/min)  —C -  -O  -D  Fisure C-7. Settling velocity distributions obtained from tests C and D.  o- -n ^ on  o  0  _>4  o  v t —  .  yu.u  O)  c c "5 E  £O.U  0)  'TS  !5  OU.U  3 C  o  o  on n  ou.u  I.  0)  0.  0.01  0.10  1.00  10.00  Settling velocity (cm/min)  - • — E - -o - F Figure C-8. Settling velocity distributions obtained from tests E and F.  128  100.00  Fisure C-10. Settling velocity distributions obtained from tests K, L, M and N.  129  APPENDIX  D:  RESULTS  OF MIXING  EXPERIMENTS  MEASURING  TURBIDITY  Table D-l. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  = 58 mg/L,  River  Turb j R  ver  = 31 NTU, and TSS Egiuem  mg/L, and Turb fp , = 6.0 NTU. E  Effluent  -  39.7  mn  River water Measured  Predicted  Percent  Volume  Volume  Turbidity  Turbidity  turbidity  (mL)  (mL)  (NTU)  (NTU)  reduction  800  200  7  11.0  36.4  750  250  8  12.3  34.7  666  334  7.5  14.3  47.6  500  500  16  18.5  13.5  333  667  21  22.7  7.3  250  750  25  24.8  -1.0  200  800  24  26.0  7.9  ..  Table D-2. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  = 58 mg/L, Turb  Distilled H 0  River water  Measured  Predicted  Percent  Volume  Volume  Turbidity  Turbidity  turbidity  (mL)  (mL)  (NTU)  (NTU)  reduction  800  200  6.5 .:  6.2  -4.8  750  250  8.5  7.75  -9.7  666  334  9.5  10.3  7.8  500  500  15  15.5  3.2  333  667  21  20.6  -1.9  250  750  25  23.25  -7.5  200  800  25  24.8  -0.8  2  130  River  = 31 NTU.  Table D-3. Summary of volumes used, measured, turbidities, predicted turbidities and calculated relative percen reduction in turbidity for tests with initial parameters TSS  River  mg/L,  and  TurbEgtuent = 5.5  = 141 mg/L, Turb  River  = 52 NTU, and TSS 40  Effluent  =  NTU.  Effluent  River water  Measured  Predicted  Percent  Volume  Volume  Turbidity  Turbidity  turbidity  (mL)  (mL)  (NTU) '  (NTU)  reduction  800  200  16  14.8  -8.1  750  250  15  17.1  12.3  666  334  18  21  14.3  500  500  24  28.8  16.7  333  667  32.5  36.5  11.0  250  750  35  40.4  13.4  200  800  38  42.7  11.0  Table D-4. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS j = 141 mg/L, and Turb j — 52 NTU. R ver  Distilled H 0  River water  Measured  Predicted  Percent  Volume  Volume  Turbidity  Turbidity  turbidity  (mL)  (mL)  (NTU)  (NTU)  reduction  800  200  12  10.4  -15.4  750  250  14  13  -7.7  666  334  18  17.3  -4.0  500  500  25  26  3.8  333  667  35  34.6  -1.2  250  750  40  39  -2.6  200  800  44  41.6  -5.8  2  131  R ver  Table D-5. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  = 286 mg/L, Turb  = 95 NTU, and TSS  River  Effluent -  47.8 mg/L, and Turb ffl em = 5.2 NTU. E  Effluent  U  River water Measured  Predicted  Percent  Volume  Volume  Turbidity  Turbidity  turbidity  (mL)  (mL)  (NTU)  (NTU)  reduction  800  200  20  23.4  14.5  750  250  22  27.9  21.1  666  334  30  35.3  15.0  500  500  36  50.2  28.3  333  667  54  65.1  17.1  250  750  64  72.6  11.8  200  800  67  77.1  13.1  600  400  33  41.3  20.1  400  600  46  59.2  22.3  Table D-6. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSS  River  Distilled H 0 2  River water Measured  = 286 mg/L, and Turb  Predicted  Percent  Volume  Volume  Turbidity  Turbidity  turbidity  (mL)  (mL)  (NTU)  (NTU)  reduction  800  200  21  19  -10.5  750  250  24  23.7  -1.1  666  334  32  31.6  -1.3  500  500  50  47.5  -5.3  333  667  66  63.3  -4.3  250  750  73  71.2  -2.5  200  800  76  77.0  1.3  132  River  = 95 NTU.  Table D-7. Summary of volumes used, measured turbidities, predicted turbidities and calculated relative percent reduction in turbidity for tests with initial parameters TSSm = 156 mg/L, Turb/a = 62 NTU, and TSS ver  40.2 mg/L, and TurbEffiuem = 5.0 NTU. Effluent  River water  Measured  Predicted  Percent  Volume  Volume  Turbidity  Turbidity  turbidity  (mL)  (mL)  (NTU)  (NTU)  reduction  800  200  16.0  16.4  2.4  750  250  17.0  19.3  11.7  666  334  21.0  24.0  12.6  600  400  24.0  27.8  13.7  500  500  27.0  33.5  19.4  400  600  31.0  39.2  20.9  333  667  33.0  43.0  23.3  250  750  41.0  47.8  14.1  200  800  43.5  50.6  14.0  100  900  53.0  56.3  5.9  133  ver  Effluent -  APPENDIX E: RESULTS OF SETTLING COLUMN EXPERIMENTS MEASURING TSS  134  52.9  Tf  VD  64.4 60.6 56.7  >/-> r-  64.1 41.7  >n oo TtTf  vo  r-  cn cn  oo TT Tf TT  co  CN CN  SO  oo  o  oo  TT  CN  135 m  CN  TT  u->  VD  Tf  o CN  CO CN  CN  VD  CO •t  •*  TT  Tt-  TT  CN  Tt  co  u->  1447  t  vo  Ul  CN  1465  s cn  64.4 60.6 >/->  26.1  vo t--  68.3  52.9  72.1  oo  r-  co  VO 00  vo 1447  r-  56.7  CN  CO CN  52.9  vo oo od  35.7  o\  t-  3575  56.7  14.7  VD  68.3  d  25.6  60.6  18.6  oo  78.8  ON  38.4  64.4  21.5  >o  72.1  r-  65.5  68.3  18.9  r-  79.4  VD  104.5  72.1  21.6  cn  80.6  od co  106.2  B,  21.6  i>  od  79.7  S,  OA  104.5  (cm)  (min)  (cm)  (mg/L)  SS cc Ov  (min)  (min)  (cm)  (mg/L)  (min)  S. H.  TS  Time  S. H.  TSS  Tesl  Time  TestC  s  TSS  TestB Q  (cm)  Time  S. H.  TSS  Time  Test A  /  ON  TT  r~  ON  TT  CN  o\  TT  VO  CN >/->  oo •*  o  1453  66.3  38.3  26.6 1453  CO CO  p-  pd VO  cs  vq  ON  m  r> m  p-  42.8  CO  55.8  co P~  oo oo  CO  o  CO  m  00  m  in  vd  d VO  od m  23.8  VO  co  cs  co  co pco  co co p~  1453  vo  77.6  co  in  vd  CO  70.6  CO  p-  m  10.4  ON  in in  Tj-  1453  CO ON  oo  1453  co o  oo  oo  co VO  112.5  00  vq  CO  122.9  in  od m  Tt  o  122.7  ON  CO  VO  128.1  VO  CO ON  p-  136  00  VD  O  in VO  in  in  CO  CO  cs p-^ VO  co  o  CO  oo  ON  CO CO  co p~  00  pd p-  CO CS  cs o cs  co p~  in cs p-  m  CO  CO  Tf  p-  O  VO  vo  oo  _  oo oo  vo  <n  -  vd  co cs VO  CO  ON  od in  c VO  00  CO  00  vo  VO  p-  vo  d  VO  16.9  vq  d  co p-  OO  p~  25.5  CN  p-  CO  in co VO  co  1230  VO  CO CO  27.7  26.2 co m  ON CN  24.9  29.6  VO  ON ON  ON  in d cs  82.6  cs  VO  co  66.8  cs  00  co o  r-  79.7  co co  in  vd  94.1  78.3  co cs  cs od  76.9  81.5  oo  co p-  108.5  81.6  84.5  oo  r-  co VO  co  (mg/L)  (min)  co CO  in  VO  (min)  (cm)  TSS Time Time  TSS  tz5  p-  VO  vo  CO  vo  CO CO  in  ON  ON CS  10Z.  33  -  ON  ON  vd  76.8  32.32  in VO p-  X  cs  ON  32.2  32.6  (cm)  124.1  CO  32.2  (mg/L) (min)  oo  (mg/L)  TSS Time  p-  p-  in od  vo  vo  Tj-  97.9  vq  oo oo  m  m p^ m  pcs  CO  73.4  CO CO  co p-  CO  (  100.8  m p-  VO  oo  vo  ^ ON  cs od m  oo  26.5  X  oo  VO d  102.1  CO  CO  d  00  vo  cs VO  24.5  co  CS  „  co  107.1  co cs  ON  in  Tf CO  110.5  oo  CO  in  CO  30.7  P-  P-  od  VO  P-;  23.9  vo  CS  ON  vo  cs in VO  41.2  co in p-  123.9  CO  125.9  CO  co  121.8  CO  cs  (min)  CO  (cm)  00  (mg/L)  ON  (min)  ON  (cm)  ON  oo in oo  TS  vq  Time  co  S. H.  VO  TSS  vo  00  vd  57.3  CS  r-  p-  co od  113.7  (cm)  S. H.  ON  Time  TestD TestC  00  CS  p~  «3  TestB  ON  oo  t/5  Test A  oo  in TJ-' p-  Sui)  Tesi  w  VO p-  ON  74.4 72.8 71.2 69.7 68.1 66.5 64.9 63.3 61.8 60.2 58.6  22.2 22.8 23.3 25.7 23.1 25.1  23.6  14.2  28.9 28.3 28.2 31.1 27.9 32.1 27.3 27.8 27.2 22.1 16.1 12.9  so cn  cn rcn oo cn O  o  CN CN cn e'en cn cn r-  1453  (cm)  (mg/L)  (mg/L)  23.5  21.2  15.7  11.9  10.5  57.4  47.1  31.4  24.3  19.1  69.6  00  rCN  25.2  71.2  CN  58.1  cn  26.3  cn  65.7  cn cn  25.8  m CN  67.5  31.2  oo  27.7  in  74.5  44.6  o  72.8 cn CN  68.8 67.2 65.7 64.2 62.6 61.1  rCN CN CN CN cn  -0.9  -  25.4 20.6 18.9 20.5  so cn  cn roo oo cn  cn CN  cn cn  cn  cn  cn  cn  oo oo  cn O  cn  cn rcn  o o  TT  137  IT) NO  oo  NO  r-  o 11.5  ON  ON  58.5  59.5  CN CN Tt  60.1  r-; cn  70.3  Tt-  61.7  71.9  oo </->  in  cn  63.2  r-  CN  64.8  </-> Tf  66.4  in  73.4  CN  77.7  46.4  NO  74.3  10.8  CN  29.3  r-  86.2  61.1  75.9  25.9  31.7  m  76.5  29.9  cn  81.4  cn cn  S. H. (cm)  (mg/L)  (mg/L)  cn CN  82.5  ON  111  29.4  123.9  77.5  (mg/L)  (mg/L  (min)  oo  TSS  TSS  Time  -  en  31.9  (cm)  VSS  TesitC  od  ON  106.8  Time  X  c«  TSS  TestB  cn CN  NO  cn od SO  cn  cn rcn  OIL  Time (UIUI)  Test A  cn CN  (UIUI)  SSA i/->  SSA I w3 r~ in  oo d  ON  r-  X  a  09  in  > s H  in in p-  ON  Ti-  CO  cs  oo d  p-  cs  vd  vd  CO  ON  ^  Vi Vi  vi vi  pr-  ON  CS  cs  4  & 1 H  3  X  g  Vi  vi  co  cs  >  S  Vi Vi  i  H  E  in in  oo  CO  co  CO CO  CO  co m  rt  cs  CO  oo cs  p~  Tt  ON  oo  Pp-  P-^  p~  in in r-  CO  T*  Tf  CS  p*  cs  d  CN CO  _ _  ON  cs cs co  od cs  CO  p-  pd  p-  p-  ON  co cs  CO  co  co m  p-  vd  p-  >—'  '  od cs  cs  co  pO  ON  CN OO  co CN  co co  CO  Tl  vo  co  CO  cs  CO  d  CS  p-  CO vd cs  VO  vo  00  in  — rf  CO  oo  T]  vq  00  co  cs  co o  co  CO CO PCO  co  co  p-  Tf  Tf  r-; od  pin  in  cs d  ON  vo  vo  in  p-  Tt  ON  cs od VO  in  cs  CO  CO  CS  CS  d  VO  cs  d  ON  P-;  oo  CS  co  CS  CO  00 00  p-  oo  co  vd vo  in  CO in  cs  oo  r>  00  co  Tl  cs  in in  VO  cs  od  od cs  ON  VO  r~  co  ON VO  vd  O  VO  VO  P-; P^  ON  in Tf  p-  cs  ON  Ti  p~  CS  cs  T,  cs  cs  cs  00  cs  00  oo cs  P-;  vq  ON  co  ON VO  p~  3  VO  VO  co  cs  vd vo  p-  CO  vq  in p-  VO  cs  CS  cs  oo  ON VO  in  oo cs  00  vd  p~  P-;  Ti  2  p-  cs  co cs  -a  P-  cs  od cs  S  P-  co  vo  CO o  ON  co  co ON  —i  in  d  VO  m co  OO  Tt  VO  vd  CO co  r-  in  cs  ON  vo  vo  oo d  CO vd  oo  Tt  — d co  in  T)  CO  ON  CO CS  CO  ON  cs  CN  m  <—'  «5  H  vq od  P-  •<fr  138  co  co  in  oo  VO  VO  co  Ti  VO  P-  vd  •>*.  co od  o  CO  CO  00  in  ON  in  o  CO  ON  —  OO  co  d  CN  co  p-  APPENDIX F: CROSS-SECTIONS IN THE VICINITY OF THE NORTHWOOD DIFFUSER  -6 + ^  -8 -  > -10-12 -14 --16 -18 J  —  :  —  X(m)  Figure F-l. Cross-section and vertically averaged velocities 2000 m upstream of the diffuser.  139  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0050356/manifest

Comment

Related Items