Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Factors influencing oxygen transfer in diffused aeration systems and their application to hypolimnetic… Ashley, Kenneth Ian 1989

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

Item Metadata

Download

Media
831-UBC_1989_A7 A83.pdf [ 6.36MB ]
Metadata
JSON: 831-1.0062903.json
JSON-LD: 831-1.0062903-ld.json
RDF/XML (Pretty): 831-1.0062903-rdf.xml
RDF/JSON: 831-1.0062903-rdf.json
Turtle: 831-1.0062903-turtle.txt
N-Triples: 831-1.0062903-rdf-ntriples.txt
Original Record: 831-1.0062903-source.json
Full Text
831-1.0062903-fulltext.txt
Citation
831-1.0062903.ris

Full Text

FACTORS INFLUENCING O X Y G E N TRANSFER IN DIFFUSED AERATION SYSTEMS AND THEIR APPLICATION TO HYPOLIMNETIC AERATION  By Kenneth Ian Ashley B. Sc. (Zoology) University of British Columbia, 1976 M. Sc. (Zoology) University of British Columbia, 1981 A  THESIS  THE  SUBMITTED  IN P A R T I A L  REQUIREMENTS  MASTER  OF  FOR  FULFILLMENT  THE  APPLIED  DEGREE  OF  SCIENCE  in THE  FACULTY  OF  CIVIL  GRADUATE  STUDIES  ENGINEERING  We accept this thesis as conforming to the required standard  THE  UNIVERSITY  OF  BRITISH  COLUMBIA  June 1989 © Kenneth Ian Ashley, 1989  OF  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 department  or  by  his or  her  representatives.  be  It is understood  publication of this thesis for financial gain shall not be permission.  Department of  Civil  Engineering  The University of British Columbia Vancouver, Canada Date  DE-6  (2/88)  granted by  the head of  my  that  or  copying  allowed without my  written  Abstract  A series of laboratory and field experiments were conducted to examine the effect of several design variables on the oxygenation capacity of hypolimnetic aeration systems. The laboratory experiments used non-steady state gas transfer methodology to examine the effect of air flow rate, air flow rate per diffuser, orifice size and reduced tank surface area on the overall oxygen transfer coefficient  standard oxygen  (.Kz, 20) hr ); -1  A  transfer rate ( 0 T , g Oa/hr); energy efficiency ( E , g C^/kW-hr) and transfer efficiency P  5  ( E , % ) . The field experiments examined the effect of diffuser depth, orifice size and D  reduced separator box surface area on the oxygen input per cycle (mg/L), daily oxygen load (kg C^/day), transfer efficiency ( E , % ) , energy efficiency ( E , kg 02/kW-hr) and 0  P  water velocity (m/sec) in a full lift hypolimnetic aerator. The laboratory experiments demonstrated that -Kz,a 0) OT,, E and E increased with air flow rate in the orifice range 2  P  D  of 397 fi to 3175 \L diameter. In the 40 fi and 140 ft diameter orifice range,  KLO, O 2  and  OT increased with air flow rate; however, E and E were not affected. A decrease in D  a  P  orifice size from 3175 fi to 140 /z diameter increased  KLCL  2 0  , OT , t  E  P  and  E ; D  however,  there was no significant difference between the 140 fi and 40 fi diameter silica glass diffusers. Reducing the air flow rate per silica glass diffuser (40 fi and 140 fi diameter) significantly increased minimal effect on K L < I  Kia , 2Q  2  O,  OT , s  E  P  and  E . A C  reduction in tank surface area had a  OT„, E and E in two tank configurations with different surp  D  face area to volume ratios (0.94 and 2.2 m ). Thefieldexperiments demonstrated that -1  increased depth of air release increased the oxygen input per cycle and water velocity, which, in turn increased the daily oxygen load, E and E „ . Orifice size in the 140 fi range P  significantly increased oxygen input per cycle, daily O 2 load, E and E ; however, the size P  11  D  range from 794 u to 3175 a exhibited similar but reduced gas transfer characteristics. A reduction in surface area in the separator box had no effect on the oxygenation capacity of the hypolimnetic aerator. Design criteria for hypolimnetic aerators are discussed including several modifications which should increase the oxygenation capacity of full lift hypolimnetic aeration systems.  m  Table of Contents  Abstract  ii  List of Tables  ix  List of Figures  xi  Acknowledgement  xii  1 Introduction  1  2 Methods  6  2.1 Lab Experiments  6  2.1.1 Air Supply  6  2.1.2 Air Flow Rate Measurement  6  2.1.3 Oxygen and Temperature Measurements  7  2.1.4 Deoxygenation-Oxygenation Procedure  7  2.1.5 Tank Size and Geometry  8  2.1.6 Diffuser Type and Orifice Size  8  2.1.7 Experimental Design  10  2.1.8 Bubble Size  14  2.2 Field Experiments  15  2.2.1 Air Supply  16  2.2.2 Air Flow Rate Measurement  16  2.2.3 Oxygen, Temperature and Current Measurements  16  iv  2.3  3  2.2.4  Oxygenation Procedure  17  2.2.5  Diffuser Type and Orifice Size  18  2.2.6  Experimental Design  20  Parameter Calculation  21  2.3.1  Lab Experiments  21  2.3.2  Field Experiments  25  2.3.3  Statistical Analysis  26  Results 3.1  3.2  3.3  28  Group 1: K a L  2 0  and OT„  28  3.1.1  Air Flow Rate  28  3.1.2  Orifice Size  30  3.1.3  Air Flow Rate by Size Interaction  30  3.1.4  Replication  30  3.1.5  Surface Cover  33  Group 1: E and E 0  33  p  3.2.1  Orifice Size  34  3.2.2  Air Flow Rate  34  3.2.3  Replication  37  3.2.4  Surface Cover  37  Group 2: K a L  2 0  and O T ,  37  3.3.1  Air Flow Rate  38  3.3.2  Number of Diffusers  38  3.3.3  Air Flow Rate by Number of Diffusers Interaction  41  3.3.4  Surface Cover  41  3.3.5  Orifice Size and Replicate Days  41  v  3.4 Group 2: E and E 0  3.4.1  41  p  Number of Diffusers  43  3.5 Group 3: K a and OT, L  43  20  3.5.1  Orifice Size  44  3.5.2  Surface Cover  44  3.6 Group 3: E and E D  45  p  3.6.1  Orifice Size  45  3.6.2  Surface Cover  45  3.7 Bubble Size  46  3.8 Group 4: Water Velocity, Oxygen Increase Per Cycle, Daily Oxygen Load, Ep  48  3.8.1  Depth of Air Diffuser  48  3.8.2  Orifice Size  48  3.8.3  Replicate Days and Interactions  52  E  0  and  3.9 Group 5: Water Velocity, Oxygen Increase Per Cycle, Daily Load, E and D  E  52  p  3.9.1  Orifice Size  52  3.9.2  Surface Cover, Replicate Days and Interaction  53  4 Discussion: Gas Transfer Theory  54  4.1 Gas Transfer Theory  54  4.2 Non-Steady State Reaeration Test  57  5 Discussion: Group 1-3 Laboratory Experiments  60  5.1 Group 1  61  5.1.1  Air Flow Rate  61  5.1.2  Orifice Size  64 vi  5.1.3  Surface Cover  66  5.1.4  Interaction  66  5.1.5  Replication  67  5.2 Group 2  67  5.2.1  Air Flow Rate  68  5.2.2  Number of Diffusers  69  5.2.3  Surface Cover  70  5.2.4  Orifice Size  70  5.2.5  Interaction  71  5.2.6  Replicate Days  72  5.3 Group 3  72  5.3.1  Orifice Size  72  5.3.2  Cover  72  5.4 Summary Analysis: Group 1-3  6  74  5.4.1  Comparison of Ki,a2o, OT„, E and E„  74  5.4.2  Ka  75  5.4.3  E  p  75  5.4.4  E„  76  5.4.5  Optimum Bubble Size  76  p  L  20  Discussion: Group 4 and 5 Field Hypolimnetic Aeration Experiments  80  6.1 Group 4 Field Experiments  80  6.1.1  Depth of Air Release  80  6.1.2  Orifice Size  82  6.2 Group 5 Field Experiments 6.2.1  85  Orifice Size  85 vu  6.2.2  Surface Cover  85  6.3 Summary Analysis: Group 4 and 5  86  6.3.1 Diffuser Depth  86  6.3.2  Surface Cover  87  6.3.3  Air Flow Rate  88  6.3.4  Air Flow Rate per Diffuser and Orifice Size  88  6.3.5  Comparison to Literature E  89  6.3.6  Retrofitting Undersized Systems  89  6.3.7  New Designs  90  p  7 Conclusions  93  Bibliography  96  Appendices  105  A Group 1 ANOVA Results: K a , O T , E and E  p  105  B Group 2 ANOVA Results: K a , O T , E and E  p  107  C Group 3 ANOVA Results: K a , O T „ E and E  p  110  L  L  L  2 0  s  2 0  a  2 0  0  0  0  D Group 4 Results: Water Velocity, Oxygen Input, Daily Oxygen Load, E and E 0  112  p  E Group 5 Results: Water Velocity, Oxygen Input, Daily Oxygen Load, E and E 0  114  p  vui  List of Tables  2.1 Airflowrate through an orifice at 2 psig (0.136 kg/cm )  9  2  2.2 Group 1 experimental treatments  11  2.3 Group 2 experimental treatments  12  2.4 Group 3 experimental treatments  13  2.5 Oxygen-temperature profiles at Black Lake during field experiments.  ..  17  2.6 Discharge of air through an orifice at 20 psig (1.4 kg/cm )  18  2.7 Group 4 experimental treatments  20  2.8 Group 5 experimental treatments  21  2  3.9 Effect of air flow rate on Group 1 Ki,a o and OT,  28  3.10 Effect of orifice size on Group 1 Kia  30  2  OT,  20  3.11 Effect of replication on Group 1 K£,a and OT,  33  3.12 Effect of surface cover on Group 1 Kia o and OT,  33  3.13 Effect of orifice size on Group 1 E and E  34  20  2  Q  p  3.14 Effect of air flow rate on Group 1 E and E D  3.15 Effect of surface cover on Group 1 E and E 0  34  p  37  p  3.16 Effect of airflowrate on Group 2 Kx,a and OT,  38  3.17 Effect of diffuser number on Group 2 Kt,a o and OT,  38  3.18 Effect of surface cover on Group 2 Kia and OT,  41  20  2  20  3.19 Effect of orifice size and replication on Group 2 K£a and OT,  43  3.20 Effect of diffuser number on Group 2 E and E  43  20  0  p  3.21 Effect of orifice size on Group 3 Ki,a o and OT, 2  ix  .  44  3.22 Effect of surface conditions on Group 3 Kj,a2o ^ 0T„  44  3.23 Effect of orifice size on Group 3 E and E  45  a n  D  p  3.24 Effect of surface conditions on Group 3 E and E D  46  p  3.25 Equivalent bubble diameter as a function of air flow and orifice size. . . .  46  3.26 Effect of diffuser depth on Group 4 (field experiments) water velocity, oxygen input and daily load  48  3.27 Effect of diffuser depth on Group 4 (field experiments) E and E D  p  48  3.28 Effect of orifice size on Group 4 (field experiments) water velocity, oxygen input and daily load  51  3.29 Effect of orifice size on Group 4 (field experiments) E and E D  p  51  3.30 Effect of orifice size on Group 5 (field experiments) water velocity, oxygen input and daily load  52  3.31 Effect of orifice size on Group 5 (field experiments) E and E D  x  p  52  List of Figures  3.1 Effect of airflowrate on Group 1 Kr,a o, E and E 2  3.2 Effect of orifice size on Group 1 Kjr,a  p  29  e  31  20  3.3 Group 1 airflowrate by size interaction effect for K£a  32  20  3.4 Effect of orifice size on Group 1 E  Q  35  3.5 Effect of orifice size on Group 1 E  p  36  3.6 Effect of airflowrate on Group 2 K^a^, E and E p  3.7 Effect of diffuser number on Group 2 Kia and E 20  39  D  40  D  3.8 Group 2 airflowrate by number of diffusers interaction effect on Kr,a . .  42  3.9 Equivalent bubble diameter as a function of air flow rate and orifice size.  47  20  3.10 Effect of diffuser depth on Group 4 (field experiments) water velocity, oxygen input and E  49  0  3.11 Effect of Group 4 (field experiments) orifice size on water velocity, oxygen input and E  50  D  5.12 Surface area vs bubble diameter  78  6.13 Conceptual drawing of co-current upflow and counter-current downflow hypolimnetic aerator  92  xi  Acknowledgement  This thesis project would not have been possible without the assistance of many individuals. I am especially grateful to Dr. K.J. Hall, Dr. D.S. Mavinic and Mr. B.R. Ward for their continuous support throughout this thesis program. Special thanks to Dr. W.K. Oldham for suggesting how to approach an engineering degree, giving thoroughly entertaining and informative lectures and providing excellent direction to the Civil Engineering Department. Thanks to Dr. G.A. Lawrence for deciphering various empirical mathematical formulae and providing advice on numerous occasions. Special thanks to Angelo Facchin for assistance with data analysis and computing services. Dr. J. Berkowitz (Co-ordinator of the Statistical Consulting and Research Laboratory) and Dr. N.T. Johnston provided advice on experimental design and statistical interpretation. Mr. M. Weis kindly provided the SEM's of the ceramic diffusors. This study was supported by the Fisheries Research and Development Section (Ministry of the Environment, Province of British Columbia) and an NSERC grant to Dr. K.J. Hall. I would like thank the dedicated employees of the Fisheries Branch (Ministry of Environment), as this project would not have been possible without their enthusiasm and interest in designing a world class hypohmnetic aeration system. In particular, I would like to thank Shawn Hay, Peter Law, George Reid and George Scholten of Region 1 and Jim Bomford (P.Eng.) from Victoria Headquarters.  xii  Chapter 1 Introduction  Cultural eutrophication is caused by excessive addition of limiting nutrients such as phosphorus and nitrogen to lakes, streams, rivers, estuaries and coastal waters (Wetzel, 1975). In lakes, these additions result in increased aquatic plant growth, undesirable changes in species composition, oxygen depletions, fish kills and decreased water quality for recreational use (Lee and Jones, 1988). In terms of domestic and industrial impact, cultural eutrophication decreases raw water quality (Dorin, 1981; Walker, 1983), increases treatment costs (Clark and Dorsey, 1980) and introduces possible carcinogens (eg. trihalomethanes) into the distribution system (Cantor et al., 1978; Tuthill and Moore, 1980; Jones and Lee, 1982). Following the limiting nutrient controversy of the late 1960's (see Vallentyne, 1974 for review), attention in the 1970's focused on reducing nutrient inputs (Lee and Jones, 1988) and rehabilitating culturally eutrophied lakes (Dunst et al., 1974). Many lakes recovered naturally from excessive nutrient loading after nutrient diversion eg. Lake Washington (Edmondson and Lehman, 1981); however, in some lakes the eutrophic status remained unchanged following nutrient diversion eg. Lake Trummen (Bjork et al., 1972). Lakes of this type were sufficiently eutrophic to maintain their present state via internal nutrient cycling after external nutrient sources were reduced. In addition, many lakes receive nutrients from non-point sources, which may prove difficult, if not impossible, to control (Lee and Jones, 1988). As a result, the interdisciplinary field of lake restoration emerged as limnologists and 1  Chapter 1.  Introduction  2  engineers began to develop techniques for restoring eutrophic lakes. Lake restoration refers to "... the manipulation of a lake ecosystem to effect an in-lake improvement in degraded or undesirable conditions" (Dunst et al., 1974). Hypolimnetic aeration is a lake restoration technique that has received widespread application. Originally developed in postwar Switzerland (Mercier and Perret, 1949), and rediscovered in West Germany (Bernhardt, 1967), hypolimnetic aeration is now used throughout Western Europe and North America (Verner, 1984). At least three multinational companies (Atlas Copco AB, Locher and Kobe Steel) are now actively marketing hypolimnetic aeration systems. Two of the main difficulties associated with hypolimnetic aeration are (1) estimating the oxygen consumption of the water body and (2) estimating the oxygen input capacity of the aeration system. This illustrates the true interdisciplinary nature of lake restoration, as the first problem lies within the realm of limnology, while the second is in the field of civil and environmental engineering. The lack of interaction between these traditional disciplines is responsible in part for the current paucity of information on factors influencing the oxygen transfer capabilities of various hypolimnetic aeration systems. A fair amount a basic (eg. Cornett and Rigler, 1984; Babin and Prepas, 1985) and applied research (eg. Ashley, 1983; McQueen et al., 1984) has recently been conducted on whole-lake oxygen consumption. The general consensus from the applied research is that estimates of hypolimnetic oxygen depletion should be calculated from well oxygenated hypolimnia to ensure that maximum depletion rates are obtained (McQueen and Lean, 1986; Ashley et al., 1987). Estimating the oxygen input capacity of hypolimnetic aeration systems has not received the same amount of attention. A wide range of oxygen input capacitites have been recorded (Taggart and McQueen, 1982) and an equally wide range of hypolimnetic aeration systems are available (Fast and Lorenzen, 1976). Although some attempt has  Chapter 1.  3  Introduction  been made to standardize aerator design specifications (Taggart and McQueen, 1982; Ashley, 1985; Ashley, 1988) this has not addressed the problem of variable oxygen input. Aside from the obvious influence of variable hypolimnetic BOD's and aerator volumetric water flow rates, few researchers have experimentally examined the effect of different diffuser designs, airflowrates and separator box surface exchange areas on the oxygenation capacity of hypolimnetic aeration systems. The purpose of this experiment was to use civil engineering gas transfer methodology to address this research deficiency. Specifically, its objectives were to investigatefivedesign variables which this researcher felt, after considerable literature review, were poorly understood in terms of their contribution to the oxygenation capacity of hypolimnetic aeration systems. These design variables were as follows: 1. Depth of air injection. The co-current method of bubble-water transport in the inflow tube of full lift hypolimnetic aerators becomes progressively less efficient at oxygen transfer throughout a given rise. Decreasing hydrostatic pressure is partly responsible for this decline; however, the decreasing oxygen content of rising air bubbles and the additive effect of vertical water velocity and buoyant bubble velocity contribute to poor oxygen transfer efficiency (Speece, 1975). The few field measurements available support this conclusion as most oxygen transfer has been found to occur in the lower half of inflow tubes (Bernhardt, 1967; Smith et al., 1975). The effect of diffuser depth on water velocity has received little research attention. Small changes in water velocity can result in significant changes in induced volumetric flow, daily oxygen load, transfer efficiency (E ; %) and energy efficiency c  (E ; kg O^/kW-hr). This aspect of hypolimnetic aeration oxygen transfer was exp  amined by inserting various diffusers into a full lift hypolimnetic aeration system  Chapter 1.  4  Introduction  at different depths and measuring changes in water velocity and dissolved oxygen concentration in the outflow tube. 2. Water surface exchange area. Neilson (1974) examined oxygen transfer under laboratory conditions, and concluded the surface area of the tank available for gas transfer directly influenced the oxygenation rate of his laboratory system. Neilson (1974) also estimated that only 6 to 12 % of total oxygen transfer in natural systems originates from bubble formation, rise and bursting. However hypolimnetic aerators, with their relatively small degassing chambers, are critically dependent on oxygen transfer during bubble formation, rise and bursting. A floating surface cover was used in the laboratory and field experiments to vary the surface exchange area and determine the relative importance of water surface area and gas transfer in relation to the overall oxygen transfer in a diffused aeration system. 3. Air flow rate. The volume of air injected per unit time is an important factor influencing the rate of oxygen transfer in diffused aeration systems. Higher air flows increase the turbulence at the air-water interface and the total interfacial area available for oxygen transfer to the surrounding liquid (Mavinic and Bewtra, 1974). As a result, the overall oxygen transfer coefficient (K£,a ; hr ) usually -1  20  increases with airflowrate. However, the effect of increased airflowrate on E and p  E is dependent on orifice size. In these experiments, the airflowrate was varied D  by a factor of two to examine its influence on Kx,a o, E and E . 2  D  p  4. Air flow rate perfinebubble diffuser. The effect of varying the airflowrate per fine bubble diffuser was examined as a number of researchers have demonstrated an increase in the Kxa o, E and E by reducing the airflowrate perfinebubble /  2  c  p  diffuser (eg. Doyle et al., 1983; Morgan and Bewtra, 1960; Ippen and Carver, 1954). The purpose of this experiment was to determine if the 40 p and 140 p  5  Chapter 1. Introduction  silica glass diffusers responded in a similar manner, and determine which factors were responsible for this behaviour. 5. Orifice size. The size of an orifice is one of the most important factors influencing the rate of oxygen transfer in diffused aeration systems, due to its influence on bubble size, contact time of the bubble in the liquid and turbulence in and around the gas-liquid interface (Bewtra and Mavinc, 1978). Large bubbles have a higher liquid film coefficient (KL)  than small bubbles; however, their larger size reduces  their contact time in the liquid and their surface area to volume ratio. Small bubbles (ie. less than 0.5 mm diameter) have higher surface area to volume ratios for improved gas exchange; however, their slower rise velocity results in a lower liquid film coefficient (K^) but a longer contact time. This aspect of the experiment was examined by using a range of orifice sizes from 40 p to 3175 (i diameter to determine which orifice size generated bubbles with the highest  Kia o, 2  E and E . D  p  A combination of laboratory andfieldtesting was selected for this research project. This allowed for a detailed examination of several factors capable of influencing gas transfer under controlled conditions, followed by the selection of appropriate variables for further evaluation and testing under actualfieldconditions. Although this increased the cost and complexity of this project, it significantly improved the reliability and robustness of the conclusions. The transition from bench scale to pilot scale is a crucial step in engineering development, and provides considerable insight into factors influencing the scale-up process.  Chapter 2 Methods  2.1  Lab Experiments  The laboratory experiments were conducted at the South Campus Fisheries Compound at U.B.C. 2.1.1  Air Supply  Air was supplied by a 1/6 hp (0.12 kW) Gast rotary vane vacuum-pressure pump (model 0211-V36A-G8CX), rated at 1.3 ft /min (36.8 L/min) ® 0 psig (0 kg/cm ). The com3  2  pressor was oil lubricated and fitted with a 10 \i oil removing element to prevent oil mist from contaminating the delivered air. The compressor was run several times for extended periods (4-5 hrs) and no oil film was detected in the test tank water. A pressure gauge was attached to the compressor to monitor air pressure in the discharge line (0-30 psig or 0-2.1 kg/cm ). 2  2.1.2  Air Flow Rate Measurement  Air flow rate was measured by a Brooks flow meter (Sho-Rate 50 Purgemeter), specifically manufactured for the experiments. The meter was equipped with a pressure gauge (0-30 psig or 0-2.1 kg/cm ) at both inlet and outlet nipples, and calibrated to read 4.7 to 56.6 2  L/min (0.2-2.0 ft /min) at S.T.P. (1.0 kg/cm , 21 degrees C). 3  2  6  Chapter 2.  2.1.3  7  Methods  Oxygen and Temperature Measurements  Dissolved oxygen and temperature in the test tanks was measured with a YSI 54 ARC oxygen-temperature meter. The oxygen meter was calibrated with two replicate Winkler titrations (Azide modification) (Lind, 1979) at the start of each experimental period. The temperature probe was checked against two mercury thermometers and was accurate within 0.5 degrees C. The oxygen-temperature probe was suspended in the center of each test tank (70 L and 239 L) approximately 5 cm below the suspension point for the diffuser being tested. The probe was weighted so it hung vertically and did not contact the sides of the tank when the experiments were in progress. 2.1.4  Deoxygenation-Oxygenation Procedure  The deoxygenation-oxygenation procedure used was the non-steady state reaeration test as outlined in APHA et al. (1980). Basically, the test involves deoxygenating a known volume of water with sodium sulfite (Na S0 ) and cobalt chloride (CoCl .6H 0), and 2  3  2  2  measuring the rate of reoxygenation. The chemical reaction is (Beak, 1977): Na SO 2  z  + 0  2  — • Na SO 2  A  (2.1)  The test water was deoxygenated with 0.1 mg/L cobalt chloride as a catalyst and 10.0 mg/L of sodium sulfite for each 1.0 mg/L of dissolved oxygen present in the water (Boyd, 1986). Theoretically only 7.9 mg/L of sodium sulfite is required for each mg/L of dissolved oxygen; however, due to partial oxidation during mixing, it is necessary to add up to 1.5 times the theoretical amount (Beak, 1977). The cobalt chloride was added first and thoroughly mixed into the test water. Sodium sulfite was mixed into a slurry in a 1 L flask, then added to the tank water and thoroughly mixed by a large paddle. The oxygen meter confirmed the tank water was rapidly deoxygenated as the dissolved oxygen  Chapter 2.  Methods  8  concentration usually declined to 0.2-0.3 mg/L within 30 seconds. The air compressor was then turned on, and oxygen concentrations recorded every 30 seconds until the dissolved oxygen reached 6-7 mg/L. 2.1.5  Tank Size and Geometry  Two sizes of tanks were used in the experiments. One tank was a clear plexiglass cylinder, with an inside diameter of 0.29 m and a height of 1.06 m. This tank wasfilledwith 70 L of water during the experiments. The second tank was a rectangular translucent polyethylene tub, 0.89 m L x 0.59 m W x 0.57 m H filled with 239 L of water. Afloatingsurface cover of 2.5 cm polystyrene foam was fabricated for each tank. The foam was cut with sufficient clearance (1 cm) to allow rapid installation and removal, but cover as much of the water surface area as possible. 2.1.6  Diffuser Type and Orifice Size  Two types of air diffusers were used in these experiments; coarse bubble diffusers and silica glass diffusers. The coarse bubble diffusers were constructed of 1.27 cm Schedule 40 white PVC irrigation pipe. The diffusers were cross shaped, with 4 arms joining into a common center. The center wasfittedwith a 0.64 cm nipple for attaching 0.64 cm tygon tubing airfine.The outside diameter of the diffusers was 25 cm , and they fit inside the plexiglass cylinder with 2 cm clearance on either side. The coarse bubble diffusers were fabricated to cover the orifice diameter range normally encountered in shop built diffusers ie. 1/8" (3175 u); 1/16" (1588 u), 1/32" (794 u) and 1/64" (397 p). The surface area of a circle increases 4x as the diameter doubles, so the number of holes drilled in each diffuser was as follows: • 1/8" (3175 u) - 1 hole  Chapter 2.  Methods  9  Orifice Size 1/8" (3175 fi) 1/16" (1588 p) 1/32" (794 p) 1/64" (397 p)  1.586 0.421 0.099 0.025  Air Flow Rate ft /min (44.9 L/min) ft /min (11.9 L/min) ft /min (2.8 L/min) ft /min (0.7 L/min) 3  3  3  3  Table 2.1: Air flow rate through an orifice at 2 psig (0.136 kg/cm ). 2  • 1/16" (1588 p) - 4 holes (one on each arm) • 1/32" (794 fi) - 16 holes (four on each arm) • 1/64" (397 p) - 64 holes (sixteen on each arm) Standard tables of air discharge through an orifice at 2 psig (0.136 kg/cm ) (assuming 2  a discharge coefficient of 0.65 for a sharp edged orifice) are shown in Table 2.1. The compressor was capable of a maximum output of 36.8 L/min @ 0 psig (0 kg/cm ). Each 2  diffuser was capable of passing the following volumes of air: • 1/8" (3175 p) 1 x 1.586 ft /min = 1.59 ft /min (44.9 L/min) @ 2 psig (0.14 kg/cm ) 3  3  2  • 1/16" (1588 p) 4 x 0.421 ft /min = 1.68 ft /min (47.6 L/min) @ 2 psig (0.14 3  3  kg/cm ) 2  • 1/32" (794 p) 16 x 0.099 ft /min = 1.58 ft /min (44.8 L/min) @ 2 psig (0.14 3  3  kg/cm ) 2  • 1/64"  (397 p) 64 x 0.025 ft /min = 1.60 ft /min (44.8 L/min) @ 2 psig (0.14 3  3  kg/cm ) 2  Therefore, orifice size and number were not considered restrictive to air flow at the experimental flow rates.  Chapter 2.  Methods  10  The silica glass diffusers were obtained from Aquatic Eco-Systems Inc. (Apopka, Florida) in two pore sizes: 140u maximum pore size (Model AS-8) and 40//. maximum pore size (Model AS-8-0). The external dimensions of both diffuser groups were identical, 3" L x 1.5" W x 1.5" D (7.62 cm L x 3.8 cm W x 3.8 cm D). Each diffuser weighed 0.39 lbs (0.18 kg) and was fitted with a 1/4" (0.64 cm) hose nipple. Both groups of diffusers (coarse and silica glass) were suspended in the center of each test tank (cylinder or rectangular tank) by tygon tubing air line. The diffusers hung 0.80 m below the water surface in the cylinder and 0.34 m below the surface in the rectangular tank. A 0.45 kg weight was attached to the coarse bubble diffusers to counter their positive buoyancy and stop them from swinging about when discharging air. The silica glass diffusers were sufficiently heavy and did not require additional weighting. The coarse bubble diffusers were quickly lowered into position during testing with the compressor running. This avoided flooding the diffusers with water which caused an uneven discharge of air while the diffuser was purged of water. The silica glass diffusers did not have this problem and immediately purged themselves of water when the compressor was turned on. Dye was added on several occasions to each test tank to examine circulation patterns and determine if any stagnant zones existed. The water in the 70 L cylinder was completely mixed within an average of 17 seconds (n=16), and the 239 L tank required an average of 30 seconds (n=3) to mix completely. This confirms initial observations that complete mixing was quickly achieved, and the oxygen-temperature probe was adequately positioned to measure the rate of oxygen increase in the water column. 2.1.7  Experimental Design  The lab experiments were divided in three groups: Group 1 - coarse bubble diffusers in the 70 L cylinder; Group 2 - silica glass diffusers in the 70 L cylinder; and Group 3 -  Chapter 2.  11  Methods  No. Air Flow Rate (L/min) Orifice Size (p) Cover 1. 9.4 397 no 2. 9.4 397 yes 3. 18.8 397 no 4. 18.8 397 yes 5. 9.4 794 no 6. 9.4 794 yes 7. 18.8 794 no 8. 18.8 794 yes 9. 9.4 1588 no 10. 9.4 1588 yes 11. 18.8 1588 no 12. 18.8 1588 yes 13. 9.4 3175 no 14. 9.4 3175 yes 15. 18.8 3175 no 16. 18.8 3175 yes Table 2.2: Group 1 experimental treatments. coarse bubble and silica glass diffusers in the 239 L tank. The treatments examined in Group 1 were: the effect of air flow rate (9.4 L/min or 18.8 L/min); the effect of surface cover (present or-absent), and the effect of orifice size (397 p, 794 //, 1588 p and 3175 p). This resulted in 16 combinations offlow,cover and orifice size (Table 2.2). The experiments were carried out in a randomized complete block design. Each of the treatments was assigned a number from 1 to 16, and the order in which the treatments were tested were selected from a 10,000 digit random number table (Rohlf and Sokal, 1969). Each set of 16 treatments was completed in one day, then repeated the next day with a new set of random numbers. For example, on Day 1 the sequence of testing was: 7, 4, 11, 3, 1, 10, 12, 9, 15, 6, 2, 5, 14, 13, 8 and 16. On Day 2 the sequence was: 4, 12, 5, 16, 3, 10, 6, 7, 1, 13, 11, 14, 8, 9, 2 and 15. The purpose of this design was to remove random error that may occur during any  Chapter 2. Methods  12  No. Air Flow Rate (L/min) Orifice Size (u) Cover Number of 1. 9.4 40 no 1 2. 9.4 40 yes 1 3. 9.4 40 no 2 4. 9.4 40 yes 2 5. 18.8 40 no 1 6. 18.8 40 yes 1 7. 18.8 40 no 2 8. 18.8 40 yes 2 9. 9.4 140 no 1 10. 9.4 140 yes 1 11. 9.4 140 no 2 12. 9.4 140 yes 2 13. 18.8 140 no 1 14. 18.8 140 yes 1 15. 18.8 140 no 2 16. 18.8 140 yes 2 Table 2.3: Group 2 experimental treatments. given treatment day and block the treatments over time (days) to remove any systematic error introduced over time. Each treatment was replicated 5 times, always on a different day. The Group 1 experiments were conducted on September 2, 3, 4, 9 and 10, 1987. The treatments examined in Group 2 were: the effect of air flow rate (9.4 L/min or 18.8 L/min); the effect of surface cover (present or absent); the effect of orifice size (40 p or 140 a); and the effect of diffuser numbers (1 or 2). This again resulted in 16 combinations of flow, cover, orifice size and number of diffusers (Table 2.3). The Group 2 experiments were conducted in the same randomized complete block design as Group 1. The Group 2 experiments were conducted on October 20, 21, 22, 23 and 26, 1987. The treatments examined in Group 3 were: the effect of orifice size (40 u, 397 p and 1588 u) and the effect of surface cover in the 239 L tank. Three surface conditions were examined: no cover, cover and no cover plus wind generated from a 23 L vacuum exhaust  Chapter 2.  13  Methods  No. Air Flow Rate (L/min) Orifice Size-(^i) Cover ~T 2873 1588 ye~s 2. 28.3 1588 no 3. 28.3 1588 no + wind 4. 28.3 397 yes 5. 28.3 397 no 6. 28.3 397 no + wind 7. 28.3 40 yes 8. 28.3 40 no 9. 28.3 40 no + wind Table 2.4: Group 3 experimental treatments. port. The velocity of the wind was measured with a hot wire anemometer (Thermo-Air 1) and was sufficient to create 0.5 to 1.0 cm waves on the 239 L tank and circulate dye across the surface of the tank in 10-15 seconds. This resulted in 9 combinations of orifice size and surface conditions (Table 2.4). The Group 3 experiments were conducted in the randomized complete block design, and performed on November 6, 7, 8, 9 and 11, 1987. At the start of each experimental day, the barometric pressure was obtained from the Vancouver Weather Office. A series of data sheets were designed and the following information was recorded for each oxygen transfer test: 1. Date 2. Time 3. Orifice size 4. Barometric pressure 5. Water temperature  Chapter 2.  14  Methods  6. Air temperature 7. Tank configuration 8. Tank volume 9. Type of surface cover 10. Number of diffusers 11. Air flow 12. Grams of Na2SC"3 added 13. Random number order 14. Oxygen concentration at 30 second intervals 15. Compressor discharge pressure 16. Pressure at inlet of airflowmeter 17. Pressure at outlet of airflowmeter A digital electronic balance (Ohaus, C-300 M) was used to weigh the deoxygenation chemicals required for the experiments. The Na S03 and C0CI2.6H2O were Reagent 2  grade and were obtained from local suppliers (BDH Chemicals and Chemonics Scientific). 2.1.8  Bubble Size  Bubble size was determined by photographing rising bubbles in the 70 L column with a Pentax ME camera andflashattachment, synchronized at 1/100 second. A meter stick  Chapter 2.  Methods  15  graduated with 1 mm increments was suspended in the cylinder and bubbles were photographed against the meter stick for scale. The slide photographs were then examined with a Bausch and Lomb dissecting microscope at 60-70x to determine bubble size. Approximately 20 bubbles were measured for each orifice size and air flow setting. A spreadsheet program was then written to calculate the volume of the bubbles. Since most of the bubbles were oblate spheroid in shape, the following formula was used to calculate volume: V = 4/37ra  (2.2)  where: V = volume in mm  3  a = 1/2 long axis of the bubble (mm) b = 1/2 short axis of the bubble (mm) The spreadsheet program then calculated the equivalent diameter of the bubbles according to: (2.3) where: d = equivalent diameter (mm) V = volume in mm  3  Finally, the program calculated the mean bubble size and standard deviation for each orifice size and air flow rate. 2.2  Field Experiments  The field experiments were conducted at Black Lake, midway between Keremeos and Kaledon B.C. on Highway 3A. Black Lake is a small (max. depth = 9.0 m; volume = 178,500 m ) naturally eutrophic lake that was formed approximatley 8900 years ago by 3  Chapter 2.  16  Methods  a major meltwater outflow which drained the Kaledon tongue of the main Okanagan ice lobe (Nasmith, 1962). A full lift hypohmnetic aeration system was installed in Black Lake in 1978 (Ashley, 1981). The aerator consists of an open box (2.4 m L x 1.2 m W x 0.9 m D) with two 0.76 m x 7.3 m galvanized steel pipes attached through the bottom of the box. 2.2.1  Air Supply  Air was provided by a 7.5 kW rotary vane compressor (Hydrovane SR 4000 rated at 1.13 m /min free air delivery (FAD) @ 100 psig or 7.0 kg/cm ). The compressor was oil 3  2  lubricated and fitted with a 3 u oil removing absorbent element. A pressure gauge was attached to the compressor to monitor discharge pressure at the outlet valve. 2.2.2  Air Flow Rate Measurement  Air flow rate was measured by a Brooks (1305 0 Ring Seal)flowmeter,specifically manufactured for the experiments. The meter was equipped with a pressure gauge (0-8 kg/cm ) at both inlet and outlet ports, and calibrated to read 113-1133 L/min (4-40 2  ft /min) at 7.0 kg/cm (100 psig) and 21 degrees C. 3  2.2.3  2  Oxygen, Temperature and Current Measurements  Dissolved oxygen and temperature were measured with the YSI 54 ARC oxygen-temperature meter used in the lab experiments. Oxygen-temperature profiles were taken approximately 3 m from the aerator at the start of the experiment to establish the oxygen concentration and temperature of the lake. When the various diffusers were being tested, the oxygen-temperature probe was suspended at 3 m in the outflow tube to measure the oxygen concentration and temperature of the outflow water. A General Oceanics current  Chapter 2.  17  Methods  z  0 1 2 3 4 5 6 7 8  September 24, 1987 Before After 02 T 02 T 9.4 16.0 8.4 17.0 9.0 16.0 8.6 16.4 8.8 16.0 8.4 16.0 8.8 16.0 8.5 16.0 8.1 15.6 8.1 15.7 3.8 15.0 8.2 15.0 0.4 10.5 0.4 11.4 0.4 9.0 0.4 9.8 0.4 8.8 0.4 9.5  September Before 02 T 8.2 16.0 7.9 16.0 7.9 16.0 7.9 16.0 6.5 15.9 2.2 15.1 0.4 11.2 0.4 9.5 0.4 9.0  25, 1987 After 02 T  7.9 8.0 7.8 7.7 7.7 4.8 0.4 0.4 0.4  16.0 16.0 16.0 16.0 15.8 15.0 10.9 9.7 9.6  Table 2.5: Oxygen-temperature profiles at Black Lake during field experiments, meter (Model 2035) was also suspended at this depth to measure outflow water velocity.  2.2.4  Oxygenation Procedure  The procedure used to measure oxygen input from various diffusers involved lowering a diffuser to 3 m or 7 m i n the inflow tube and measuring the oxygen concentration at 3 m i n the outflow tube.  T h e difference i n oxygen concentration between inflow and  outflow water is the amount of dissolved oxygen transferred by a specific diffuser. T h e loading rate of each diffuser was calculated by multiplying the oxygen differential by the volumetric flow of the aerator as measured by the current meter. Each diffuser was operated for several minutes before a final oxygen and current measurement were recorded. T h e temperature of the hypolimnetic water changed very little during the experiment, and the influent oxygen concentrations were essentially constant at 0.4 m g / L (Table 2.5).  Chapter  2.  18  Methods  Orifice Size 1/8" 1/16" 1/32"  (3175 p) (1588 ft) (794 fi  Air Flow Rate 5.005 ft /min 1.255 ft /min 0.351 ft /min 3  3  3  (141.7 L/min) (35.5 L/min) (9.9 L/min)  Table 2.6: Discharge of air through an orifice at 20 psig (1.4 kg/cm ). 2  2.2.5  Diffuser Type and Orifice Size  Two types of diffusers were used in the field experiments; coarse bubble diffusers and silica glass diffusers. The coarse bubble diffusers were constructed of 3.8 cm ID Schedule 80 P V C pipe.  The diffusers were cross-shaped, with 4 arms joining into a common  center. The center of the cross was fitted with a 1.9 cm hose nipple for attaching 1.9 cm ID compressed air hose. The outside dimensions of the diffusers were 69 cm and they fit inside the 0.76 m outlet tube with 3.5 cm clearance on either side. The coarse bubble diffusers were fabricated in three orifice sizes: 1/8" (3175 p); 1/16" (1588 p); and 1/32" (794 u). A 1/64" (397 p) diffuser was not possible in this size range due to the large number of orifices required (1280) and the high breakage rate of 1/64" bits when drilling schedule 80 P V C pipe. The number of holes drilled in each diffuser was as follows: • 1/8" (3175 p) - 20 holes (5 on each arm) • 1/16" (1588 ii) - 80 holes (20 on each arm) • 1/32" (794 fi) - 320 holes (80 on each arm) Standard tables of air discharge through an orifice at 20 psig (1.4 kg/cm ) (assuming a 2  discharge coefficient of 0.65 for a sharp edged orifice) are shown in Table 2.6. The compressor was rated for a maximum output of 40 ft /min FAD (1.13 m /min) @ 100 psig 3  3  Chapter 2.  19  Methods  (7.0 kg/cm ), and was operated at 10 ft /min (0.28 m /min) @ 100 psig (7.0 kg/cm ) 2  3  3  2  throughout the experiments. According to Boyle's Law, PiVi = P2V2 at constant temperature (Atlas Copco, 1978). Therefore, 10 ft /min (0.28 m /min) @ 114.7 psia (8.1 3  3  kg/cm ) is equivalent to 33 ft /min (0.93 m /min) @ 34.7 psia (2.4 kg/cm ). Each diffuser 2  3  3  2  was capable of passing the following amounts of air: • 1/8" (3175  - 20 x 5.005 ft /min = 100 ft /min (2.8 m /min) @ 20 psig (1.4 3  3  3  kg/cm ); 2  • 1/16" (1588 p) - 80 x 1.255 ft /min = 100 ft /min (2.8 m /min) @ 20 psig (1.4 3  3  3  kg/cm ); 2  • 1/32" (794 ft) - 320 x 0.351 ft /min = 112 ft /min (3.2 m /min) @ 20 psig 1.4 3  3  3  kg/cm ). 2  Therefore, orifice size and number were not considered restrictive to air flow at the experimental flow rates. The silica glass diffusers were obtained from Aquatic Ecosystems Inc. (Apoka, Florida) in the 140 fi maximum pore size (Model ALR-23). The external dimensions of the diffusers were 23 cm L x 3.8 cm W x 3.8 cm D. Each diffuser weighed 1.35 lb (0.61 kg) and was fitted with a 0.5" (1.27 cm) NPT fitting. The diffusers were connected to a 1.5" (3.81 cm) PVC 4 way center, and arranged in a spiral pattern. As mentioned previously, the diffusers were lowered into the inflow tube to a depth of 3 m or 7 m, and suspended by the 3/4" (1.9 cm) air hose attached to the center hose nipple. The silica glass diffusers were sufficiently heavy to remain submerged when operating; however, the coarse bubble diffusers required 10 lbs (4.5 kg) of additional weight to remain submerged. Both coarse and glass diffusers were lowered into position with the compressor running, to avoidfloodingthe coarse bubble diffusers and causing an uneven discharge of air.  Chapter 2.  20  Methods  No. Orifice Size (p) Depth (m) 1. 140 3 2. 140 7 3. 794 3 4. 794 7 5. 1588 3 6. 1588 7 7. 3175 3 8. 3175 7 Table 2.7: Group 4 experimental treatments. 2.2.6  Experimental Design  Thefieldexperiments were divided into two groups; Group 4 and Group 5. The treatments examined in Group 4 were: the effect of orifice size (3175 p, 1588 p, 794 p and 140 p) and depth of air release (3 m or 7 m). This resulted in 8 combinations of depth and orifice size (Table 2.7). The Group 4 experiments were conducted in the same randomized complete block design as Groups 1-3. Each treatment was replicated 5 times. The Group 4 experiments were conducted on September 24, 1987. The treatments examined in Group 5 were: the effect of orifice size (140 p and 3175 p) and the effect of afloatingsurface cover of 2.5 cm styrene foam board (present or absent). This resulted in 4 combinations of orifice size and cover (depth fixed at 7. m) (Table 2.8). The Group 5 experiments were also conducted in a randomized complete block design, replicated five times and conducted on September 25, 1987.  Chapter 2. Methods  21  No. Orifice Size (p) Cover 1. 140 No 2. 140 Yes 3. 3175 No 4. 3175 Yes Table 2.8: Group 5 experimental treatments. 2.3  Parameter Calculation  2.3.1  Lab Experiments  Calculation of the oxygen transfer coefficient (K^a) is usually done by plotting the dissolved oxygen (DO) deficit vs time data on one-cycle semi-logarithmic graph paper, and the slope of the line is the overall oxygen transfer coefficient (K^ar) (APHA et al., 1980). The formula for calculating K^ay is:  ti£^m^i  KiaT=  t-7 — ti  (2  .4)  where: In = natural logarithm K^ay — oxygen transfer coefficient at the temperature of the testwater, (hr ) -1  Ci = DO (mg/L) at point 1 on the graph C = DO (mg/L) at point 2 on the graph 2  C = DO saturation concentration (mg/L) g  t — time at point 1 on the graph (hr) 1  t = time at point 2 on the graph (hr) 2  Ti and t are usually chosen as the times at which the measured oxygen concentra2  tion is 20% (ti) and 80% (t ) of the saturation value for the test water, corrected for 2  temperature and barometric pressure.  Chapter 2.  22  Methods  The dissolved oxygen saturation on test days was adjusted to current barometric pressure by: C, = C  e 7 6 0  j^  (2.5)  where: C = DO saturation concentration during test (mg/L) g  C« 760 — DO saturation at 760 mm Hg total pressure (mg/L) Pb = barometric pressure during test (mm Hg) Since the tests were conducted below 1000 m elevation and 25 degrees C, there was no correction in C„ for the vapor pressure of water (APHA et al., 1980). The saturation pressure was not corrected for mid-depth oxygen partial pressure as the test tanks were very shallow (0.8 m in the cylinder and 0.34 m in the square tank). The problems with the graphical method of K^a estimation are threefold: 1. Errors generated when reading numbers off the x and y axis; 2. Errors generated when plotting the line when the data points do not fall in a straight line; 3. The time required to draw the graphs, the main purpose of which is to obtain the time estimates of t and t , corresponding to 20% and 80% of oxygen saturation. x  2  When this researcher began analyzing the data, it was clear that human error was introduced from the graphical determination of t and t , and the time required to draw x  2  205 graphs was a major impediment to completing this project in a timely and unbiased manner. As a result, it was decided to eliminate the graphical procedure, and estimate t  x  and t directly from a simple linear regression of time vs natural logarithm of the oxygen 2  deficit.  Chapter 2.  23  Methods  This researcher used 10% and 60% saturation values to calculate ti and t , as the time 2  required to reach 80% saturation in the 239 L tank exceeded the 30 minute time limit recommended by Beak (1977), especially when testing the larger coarse bubble diffusers. The regression of time vs In oxygen deficit and subsequent calculation of K^a was done on a DEC Professional 350 microcomputer with the Pro 20/20 spreadsheet. The time required to write the spreadsheet template and command procedure, enter and check the data and calculate K^a2o  KL&T  for 205 separate tests was 4 days. K^ar was converted to  according to: Ka L  = K a /9 T  20  L  (2.6)  20  T  where: 8 = 1.024 and T = water temperature in degrees C (Boyd, 1986). The spreadsheet program also calculated the standard oxygen transfer rate (OT„, in grams 0 /hr) and the energy efficiency (E , in grams 0 /kW-hr). OT, was calculated 2  p  2  as follows (Boyd, 1976): OT, = K a L  20  DO  20  V  (2.7)  where: 0T = standard oxygen transfer rate (g 0 /hr); 8  2  D0 o = DO concentration (mg/L) at saturation for 20 degrees C and standard pressure 2  (760 mm Hg); V = volume of water in tank, m . 3  E was calculated as follows (APHA et al, 1980): p  E  p  = OT./P  where: E = energy efficiency in grams 0 /kW-hr; p  P = power input, kW.  2  (2.8)  Chapter 2.  24  Methods  The power input (P) for each airflowrate was adjusted to reflect the fraction of the compressor's energy consumption required to deliver a given airflowrate. The compressor was rated at 36.8 L/min @ 0 kg/cm (psig) with a nameplate horsepower of 0.1243 kW. 2  The following power inputs were used for the Group 1-3 E calculations: p  1. 9.4 L/min = ^fj™ x 0.1243 kW = 0.0317 kW; '  36.8L/rnin  2. 18.8 L/min = '  '  "S"?  1 1  x 0.1243 kW = 0.0635 kW;  36.8L/mtn  '  3. 28.3 L/min = § # 7 ^ x 0.1243 kW = 0.0956 kW. The oxygen transfer efficiency (E %) (weight of oxygen dissolved / weight of oxygen OJ  supplied x 100) was calculated as follows: Each airflowrate ie. 9.4 L/rnin, 18.9 L/min and 28.3 L/min was multiplied by 0.21 to reflect the percent oxygen composition by volume of air (Atlas Copco, 1978). These values were multiplied by 60 min/hr to derive the L/hr of 0 gas supplied to the test tank, 2  then divided by 22.4 L/mole to obtain the number of moles supplied, then multiplied by the molecular weight of 0 (32 g/mole) to obtain the grams of O2 supplied per hour. 2  This value was divided into the OT, (grams 0 /hr) and multiplied by 100 to derive a 2  percent transfer efficiency (E ). An example is: D  18.9 L/min of air supplied x 0.21 = 3.97 L/min O2 supplied; and multiplied by 60 min/hr = 238.14 L/hr O2 supplied; and divided by 22.4 L/mole = 10.631 moles supplied; • and multiplied by 32 g/mole = 340.19 g 0 /hr supplied. 2  OT, = 16.16 g 0 /hr, therefore E = 16.16 g 0 /hr / 340.19 g 0 /hr x 100 = 4.75%. 2  D  2  2  Chapter 2.  2.3.2  25  Methods  Field Experiments  Four parameters were calculated or measured for thefieldexperiments. Oxygen input per cycle (mg/L) is the measured difference in oxygen concentration between outflow water at 3 m in the outlet tube and inflow water at the entrance of the intake tube (8 m). Daily oxygen load (kg 0 /day) is the product of the oxygen input per cycle (g/m ) 3  2  and the daily volumetric flow (Q™) (m /day) as calculated by: 3  Q  w  (2.9)  = irr v 2  where: r — radius of outflow tube (m) v = velocity of outflow water (m/sec) The oxygen transfer efficiency (E ) for thefieldexperiments was calculated as follows: 0  The compressor was operated at a constant airflowof 0.28 m /min (10 ft /min) @ 8.1 3  kg/cm  2  (114.7  3  psia) throughout the experiments, as measured by aflowmeterand a series  of pressure gauges. The volume of intake air was then calculated to be 2.21 m /min 3  (78  ft /min) at 3  1.03  kg/cm  2  (14.7  psia) using Boyle's Law ( P i V i  = P V 2  at constant  2  temperature). The weight of oxygen supplied per day was then calculated by the same procedure used in the laboratory oxygen transfer efficiency calculations ie. 2.21 m /min x 1000 3  L/m x 60 min/hr x 24 hr/day x 0.21 (% 0 ) / 22.4 L/mole x 32 g/mole 0 / lOOOg/kg 3  2  2  = kg 0 supplied per day. The weight of oxygen transferred per day divided by the 2  weight of oxygen supplied to the lake x 100 = the oxygen transfer efficiency for the field experiments (E ). D  The aeration efficiency (E ) was calculated as the daily load divided by the daily p  power input. The power input was adjusted to reflect the fraction of the compressor's  Chapter 2.  26  Methods  energy consumption required to deliver a fixed air flow rate. The compressor was rated at 1.13 m /min FAD at 7.0 kg/cm with a nameplate horsepower of 7.5 kW. The compressor 3  2  delivered 0.28 m /min @ 7.0 kg/cm for the Group 4 and 5 experiments, so the adjusted 3  2  power input was 0.28/1.13 x 7.5 = 1.858 kW.  2.3.3  Statistical Analysis  The statistical procedure used to analyze experimental and field data was an analysis of variance program; Manova) in the SSPS statistical package. The level of significance was set at a = 0.01 for each statistical test. The Anova was conducted on K ^ o , O T , s  E  p  and E„ for Groups 1, 2 and 3; and on water velocity, oxygen input per cycle, daily  oxygen load, E and E for Groups 4 and 5. The arc sin square root transform was used D  p  on the E Anova's to reduce the skewness of the percentage values (Larkin, 1975). 0  In situations where the null hypothesis was rejected, a comparison among means test was conducted using Scheffe's test. The level of significance was also set at a = 0.01 for Scheffe's test.  This is the most rigorous a posteriori test and is recommended by  statistical purists when doing comparison among means tests (Larkin, 1975). The initial Anova was set up to show the significance of the first, second and third order interaction effects. The standard procedure was to examine the second and third order interactions for significant results, then add the second and third order interaction sum of squares into the residual sum of squares and recalculate the Anova if no significant results were obtained. This leaves the main and first order interactions, which are easier to explain, and avoids the convoluted statements associated with explaining higher order interactions.  This procedure was recommended by Dr. J . Berkowitz, Co-ordinator of  the Statistical Consulting and Research Laboratory (SCARL) at U B C . The overall experimental design was as follows:  Chapter 2.  Methods  Group  3 4 5  Experimental System Laboratory (70 L cylinder)  Variables Airflowrate (9.4, 18.8 L/min) Orifice size (397, 794, 1588, 3175 p) Cover (yes, no) Laboratory Airflowrate (9.4, 18.8 L/min) (70 L cylinder) Orifice size (40, 140 p) No. of diffusers (1,2) Cover (yes, no) Laboratory Orifice size (40, 397, 1588 p) (239 L tank) Cover (yes, no, no + wind) Field Orifice size (140, 794, 1588, 3175 p) (Hypolimnetic aerator) Diffuser depth (3, 7m) Field Orifice size (140, 3175 p) (Hypolimnetic aerator) Cover (yes, no)  27  Chapter 3 Results  3.1  Group 1: Kz,a and OT, 20  The results for the Anova on Group 1 Kx,a and OTs are shown in Appendix A. Signif20  icant results (a < 0.01) were obtained for air flow rate, orifice size, air flow rate by size interaction and replicate days. There was no significant effect of surface cover on Kjr ao 2  or OT,. The significant and non-significant results are similar for Kj,a and OT, as the values of OT, are derived from the initial Kx,ao calculation. 2  3.1.1  Air Flow Rate  The airflowrate supplied to the diffusers produced the most significant result (Figure 3.1). The cell means for Kia  20  and OT, (Standard Deviation in brackets) for the two  airflowrates studied, lowflow(9.4 L/min) and mediumflow(18.8 L/min) are shown in Table 3.9. The net result is that doubling the airflowrate produced a 122% increase in K^azo and OT,. Treatment K a (hr" ) OT, (g Q /hr) n Low Flow 4.5 (1.7) 2.8 (1.1) 40 High Flow 10.0 (3.7) 6.2 (2.3) 40 1  L  20  2  Table 3.9: Effect of airflowrate on Group 1 Kjr ao and OT,. 2  28  29  Chapter 3. Results  12 -  KLa (l/hr)  Ep (g 02/kWh x 10)  Eo  (%)  • • 9.4 L/min EE2 18.8 L/min Figure 3.1: Effect of airflowrate on Group 1 K/,a o, E and E . 2  p  p  Chapter 3.  30  Results  Treatment 397 p 794 p 1588 p 3175 p  Ka L  20  (hr  J  )  11.5 (4.5) 7.3 (2.9) 5.5 (2.2) 4.8 (2.1)  OT, (g 0 /hr) 7.1 (2.8) 4.5 (1.8) 3.4 (1.4) 3.0 (1.3) 2  n 20 20 20 20  Table 3.10: Effect of orifice size on Group 1 Kr,a OT,. 20  3.1.2  Orifice Size  Orifice size produced the next most significant result (Figure 3.2). The cell means for K/,a and OT, (Standard Deviation in brackets) for the four orifice sizes studied (397 20  p, 794 p, 1588 p and 3175 p) are shown in Table 3.10. Scheffe's test was used in the comparison among means test for the four orifice sizes. The Scheffe's test indicates that the 397 p orifice diffuser is significantly greater (a = 0.01) for Kia  20  and OT , and the 20  remaining three diffusers (794 p, 1588 p and 3175 p) are not significantly different from each other. 3.1.3  Air Flow Rate by Size Interaction  A significant air flow rate by size interaction effect was detected for Kia and OT,. A 20  graphical plot of the data (K^a vs orifice size) (Figure 3.3) reveals the magnitude of the difference between cell means for K£,ao at low and medium air flow rates increases as 2  orifice size decreases. 3.1.4  Replication  A significant effect was observed for the replication procedure used in the experiments. The cell means for Kr,a and OT, (Standard Deviation in brackets) for the five replicate 2u  days are shown in Table 3.11. A comparison among means using Scheffe's test was  Chapter 3.  Results  KLa (1/hr)  71  z 397 u  794 u  1588 u  3175 u  Figure 3.2: Effect of orifice size on Group 1 K2,a n. 2  32  Chapter 3. Results  KLa20  397 u  794 u  — - 9.4 L/min  1588 u  3175  ~ r ~ 18.8 L/min  Figure 3.3: Group 1 airflowrate by size interaction effect for Kr,a  2u  Chapter 3.  33  Results  Treatment K a (hr- ) OT. (g 0 /hr) Day 1 6.6 (3.6) 4.1 (2.2) Day 2 7.8 (4.4) 4.8 (2.8) Day 3 7.6 (4.1) 4.7 (2.5) Day 4 7.2 (4.1) 4.4 (2.5) Day 5 7.1 (4.1) 4.4 (2.5)  n 16 16 16 16 16  1  L  20  2  < Group 1 Kia and OT, Table 3.11: Effect of replication on 20  Treatment K a (hr" ) OT, (g 0 /hr) n Cover 7.2 (4.0) 4.4 (2.5) 40 No Cover 7.4 (4.0) 4.6 (2.5) 40 1  L  2  20  Table 3.12: Effect of surface cover on Group 1 K i a and OT,. 20  unable to distinquish any significant differences among the five replicate days on which the experiments were conducted. 3.1.5  Surface Cover  The presence or absence of a surface cover did not exert a significant effect. The cell means for K£,a and OT, (Standard Deviation in brackets) for the cover treatment are 20  shown in Table 3.12. The data suggest that the presence of a surface cover may reduce Ki,a 20  a n  d OT,; however, a larger sample size is required to adequately assess this  treatment. 3.2  Group 1: E and E D  p  The results for the Anova on Group 1 E„ and E are shown in Appendix A. Significant p  results (a < 0.01) were obtained for orifice size, airflow,replicate days and surface cover. There was no significant interaction effect on E and E . p  D  Chapter 3.  34  Results  Treatment 397 p 794 p 1588 p 3175 p  E (%) E (g 0 /kW-hr) 2.8 (0.21) 147.5 (11.4) 1.7 (0.16) 93.0 (8.4) 1.3 (0.13) 70.4 (7.3) 1.1 (0.14) 60.5 (7.6) p  p  n 20 20 20 20  2  Table 3.13: Effect of orifice size on Group 1 E and E . D  p  Treatment E (%) E (g Q /kW-hr) n Low Flow 1.6 (0.63) 87.6 (33.9) 40 Medium Flow 1.8 (0.67) 98.2 (35.7) 40 p  p  2  Table 3.14: Effect of air flow rate on Group 1 E and E . D  3.2.1  p  Orifice Size  Orifice size produced the most significant result (Figure 3.4 and 3.5). The cell means for E and E (Standard Deviation in brackets) for the four orifice sizes examined (397 0  p  p, 794 p, 1588 p and 3175 p) are shown in Table 3.13. Scheffe's test was used in the comparison among means test for the four orifce sizes. The test indicates each orifice size is significantly different from each other (a = 0.01) for E and E , and that both c  p  transfer efficiency and energy efficiency decrease with increasing orifice size. 3.2.2  Air Flow Rate  The air flow rate to the diffusers produced the next most significant result (Figure 3.1). The cell means for E and E (Standard Deviation in brackets) for the two air flow rates 0  p  studied, low flow (9.4 L/min) and medium flow (18.8 L/min) are shown in Table 3.14.  Chapter 3.  Results  Figure 3.4: Effect of orifice size on Group 1 E . c  Chapter 3.  160  36  Results  Ep (g 02/kWh)  140  120  100  80  60  40  20  0 397 u  794 u 1588 u 3175 u Figure 3.5: Effect of orifice size on Group 1 E . p  Chapter 3.  37  Results  Treatment E (%) E (g 0 /kW-hr) n Cover 1.7 (0.66) 91.4 (35.6) 40 No Cover 1.8 (0.65) 94.3 (34.8) 40 p  p  2  Table 3.15: Effect of surface cover on Group 1 E and E . D  3.2.3  p  Replication  A significant result was observed for the replication procedure used in the experiments. The F value of the replication effects is small (« 20) in relation to the main experimental effects and does not change the overall conclusions of the experiment. 3.2.4  Surface Cover  A marginally significant result was observed for the cover-no cover treatment. The cell means for E and E (Standard Deviation in brackets) are shown in Table 3.15. The c  p  F value of the cover-no cover treatment is very small (« 7-11) in relation to the main experimental effects. The presence of a surface cover may reduce E and E ; however, a D  p  larger sample size is required to adequately assess this treatment. 3.3  Group 2: K a L  2 0  and O T  a  The results for the Anova on Group 2 Kz,a and OT are shown in Appendix B. Signifi20  s  cant results were obtained for air flow rate, number of diffusers, air flow rate by number of diffuser interaction and surface cover. There was no significant effect of orifice size or replicate days on K£a o or OT . 2  s  Chapter 3.  38  Results  Treatment K a (hr" ) OT, (g Q /kW-hr) n Low Flow 11.6 (1.7) 7.2 (1.0) 40 Medium Flow 22.9 (3.8) 14.2 (2.3) 40 1  L  20  2  Table 3.16: Effect of air flow rate on Group 2 K^a^ and OT,.  Treatment K^a (hr" ) OT, (g Q /kW-hr) n One Diffuser 15.3 (5.5) 9.5 (3.4) 40 Two Diffuser 19.1 (6.7) 11.8 (4.1) 40 1  20  2  Table 3.17: Effect of diffuser number on Group 2 Kia and OT,. 20  3.3.1  Air Flow Rate  The airflowrate supplied to the diffusers produced the most significant result (Figure 3.6). The cell means for Kx,a and OT, (Standard Deviation in brackets) for the two air 20  flow rates examined, lowflow(9.4 L/min) and mediumflow(18.8 L/min) are shown in Table 3.16. The net result is that doubling the airflowrate produced a 90% increase in Ka L  20  3.3.2  and OT,. Number of Diffusers  The number of silica glass diffusers used in the experiment (1 or 2) produced the next most significant result (Figure 3.7). The cell means for Kia o and OT, (Standard Deviation 2  in brackets) are shown in Table 3.17. The net result is that increasing the numbers of diffusers from 1 to 2 at a constant airflowrate produced a 25% increase in K^,ao and 2  OT,.  Chapter 3.  Results  SI  9.4 L/min E223 18.8 L/min Figure 3.6: Effect of air flow rate on Group 2 K£,a o, E and E . 2  p  D  Chapter 3.  Results  BM KLa Eo Figure 3.7: Effect of diffuser number on Group 2 K ^ o and E . D  Chapter 3.  41  Results  Treatment Cover No Cover  K a (hr- ) 17.8 (6.8) 16.6 (5.9) 1  L  2 0  O T , (g Q /kW-hr) 11.0 (4.2) 10.3 (3.7) 2  n 40 40  Table 3.18: Effect of surface cover on Group 2 Ki,a o and OT». 2  3.3.3  Air Flow Rate by Number of Diffusers Interaction  A significant air flow rate by number of diffusers interaction effect was observed for K £ a  20  and O T , . A graphical plot of the data reveals that the magnitude of the difference, between the cell means for Kx,a o at one or two diffusers, increases with increased air 2  flow rate (Figure 3.8).  3.3.4  Surface Cover  A barely significant inverse result was observed for the cover-no cover treatment. The cell means for Kx,a  3.3.5  20  and O T , (Standard Deviation in brackets) are shown in Table 3.18.  Orifice Size and Replicate Days  There was no significant effect of orifice size or replicate days on K i a  2 0  and O T , in  the Group 2 experiments. The cell means for orifice size and replicate days (Standard Deviation in brackets) are shown in Table 3.19.  3.4  Group 2: E and E D  p  The results for the Anova on Group 2 E and E are shown in Appendix B. A significant c  p  result (a < 0.01) was obtained for the number of diffusers treatment. There were no significant results from the air flow rate, orifice size, surface cover, interaction or replication  Chapter 3.  Results  —— 9.4 L/min  H ~ 18.8 L/min  Figure 3.8: Group 2 air flow rate by number of diffusers interaction effect on Kr,a o 2  43  Chapter 3. Results  Treatment K a (hr" ) OT, (g Q /kW-hr) 40 p 17.5 (6.6) 10.8 (4.1) 140 p 17.0 (6.2) 10.5 (3.9) Day 1 17.5 (8.2) 10.8 (5.1) Day 2 17.7 (6.4) 11.0 (4.0) Day 3 17.3 (6.0) 10.7 (3.7) Day 4 16.6 (5.8) 10.3 (3.7) Day 5 17.1 (5.9) 10.6 (3.7) 1  x  20  2  n 40 40 16 16 16 16 16  Table 3.19: Effect of orifice size and replication on Group 2 K£a and OT,. 20  Treatment One Diffuser Two Diffuser  E (%) E (g Q /kW-hr) n 3.8 (0.52) 201.1 (27.7) 40 4.6 (0.41) 248.8 (21.8) 40 c  p  2  Table 3.20: Effect of diffuser number on Group 2 E and E. D  treatments. 3.4.1  Number of Diffusers  The number of silica glass diffusers (1 or 2) used produced the only significant effect on E and E from the Group 2 treatments (Figure 3.7). The cell means for E and E D  p  D  p  (Standard Deviation in brackets) are shown in Table 3.20. 3.5  Group 3: K z , a  20  and OT,  The results for the Anova on Group 3 Kx,a and OT, are shown in Appendix C. Signif20  icant effects (a < 0.01) were observed for orifice size and surface cover. There were no significant effects of replication or interaction.  Chapter 3.  44  Results  Treatment 40 p 397 p 1588 p  K a (hr" ), OT, (g Q /hr) 6.3 (0.5) 13.3 (1.0) 3.7 (0.3) 7.8 (0.6) 2.0 (0.1) 4.3 (0.3) 1  L  20  2  n 15 15 15  Table 3.21: Effect of orifice size on Group 3 K.&ao and OT,. 2  Treatment K (hr" ) OT, (g 0 /hr) n Cover 3.8 (1.8) 8.1 (3.8) 15 No cover 4.3 (2.0) 9.0 (4.1) 15 No cover + wind 3.9 (1.8) 8.3 (3.8) 15 1  i a 2 0  2  Table 3.22: Effect of surface conditions on Group 3 K£a and OT,. 20  3.5.1  Orifice Size  Orifice size was highly significant. The cell means for Kx,a and OT, (Standard Deviation 20  in brackets) for the three orifice sizes studied (40 p, 397 p and 1588 /z) are shown in Table 3.21. Scheffe's test was used in the comparison among means test for the three orifice sizes examined. Scheffe's test indicates each diffuser is significantly different (a = 0.01) from each other for K/,a and OT,. 20  3.5.2  Surface Cover  A marginally significant effect was produced by the surface cover treatment. The three treatment (cover, no cover, no cover plus wind) cell means for K£a o and OT, (Standard 2  Deviation in brackets) are shown in Table 3.22. Scheffe's test was used in the comparison among means test for the three surface treatments. The test indicates no two treatments were significantly different at the a = 0.01 level.  Chapter 3.  45  Results  Treatment E (%) E (g 0 /kW-hr) n 40 p 2.6 (0.2) 140.1 (10.4) 15 397 fi 1.5 (0.1) 82.1 (6.6) 15 1588 fi 0.8 (0.1) 44.7 (3.1) 15 p  p  2  Table 3.23: Effect of orifice size on Group 3 E and E . D  3.6  Group 3: E and E 0  p  p  The results for the Anova on Group 3 E and E are shown in Appendix C. Signifi0  p  cant results were obtained for orifice size and surface cover. There were no significant replication or interaction effects. 3.6.1  Orifice Size  Orifice size was highly significant, and the cell means for the three orifice sizes studied (40 p, 397 fi and 1588 fi) (Standard Deviation in brackets) are shown in Table 3.23. Scheffe's test indicates each diffuser orifice size is significantly differentia = 0.01) from each other for E„ and E , and that E„ and E increase with decreasing orifice size. p  3.6.2  p  Surface Cover  A marginally significant effect was produced for the surface cover treatment, and the cell means for the three treatments (Standard Deviation in brackets) are shown in Table 3.24. Scheffe's test was unable to distinguish a significant difference between the treatments (a = 0.01) in the comparison among means test.  Chapter 3.  46  Results  Treatment E (%) E (g 0 /kW-hr) n Cover 1.6 (0.8) 85.2 (40.2) 15 No cover 1.8 (0.8) 94.5 (43.2) 15 No cover + wind 1.6 (0.7) 87.3 (39.8) 15 0  p  2  Table 3.24: Effect of surface conditions on Group 3 E and E . D  p  Orifice (u) Air Flow (L/sec) Mean Equivalent diameter (mm) n CV 40 p 18.8 3.8 20 15.5 40 p 9.4 4.2 20 23.1 140 p 18.8 4.5 20 26.9 140 p 9.4 3.8 20 21.8 397 p, 18.8 5.0 20 16.6 397 p 9.4 4.4 20 21.4 794 p 18.8 4.4 20 42.5 794 p 9.4 7.6 7 21.5 1588 /x 18.8 7.1 20 31.8 1588 p 9.4 8.2 20 44.3 3175 /x 18.8 7.8 20 31.0 3175 ^ 9.4 6.6 20 30.3 Table 3.25: Equivalent bubble diameter as a function of airflowand orifice size. 3.7  Bubble Size  The results of the bubble size analysis from the lab experiments are shown in Table 3.25. The results confirm the visual observations that smaller orifice sizes generate smaller bubbles. A clear trend toward increasing bubble size with increasing orifice diameter was obtained (Figure 3.9). There was no obvious trend for the effect of airflowon bubble size for a given orifice size. The coefficient of variation increased with increasing bubble size. This is also in agreement with visual observations as orifice sizes larger than 397 p generated a wider range of bubble sizes, which continuously coalesced and fragmented during their ascent through the water column.  Chapter 3.  47  Results  Bubble size (equivalent diameter, mm)  w Wv//<  WA  /A?  140 u  SH  397 u  18.8 L / m i n  794 u  1588 u  3175  u  9.4 L / m i n  Figure 3.9: Equivalent bubble diameter as a function of air flow rate and orifice size.  Chapter 3.  Results  48  Treatment Water Velocity (m/sec) Oxygen Input (mg/L) Daily Load (kg) n 3m 0.17 (0.04) 0.62 (0.19) 4.0 (1.2) 20~ 7m 0.38 (0.02) 0.73 (0.15) 10.9 (2.1) 20 Table 3.26: Effect of diffuser depth on Group 4 (field experiments) water velocity, oxygen input and daily load.  Treatment 3m 7m  E (%) E (kg Q /kW-hr) n 0.4 (0.13) 0.09 (0.03) 20 1.2 (0.23) 0.24 (0.05) 20 p  p  2  Table 3.27: Effect of diffuser depth on Group 4 (field experiments) E and E . D  3.8  p  Group 4: Water Velocity, Oxygen Increase Per Cycle, Daily Oxygen Load, E and E 0  p  The results for the Anova on Group 4 water velocity, oxygen increase per cycle, daily oxygen load, E and E are shown in Appendix D. D  3.8.1  p  Depth of Air Diffuser  The depth of the air diffuser in the inflow tube had a significant effect on water velocity in the outflow tube, oxygen increase per cycle, daily oxygen load, E„ and E (Figure p  3.10). The cell means (Standard Deviation in brackets) for the two depths of air release (3 m and 7 m) are shown in Table 3.26 and 3.27. 3.8.2  Orifice Size  Orifice size had a significant effect on oxygen increase per cycle, daily oxygen load, E and e  E , but did not significantly affect water velocity in the outflow tube (Figure 3.11). The p  cell means for water velocity, oxygen input, daily load, E and E (Standard Deviation in D  p  Chapter 3.  ^4  49  Results  Water velocity, oxygen input and Eo .  Velocity (m/sec) Oxygen (mg/L)  Eo (%)  SUB 3 meters 7 meters Figure 3.10: Effect of diffuser depth on Group 4 (field experiments) water velocity, oxygen input and E . 0  Chapter 3.  50  Results  Water velocity, oxygen input and Eo 1.6 n  :  Velocity (m/sec) Oxygen (mg/L) K B 140 u  Ez23 794 u  ESU 1599 u  Eo (%) ESS] 3175 u  Figure 3.11: Effect of Group 4 (field experiments) orifice size on water velocity, oxygen input and E„.  Chapter 3.  Treatment 140 p 794 n 1588 p 3175 p  Results  51  Water Velocity (m/sec) 0.27 (0.12) 0.30 (0.11) 0.27 (0.12) 0.27 (0.12)  Oxygen Input (mg/L) Daily Load (kg) 0.89 (0.12) 9.4 (4.8) 0.60 (0.14) 7.1 (3.1) 0.59 (0.10) 6.6 (3.6) 0.61 (0.12) 6.7 (3.7)  n lO 10 10 10  Table 3.28: Effect of orifice size on Group 4 (field experiments) water velocity, oxygen input and daily load.  Treatment E (%) E (kg 0 /kW-hr) 140 fi 1.0 (0.52) 0.21 (0.11) 794 fi 0.8 (0.34) 0.16 (0.07) 1588 fi 0.7 (0.40) 0.15 (0.08) 3175 p 0.7 (0.41) 0.15 (0.08) 0  p  2  n 10 10 10 10  Table 3.29: Effect of orifice size on Group 4 (field experiments) E and E . 0  p  brackets) for the four orifice sizes examined (140 fi, 794 p, 1588 fi and 3175 p) are shown in Table 3.28 and 3.29. Scheffe's test was used in the comparison among means test for the four orifice sizes examined. The test indicated that the 140 fi orifice diffuser was significantly different (a — 0.01) from the other three diffusers for oxygen input, and the remaining three diffusers (794 fi, 1588 fi and 3175 fi) were not significantly different from each other. Scheffe's test was unable to distinguish any significant differences among the four orifice sizes when applied to the cell means for daily oxygen load, E and E . The D  F values for daily oxygen load (« 12), E (« 10) and E D  p  p  12) were barely significant  in the Anova, however Scheffe's test is conservative by design and just misses separating the 140 fi diffuser from the other three diffusers.  Chapter 3.  Results  52  Treatment Water Velocity (m/sec) Oxygen Input (mg/L) Daily Load (kg) n 140> 0.40 (0.04) 0.92 (0.09) 14.5 (1.9) IF 3175 fi 0.43 (0.01) 0.24 (0.02) 10.5 (1.0) 10 Table 3.30: Effect of orifice size on Group 5 (field experiments) water velocity, oxygen input and daily load.  Treatment 140 ft 3175 p  E (%) E (kg Q /kW-hr) n 1.6 (0.21) 0.32 (0.04) 10 1.2 (0.27) 0.63 (0.05) 10 p  p  2  Table 3.31: Effect of orifice size on Group 5 (field experiments) E and E . D  3.8.3  p  Replicate Days and Interactions  There were no significant effects of replicate days or interactions on water velocity, oxygen increase per cycle, daily oxygen load, E and E in the Group 4 experiments. D  3.9  p  Group 5: Water Velocity, Oxygen Increase Per Cycle, Daily Load, E and E  D  p  The results for the Anova on Group 5 water velocity, oxygen increase per cycle, daily oxygen load, E and E are shown in Appendix E. c  3.9.1  p  Orifice Size  Orifice size had a significant effect on oxygen increase per cycle, daily oxygen load, E  D  and E but did not significantly influence water velocity in the outflow tube. The cell p  means (Standard Deviation in brackets) for the two orifice sizes (140 fi and 3175 p) are shown in Table 3.30 and 3.31.  Chapter 3.  3.9.2  53  Results  Surface Cover, Replicate Days and Interaction  Surface cover, replicate days and interactions had no significant effect on water velocity, oxygen increase per cycle, daily oxygen load, E and E in the Group 5 experiments. AlD  p  though the cover-no cover treatment was not significant, the data trend suggests slightly higher values for all measured parameters occurred when the surface cover was absent.  Chapter 4 Discussion: Gas Transfer Theory  A brief review of gas transfer theory and the non-steady state aeration test is required before discussing the experimental results. 4.1  Gas Transfer Theory  Several theories have been proposed to describe the mass transfer of a sparingly soluble gas such as oxygen to water (Krenkel and Orlob, 1962). Although these theories are approximations of the actual physical process, the theory which best describes the transfer mechanism was proposed by Dobbins (1964). This theory is a combination of the classic Whitman Two Film theory (Lewis and Whitman, 1924) and the Surface Rejuvenation theory (Dankwertz, 1951). The Two Film theory was developed in 1924 (Lewis and Whitman, 1924) and later revised by Ippen et al. (1952) for the computation of oxygen absorption rates in water (Mavinic and Bewtra, 1974). According to this theory, a gas passes through two films of gas and liquid, respectively, by molecular diffusion and the mass transfer is driven by a partial pressure gradient in the gas phase and a concentration gradient in the liquid phase. Initially oxygen molecules from the gas phase are transported to the liquid film surface, resulting in saturation conditions at the interface. For slightly soluble gases such as oxygen, the gas film offers very little resistance and this phase proceeds rapidly. The liquid interface or film is estimated to be at least three molecules thick and composed of water molecules oriented with their negative sides (ie. oxygen) facing the gas phase 54  Chapter 4.  Discussion:  55  Gas Transfer Theory  (Eckenfelder and Ford, 1968). In the next phase, oxygen molecules slowly pass through this film by molecular diffusion. All of the resistance to the passage of oxygen into the water is due to molecular diffusion across the liquid film (Eckenfelder, 1959). In the final stage, oxygen is mixed into the water by diffusion and convection currents. At low mixing levels (ie. laminar flow conditions) the rate of oxygen absorption is regulated by the rate of molecular diffusion through the undisturbed liquid film and the Two Film theory holds true. However, as turbulence levels increase, the surface film is disrupted and renewal of the film becomes responsible for transferring oxygen to the liquid (Eckenfelder, 1969). Dobbins (1964) resolved this dilemma by developing the following equation to describe the liquid film coefficient (KL) (King, 1970): L  (4.10)  where: KL = liquid film coefficient (m/hr) DL — diffusion coefficient for oxygen (m /hr) 2  r — rate of surface renewal P = liquid film thickness (m). When the rate of surface renewal (r) is near zero (ie. laminarflowconditions), equation [4.10] reduces to: K  L  =  D /l L  (4.11)  and the transfer is controlled by molecular diffusion through the liquid film according to the Two Film theory. As the rate of renewal increases equation [4.10] reduces to: K  L  = VD r L  (4.12)  Chapter 4.  Discussion:  56  Gas Transfer Theory  and the mass transfer becomes a function of the rate of surface renewal as described by the Dankwertz Surface Rejuvenation theory. These two models may be regarded as the limits between which both transfer mechanisms contribute to the overall oxygen transfer process. In the transitional zone between molecular diffusion and turbulent mixing, the process may be visualized as a transfer, in series, through the diffusional sublayer, at the interface whose boundary layer is subjected to turbulence and subsequent surface renewal (O'Connor, 1982). Mathematically, this theory can be expressed as follows (Mavinic and Bewtra, 1974): (4.13)  dm/dt = K A{Ci-C ) L  L  where: = time rate of mass transfer (g/hr)  dm/dt  KL = liquid film coefficient (m/hr) A = interfacial or absorbing surface area of air (m ) 2  d — saturation value of dissolved oxygen at the interface between liquid and air bubble (mg/L) CL = average concentration of dissolved oxygen in the bulk liquid (mg/L). This mass equation may be expressed in concentration units by introducing the volume of the liquid (V): dc/dt = 1/V dm/dt = K ^(Ci  - C ) = K a(Ci  L  L  L  (4.14)  - C) L  where: dc/dt  = rate of oxygen transfer (g/L/hr)  V = volume of the liquid m  3  A/V = a = the interfacial surface area of the air through which diffusion can occur generated by the particular aeration system per unit volume of water (m /m ) 2  KL&  = overall oxygen transfer coefficient  (hr' ) 1  3  Chapter 4.  Discussion:  57  Gas Transfer Theory  In practice, it is difficult to measure KL and (a) directly due to the hypothetical nature of 1 in KL (according to the Two Film theory) and the technical difficulties of measuring (a) (ie. A/V). Therefore, it has become standard practice to consider the aeration process in terms of the overall oxygen transfer coefficient  (KLO.)  when evaluating  aeration equipment (Nienow, 1980). 4.2  Non-Steady State Reaeration Test  The experimental procedure used to obtain KLU is the clean water non-steady state reaeration test as defined by APHA et al., (1980). As mentioned earlier, the non-steady state reaeration test involves deoxygenating a known volume of water and measuring the rate of reoxygenation. The slope of the DO deficit vs time data when plotted on onecycle semi-logarithmic paper is the overall oxygen transfer coefficient  (KLO,T)-  Although  several different aeration tests exist (Beak, 1977), the non-steady state test is regarded as the most accepted method for comparing the performance of various aeration systems (Lakin and Salzman, 1979; Ewing et al., 1979). The reason this test has become an industry standard is due to its relative simplicity and ability to generate reasonably accurate results. The use of clean water ensures the transfer process involves diffusion only and is not confounded by chemical or biological reactions. However, a number of assumptions and proper test procedures are required to ensure the test results are meaningful. The following assumptions are implied by the non-steady state test: 1. The overall mass transfer process occurs according to the Whitman Two Film theory, and the main resistance to gas transfer is due to molecular diffusion across the liquid film. 2. The test basin is completely mixed throughout the test period.  Chapter 4.  Discussion:  Gas Transfer Theory  58  3. The Ki,a of the aeration system is constant throughout the test and independent of test duration and dissolved oxygen concentration. 4. Environmental conditions of air temperature, wind velocity and relative humidity may be ignored during the test runs, due to the assumption that the main resistance to gas transfer is due to the liquid film. 5. The dissolved oxygen saturation concentration remains constant throughout the test duration, and is influenced only by changes in total dissolved solids, water temperature and atmospheric pressure (Landberg, et al. 1969). The general acceptance of the non-steady state reaeration test by the scientific and industrial community indicates the Whitman Two Film theory is adequate for describing the gas transfer process and the stated assumptions are valid for the test procedures. However, a number of precautions must be taken to ensure the non-steady state test is properly conducted. These include: 1. Limiting the test time to periods of 10-30 minutes. 2. Using an oxygen probe to avoid measurement of entrained gas bubbles. 3. Collecting at least 6 data points between 10% and 80% saturation. 4. Adding the cobalt chloride first, only once, and maintaining its concentration below 0.05 mg/L. 5. Conducting a maximum of 10 tests on a single batch of water. 6. Using linear least squares analysis to eliminate hand drawn line errors (Beak, 1977).  Chapter 4.  Discussion:  Gas Transfer  Theory  59  Air temperatures should be less than 10 degrees C different from water temperatures during the experiments and the sodium sulfite should be premixed in a slurry before adding it to the test water, as recommended by APHA et al. (1980). In summary, although the non-steady state test is not perfect, it remains the best method for realistically comparing various aeration devices. Provided the proper test procedures are followed, the non-steady state test is capable of an accuracy of plus or minus 5% of the true Kia value (Beak, 1977), and is reproducible within 8% for tests involving a single aerator in a single basin (APHA et al., 1980).  Chapter 5 Discussion: Group 1-3 Laboratory Experiments  A variety of factors influence the rate of oxygen transfer into water in diffused aeration systems. These include: 1. Oxygen concentration gradient (C—Cx,). 2. Temperature of the liquid. 3. Turbulence in and around the gas-liquid interface. 4. Depth of the liquid. 5. Contact time of the air bubble in the liquid. 6. Size of the air bubble. 7. Rate of air flow. 8. Type of diffuser. 9. Position of the diffuser. 10. Tank geometry (Mavinic and Bewtra, 1974). The individual or collective effect of these factors directly influences the liquid film coefficient (KL), the interfacial area for gas transfer (a), the overall oxygen transfer coefficient (Ki,a) and the oxygen concentration gradient (C,—Cx) (Mavinic and Bewtra, 1976). 60  Chapter 5.  Discussion:  Group 1-3 Laboratory  61  Experiments  Although all of these factors are important, the three key factors controlling the mass transfer of oxygen are those included in equation [4.14]: dc/dt  — Ki,a{Ci  — CL)  1. the hydrodynamics of the system which influence K^; 2. the area of contact between the gas and the liquid (a); 3. the concentration gradient between the gas and liquid phase  (C—CL)  (Nienow,  1980). The rationale of the laboratory experiments (Groups 1 to 3) was to maintain a relatively constant concentration gradient  (C;-CL)  via the non-steady state deoxygenation  procedure, and examine the effects of air flow rate, orifice size, diffuser number and surface conditions on Kia o, OT , E and E . Although Kxa is the parameter which 2  s  p  D  20  provides the basic information on the characteristics of each experimental configuration, the two most relevant parameters for comparing aeration systems are E and E (Bewtra p  0  and Mavinic, 1978; Beak, 1977). The effect of varying these factors on Kx,a , OT,, E 20  p  and E are discussed below. D  5.1  Group 1  The treatments examined in Group 1 were: airflowrate (9.4 L/min or 18.8 L/min); surface cover (present or absent) and orifice size (397 p, 794 p, 1588 p and 3175 p). 5.1.1  Air Flow Rate  The airflowrate (9.4 or 18.8 L/min) to the diffusers produced the most significant result, increasing Kj^a and OT„ 122% with a 100% increase in airflowrate (Q ). As mentioned a  earlier, the values for OT„ are derived from the Kjra , so these values respond similarly 20  to treatment effects.  Chapter 5.  Discussion:  Group 1-3 Laboratory  62  Experiments  A number of researchers have observed a similar effect of flow on Kx,a (eg. Bewtra 20  et al., 1970; Schmit et al., 1978) and at least two mechanisms are responsible for this result. Firstly, the increased volume of air injected into the column greatly increases the turbulence of the air-water interface. As mentioned earlier, mass transfer becomes a function of the rate of renewal of the liquid film under highly turbulent conditions, as described by the Surface Rejuvenation theory (Dobbins, 1964) [4.12]:  KL  —  y/Di,r.  As r  (the rate of surface renewal) increases, K/_, increases. Therefore, one mechanism by which higher air flow rates generate larger K£,ao values is through their effect on KL2  Secondly, higher air flow rates increase the number of bubbles present in the water column per unit time. This increases the total interfacial area available for the transfer of oxygen to the surrounding liquid (Mavinic and Bewtra, 1974). Therefore higher air flow rates also influence Kj,a via their effect on (a). 20  Higher airflowrates produced a 12% increase in Eo and Ep. This effect is also a result of higher turbulence and A/V ratios in the aeration column as previously discussed, and the size of bubbles and type of circulation pattern in the aeration column. Some of the literature surveyed reported decreased transfer efficiency with increased air flow rate (eg. Bewtra and Nicholas, 1964; Ellis and Stanbury, 1980; Mavinic and Bewtra, 1976). The general explanation for this response is that air bubbles become larger with an increase in Q . This results in less oxygen transfer due to the reduced a  ratio of interfacial area to bubble volume, and decreased bubble-water contact time resulting from increased rise velocities (Mavinic and Bewtra, 1976). This effect was not seen in the Group 1 experiments for several reasons: 1. The type of water column used in the experiments has a low circulating water velocity, similar to the System I described by Bewtra and Mavinic (1978). As a result, there is less additive effect of water velocity and the bubbles are assumed  Chapter 5.  Discussion:  Group 1-3 Laboratory  63  Experiments  to be rising at their terminal velocity. This results in longer contact times than in systems with a circulating water velocity (eg. System II; Mavinic and Bewtra, 1976), and is less subject to decreasing E with increased air flow rates. Among the D  4 types of systems investigated by Mavinic and Bewtra (1976), the simple column of water (System I) exhibited the least E response to increasing air flow rate at D  shallow diffuser depths. 2. There was no obvious effect of air flow rate on bubble size (see Table 3.25). This negated the usual effect of decreased interfacial area and contact time resulting from increased air flow rate. 3. The type of diffusers and resultant bubble sizes generated in these experiments are less influenced by increased air flow rate than fine bubble diffusers. Bewtra and Nicholas (1964) observed no change with E„ with increased airflowrate when using coarse bubble diffusers (Spargers), but observed a decline in E with Q when D  n  using fine bubble diffusers (Saran tubes). Ellis and Stanbury (1980) reported no significant change in E with increasing Q for coarse bubble diffusers at depths D  a  less than 4 m. However, at depths exceeding 4 m, coarse bubble E increased with D  Q , due to longer contact time-and increased bubble shear and turbulence. Schmit Q  and Redmon (1975) and Schmit et al. (1978) also reported increased E with Q , D  a  when testing coarse bubble diffusers in deep tanks. 4. The type of circulation pattern in the simple column generates more turbulence than circulating systems. In a confined column, there is considerable turbulence generated between the centrally rising air-water mixture and the adjacent water flowing in the reverse direction, to replace the water which is carried up in the air-water stream (Morgan and Bewtra, 1960). Dye additions showed considerable turbulence and eddy formation in the zone between the opposing flows. Thus,  Chapter 5.  Discussion:  Group 1-3 Laboratory  64  Experiments  Kxa o values would tend to increase with Q . Mavinic and Bewtra (1974) also 2  a  concluded that the simple column (System I) generated the highest turbulence and Kz,a values. 20  In summary, E and E increased with air flow rate over the range of air flows tested. D  p  Although this appears contrary to the usual response, further investigation of the literature reveals a number of instances where E remains constant or increases with increasing D  Q , when using coarse bubble diffusers in deep tanks (greater than 4 m) or in simple 0  columns. 5.1.2  Orifice Size  Orifice size exerted a significant effect on Kt,a o, OT,, E and E . Each parameter 2  p  D  increased with a decrease in orifice size. An examination of Table 3.25 (Bubble Size) provides the explanation for this result. A clear trend towards increasing bubble size with increasing orifice diameter was obtained for the 397 \i to 3175 fi diameter orifice range. The mean bubble size for the 3175 \i diameter orifice appears slightly smaller than the 1588 p diameter orifice (7.2 mm vs 7.7 mm); however, the bubble measuring procedure was unable to accurately estimate the high end of the 3175 fi diameter orifice bubbles due to their extremely large size. A reduction in bubble size produces three distinct results: 1. An increase in surface area per unit bubble volume (Eckenfelder, 1969). . 2. A decrease in terminal rise velocity (Stenstrom and Gilbert, 1981) 3. A decrease in the liquid film coefficient (K^) (Bewtra and Nicholas, 1964). An increase in bubble surface area per unit volume increases the (a) in K^a and acts to increase Ki,a o, OT,. A decrease in terminal rise velocity increases the bubble contact 2  Chapter 5.  Discussion:  Group 1-3 Laboratory  65  Experiments  time which acts to increase E and E . However, a decrease in terminal rise velocity D  p  decreases the liquid film coefficient (Kx,), which will decrease Kx,a, OT,, E and E . D  p  The interaction between these opposing factors determines the net effect on Kx,a , 20  0T , E and E of a decrease in bubble size. In this case, the net effect was an increase S  p  Q  in Kx,a o, 0T„, E and E indicating that the effect of increased surface area and contact 2  p  OJ  time more than compensated for the reduction in  due to lower terminal rise velocities  and liquid film coefficients. This suggests that in a simple column, interfacial area and contact time are important parameters for oxygen transfer in 4 to 8 mm diameter bubbles. A number of researchers have reported a similar effect of increased E„ with decreasing bubble size. For example, Morgan and Bewtra (1960) and Bewtra and Nicholas (1964) both observed increased E with fine bubble diffusers (Saran tubes) as compared to coarse 0  bubble diffusers (Spargers). In contrast, few papers were located which examined the effect of bubble size on Kx,a . 20  Mavinic and Bewtra (1976) examined Kx,a in a simple column, however their diffuser 20  orifice size was fixed at 1600 p diameter, so a comparison with the bubble sizes generated in these experiments is not possible. Bernhardt (1969) was the only paper located which examined this aspect of gas transfer, and his study showed a clear decrease in K^a with increasing bubble diameters of 2 mm and larger. The Scheffe's test used in the statistical analysis of Group 1 results indicated each orifice size was significantly different from each other with respect to E and E . HowD  p  ever, Scheffe's test separated only the 397 p diameter orifice from the rest of the orifice diameters (794 p, 1588 u and 3175 p) with respect to Kia and OT.. Scheffe's test is 20  conservative by design, and is the most rigorous a posterior test for performing comparisons among means (Larkin, 1975). Although Scheffe's Test indicates Kx,ao and OT, 2  are not significantly different for the 794 p, 1588 p, and 3175 p orifice diameters, a clear trend is present and a larger sample size should show significant separation at the a =  Chapter 5.  Discussion:  Group 1-3 Laboratory  66  Experiments  0.01 level. 5.1.3  Surface Cover  The presence of a floating surface cover did not exert a significant effect on Kx,a and 20  OTs, but a marginally significant effect was detected for E and E . The cell means for D  p  Kx,a o, OT,, E and E suggest that slightly lower values occur when a floating surface 2  p  D  cover is present. However, due to the non-significance of Kx,a and OT, and the marginal 20  significance of E and E„ (F « 7-11), a larger sample size is required to adequately assess p  this treatment effect. Theoretically, lower K^a, OT,, E and E should occur with a floating surface cover, p  D  as this would tend to decrease the transfer of atmospheric oxygen at the turbulent airwater interface generated by the bursting air bubbles. Nielson (1974) observed reduced rates of oxygenation in similar laboratory experiments, using floating styrene foam. One possible explanation for the marginal result observed in these experiments is the relatively small surface area to volume ratio of the experimental column (A/V = 0.94 m ) as compared to Nielson's (1974) tank (A/V = 1-10 rn ). As the surface area -1  -1  to volume ratio increases, the effect of reducing the surface component of gas transfer should become more apparent. Nielson (1974) estimates that natural surface aeration induced by a rising bubble plume should account for 88 to 94% of the oxygen transfer to small water supply reservoirs; however, the surface area to volume ratio of lakes is considerably larger than the 70 L experimental tank. 5.1.4  Interaction  The significant airflowrate x orifice size interaction effect detected for Kx,a is graphi20  cally presented in Figure 3.3. This researcher's interpretation of this interaction effect is that higher turbulence generated by larger airflowrates caused small bubbles to remain  Chapter 5.  Discussion:  Group 1-3 Laboratory  67  Experiments  trapped in vortices near the top of the 70 L cylinder. This effect was quite noticeable during the experiments, as the small bubbles generated from the 397 p and 794 p diffusers persisted much longer at the higher flows. This resulted in a longer contact time in the water column, which increased the Kia o 2  and OT, above what would be normally be expected for a given air flow and orifice size. Although the effect is significant, its F value (« 100) is small in relation to the main effects of the experiment (air flow rate and orifice size), and it does not change the principle conclusions of the experiment. Bewtra and Nicholas (1964) observed a similar entrainment effect when downward water velocitites exceeded the rise velocities of small bubbles, resulting in longer detention times and increased oxygen transfer. 5.1.5  Replication  A significant replication effect was detected for Kj,a2o, OT,, E and E . The F value of p  0  the replication effect is small (« 15-20) in relation to the main experimental results and does not change the overall conclusions of the experiment. The result simply indicates that the experimenter was somewhat variable in his experimental procedure during the first few days of the experimental period. 5.2  Group 2  The treatments examined in Group 2 were: air flow rate (9.4 L/min or 18.8 L/min), surface cover (present or absent), orifice size (40 p or 140 p) and diffuser number (1 or 2)-  Chapter 5.  5.2.1  Discussion:  Group 1-3 Laboratory  68  Experiments  Air Flow Rate  The rate of air flow (9.4 or 18.8 L/min) to the diffusers produced the most significant result, increasing Kr,a o and OT, by 97% with a 100% increase in Q . The mechanisms 2  a  responsible for this result are the same as in Group 1, ie. increased turbulence and interfacial area. However, doubling the air flow rate had no effect on E„ and E . Unlike p  the Group 1 response where a 100% increase in air flow rate produced a 122% increase in Kz,ao and OT, and a 12% increase in E and E , the Group 2 response showed only 2  D  p  a 97% increase in K£,ao and OT, and no increase in E and E . 2  D  p  This differential response is due to the smaller bubbles produced by the Group 2 silica glass diffusers. As shown in Table 3.25, the mean bubble diameter produced by the 40 \i and 140 \i silica glass diffusers was 4.0 and 4.2 mm respectively. A decline in transfer efficiency with increasing air flow is the usual response with fine bubble diffusers. Morgan and Bewtra (1960), Bewtra and Nicholas (1964) and Ellis and Stanbury (1980) observed decreased E with increasing Q . This response is most likely due to a combination D  a  effect of decreased oxygen absorption during bubble formation and interference from adjacent rising bubbles (Ellis and Stanbury, 1980). Increased air flow rates create a greater concentration of bubbles with relatively restricted lateral diffusion. This causes the so called "chimney effect," where oxygen transfer does not increase in proportion to Q , due to the resulting increase in resistance to lateral diffusion (Ippen and Carver, a  1954). The less than 1:1 response of increased Q on Kia o and OT, is also due to the smaller a  2  bubble size of the 40 fi and 140 /x diameter diffusers. As shown in Table 3.25, there was no clear effect of airflowrate on bubble size. As a result, higher airflowrates combined with the chimney effect did not produce a proportional increase in K L , the net effect being a less than 1:1 response of K£,a and OT, to Q . 20  a  Chapter 5.  5.2.2  Discussion:  Group 1-3 Laboratory  69  Experiments  Number of Diffusers  The purpose of this treatment was to examine the effect of air flow rate per diffuser, which is different from air flow rate per se. Increasing the number of silica glass diffusers from 1 to 2 resulted in a 25% increase in Kr_,ao d 0T„ and a 21-24% gain in E and E . a n  2  D  p  This response is well documented throughout the civil engineering literature (eg. Doyle et al., 1983; Morgan and Bewtra, 1960; Leary et al., 1969; Ippen and Carver, 1954). Bewtra and Nicholas (1964) concluded that this response was a combined effect of (1) an increase in oxygen absorption during bubble formation, (2) a change in bubble rise velocity with Q , (3) a change in bubble diameter and K L with Q„ and (4) a decrease a  in air-bubble entrainment with reduced Q . There was no obvious effect of air flow a  rate on bubble size in this set of experiments (Table 3.25), therefore reasons (2) and (3) from Bewtra and Nicholas (1964) may not be as important in this particular situation. It should be noted, however, that a maximum of 20 bubbles were measured for each combination of air flow and orifice size. Given the thousands of bubbles in the aeration column at any time, it is possible that the sample size estimation procedure was unable to detect an increase in bubble size with Q„. This researcher believes the explanation for  Kf,a2o,  OT„, E and E increasing with p  0  reduced air flow rate per diffuser is related to (1) and (4) above ie. increased gas transfer during bubble formation and reduced air-bubble entrainment. A high rate of gas transfer occurs at this stage, due to the continued expansion of the fresh gas-liquid interface and subsequent steep oxygen concentration gradient across the gas-liquid interface (Mancy and Okun, 1960). A reduction in gas flow rate per diffuser results in the production of smaller bubbles, reduces the likelihood of coalescence and allows better lateral diffusion through more uniform bubble dispersion (Ippen and Carver, 1954). The combined effect of these factors results in more interfacial area and contact time which increases Kx,a o, 2  Chapter 5.  Discussion:  Group 1-3 Laboratory  70  Experiments  OT , E and E . s  5.2.3  p  0  Surface Cover  The presence of a floating surface cover exerted a marginally significant inverse effect on Kx,ao and OT , and no effect on E„ and E . The cell means for K£,ao and OT, 2  a  p  2  suggest that slightly higher values occur when a floating surface cover is present. One possible explanation for this result is due to the longer path length that bubbles must take through the water due to the presence of a surface cover. Markofsky (1979) noticed a slight increase in E when a surface cover was present and attributed this to increased D  contact time. However, due to the non-significance of E and E and marginal significance D  p  of Kx,a and OT, (F % 7) a larger sample size is required to adequately assess this 20  treatment effect. 5.2.4  Orifice Size  The size of the silica glass diffuser orifices examined in the Group 2 experiments (40 and 140 u) had no effect on Ki,a o, OT„, E or E . The bubble size analysis supports 2  p  Q  this conclusion, as the mean bubble size created by the 40 p and 140 p diameter orifice diffusers were similar (Table 3.25). A scanning electron micrograph (SEM) of each diffuser surface revealed a distinct difference in the pore size and size of the silica granules that comprise the diffuser body. The SEM shows the orifices are not round holes, but rather irregular shaped openings in a bonded matrix of variable sized silica granules. The size of the bubble produced by a porous diffuser depends on the surface tension of the air-liquid interface, the porosity of the diffuser medium and the air flow rate through each diffuser, in addition to the pore size of the diffuser (Bewtra and Nicholas, 1964).  Chapter 5.  Discussion:  Group 1-3 Laboratory  Experiments  71  The size of a single bubble formed at low air flow rates from a single orifice is the result of a balance between the buoyant force of the bubble and the surface tension holding the bubble to the orifice. The bubble increases in size until its buoyancy exceeds the surface tension forces and it then detaches itself (Bowers, 1955). At low air flow rates, bubbles tend to emerge in single formation, with a relatively constant diameter of approximately 11 times the orifice diameter (Haney, 1954). However, as air flow rates increase, single bubbles cannot carry the gas away quickly enough so the bubbles become larger and leave the orifice in the form of a chain, with adjacent bubbles just touching (Bowers, 1955). The gas flow rate at which chain formation occurs is known as the critical point, above which bubble size becomes dependent on gas flow rate and is independent of orifice diameter. The mathematical derivation of the critical gas flow point for various sized bubbles is described by Bowers (1955). Visual observations of bubble formation with the 40 p. and 140 p silica glass diffusers indicated bubbles emerged in chain formation, so the gas flow rate per orifice was above the critical rate for single bubble formation. As a result, bubble size was dependent on gas flow rate and the resulting bubble sizes were similar for both the 40 u and 140 p. diffusors. Markofsky (1979) observed a similar effect with 90 p and 180 p porous diffusers and concluded there was no significant difference in transfer efficiency between the two orifice sizes at the experimental gas flow rates.  5.2.5  Interaction  The significant air flow rate x number of diffusers interaction effect is shown in Figure 3.8. The cause of this interaction effect is the same as in Group 1, ie. at higher air flow rates, smaller bubbles became trapped in the vortices near the top of the 70 L cylinder and inflated the Kta o and OT„ above what would be normally expected for a given air 2  flow and number of diffusers.  Chapter 5.  Discussion:  Group 1-3 Laboratory  72  Experiments  Although the effect is significant, its F value  13) is small in relation to the main  effects of the experiment (air flow rate and number of diffusers), and it does not alter the main conclusions of the experiments. 5.2.6  Replicate Days  There was no significant effect of replicate days on Kx,a o, OT,, E and E . This indicates 2  p  c  the experimenter was becoming more proficient in his experimental technique and/or the experimental equipment had stabilized from the initial Group 1 experiments. 5.3  Group 3  The treatments examined in Group 3 were: effect of orifice size (40 /x, 397 fi and 1588 /x) and effect of surface conditions in the 239 L tank (cover, no cover, no cover and wind). 5.3.1  Orifice Size  Orifice size exerted a significant effect on Kz,a , OT,, E and E with each parameter 20  p  OJ  increasing in value with decreasing orifice size. Scheffe's test indicates each diffuser size was significantly different from each other for Kx,a o, OT,, E and E . 2  D  p  These results are similar to the Group 1 and Group 2 results, and the factors responsible are the same ie. smaller bubbles increased interfacial area and contact time, thus increasing Kjr,a o, OT,, E and E . 2  5.3.2  p  D  Cover  A marginally significant result for K£,a , OT,, E and E was produced by the surface 20  p  0  cover treatment. The results (Table 3.22 and 3.24) indicate parameter values were highest  Chapter 5.  Discussion:  Group 1-3 Laboratory  73  Experiments  with no surface cover, and that the no cover and wind treatment effect was similar to thefloatingcover effect. Scheffe'5 test was unable to detect a significant difference between the three treatment  effects in the comparison among means test. Given the conservative nature of Scheffe's test and the marginally significant effect of the cover treatments, this result was expected. Therefore, a larger sample size is required to adequately assess the effect of surface cover. Regardless, it is interesting to speculate on the factors responsible for the observed result. Logically, the cover-no cover effect makes sense as the increasing surface area to volume ratio of the 239 L tank (A/V = 2.2 m  -1  for the 239 L tank, 0.94 m for the -1  70 L cylinder) should result in a more noticeable effect as the surface component of gas transfer increases in relative importance (eg. Nielson, 1974). However, the negative effect of the no cover and wind treatment effect is puzzling. One would expect an increase in gas transfer from the wind and wave action (Downing and Truesdale, 1955). One possible explanation is that the velocity and direction of the wind generated circulation currents changed the circulation within the 239 L tank to a less efficient pattern. For example, the maximum wind speed measured in the 239 L tank was 4.1 m/sec at a distance of 10 cm from the nozzle. The velocity of wind induced surface currents are approximatley 3% of wind speed (O'Connor, 1982), therefore a surface velocity of 13 cm/sec was possible. The velocity of the rising air-bubble mixture should approximate the rise velocity of individual bubbles, which ranged in size from 4 to 8 mm diameter. Therisevelocity of bubbles in this size range is described by: « = 1.02^1' where: u — terminalrisevelocity (cm/sec)  (5.15)  Chapter 5.  Discussion:  Group 1-3 Laboratory  74  Experiments  g = acceleration of gravity (980 cm/sec ) 2  r = equivalent bubble radius (cm) e  (Haberman and Morton, 1954). This formula predicts a rise velocity of approximately 20-29 cm/sec, hence the velocity of the outward flowing surface current should be similar. The effect of the outward flowing surface current meeting the wind-induced circulation current would be a 40 to 60% reduction in velocity for the surface current flowing directly into the wind-induced current in the upwind half of the tank. The net effect of reduced surface current may reduce the entrainment of small bubbles and reduce the Kj,a o, OT,, E and E . Bewtra 2  D  p  and Nicholas (1964) observed a similar "stilling phenomena" during their experiments on diffuser arrangements. They attributed the decline in transfer efficiency under certain diffuser arrangements to decreased water velocities and less bubble entrainment when two opposing air-water mixture.streams met. 5.4 5.4.1  Summary Analysis: Group 1-3 Comparison of K£a , O T , , E and E 20  p  D  The results from the Group 1-3 laboratory experiments are in agreement with the civil engineering literature, with respect to the effects of air flow, orifice size, air flow rate per diffuser and surface cover on K£,a , OT,, E and E„. An examination of the data sum20  p  maries in the Appendix (Appendix A, B and C) clearly shows that, among comparable performance variables (ie. E and E ), the highest values were obtained by discharging p  c  air through the largest number of diffusers, with the smallest (ie. 40-140 fi) orifice size. Within a given tank configuration (ie. 70 L or 239 L), the tank geometry sets the limits by which diffuser number and orifice size influence the hydrodynamics and contact time. This ultimately determines the Kj,a, OT,, E and E for each combination of diffuser p  Q  Chapter 5. Discussion:  Group 1-3 Laboratory  75  Experiments  number and orifice size.  Kz,a  5.4.2  20  The highest Kx,a o values were obtained i n the Group 2 experiments (ie. Treatment 16; 2  26.4 hr ),  and the lowest values were found i n the Group 3 experiments (ie. Treatment  -1  1; 1.9 hr~ ).  T h e main reason for this effect is the increased interfacial area of the  1  smaller bubbles, combined with the increased turbulence of the higher air flow rate i n the 70 L column. It is interesting to note that despite receiving the highest air flows (ie. 28.3 L/min), the Group 3 Kx,a o values were quite low. This suggests the turbulence 2  generated by the flow pattern i n the 239 L tank was considerably less than i n the 70 L cylinder.  Morgan and Bewtra (1960) and Bewtra and Mavinic (1976) also reported  considerable turbulence and high K i , a  5.4.3  E  2 0  values i n simple columns.  p  T h e highest E  p  values were also achieved i n the Group 2 experiments (ie. Treatment 4;  263 g 0 / k W - h r ) . T h e combination of multiple small orifice diffusers is clearly the most 2  efficient in terms of energy efficiency. Despite having the lowest K £ , a ments 7, 8 and 9) generated E  p  20  values, the fine bubble diffusers i n Group 3 (Treat-  values greater than 100 g 0 / k W - h r . 2  This is due to  the larger volume of the 239 L tank, which is taken into account i n the calculation for Ep. This demonstrates the utility of using E , rather than K £ , a , when comparing the p  20  performance of aeration systems. The range of E  p  values from the Group 1-3 experiments were much lower than the  values reported by Mavinic and Bewtra (1976) for a simple column system (53-263 g 0 / k W - h r vs 1,203-1,657 g 0 / k W - h r ) . There are several reasons for this wide range i n 2  2  Ep, including different diffuser submergence and size of aeration columns; however, one  Chapter 5. Discussion: Group 1-3 Laboratory Experiments  76  of the main reasons is the method in which energy consumption was calculated. Mavinic and Bewtra (1976) used the theoretical power required to compress the air, which does not include the efficiency of the electric motor and compressor. The power input for the Group 1-3 experiments was simply taken as the nameplate horsepower of the compressor, and adjusted for the fraction of air delivered. Pasveer (1966) estimates only 10% of the theoretical power is actually used in the transfer of oxygen, after correcting for compressor efficiency, motor efficiency, air line losses and diffuser resistance. 5.4.4  E  0  The highest E values were achieved in the Group 2 experiments (ie. Treatments 4, 8 D  and 16; 4.8-4.9%) while the lowest values were found in the Group 3 experiments (ie. Treatment 1, 2, and 3; 0.3%). The combination of multiple diffusers with small orifices was the most efficient at gas transfer.  As previously mentioned, air flow rate had no  significant effect on E in the Group 2 experiments. D  The low E values in Group 3 were mainly a result of the reduced turbulence as D  previously mentioned and the shallow depth of air release (ie. 0.34 m); this reduces contact time and the driving force (Ci - CL) for oxygen transfer. The fine bubble diffuser used in the Group 3 experiments were nearly 3 times more efficient at oxygen transfer than the 1588 \i diameter diffuser; however their overall E„ was still quite low compared to their performance in the 70 L column in the Group 2 experiments. 5.4.5  Optimum Bubble Size  Given the observed increase in K i ^ o , O T , E and E with decreasing bubble size, s  p  D  and the competing effects of large and small bubbles on rise velocity, contact time, KL and interfacial area, is there an optimum bubble size which generates the highest overall oxygen transfer coefficient? An examination of a rise velocity vs equivalent bubble  Chapter 5.  Discussion:  Group 1-3 Laboratory  Experiments  77  diameter curve (eg. Figure 2; Andeen, 1974) reveals a distinct pattern and the solution to this question. Initially, bubble velocity increases linearly with increasing size, to a local maximum of approximately 32 cm/sec for single bubbles of 1.2-1.4 mm diameter (Haney, 1954). Bubble velocity then decreases in the diameter range from 1.4 to 6 mm before increasing again with an increase in bubble diameter. This characteristic curve is due to the interaction between the hydrodynamic, viscous and interfacial forces acting on the bubbles (Haberman and Morton, 1954). Small bubbles assume a spherical shape as surface tension reduces the surface area to a minimum for a given volume. This spherical shape dictates that small bubblesriseaccording to Stoke's Law, and the viscosity of the liquid is the most important parameter influencing theirrisevelocity. As bubble size increases, the viscous and hydrodynamic forces acting on the bubble become more important andflatteningof the bubble occurs. This results in an oblate spheriod shape, which has a higher drag than a sphere of the same volume, and the rise velocity declines. As bubble size continues to increase, the viscous and surface tension forces become small relative to hydrodynamic forces and the bubble assumes a spheriod cap shape. The upper surface of these bubbles is essentially spherical, and they rise independently of the properties of the liquid (Haberman and Morton, 1954). An examination of a liquid film coefficient (ie. K^) vs bubble diameter graph reveals a sharply increasing Ki, to a peak at 2-2.5 mm diameter, then a gradual decline with increasing bubble size (eg. Figure 6; Bernhart, 1969). A plot of surface area vs bubble diameter results in an exponential decay type of curve (Figure 5.12). The net result of these three effects on Kx,a is a response profile similar to the Kj, vs bubble size curve. The Kxaincreases sharply to apeak at 2.0-2.5 mm diameter, then declines exponentially with increasing bubble size (see Figure 11; Barnhart, 1969). Therefore, the optimum bubble size for maximum K^a is 2.0-2.5 mm diameter.  Chapter 5.  Discussion: Group 1-3 Laboratory  Experiments  78  Chapter 5.  Discussion:  Group 1-3 Laboratory  79  Experiments  This confirms the observed response of orifice size on K^a o i the Group 1-3 expern  2  iments, and suggests that the bubble sizes generated by the silica glass diffusers in the Group 2 experiments were approximately twice as large as the optimum size. Eckenfelder and Ford (1968) state that silicon dioxide, aluminum oxide and Saran diffusers generally produce bubbles in the 2.0-2.5 mm range, which explains why these diffusers are so efficient at oxygen transfer. However, Downing (1966) states "... that practical difficulties, such as the clogging of orifices of air diffusers, limit the smallest size of bubble that it is feasible to produce by air diffusion alone to a diameter of about 2-3 mm." Therefore, the lower limit of practical bubble size formation from ceramic diffusers is also the optimum bubble size for oxygen transfer.  Chapter 6 Discussion: Group 4 and 5 Field Hypolimnetic Aeration Experiments  6.1  Group 4 Field Experiments  The treatments examined in the Group 4 field experiments were the effect of orifice size (140 a, 794 a, 1588 fi and 3175 u) and depth of air release (3 m or 7 m). 6.1.1  Depth of Air Release  The depth of air release in the inflow tube had a significant effect on water velocity in the outflow tube, daily oxygen load, E , E and oxygen increase per cycle. The parameter 0  p  most significantly influenced by the depth of air release was the water velocity in the outflow tube. The cell means (Table 3.26) indicate the water velocity at the 7 m release depth was 124% greater than the 3 m release depth (0.17 m/sec vs 0.38 m/sec). This effect is due to increased contact time in the inflow tube which influences the rise velocity of the air-water mixture. When air is injected at the 7 m depth, the period of contact is longer, thus allowing the hypolimnetic water additional time to accelerate and approach the rise velocity of the ascending bubbles. When air is injected at 3 m, the bubbles reach the surface and escape before the water mass has had enough time to approach the bubble rise velocity. This principle is known as riser efficiency and is used to describe the efficiency of air-lift pumps (Andeen, 1974): ' = vTv  n  80  ( r  6  1  6  )  Chapter 6. Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration  Experiments  81  where: n  = riser efficiency (%)  T  V = liquid velocity (m/sec) V  T  = velocity difference between liquid and gas (m/sec) Clearly it is desirable to maintain a minimum velocity difference to achieve high riser  efficiency. There is a paucity of published data to compare with these results, as the hypolimnetic aeration literature has not examined the effect of diffuser depth on water velocity. There are a number of papers which include water velocity as a function of air flow rate for a single fixed depth (eg.  Smith et al., 1975); however, the effect of diffuser depth on  water velocity per se cannot be determined, as each system has a different depth of air injection, rate of air flow and diffuser design.  There are a number of reports in the  civil engineering literature in which air was injected at different depths (eg. Morgan and Bewtra, 1960; Ippen and Carver, 1954; Schmit et al., 1978); however, these studies were generally concerned with the effect of diffuser submergence on transfer efficiency. Bewtra and Nicholas (1964) mention "that for a given liquid depth, the velocities are decreased when the diffuser submergence is decreased", but no data or explanation is given. The oxygen increase per cycle, E , E and daily 0 p  by the depth of air injection.  D  2  load were significantly influenced  Although the F value for oxygen increase per cycle is  marginal (F=8), the trend is in the expected direction and a larger sample size or a greater range of injection depths would undoubtedly result in a more significant effect. The F value for transfer efficiency (E ) (RS 327) is much larger as E is calculated on a 0  D  daily basis; therefore, the depth effect on velocity and subsequent induced flow is included in the E analysis. The E , E and daily load calculations are related and their statistical D  D  p  behaviour is similar. This effect is well documented in the civil engineering literature. For example, Morgan  Chapter 6.  Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration  Experiments  82  and Bewtra (1960), Bewtra and Nicholas (1964), Schmidt et al. (1978) and Doyle et al. (1983) all observed increased E with an increase in diffuser submergence. D  This effect  is a result of increased contact time due to an increase in bubble travel distance.  In  addition, increased depth of air injection results in increased oxygen solubility due to greater hydrostatic pressure, which increases the driving force for gas transfer (C,—CL) (Mavinic and Bewtra, 1976). Unfortunately, there are no published experimental studies in the hypolimnetic aeration literature to compare with these results. In a discussion of hypolimnetic aerator design, Taggart and McQueen (1982) suggested that greater rise distances would result in longer bubble-water contact periods and increased oxygen transfer.  Given the uni-  versality of this response in the civil engineering literature, there is little doubt that increased depth of air injection will generate a similar response in hypolimnetic aeration systems, and that increased contact time, induced flow and hydrostatic pressure will be the causitive factors.  6.1.2  O r i f i c e Size  The size of orifice had a significant effect on oxygen increase per cycle, daily oxygen load, E  e  and E . Although the F values are marginal (F  10-15), the trend is in the expected  p  direction and a larger sample size should result in a more significant effect. Bubble size was not measured in the field experiments. However, based on the results of the laboratory experiments, the bubble size emerging from the 140 p diffuser should be smaller than from the coarse bubble diffusers. As mentioned in the Group 1 discussion, a reduction in bubble size increases interfacial area, decreases terminal rise velocity and decreases the liquid film coefficient (K^). Since orifice size had no effect on water velocity (and subsequently K^) in this test group (Table 3.28), the factor most likely responsible for the increase in oxygen per cycle, daily load, E and E was an increase in interfacial p  D  Chapter 6.  Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration Experiments  83  area. However, since the cell means for bubble velocity were similar (Table 3.28), and smaller bubbles should have a lower rise velocity (Andeen, 1974), the smaller bubbles must have coalesced soon after forming into a heterogeneous mixture of bubble sizes, in order to generate the same water velocity for each orifice size. Field observations lend support to this hypothesis, as it was difficult to distinguish which diffuser was being tested on the basis of surface bubble size. How then are small bubbles more efficient at oxygen transfer when their existence is so ephemeral? This researcher believes the answer lies with the life span of a bubble, in which there are three distinct phases with three different rates of oxygen transfer.  The  first phase occurs during bubble formation at the interstitial openings of the diffuser. During formation and growth of the bubble, the liquid-gas interface surrounding the bubble is continually expanding and the concentration gradients in the liquid film remain high, resulting in an unusually high rate of oxygen absorption (Ippen and Carver, 1954). In the second stage, known as the intermediate steady-state phase, the concentration gradients in the surrounding liquid film attain lower values, hence a reduced transfer of oxygen occurs during the bubble's ascent through the water column. This phase is subject to deviations from the steady state, depending on shear and turbulence in the water column (Mancy and Okun, 1960). In the final phase, the bubble bursts when it reaches the surface and releases its gas contents to the atmosphere.  The liquid film  surrounding the bubble containing fairly high concentrations of dissolved oxygen is left behind, and the disturbance of the interface by the bursting bubbles tends to enhance atmospheric oxygen exchange at this point. Therefore, the increase in transfer efficiency arising from smaller orifices occurs during the bubble formation phase, prior to the bubbles coalescing into a heterogenous mixture. Although this period of bubble formation is quite brief, it is sufficient to allow for increased gas transfer. Pasveer (1966) states that the liquid film becomes saturated with  Chapter 6.  Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration  Experiments  84  oxygen in a time span of 1 x 10 seconds, which is considerably faster than the time -7  span of bubble formation. Scheffe's test indicated that the 140 p diffuser was significantly more efficient in terms of oxygen increase per cycle than the 794, 1588 and 3175 p diffusers, and that the 794, 1588 and 3175 u diffusers were not significantly different from each other. Given the aforementioned discussion on the transient nature of increased oxygen transfer during bubble formation, this analysis appears correct. In other words, the difference in bubble size generated by orifice diameters in the 794 to 3175 p range is too small to have a significant effect on oxygen transfer in a hypolimnetic aerator, because of their smaller surface area to volume ratio and the coalescing nature of the air-water mixture in the inflow tube. Scheffe's test was unable to distinguish any significant differences between the four orifice sizes, with respect to the cell means for daily oxygen load, E and E . The F values p  for daily oxygen load (=s 12), E  D  10) and E  p  D  12) are significant in the ANOVA;  however, Scheffe's test is conservative by design and just misses separating the 140 p diffuser effects from the remaining diffusers. Given a larger sample size, this researcher believes Scheffe's test would generate the same result as was observed for oxygen input per cycle ie. the 140 p diffuser would be significantly more efficient, and the 794 to 3175 p range diffusers would be statistically similar in terms of daily oxygen load, transfer efficiency and aeration efficiency. As mentioned in the Group 1 discussion, the effect of reduced orifice size and bubble size on transfer efficiency is well known in the civil engineering field (eg. Morgan and Bewtra, 1960; Bewtra and Nicholas, 1964). In the hypolimnetic aeration field, several researchers have discussed the theoretical effect of orifice size and bubble size on transfer efficiency (Smith et al., 1975; Speece, 1975; Ashley, 1985; Ashley et al., 1987); however most hypolimnetic aeration installations pay little attention to diffuser design, and if  Chapter 6.  Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration  Experiments  85  mentioned, it is usually in terms of orifice spacing for pressure loss, rather than transfer efficiency (eg. Fast, 1971).  6.2  Group 5 Field Experiments  6.2.1  Orifice Size  Orifice size significantly influenced oxygen increase per cycle, daily oxygen load, E and D  E . The mechanisms responsible for this result are identical to those mentioned in the p  Group 4 discussion, ie.  reduced bubble size and increased interfacial area due to the  smaller orifice size of the silica glass diffusers. In addition, since orifice size had no effect on water velocity, the enhanced transfer must have occurred during the brief bubble formation phase as previously described. It is useful to note how sensitive the area to volume ratio is to changes in bubble size. Since the area of a sphere of diameter d is ird and its volume is 7rd /6, 2  3  the A / V ratio =  (  6  -  1  7  »  or 6/d. A graphical plot of this ratio (Figure 5.12) illustrates how quickly the A / V ratio declines with increasing bubble size. Since the rate of gas transfer is directly proportional to A / V (ie. a), the efficiency of gas transfer will decrease in approximately the same way as the A / V ratio in Figure 5.12.  Haney (1954) discusses this aspect of gas transfer in  more detail and shows that if only 1 % of a group of bubbles are oversized by a factor of 5, the rate of gas transfer will be decreased by over 50%.  6.2.2  Surface Cover  The presence or absence of a floating styrene cover had no effect on water velocity, oxygen input, daily 0  2  load, E and E . Given the small size of the separator box (2.4 m x 1.2 p  D  Chapter 6.  Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration  Experiments  86  m) and the resulting low A/V ratio (ie. 0.36 m ), this result is not surprising. An A/V _1  ratio of 0.94 m in the 70 L cylinder was also insignificant, and it was not until the 239 -1  L tank was used (A/V = 2.2 m ) that surface cover exerted a marginally significant _1  effect on oxygen transfer. It is clear from these experiments that the majority of oxygen transfer occurs in the inflow tube, and aerator design modifications would be required to increase the surface component of the overall oxygen transfer process. 6.3  Summary Analysis: Group 4 and 5  6.3.1  Diffuser Depth  The effect of diffuser depth on oxygen increase per cycle was significant, however the effect was less than expected (ie. 0.62 mg/L at 3 m vs 0.73 mg/L at 7 m). The two observations from the literature (Bernhardt, 1967 and Smith et al., 1975) indicate most oxygen transfer occurs in the lower half of the inflow tube, and that declining hydrostatic pressure, decreasing oxygen content of rising bubbles and the additive effect of bubble and water velocity are responsible for this effect (Speece, 1975). The small difference between the 3 m and 7 m oxygen input suggests additional factors may be involved. For example, the amount of bubble coalescence in the inflow tube should influence oxygen input per cycle. Downing (1966) suggested the dissolved oxygen in the interstitial liquid rising with the dense bubble clouds becomes saturated very quickly, and is not dispersed rapidly enough into the main body of the liquid. This results in a lower rate of oxygen transfer than would be obtained from a single free rising bubble. This may occur in the inflow tube and the observed minor effect of depth on oxygen input may be partially explained by rapid saturation of the interstitial liquid. The significant effect of diffuser depth on E , E and daily load was mainly a result of 0  p  increased water velocity and subsequent volumetricflow,rather than increased oxygen  Chapter 6.  Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration Experiments  87  input per cycle. This is an interesting result that suggests hypolimnetic aerator design criteria should focus on obtaining maximum induced volumetric flow, in addition to achieving high oxygen inputs per cycle.  6.3.2  Surface Cover  The results of the Group 4 and 5 field experiments agree with the hypolimnetic aeration literature in that most oxygen transfer occurs within the inflow tube. There are at least two design modifications which could increase oxygen input per cycle by increasing the surface oxygen transfer component. The obvious approach is to increase the surface area of the separator box so that more surface area is available for gas transfer. This approach may be theoretically feasible, however the practical considerations of separator box size, cost and installation difficulties would probably invalidate this approach. A second approach would be to increase the turbulence within the separator box by installing additional aeration equipment. LaBaugh (1980) installed an electric surface aerator inside the separator box in the Spruce Knob Lake hypolimnetic aerator in an attempt to increase its oxygenation capacity. However, electric aerators are limited to situations where the power cable length does not exceed the voltage and phase restrictions of the motor. A more flexible approach would be to use an air driven aerator. A unit of this type could be powered by the main compressed air supply to the diffusor. An experimental program would be required to assess the cost effectiveness and net impact on E and E from this type of modification. D  p  A third approach would be to place a series of baffles in the separator box to cause additional shearing and mixing. This modification would not require any additional power.  Chapter 6.  6.3.3  Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration  Experiments  88  Air Flow Rate  The rate of air flow to a hypolimnetic aerator has an upper and lower limit. The lower limit is the amount of air required to overcome frictional losses in the distribution system and start the air-lift flow. The upper limit is determined by the transition from bubbly flow to plugflow,which usually occurs above an air void fraction (ie. air volume / water volume) of 10% (Andeen, 1974). Within these two limits, the induced waterflowis a function of depth of air release and airflow(Taggart and McQueen, 1982): Q  (6.18)  = (Z )(Q° )(lM) ll6  w  66  a  where: Q = waterflowin L/sec w  Q = airflowin L/sec a  Z = depth of air release in meters. Therefore, the basic design guidelines for airfloware to inject air as deep as economically possible, up to a void fraction of 10% of the induced water flow. To obtain more accurate estimates of required airflow,an empirical sizing method should be used (eg. Lorenzen and Fast, 1977; Taggart and McQueen, 1982; Ashley, 1985). 6.3.4  Air Flow Rate per Diffuser and Orifice Size  The effect of orifice size and airflowrate per diffuser on E and E has been previously p  D  discussed. In terms of hypolimnetic aerator design, the solution is quite simple: install as many fine bubble diffusers as is physically and economically possible. Ceramic and Saran diffusers generally produce bubbles in the optimum 2.0-2.5 mm diameter range (Eckenfelder and Ford, 1968) and are well suited for this type of application. Speece (1975) also states that bubble diameter should be in the 2.0-2.5 mm range to achieve optimum E and E . c  p  Chapter 6.  6.3.5  Discussion:  Group 4 and 5 Field Hypolimnetic  Comparison to Literature E  Aeration  Experiments  89  p  The highest E obtained was 0.34 kg 0 /kW-hr when using the 140 p diffusers at the 7 m p  2  injection depth (Appendix D). Published values for full lift hypolimnetic aerators range between 0.18 to 1.1 kg 02/kW-hr for very deep installations (eg. Wahnbach Reservoir) (Lorenzen and Fast, 1977). The value of 0.34 kg 0 /kW-hr was similar to the E reported 2  p  from Larsen and Mirror Lakes (0.32 kg 0 /kW-hr) which used a similar design of aerator 2  (Smith et al., 1975). 6.3.6 Retrofitting Undersized Systems  Once a system has been installed and found to be undersized, there are at least three solutions to increase oxygen input. Thefirstmethod is to supply additional air to the system. This generally increases the oxygen input per cycle due to increased turbulence and interfacial area in the inflow tube. More importantly, the velocity and induced volumetric flow increase with air flow rate (Ashley et al., 1987; Taggart and McQueen, 1982). This approach is more capital intensive ie. compressor purchase, operating and maintenance costs, and should be only used up to a void fraction (air volume/water volume) of 10%, at which point air-lift pump efficiency declines (Andeen, 1974). A second approach is to inject pure oxygen or a mixture of compressed air and oxygen into the aeration system (eg. Smith et al., 1975). This approach is suitable when lakes are located near areas where liquid oxygen is available eg. Amisk Lake, Alberta (Dr. E. Prepas, Zoology Department, University of Alberta, pers. comm.). However, this method may be logistically impractical and too expensive for remote areas or for large lakes. The third approach is to increase the oxygen transfer efficiency of the existing system.  Chapter 6.  Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration Experiments  90  This may be accomplished through improvements in diffuser design, which would be the most cost effective solution.  In addition, the idea of adding surface aerators may be a  viable option if excess air capacity is available and the diffuser is already an efficient design.  6.3.7 New Designs A number of design ideas for full lift hypolimnetic aerators have emerged as a result of exposure to the civil enginerring literature on gas transfer. T h e first idea involves a modification to improve oxygen transfer from the rising air bubbles.  This would be achieved by placing an open mesh screen in the inflow tube,  approximately 1/3 to 1/2 the distance up the inflow tube. T h e idea behind this modification is to re-fragment the coalesced bubbles in the rising water column into a smaller size and increase their surface area to volume ratio. T h e declining hydrostatic pressure and reduced oxygen content of the air bubbles will obviously limit the effectiveness of this idea; however, given the sensitivity of the A / V ratio to changes in bubble diameter (Haney, 1954) it may be possible to extract additional oxygen without further increases in energy input.  This modification would be susceptible to plugging of the mesh from  natural debris in the lake. T h e second idea involves increasing the contact time of the air-water mixture in the inflow tube, to improve oxygen transfer efficiency.  This would be achieved by placing vane  deflectors in the inflow tube to induce a corkscrew flow pattern to the rising air-water mixture. time.  This should result in a longer bubble-water rise path, hence longer contact  It may be necessary  to place vanes in the outflow tube to spiral water in the  opposite direction to reduce torque stress on the separator box. T h e idea of increasing contact time to improve oxygen transfer has been used by Pasveer (1966) in oxidation ditches; W i r t h et al.  (1975) used a spiral flow riser tube  Chapter  6.  Discussion:  Group 4 and 5 Field Hypolimnetic  Aeration  Experiments  91  in their hypolimnetic aeration experiments. The net result of the spiral helix used by Wirth et al. (1975) was a reduction in transfer efficiency, due to coalescence of bubbles on the underside of the helix plate; however, the concept of increasing transfer efficiency through increased contact time is theoretically sound. The third idea involves injecting compressed air into the downflow tube to increase transfer efficiency. This idea would involve injecting very fine bubbles (1-2 p diameter) near the outflow tube so, that the outflowing water velocity would entrain the small bubbles into the outflow tube. The bubbles would then be subjected to increasing hydrostatic pressure as they moved down the outflow tube; this should result in efficient gas transfer. Using this configuration, a hypolimnetic aerator would function as a co-current upflow, on the inflow side, and a counter-current downflow on the outflow side (Figure 6.13). This is roughly analogous to Mavinic's and Bewtra's (1976) System III, or Speece's (1971) U-Tube hypolimnetic aeration system. Potential problems with this design include residual bubble escape, which may cause local destratification, and N supersaturation in the hypolimnion, which could be detri2  mental to in-lake and downstream (if installed in a reservoir) populations of salmonids (Rucker, 1972). Further research on the N saturation aspect of this idea is required. 2  In conclusion, there are a number of modifications which may increase the oxygen transfer efficiency and aeration efficiency of full lift hypolimnetic aeration systems. Some of these ideas may prove impractical; however, they certainly warrant further investigation in laboratory and pilot scale experiments, considering the in-situ cost of hypolimnetic aeration.  Chapter 6. Discussion: Group 4 and 5 Field Hypolimnetic Aeration Experiments 92  WASTE AIR  Figure 6.13: Conceptual drawing of co-current upflow and counter-current downflow hypolimnetic aerator  Chapter 7  Conclusions  1. Increased airflowrates through 397 zz to 3175 \i diameter orifices resulted in increased K£,a , 0T , E and E , due to increased turbulence and interfacial area in 20  S  p  0  the water column. 2. A decrease in orifice size from 3175 /z diameter to 397 \i caused an increase in Kx,a o, OT,, E and E due, to smaller bubble size and a corresponding increase in 2  p  0  interfacial area and contact time. 3. A floating surface cover exerted a minimal effect on K£,a , OT,, E and Eo, indi20  p  cating surface oxygen exchange in low A / V ratio (0.94-2.2 m ) tanks is a small -1  componient of the overall oxygen transfer process. 4. Increased airflowrates through 40 /z and 140 /z orifice diameter silica glass diffusers caused a linear increase in Kx,a and OT,, due to increased turbulence and 20  interfacial area, but had no effect on E and E . p  D  5. Reducing the airflowrate per fine bubble diffuser (40 zz and 140 /z diameter orifice) increased Kxa o, OT,, E and E via smaller bubble size; it reduced the likelihood 2  D  p  of bubble coalescence and increased bubble dispersion. 6. Orifice size in the range of 40 zz and 140 /z diameter did not influence Kx,a o, OT,, 2  Ep and E , as the gas flow rate was above the critical rate and the bubble size D  generated was similar for both orifice sizes. 93  Chapter 7.  94  Conclusions  7. A greater depth of air release enhanced the oxygenation capacity of the hypolimnetic aerator through a combined effect of hydrostatic pressure and contact time related gains in oxygen increase per cycle, and an increase in water velocity and induced volumetric flow. 8. An orifice size of 140 p diameter increased the oxygenation capacity of a hypolimnetic aerator; however, 794-3175 p diffusers had no effect due to the coalescing environment in the inflow tube. 9. Afloatingsurface cover had no effect on oxygenation capacity, indicating little oxygen transfer occurs in the separator box of a standard design, full lift hypolimnetic aerator. 10. Hypolimnetic aerator design criteria should focus on obtaining maximum volumetricflows,in addition to achieving high oxygen input per cycle values. 11. The aeration efficiency (E ) of hypolimnetic aerators may be increased by enhancing p  the surface exchange component through design modifications, involving increased separator box size or additional mechanical surface aeration; however, these modifications may not be practical. 12. The design guidelines for diffuser flow rate and orifice size are to install as many fine bubble  140 p diameter) diffusers as physically and economically feasible.  13. Undersized hypolimnetic aeration systems may be upgraded by injecting additional air, increasing the oxygen content of the injected air or improving the oxygen transfer efficiency of the existing system. 14. Three design modifications for full lift hypolimnetic aeators, which may increase  Chapter 7.  Conclusions  transfer a n d energy efficiency,  95  are b u b b l e - b r e a k e r s  i n the inflow tube,  counter-  r o t a t i n g s p i r a l flows i n t h e i n f l o w a n d o u t f l o w t u b e s a n d fine b u b b l e d o w n - f l o w a i r injection i n the outflow tube.  Bibliography  [1] Andeen, G. B. 1974. Bubble pumps. Compressed Air Magazine 79:16-19. [2] American Public Health Association, American Water Works Association and Water Pollution Control Federation. 1980. Standard methods for the examination of water and wastewater, 15th edition. American Public Health Association, Washington, D. C. [3] Ashley, K. I. 1981. Effects of hypolimnetic aeration on functional components of the lake ecosystem. M. Sc. Thesis, Dept. of Zoology, University of B. C. Vancouver, B. C. 99p. [4] Ashley, K. I. 1983. Hypolimnetic aeration of a naturally eutrophic lake:physical and chemical effects. Can. J. Fish. Aquat. Sci. 40(9):1343-1359. [5] Ashley, K. I. 1985. Hypolimnetic aeration:practical design and application. Water Research 19(6):735-740. [6] Ashley, K. I., S. Hay and G. H. Scholten. 1987. Hypolimnetic aeration:field test of the empirical sizing method. Water Research 21(2):223-227. [7] Ashley, K. I. 1988. Hypolimnetic aeration research in British Columbia. Verh. Internat. Verein. Limnol. 23:215-219. [8] Atlas Copco Manual, Third Edition. 1978. Stockholm, Sweden. 591 p. [9] Babin, J. and E. E. Prepas. 1985. Modelling winter oxygen depletion rates in icecovered temperate zone lakes in Canada. Can. J. Fish. Aquat. Sci. 42(2):239-249. 96  Bibliography  97  [10] Barnhardt, E. L. 1969. Transfer of oxygen in aqueous solutions. J. Sanit. Eng. Div. Proc. ASCE 95 No. SA3:645-661. [11] Beak Consultants Ltd. 1977. CPAR Project Report 542-1. State of the art review Aeration. Beak Consultants Ltd., Montreal, Quebec. July, 1977. [12] Bernhardt, H. 1967. Aeration of Wahnbach Reservoir without changing the temperature profile. J. American Water Works Assoc. 59:943-964. [13] Bewtra, J. K. and W. R. Nicholas. Oxygenation from diffused air in aeration tanks. J. Water. Poll. Control Fed. 36(10):1195-1224. [14] Bewtra, J. K., W. R. Nicholas and L. B. Polkowski. 1970. Effect of temperature on oxygen transfer in water. Water Research 4(1):115-122. [15] Bewtra, J. K. and D. S. Mavinic. 1978. Diffused aeration systems from theory to design. Can. J. Civil Eng. 5(1):32-41. [16] Bjork, S., L. Bengtsson, H. Berggren, G. Cronberg, G. Digerfelt, S. Fleischer, C. Gelin, G. Lindmark, N. Maimer, F. Plejmark, W. Ripl and P. 0. Swanberg. 1972. Ecosystem studies in connection with the restoration of lakes. Verh. Int. Verein. Limnol. 18: 379-387. [17] Bowers, R. H. 1955. The mechanics of bubble formation. J. Applied Chemistry 5:542-548. [18] Boyd, C. E. 1986. A method for testing aerators for fish tanks. Prog. Fish Cult. 48:68-70. [19] Cantor, K. P., R. Hoover, T. J. Mason and L. J. McCabe. 1978. Associations of cancer mortality with halomethanes in drinking water. J. National Cancer Institute  Bibliography  98  61(4):979-985. [20] Clark, R. M. and P. Dorsey. 1980. The costs of compliance: An EPA estimate for organics control. J. American Water Works Assoc. 72(8):450-457. [21] Cornett, R. J. and F. H. Rigler. 1980. The areal hypolimnetic oxygen deficit: An empirical test of the model. Limnol. Ocean. 25(4):672-679. [22] Danckwerts, P. V. 1951. Significance of liquid film coefficient in gas absorption. Industrial and Engineering Chemistry 43(6):1460. [23] Dobbins, W. E. 1964. Mechanisms of gas absorbtion by turbulent liquids. Advances in Water Research Proc. 1st International Conf. Water Pollution Research. Pergamon Press Ltd., London, England. Vol. 2, p. 61. [24] Dorin, G. 1981. Organochlorinated compounds in drinking water as a result of eutrophication. In Restoration of Lakes and Inland Waters. U. S. EPA. Washington, D. C. EPA 440/5-81-010. [25] Downing, A. L. and G. A. Truesdale. 1955. Some factors affecting the rate of solution of oxygen in water. J. Applied Chemistry 5:570-581. [26] Downing, A. L. 1966. Response to Pasveer, A. "Considerations on the efficiency of the aeration process." Air and Water Pollution International Journal. 10:448-449. [27] Doyle, M. L., W. C. Boyle, T. Rooney and G. L. Huibregtse. 1983. Pilot plant determinations of oxygen transfer in fine bubble aeration. J. Water Poll. Control Fed. 55(12):1435-1440. [28] Dunst, R. C, S. M. Born, S. A. Smith, S. A. Nichols, J. 0. Peterson, D. R. Knauer, S. L. Serns, D. R. Winter and T. L. Wirth. 1974. Survey of lake rehabilitation  Bibliography  99  techniques and experiences. Tech. Bull. No. 75. Wisconsin Dept. Natural Resources, Madison, Wisconsin. 179 pp. [29] Eckenfelder, W. W. 1959. Absorption of oxygen from air bubbles in water. J. Sanit. Eng. Div. Proc. ASCE 85 No. SA4:89-99. [30] Eckenfelder, W. W. and D. L. Ford. 1968. New concepts in oxygen transfer and aeration. In: Advances in Water Quality Improvement (Edited by E. F. Gloyna and W. W. Eckenfelder. Water Resource Symposium No. 1. [31] Eckenfelder, W. W. (Ed. ) 1969. Oxygen transfer and aeration In: Manual of Treatment Processes, Vol. 1. Environmental Sciences Services Corp., USA. [32] Edmondson, W. T. and J. T. Lehman. 1981. Effects of changes in the nutrient income on the condition of Lake Washington. Limnol. Ocean. 26:1-29. [33] Ellis, S. E. and R. Stanbury. 1980. The scale-up of aerators. Paper No. 3 Symposium on The Profitable Aeration of Waste Water, (London, April, 1980). BHRA Fluid Engineering, Cranfield, Bedford. U.K. [34] Ewing, L., D. T. Redmon and J. D. Wren. 1979. Testing and data analysis of diffused aeration equipment. J. Water Poll. Control Fed. 51(10):2384-2401. [35] Fast, A. W. 1971. The effects of artificial aeration on lake ecology. EPA Water Pollution Control Research Series 16010 EXE 12/71. [36] Fast, A. W. and M. W. Lorenzen. 1976. Synoptic survey of hypolimnetic aeration. J. Environ. Eng. Proc. ASCE Vol. 106 (EE6):1161-1173. [37] Haberman, W. L. and R.. K. Morton. 1954. An experimental study of bubbles moving in liquids. Proc. Amer. Soc. Civil. Eng. 80:379-429.  Bibliography  100  [38] Haney, P. D. 1954. Theoretical principles of aeration. J. American Water Works Assoc. 46(4):353-376. [39] Ippen, A. T., L. G. Campbell and C. E. Carver. 1952. The determination of oxygen absorbtions in aeration processes. Mass. Int. Technol. Hydrodyn. Lab. Tech. Report No. 7. [40] Ippen, A. T. and C. E. Carver. 1954. Basic factors of oxygen transfer in aeration systems. Sewage and Industrial Wastes 26(7):813—829. [41] Jones, R. A. and G. F. Lee. 1982. Recent advances in assessing impact of phosphorus loads on eutrophication-related water quality. Water Research 16:503-515. [42] King, D. L. 1970. Reaeration of Streams and Reservoirs; Analysis and Bibliography. Engineering and Research Center, U. S. Bureau of Reclamation REC-OCE-70-55. 131 pp. [43] Krenkel, P. A. and G. T. Orlob. 1962. Turbulent diffusion and the reaeration coefficient. J. Sanit. Eng. Div., Proc. Amer. Soc. Civil Eng. 88 (SA2):53-83. [44] LaBaugh, J. W. 1980. Water chemistry changes during artificial aeration of Spruce Knob Lake, West Virginia. Hydrobiologia 70: 201-216. [45] Lakin, M. B. and R. N. Salzman. 1979. Subsurface aerator evaluation. J. Water Poll. Control Fed. 51(10):2419-2428. [46] Landberg, G. G., B. P. Graulich and W. H. Kipple. 1969. Experimental problems associated with the testing of surface aeration equipment. Water Research 3:445-455. [47] Larkin, P. A. 1975. Biometrics. Biology 300 Course Notes. Fall Term, 1975. University of British Columbia, 206 p.  Bibliography  101  [48] Leary, R. D., L. A. Ernest and W. J. Katz. 1969. Full scale oxygen transfer studies of seven diffuser systems. J. Water Poll. Control Fed. 41(3):459-473. [49] Lee, G. F. and R. A. Jones. 1988. The North American experience in eutrophication control. In: Proc. Int. Conf. Phosphates, Water and Quality of Life. Paris, France. February 1988. [50] Lewis, W. K. and W. G. Whitman. 1924. Principles of gas absorption. Ind. Eng. Chem. 16:1215-1220. [51] Lind, 0. T. 1979. Handbook of Common Methods in Limnology. Second Ed. C. V. Mosby. St. Louis, 199 p. [52] Lorenzen, M. and A. W. Fast. 1977. A guide to Aeration/Circulation techniques for lake management. U. S. Environmental Protection Agency 600/3-77-004. [53] Mancy, K. H. and D. A. Okun. 1960. Effects of surface active agents on bubble aeration. J. Water Poll. Control Fed. 32(4):351-364. [54] Markofsky, M. 1979. On the reoxygenation efficiency of diffused air aeration. Water Research 13:1339-1346. [55] Mavinic, D. S. and J. K. Bewtra. 1974. Mass transfer of oxygen in diffused aeration systems. Can. J. Civil. Eng. 1:71-84. [56] Mavinc, D. S. and J. K. Bewtra. 1976. Efficiency of diffused aeration systems in wastewater treatment. J. Water Poll. Control Fed. 48(10):2273-2283. [57] McQueen, D. J., S. S. Rao and D. R. S. Lean. Hypolimnetic aeration: changes in bacterial populations and oxygen demand. Arch. Hydrobiol. 99(4):498-514.  Bibliography  102  [58] McQueen, D. J. and D. R. S. Lean. 1986. Hypolimnetic aeration: An overview. Water Poll. Res. J. Canada 21(2):205-217. [59] Mercier, P. and J. Perret. 1949. Aeration station of Lake Bret, Monastbull, Schweiz. Ver Gas. Wasser-Fachm 29,25. [60] Morgan, P. F. and J. K. Bewtra. 1960. Air diffuser efficiencies. J. Water Poll. Control Fed. 32(10):1047-1059. [61] Nasmith, H. 1962. Late glacial history and surficial deposits of the Okanagan Valley, British Columbia. B. C. Dept. Mines and Petroleum Resources. Bull. No. 46. 49p. [62] Neilson, B. J. 1974. Reaeration dynamics of reservoir destratification. J. Amer. Water Works Assoc. 66:617-620. [63] Nienow, A. W. 1980. The aeration of waste water: Basic physical factors. Paper No. 1. Symp. on the Profitable Aeration of Waste Water, (London, April, 1980). BHRA Fluid Engineering, Cranfield, Bedford, U.K. [64] O'Connor, D. J. 1982. Wind effects on gas-liquid transfer coefficients. J. Environ. Eng., Proc. ASCE 109(3):731-753. [65] Pasveer, A. 1955. Research on activated sludge VI. Oxygenation of water with air bubbles. Sewage and Industrial Wastes 27(10): 1130-1146. [66] Rohlf, F. J. and R. R. Sokal. 1969. Statistical Tables. W. H. Freeman and Co. San Francisco, Ca. 253 p. [67] Rucker, R. 1972. Gas bubble disease of salmonids: a critical review. U.S. Fish and Wildlife Service. Tech. Paper 58.  Bibliography  103  [68] Schmit, F. L. and D. T. Redmon. 1975. Oxygen transfer efficiency in deep tanks. J. Water Poll. Control Fed. 47(11): 2586-2598. [69] Schmit, F. L., J. D. Wren and D. T. Redmon. 1978. The effect of tank dimensions and diffuser placement on oxygen transfer. J. Water Poll. Control Fed. 50(7):1750—1767. [70] Smith, S. A., D. R. Knauer and T. Wirth. 1975. Aeration as a lake management technique. Wisconsin Dept. Natural Res., Tech. Bull. No. 87. [71] Speece, R. E. 1971. Hypolimnion Aeration. J. Amer. Water Works Assoc. 64(l):6-9. [72] Speece, R. E. 1975. Ten Year's experience of reservoir aeration. Reply to H. Bernhardt. Prog. Water Technol. 7:489-494. [73] Stenstrom, M. K. and R. G. Gilbert. 1981. Effects of alpha, beta and theta factor upon the design, specification and operation of aeration systems. Water Research 15:643-654. [74] Taggart, C. T. and D. J. McQueen. 1982. A model for the design of hypolimnetic aerators. Water Research 16:643-654. [75] Tuthill, R. W. and G. S. Moore. 1980. Drinking water chlorination:A practice unrelated to cancer mortality. J. Amer. Water Works Assoc. 72(10):570-573. [76] Vallentyne, J. R. 1974. The Algal BowLLakes and Man. MSP 22. Dept. Environ., Fisheries and Marine Service, Ottawa, Canada. [77] Verner, B. 1974. Long term effect of hypolimnetic aeration of lakes and reservoirs, with special considerations of drinking water quality and preparation costs. In J. Taggart and L. Moore (Eds.) Lake and Reservoir Management:134-138. Proc. 3rd  Bibliography  104  Annual Conf. North. Amer. Lake Management Soc, October 18-20, 1983, Knoxville, Tenn., USA. U. S. EPA 440/5- 84-001. 604 p. [78] Walker, W. W. 1983. Significance of eutrophication in water supply reservoirs. J. Amer. Water Works Assoc. 75(l):38-42. [79] Wetzel, 1975. Limnology. W. B. Saunders Co., Philadelphia, Pa. 743 p. [80] Wirtli, T. L., D. R. Knauer and S. A. Smith. 1975. Total and hypolimnetic aeration of lakes in Wisconsin. Verh. Internat. Verien. Limnol. 19:1960-1970.  Appendix A Group 1 ANOVA Results: K a , O T , , E and E L  1  2 0  0  p  Source of variation DF Mean Square F Significance of F Residual 63 0.22 Constant 1 4225.47 18906.58 0.000 Air flow 1 614.83 2751.01 0.000 Surface cover 1 0.84 3.74 0.058 Orifice size 3 181.02 809.96 0.000 Flow by cover 1 0.09 0.40 0.531 Flow by size 3 22.44 100.42 0.000 Cover by size 3 0.08 0.34 0.794 Replication 4 3.43 15.34 0.000  Source of variation DF Mean Square F Significance of F Residual 63 0.09 Constant 1 1617.30 18426.20 0.000 Flow 1 235.64 2684.70 0.000 Surface cover 1 0.38 4.31 0.042 OT, Orifice size 3 69.14 787.78 0.000 Flow by cover 1 0.03 0.32 0.573 Flow by size 3 8.44 96.18 0.000 Cover by sie . 3 0.03 0.33 0.802 Replication 4 1.34 15.23 0.000  105  Appendix  E  0  A.  Group 1 ANOVA  Results: K a , L  20  0T , g  E  D  and E  p  Source of variation DF Mean Square F Significance of F Residual 63 0.03 Constant 1 4430.05 135560.54 0.000 Flow 1 3.72 113.80 0.000 Surface cover 1 0.36 11.01 0.002 Orifice size 3 46.74 1430.31 0.000 Flow by cover 1 0.00 0.07 0.794 Flow by size 3 0.04 1.37 0.260 Cover by size 3 0.05 1.53 0.215 Replication 4 0.78 23.74 0.000  Significance of F Source of variation DF Mean Square F Residual 63 23.45 Constant 1 689894.09 29416.05 0.000 Flow 1 2241.90 95.59 0.000 Cover 1 174.94 7.46 0.008 Orifice size 3 30237.38 1289.28 0.000 Flow by cover 1 0.74 0.03 0.859 Flow by size 3 25.34 1.08 0.364 Cover by size 3 16.59 0.71 0.551 Replication 4 472.04 20.13 0.000  Appendix B Group 2 ANOVA Results: K i a , O T , , E and E 20  Q  p  Source of variation DF Mean Square F Significance of F Residual 65 3.89 Constant 1 23767.95 6113.90 0.000 Flow 1 2534.22 651.88 0.000 Surface cover 1 29.00 7.46 0.008 Orifice size 1 4.72 1.21 0.274 Number of diffusers 1 287.34 73.91 0.000 Flow by cover 1 12.79 3.29 0.074 Flow by size 1 0.02 0.01 0.949 Flow by diffusers 1 50.04 12.87 0.001 Cover by size 1 8.44 2.17 0.145 Cover by diffusers 1 1.76 0.45 0.504 Size by diffusers 1 5.78 1.49 0.227 Replication 4 2.90 0.75 0.565  107  Appendix B. Group 2 ANOVA Results: K a , L  20  0T , E and E B  0  p  Source of variation DF Mean Square F Significance of F Residual 65 1.50 Constant 1 9103.64 6074.37 0.000 Flow 1 970.22 647.38 0.000 Surface cover 1 10.95 7.31 0.009 Orifice size 1 1.74 1.16 0.285 Number of diffusers 1 109.98 73.38 0.000 OT, Flow by cover 1 4.80 3.20 0.078 Flow by size 1 0.00 0.00 0.985 Flow by diffusers 1 19.40 12.95 0.001 Cover by size 1 3.20 2.14 0.149 Cover by diffusers 1 0.58 0.39 0.537 Size by diffusers 1 2.24 1.50 0.225 Replication 4 1.09 0.73 0.577  E  0  Source of variation DF Mean Square F Significance of F Residual 65 0.42 Constant 1 11115.49 26185.78 0.000 Flow 1 0.20 0.47 0.494 Surface cover 1 2.67 6.29 0.015 Orifice size 1 0.60 1.40 0.240 Number of diffusers 1 32.95 77.62 0.000 Flow by cover 1 0.44 1.04 0.311 Flow by size 1 0.08 0.18 0.672 Flow by diffusers 1 0.45 1.05 0.309 Cover by size 1 0.68 1.61 0.210 Cover by diffusers 1 0.09 0.21 0.652 Size by diffusers 1 1.63 1.63 0.206 Replication 4 0.30 0.71 0.588  Appendix  B.  Group 2 ANOVA  Results: K a , L  20  0T„, E and E 0  p  Source of variation DF Mean Square F Significance of F Residual 65 597.95 Constant 1 4048155.2 6770.10 0.000 Flow 1 152.08 0.25 0.616 Surface cover 1 3728.91 6.24 0.015 Orifice size 1 962.58 1.61 0.209 Number of diffusers 1 45557.13 76.19 0.000 Flow by cover 1 668.05 1.12 0.294 Flow by size 1 96.58 0.16 0.689 Flow by diffusers 1 477.66 0.80 0.375 Cover by size 1 1032.05 1.73 0.194 Cover by diffusers 1 112.53 0.19 0.666 Size by diffusers 1 942.02 1.58 0.214 Replication 1 355.92 0.60 0.667  Appendix C Group 3 ANOVA Results: K a , OT„, E and E L  2 0  0  p  Source of variation DF Mean Square F Significance of F Residual 32 0.07 Constant 1 724.31 10219.64 0.000 Surface cover 2 0.77 10.93 0.000 Orifice size 2 70.47 994.34 0.000 Cover by orifice 4 0.09 1.27 0.304 Replication 4 0.10 1.46 0.238  Source of variation DF Mean Square F Significance of F Residual 32 0.33 Constant 1 3232.58 9764.05 0.000 Surface cover 2 3.24 9.80 0.000 Orifice size 2 314.46 949.84 0.000 Cover by size 4 0.37 1.11 0.371 Replication 4 0.44 1.37 0.278  Source of variation DF Mean Square F Significance of F Residual 32 0.05 Constant 1 2345.62 46238.19 0.000 Surface cover 2 0.60 11.88 . 0.000 Orifice size 2 62.07 1223.58 0.000 Cover by orifice 4 0.05 1.07 0.387 Replication 4 0.07 1.41 0.254  110  Appendix  C.  Group 3 ANOVA  Results: K a , L  20  0T , e  E  D  and E  p  ource of variation DF Mean Square F Significance of F Residual 32 36.50 Constant 1 356391.60 9764.04 0.000 Surface cover 2 357.68 9.80 0.000 Orifice size 2 34699.53 949.84 0.000 Cover by orifice 4 40.38 1.11 0.371 Replication 4 48.78 1.34 0.278  111  Appendix D Group 4 Results: Water Velocity, Oxygen Input, Daily Oxygen Load, E and E  c  p  Source of variation DF Mean Square F Significance of F Residual 28 0.00 Constant 1 3.06 2946.03 0.000 Orifice size 3 0.00 2.03 0.132 Water Velocity Diffuser depth 1 0.46 445.31 0.000 Size by depth 3 0.00 0.49 0.691 Repli cation 4 0.00 0.90 0.477  Source of variation DF Mean Square F Significance of F Residual 28 0.01 Constant 1 18.09 1271.08 0.000 Oxygen Input Orifice size 3 0.21 14.82 0.000 Diffuser depth 1 0.11 7.75 0.010 Size by depth 3 0.00 0.11 0.953 Replication 4 0.00 0.24 0.915  112  Appendix D. Group 4 Results:  Water Velocity,  Oxygen Input, Daily Oxygen Load, E and  Daily Oxygen Load Source of variation DF Mean Square F Significance of F Residual 28 1.48 Constant 1 2203.74 1485.02 0.000 Orifice size 3 17.41 11.73 0.000 Diffuser depth 1 474.03 319.43 0.000 Size by depth 3 4.32 2.91 0.052 Replication 4 1.24 0.84 0.515  Source of variation DF Mean Square Significance of F F Residual 28 0.19 Constant 1 1000.13 5259.56 0.000 Orifice size 3 1.95 10.27 0.000 Diffuser depth 1 62.17 326.92 0.000 Size by depth 3 0.25 1.33 0.294 Repli cation 4 0.17 0.90 0.477  Source of Variation DF Mean Square F Significance of F Residual 28 0.00 Constant 1 1.10 1485.02 0.000 Orifice size 3 0.01 11.73 0.000 Diffuser depth 1 0.24 319.43 0.000 Size by depth 3 0.00 2.91 0.052 Repli cation 4 0.00 0.83 0.515  0  E 113 p  Appendix E Group 5 Results: Water Velocity, Oxygen Input, Daily Oxygen Load, E and E  D  p  Source of variation DF Mean Square F Significance of F Residual 12 CUX) — Constant 1 3.42 3562.13 0.000 Water Velocity Orifice size 1 0.00 3.26 0.096 Surface cover 1 0.00 1.17 0.300 Size by cover 1 0.00 0.63 0.443 Replication 4 0.00 0.54 0.708  Source of variation DF Mean Square F Significance of F Residual 12 (TOO — Constant 1 12.01 2574.11 0.000 Oxygen Input Surface cover 1 0.00 0.11 0.749 Orifice size 1 0.42 90.11 0.000 Cover by size 1 0.00 0.11 0.749 Replication 4 0.01 2.14 0.138  114  Appendix  E. Group 5 Results:  Water Velocity, Oxygen Input, Daily Oxygen Load, E and  Daily Oxygen Load iource of variation DF Mean Square F Significance of F Residual 12 2.16 Constant 1 3120.00 1443.73 0.000 Surface cover 1 1.92 0.89 0.364 Orifice size 1 76.83 35.55 0.000 Cover by size 1 0.29 0.13 0.721 Replication 4 3.11 1.44 0.280  Source of variation DF Mean Square F Significance of F Residual 12 0.15 Constant 1 897.15 5912.03 0.000 Orifice size 1 5.57 36.71 0.000 Surface cover 1 0.14 0.94 0.352 Size by cover 1 0.02 0.11 0.745 Replication 4 0.22 1.47 0.273  Source of variation DF Mean Square F Significance of F Residual 12 0.00 Constant 1 1.56 1443.72 0.000 Surface cover 1 0.00 0.89 0.364 Orifice size 1 0.04 35.55 0.000 Cover by size 1 0.00 0.13 0.721 Replication 4 0.00 1.44 0.280  Q  E 115 p  5)  PUBLICATIONS  Primary A s h l e y , K . I . 1983. H y p o l i m n e t i c l a k e . Can. J . F i s h . Aquat. S c i . Ashley, K.I. distribution.  Journals aeration of a n a t u r a l l y 40 ( 9 ) : 1 3 4 3 - 1 3 5 9 .  eutrophic  1984. H y p o l i m n e t i c a e r a t i o n and z o o p l a n k t o n C a n . J . F i s h . A q u a t . S c i 41 (12) : 1 8 5 6 - 1 8 5 7 .  Ashley, K . I . 1985. Hypolimnetic aeration:practical a p p l i c a t i o n . Water R e s e a r c h 19:735-740.  depth  design  and  Ashley, K.I., S. Hay and G. Scholten. 1987. Hypolimnetic a e r a t i o n : f i e l d t e s t o f t h e e m p i r i c a l s i z i n g method. Water R e s e a r c h 21:223-227. Ashley, K . I . 1988. Hypolimnetic aeration research Columbia. Verh. I n t e r n a t . V e r e i n . Limnol. 23:215-219.  Conference  in  British  Proceedings  A s h l e y , K . I . 1982. H y p o l i m n e t i c a e r a t i o n and f u n c t i o n a l components of the lake ecosystem:phytoplankton and z o o p l a n k t o n e f f e c t s , p p . 31-4 0 I n P r o c e e d i n g s o f t h e N o r t h A m e r i c a n L a k e Management S o c i e t y . V a n c o u v e r , B r i t i s h C o l u m b i a . U . S . EPA 4 4 0 / 5 - 8 3 - 0 0 1 . Ward, P . R . B . , W . G . D u n f o r d and K . I . A s h l e y . 1986. I c e c o n t r o l i n l a k e s by p h o t o v o l t a i c s powered w a t e r c i r c u l a t i o n , p p . 41-4 6 I n P r o c e e d i n g s o f t h e Renewable E n e r g y C o n f e r e n c e , W i n n i p e g , M a n i t o b a . Ward, P . R . B . , E . C o u s i n s , W . G . D u n f o r d and K . I . A s h l e y . 1989. A d i r e c t l y d r i v e n PV pump f o r W a t e r a e r a t i o n . P r o c e e d i n g s o f t h e 15 t h Annual Conference o f the S o l a r Energy S o c i e t y of Canada, June 1 9 - 2 1 , 1989, P e n t i c t o n , B . C . Technical  Reports  M u r p h y , T . P . , K . J . H a l l , K . I . A s h l e y , A . M u r d o c h , M . Mawhinney and H. Fricker. 1985. In-lake precipitation o f phosphorus by lime t r e a t m e n t . N a t i o n a l Water R e s e a r c h I n s t i t u t e , C o n t r i b u t i o n S e r i e s No. 8 5 - 1 6 7 , B u r l i n g t o n , O n t a r i o .  Ashley, K . I . , E . P a r k i n s o n and A . T a u t z . 1986. B r o o d s t o c k and research lake opportunities i n southern i n t e r i o r B r i t i s h Columbia. Fisheries Project R e p o r t No. RD 7. M i n i s t r y of Environment, Province of B r i t i s h Columbia. Hume, J . M . B , K . Tsumura and K . I . A s h l e y . 1986. A preliminary c o m p a r i s o n b e t w e e n P r e m i e r and Duncan r a i n b o w t r o u t s t o c k s i n a  lake containing redside shiners. F i s h e r i e s Project Report. 1. M i n i s t r y o f E n v i r o n m e n t , P r o v i n c e o f B r i t i s h C o l u m b i a .  No.  RD  Ashley, K . I . 1987. A r t i f i c i a l circulation i n British Columbia: Review and E v a l u a t i o n . F i s h e r i e s T e c h n i c a l R e p o r t N o . 7 8 . 34 p p . Fisheries Branch, M i n i s t r y o f Environment, Province of British Columbia. Ashley, Progress Project British  K . I . a n d J.M.B. Hume. 1987. H a b i t a t Conservation Fund R e p o r t (1987-1988) : S a l s b u r y Lake F e r t i l i z a t i o n . F i s h e r i e s R e p o r t N o . RD 1 5 . M i n i s t r y o f E n v i r o n m e n t , Province of Columbia.  Ashley, K . I . 1988. A preliminary economic assessment of the Salsbury L a k e f e r t i l i z a t i o n p r o j e c t . F i s h e r i e s P r o j e c t R e p o r t No. RD 1 6 . M i n i s t r y o f E n v i r o n e m t , P r o v i n c e o f B r i t i s h C o l u m b i a . Nordin, R.N., G.H. S c h o l t e n and K.I. Ashley. 1988. E f f e c t s o f hypolimnetic aeration on t h e water q u a l i t y o f S t . Mary Lake, S a l t s p r i n g I s l a n d . W a t e r Management B r a n c h , B.C. Ashley, Progress Project British  K . I . a n d N.T. J o h n s t o n . 1 9 8 9 . H a b i t a t Conservation Fund R e p o r t (1988-1989) : S a l s b u r y Lake F e r t i l i z a t i o n . F i s h e r i e s R e p o r t N o . RD 1 9 . M i n i s t r y o f E n v i r o n m e n t , Province of Columbia.  A s h l e y , K . I . 1989. The u s e o f c h l o r i n e a s a p o s s i b l e f i s h t o x i c a n t . Fisheries Project Report N o . RD 20. M i n i s t r y o f Environment, Province o f B r i t i s h Columbia. Theses A s h l e y , K . I . 1981. E f f e c t s o f h y p o l i m n e t i c components o f t h e l a k e e c o s y s t e m . M.Sc. U n i v e r s i t y o f B r i t i s h C o l u m b i a . 120 p p .  a e r a t i o n on f u n c t i o n a l thesis, Zoology Dept.,  A s h l e y , K . I . 1989. F a c t o r s influencing gas t r a n s f e r i n d i f f u s e d a e r a t i o n systems and t h e i r a p p l i c a t i o n t o hypolimnetic aeration. M.A.Sc. thesis, Civil Engineering Dept., University of British C o l u m b i a . 115 p p .  6)  PROFESSIONAL  ASSOCIATIONS  American F i s h e r i e s Society Chinook Foundation International Association f o r Theoretical N o r t h A m e r i c a n Lake Management S o c i e t y  and Applied  Limnology  

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.831.1-0062903/manifest

Comment

Related Items