Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Characterization of forestry dry land sort runoff and treatment using a sand filtration process Doig, Peter Duncan James 2005

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

Item Metadata

Download

Media
831-ubc_2005-0199.pdf [ 11.58MB ]
Metadata
JSON: 831-1.0058943.json
JSON-LD: 831-1.0058943-ld.json
RDF/XML (Pretty): 831-1.0058943-rdf.xml
RDF/JSON: 831-1.0058943-rdf.json
Turtle: 831-1.0058943-turtle.txt
N-Triples: 831-1.0058943-rdf-ntriples.txt
Original Record: 831-1.0058943-source.json
Full Text
831-1.0058943-fulltext.txt
Citation
831-1.0058943.ris

Full Text

CHARACTERIZATION OF FORESTRY DRY LAND SORT RUNOFF AND TREATMENT USING A SAND FILTRATION PROCESS by PETER DUNCAN JAMES DOIG B.Sc. Honours Biochemistry, University of Guelph, 2001 A THESIS SUBMITTED IN PARTIAL REQUIREMENT FOR THE DEGREE OF MASTER OF SCIENCE . in CHEMICAL AND BIOLOGICAL ENGINEERING THE UNIVERSITY OF BRITISH COLUMBIA April 2005 © Peter Doig, 2005 I I A b s t r a c t Runoff generated from forestry dry land sort (DLS) operations contains materials from logs, chip piles, and industrial machinery and poses an environmental threat to receiving waters due to its high suspended particle loads, potentially high metals concentration, and frequent toxicity. Since there is currently no legislation or discharge regulations specific to DLS runoff (as for pulp and paper effluents), polluting DLS operations have largely gone unnoticed and as a result few treatment technologies have been tested. In this study, twelve samples of DLS runoff were taken from three dry land sorts located on the Sunshine Coast of British Columbia. Samples had chemical oxygen demand (COD) ranging from 346 mg/L to 3690 mg/L (of twelve samples tested), pH and metals (of two samples tested) that did not comply with British Columbia (BC) Approved Water Quality Guidelines and the BC Municipal Sewage Regulations. The particles of two samples were characterized chemically and morphologically and it was determined that fresh DLS runoff contains highly aggregated inorganic and organic colloidal particles ranging from 1 to 10 pm in size. Although some of these particles sedimented during storage at 4 °C or 25 °C, most particles remained relatively stable in suspension suggesting the requirement for filtration or a secondary treatment process. An oxide-coated sand filtration process performed well over three series of' column experiments, removing up to 86% of the COD and 92% of the turbidity, when operated as a re-circulating filter for 24 hours. Batch adsorption experiments using the oxide-coated sand removed averages of 17% and 26% DLS runoff COD over two series of experiments. The same sand, however with the oxide-coating stripped (oxide-stripped sand), failed to remove COD from solution in batch adsorption experiments suggesting that the oxide-coating enhanced absorption of suspended particles. When oxide-stripped sand filters were run in parallel with oxide-coated sand filters treating successive batches of DLS runoff, the oxide-stripped sand treated on average 34% less COD and 32% less turbidity. Neither flocculation nor biological activity was determined to be an important mechanism to the sand filtration process over successive batch treatment runs. I l l Table of Contents Abstract II Table of Contents Ill List of Tables VI List of Figures IX Acknowledgements XVI Chapter 1 1 Introduction 1 1.1 Water and Wastewater Treatment Preamble 1 1.2 Research Objectives 2 1.3 Rational for Research 2 1.4 Thesis Layout 3 Background Information 4 2.1 Dry Land Sort Runoff 4 2.1.1 Dry Land Sorts and Runoff Generation 4 2.1.2 Characterization and Toxicity of DLS Runoff 6 2.2 Treatment of DLS runoff 9 2.2.1 Current Practices on Dry Land Sorts 9 2.2.2 Mechanisms of Particle Removal 10 2.3 Preliminary Runoff Characterization and Treatment Trials 20 References 22 Chapter 3 26 Characterization of Dry Land Sort Runoff 26 3.1 Introduction 26 3.2 Materials and Methods 29 I V 3.2.1 Sampling of DLS run-off 29 3.2.2 Fractionation and characterization of DLS run-off 32 3.2.3 Aging Study 34 3.3 Results and Discussion 36 3.3.1 Primary characterization of all DLS runoff samples 36 3.3.2 Fractionation and Characterization of DLS run-off 38 3.3.3 Aging study 48 3.4 Conclusions 52 3.4.1 Characterization of fresh and stored runoff 52 3.4.2 Aging Study 53 Chapter 4 56 4.1 - Background 56 4.2.1 Sand Preparation and Characterization 59 4.2.2 DLS runoff Sample Preparations 62 4.2.3 Sterilzation of DLS runoff Sample 7 63 4.2.4 Batch Adsorption Experiments 63 4.2.5 Sand Column design 64 4.2.6 Column Treatment Experiments 65 4.3 Results and Discussion 68 4.3.1 Sand Characterization.... 68 4.3.2 Sterilization of DLS runoff Sample 7 71 4.3.3 Adsorption shake flask experiments 72 4.3.4 Column Treatment Experiments 75 4.4 Conclusions 91 4.4.1 Sand Characterization and Sand Column Performance 91 V 4.4.2 Determining Key Sand Column Treatment Mechanisms 92 Conclusions and Recommendations for Future Work 98 5.1 Runoff Characterization 98 5.2 Lab-scale DLS runoff treatment trials 99 5.3 Full-scale DLS runoff treatment trials 100 V I List of Tables Table 2.1: Minimum diameters of particles that will sediment under gravitational sedimentation or centrifugation 11 Table 2.2: Typical characteristics of slow and rapid sand filters 19 Table 3.1: Runoff sample descriptions 32 Table 3.2: Results of primary characterization of DLS runoff samples 37 Table 3.3: Chemical and morphological characteristics of samples 3 and 4 39 Table 3.4: Weight percents as determined by TEM-EDX of a typical 10 pm and 2pm aggregate structure found in the fresh and stored samples respectively 41 Table 3.5: EDX analysis of fresh sample 3 colloids 43 Table 3.6: EDX analysis of 3 stored sample colloids 44 Table 3.7: Change to PSA and COD over 36 days of storage 48 Table 3.8: COD d e g raded and COD s edimented during storage of sample 9 for 13 days at 4°C 50 Table 3.9: COD d e g raded and COD s edimented during storage of sample 9 for 13 days at 25°C 50 Table 4.1: Chemical analysis provided by Target Inc. for Play S a n d ™ (coated sand) .60 Table 4.2: Characteristics of the DLS runoff samples used in sterilization batch adsorption experiments I and II and the column treatment experiments A, B, and C ....62 Table 4.4: Cation exchange capacities (CEC) for selected soils and minerals (adapted from Winegardner, 1996) 70 Table 4.5: Mineralogical content and characteristics of the sand used in column treatment experiments 71 Table 4.6: Effect of HgC _ on sterilization, particle size analysis and COD 72 VII Table 4.7: TEM-EDX analysis of Sample 3 effluent from the conditioned oxide-coated sand column 79 Table 4.8: Comparison of the COD, pH and metals analysis of untreated DLS runoff Sample 3 and conditioned oxide-coated sand column treated Sample 3 80 Table A.1: Results of Chapter 3 chemical characterization of particulate, colloidal and dissolved fractions in fresh DLS runoff. Values are reported as averages of 3 replicates as determined by ICP-AES 101 Table A.2: Results of Chapter 3 chemical characterization of particulate, colloidal and dissolved fractions in stored DLS runoff. Values are reported as averages of 3 replicates as determined by ICP-AES 102 Table A.3: Results of Chapter 4 column experiment C run 1 103 Table A.4: Results of Chapter 4 column experiment C run 2 103 Table A.5: Results of Chapter 4 column experiment C run 3 104 Table A.6: Results of Chapter 4 column experiment C run 4 104 Table A.7: Results of Chapter 4 column experiment C run 5 105 Table A.8: Results of Chapter 4 column experiment C run 6 105 Table A.9: Results of Chapter 4 column experiment C run 7 106 Table A.10: Results of Chapter 4 column experiment C run 8 ...106 Table B.1: Statistical analysis for Chapter 3 batch adsorption experiment 1 - adsorption of oxide-coated vs. oxide-stripped sand 111 Table B.2: Statistical analysis for Chapter 3 batch adsorption experiment I - adsorption of oxide-coated vs. shake runoff control 111 Table B.2: Statistical analysis for Chapter 3 batch adsorption experiment I - adsorption of oxide-stripped vs. shake runoff control 112 VIII Table B.2: Statistical analysis for Chapter 3 batch adsorption experiment II - shake control vs. static control 112 Table B.3: Statistical analysis for Chapter 3 batch adsorption experiment II - 1 g sand/runoff vs. shake control 113 Table B4: Statistical analysis for Chapter 3 batch adsorption experiment II - 5 g sand/runoff vs. shake control 113 Table B5: Statistical analysis for Chapter 3 batch adsorption experiment II - 10 g sand/runoff vs. shake control 114 Table B6: Statistical analysis for Chapter 3 batch adsorption experiment II - 1g, 5g, 10g sand/runoff 114 Table B.4: Statistical analysis for comparison of treatment 1 and 2 average COD treatment performance over eight successive batch runs 115 Table B.4: Statistical Analysis for comparison of treatment 3 and 4 COD treatment performance over eight successive batch runs 115 Table B.5: Statistical Analysis for comparison of treatments 1 and 3 COD treatment performance over eight successive batch runs 116 Table B.6: Statistical Analysis for comparison of treatments 1 and 5 COD treatment performance over eight successive batch runs 116 I X List of Figures Figure 2.1: Typical dry land sort located adjacent to Howe Sound (Howe Sort dry land sort) on Sunshine Coast, B.C 5 Figure 2.2: Common wastewater treatment sequence of mixing, flocculation and sedimentation (Hammer, 2001) 11 Figure 2.3: The potential energy diagram of the electric double layer at a negatively charged particle surface (Poortinga etal., 2002) 13 Figure 2.4: Application of a cationic polyelectrolyte to promote interparticle bridging and floe formation of negtively charged colloids 14 Figure 3.1: Colloidal aggregation processes in natural waters (Buffle et al., 1998) 28 Figure 3.2: Storm event causing runoff from log storage area at Fleatwood Forest Products (van Poppelen, 2002) 30 Figure 3.3: Chip pile producing leachate located at Terminal Forest Products (van Poppelen, 2002) 31 Figure 3.4: Fractionation and characterization scheme for samples 3 and 4 33 Figure 3.5: TEM image at 2.5K magnification of DLS run-off sample 3 (fresh) 40 Figure 3.6: TEM image at 2.5K magnification of DLS run-off sample 4 (stored) 40 Figure 3.7: TEM image at 10.0 K magnification of the colloidal fraction of DLS run-off sample 3 (fresh) 42 Figure 3.8: TEM image at 10.0 K magnification of the colloidal fraction of DLS run-off sample 4 (stored) 42 Figure 3.9: Chemical characterization of particulate, colloidal, and dissolved fractions of fresh sample of DLS runoff 45 X Figure 3.10: Chemical characterization of particulate, colloidal, and dissolved fractions in stored sample of DLS runoff 46 Figure 3.11: Particle size histograms constructed using Malvern Mastersizer (2000) software showing change in PSA over 36 storage days of storage at 4 °C 49 Figure 3.12: Change in COD w h 0 i e and CODsupernatant over storage of sample 9 at 4°C and 25°C for 13 days 51 Figure 4.1: Sand columns used in successive batch treatment experiments: from left to right columns 1 and 2 (treatment 1), columns 3,4 (treatment 2), columns 5,6 (treatment 3), columns 7,8 (treatment4), column 9 (treatment 5) 66 Figure 4.2: Batch Adsorption Experiment I comparing the adsorption capacity, for DLS runoff Sample 7 COD, of the oxide-coated sand (coated replicates 1, 2, and 3), the oxide-stripped sand (stripped replicates 1,2, and 3), and the shake runoff control (runoff, no sand) 73 Figure 4.3: Batch Adsorption Experiment II comparing the adsorption capacity for DLS runoff Sample 11 COD, of 1, 5, and 10 g of oxide-coated sand with that of the shake runoff control, no sand 74 Figure 4.4: Comparison of COD removal efficacy of oxide-coated conditioned sand and oxide-coated fresh sand during 24 treatment runs 76 Figure 4.5: Comparison of turbidity removal efficacy of conditioned oxide-coated sand and fresh oxide-coated sand during 24 treatment runs 76 Figure 4.6: Untreated fresh DLS runoff Sample 3 (left) and conditioned, oxide-coated sand 24 hour effluent (right) 77 Figure 4.7: Malvern Mastersizer pictograms of untreated fresh Sample 3 (left) and conditioned oxide-coated sand 24 hour effluent (right) 78 X I Figure 4.8: TEM image (10,000 x) of Sample 3 effluent from the conditioned oxide-coated sand column 79 Figure 4.9: Change in pH, COD and Turbidity during time-course, conditioned oxide-coated sand column treatment of Sample 3 82 Figure 4.10: 24 hour treatment of Sample 9 COD using treatment a) with two identical oxide-coated sand columns in parallel (replicate 1 and 2) 83 Figure 4.11: 24 hour treatment of Sample 9 COD using treatment b) with two identical oxide-coated, abiotic sand columns in parallel (replicate 1 and 2) 84 Figure 4.12: 24 hour treatment of Sample 9 COD using treatment c) conditioned, oxide-coated sand column (no column replicates) 84 Figure 4.13: Treated DLS runoff sample 11 from right to left respectively: treatment 1 (oxide-coated, biotic), treatment 2 (oxide-stripped, abiotic) and the static control 86 Figure 4.14: COD after the first 24 hour batch treatment run for Sample 11 for treatments: 1 - oxide-coated, biotic, 2 - oxide-coated abiotic, 3 - oxide-stripped, biotic, 4 - oxide-stripped abiotic and 5 - conditioned oxide-coated. Error bars represent the standard deviations of triplicate measurements of COD. Treatments 1 -4 are the combined results for duplicate columns 86 Figure 4.15: Turbidity after first 24 hour batch treatment run for Sample 11 for treatments 1-5. Error bars represent the standard deviations of triplicate measurements of turbidity. Treatments 1-4 are the combined results for duplicate columns ...87 Figure 4.16: Average percent COD reductions for the 5 different sand column treatments: oxide-coated, oxide-coated/abiotic; oxide-stripped; and oxide-stripped/abiotic duplicates and oxide-coated/conditioned. Averages and standard deviations are based on two column replicates for each treatment) and 3 replicate COD X I I measurements (except conditioned treatment - one column, 3 replicate COD measurements) ...88 Figure 4.17: Average percent turbidity reductions for the 5 different sand column treatments: oxide-coated, oxide-coated/abiotic; oxide-stripped; and oxide-stripped/abiotic duplicates and oxide-coated/conditioned. Averages and standard deviations are based on two column replicates for each treatment) and 3 replicate turbidity measurements (except conditioned treatment - one column, 3 replicate turbidity measurements) 89 Figure A.1 Run 2 COD treatment performance for column experiment C 107 Figure A.2 Run 3 COD treatment performance for column experiment C 107 Figure A.3 Run 4 COD treatment performance for column experiment C 108 Figure A.4 Run 5 COD treatment performance for column experiment C 108 Figure A.5 Run 6 COD treatment performance for column experiment C 109 Figure A.6 Run 7 COD treatment performance for column experiment C 109 Figure A.7 Run 8 COD treatment performance for column experiment C 110 List of abbreviations and acronyms AAS atomic adsorption spectroscopy A g 2 S 0 4 silver sulphate H 2 S 0 4 sulfuric acid BC British Columbia B O D 5 five day biochemical oxygen demand CBD citrate bicarbonate dithionite C E C cation exchange capacity DFO Department of Fisheries and Oceans DHA dehydroabietic acid DLS dry land sort FCrit critical F value FRBC Forest Renewal British Columbia GF/C glass fiber circular HCI hydrochloric acid HNO3 nitric acid HgCI2 mercuric chloride ICP-AES inductively coupled plasma - atomic emissions spectroscopy IHP inner Hemholtz radius kV kilovolts KCI potassium chloride L C 5 0 median lethal concentration wherein 50% die Mhz mega hertz NaOH sodium hydroxide NCASI National Council for Air and Stream Improvement OHP outer Helmholtz radius pHpzc pH point of zero charge PSA particle size analysis RPM revolutions per minute TEM-EDX transmission electromagnetic energy dispersive X rays TL tannins and lignins TMP thermomechanical pulping TOC total organic carbon USEPA United States Environmental Protection Agency XRD X ray diffraction +/- standard deviation d particle diameter g acceleration due to gravity nm nanometer Ps particle density P liquid density u fluid viscosity urn micrometer Vs terminal settling velocity of a particle List of units °c CFU/mL cm cm 3 g g/cm3 g/L h h/d K L m M r r i e q mg/L m/h mm mL/min nm urn wt. % degrees Celcius colony forming units per milliliter centimeters cubic centimeters grams grams per cubic centimeter grams per liter hour hour/day thousand liters meters molar milli-equivalents milligram per liter meters per hour millimeters milliliters per minute nanometer micrometer weight percent X V I Acknowledgements I would like to thank my supervisor, Dr. Sue Baldwin, for introducing me to graduate studies at the University of British Columbia and for presenting me with the opportunity to expand my field of knowledge from Biochemistry to BioResource Engieering. Sue's expertise and guidance were much appreciated throughout and my sincere thanks for the space and independence given to me when my interests took me elsewhere. Thanks goes to Dr. Andre Sobolewski, Paul van Poppelen, and Dr. Annette Muttray of Microbial Technologies Inc. for helping with project design, sharing their guidance and experience throughout, and introducing me to problem solving in commercial engineering systems. The support that I received from UBC faculty, staff and students was essential to this project. Thank you very much Anthony, Geoff, Helsa, Jason, Ken, Kurt, Lori, Ping, Royann, Sietan, Sheldon, Horace and Qi. Special thanks to Mary Mager from Metals and Materials Engineering for your instruction on TEM-EDX analysis. Funding for this project was provided by Forest Renewal British Columbia (FRBC) and the Science Council of British Columbia (GREAT) scholarship and is greatly appreciated. I would also like to thank my family and all of my loyal friends for their support and encouragement throughout. This thesis is dedicated to Yona, for sharing with me her creativity, enthusiasm and brilliance. For all the adventures and gardens we grew, I am truly thankful. 1 Chapter 1 Introduction 1.1 Water and Wastewater Treatment Preamble Fresh water is essential for life on Earth. It is the media that holds the structure of biological matter together and the media that circulates energy and materials throughout Earth's ecosystems. Unfortunately, quantity and quality of fresh water is being depleted world-wide and there seems to be no other resource more abused by humanity. In many cases, pollution of natural waters can be directly correlated with the obtainment of collective and personal wealth (whether it resource extraction or food production), which in and of itself is often created by the human notion of 'progress.' As Kurt Vonegut Jr. mused to people with nothing but paper representations of wealth (in the book Galapagos): 'Wake up you idiots! Whatever made you think paper was so valuable!' This is not progress in my opinion. In order to sustain the continuum of essential water processes for the health of Earth and humans alike, fresh water must be conserved and wastewater must be treated. Treatment processes are generally designed for the specific wastewater characteristics at hand, space and resources, and cost. All wastewater treatment processes will ideally satisfy the following relative criteria: 1) Reliable, resilient, and robust 2) Materials and energy efficient 3) Operator friendly An example of a wastewater treatment process which potentially satisfies all of the above criteria for treatment of log yard dry land sort runoff is sand filtration. This thesis evaluates a proposed technology for filter performance and attempts to define its operational mechanisms for the enhancement of the filter performance. 2 1.2 Research Objectives The first objective of this project was to characterize the wastewater generated from log storage and grading yards, termed dry land sort (DLS) runoff, in terms of its physical, chemical, and biological properties. The second objective of this project was to determine the key mechanism(s) by which one specific sand filtration process treats the suspended particles of DLS runoff, effectively lowering the organic loads and pollution potential. 1.3 Rational! for Research DLS runoff is a complex mixture of wood waste materials and leachate from chip-piles and industrial machinery. This runoff is currently being discharged into receiving waters in British Columbia (BC), often untreated and unregulated, however has exhibited water quality criteria such as chemical oxygen demand (COD), biochemical oxygen demand (BOD), total suspended solids (TSS), and pH that exceeds the BC Approved Water Quality Guidelines. In addition, many DLS runoff grab samples have exhibited acute trout toxicity, which is in contravention of the Canadian Fisheries and Waste Management Acts. It was previously determined by Microbial Technologies Inc., an environmental consulting company located in the Sunshine Coast Regional District of British Columbia, that removal of the suspended particles drastically improved water quality and removed the acute trout toxicity. Studies have yet to characterize DLS runoff particles or yet to investigate the potential properties of the particles which could be exploited in a treatment process. In this thesis, the DLS runoff particle fraction was characterized so that hypotheses could be formed as to potential mechanisms for their removal in the proposed sand filter process. The second objective of this thesis functions to characterize and test a sand filtration process proposed by Microbial Technologies Inc for treatment of DLS runoff particles. Their pilot-scale system has shown good treatment potential, sometimes excellent, however its effectiveness has been intermittent. In terms of its treatment performance, it is has been operated as a "black box." Treatment mechanisms are assessed in lab-scale sand columns in order to elucidate what factors are key 3 determinants of performance and what are the practical limitations preventing the full-scale implementation of the proposed technology. Although sand filtration is widely used for drinking water or as a tertiary polishing step for wastewater, few studies have tested its efficacy for industrial wastewaters. Therefore, this work attempted to analyze specific mechanisms within the sand filter or how they relate to the process performance. 1.4 Thesis Layout This thesis is divided into five main chapters. Chapter 1 functions to introduce the topic and give an overview of the coming chapters. Chapter 2 is the literature review which provides background information that is essential for the consideration and the rational for the experimental sections of this thesis, Chapters 3 and 4. Both Chapters 3 and 4 contain: Introduction, Methods and Materials, Results and Discussion, and Conclusions. Chapter 3 describes the DLS runoff characterization experiments. Chapter 4 reports the lab-scale sand column treatment experiments. These two chapters were organized in this manner to describe each body of work as a cohesive set of experiments, similar to papers published in scientific journals. Chapter 5 makes a final set of Conclusions and Recommendations. 4 Chapter 2 Background Information Chapter 2 provides background information on dry land sorts (DLS) and DLS runoff generation. In addition, pertinent literature characterizing DLS runoff and similar wood waste effluents and their toxic constituents, is discussed. Finally, potential treatment methods and preliminary characterization work done prior to this study is reviewed. 2.1 Dry Land Sort Runoff 2.1.1 Dry Land Sorts and Runoff Generation Dry land sorts are open areas used by the forestry industry to sort and grade logs. Most of these operations are adjacent to aquatic fish habitats into which rainwater and any materials collected on these decks flow. This effluent is termed dry land sort (DLS) run-off in this thesis. DLS run-off occurs under all precipitation conditions, however light or heavy, and in dry weather as a result of dust control and fire hazard prevention (by on deck sprinklers) and lubrication of paved areas. There are many activities conducted simultaneously on dry land sorts. These include log sorting, bark stripping, and chipping. As a result, water landing on the deck may come in contact with: metals from galvanized buildings, grease and oil from industrial machinery, and organic materials from bark, chips, wet logs (from rain or water transported logs) and leachate (Samis etal., 1999). A recent study conducted in B.C., reported runoff events on 89% of 72 surveyed log yards (Orban era/., 2002). Figure 2.1: Typical dry land sort located adjacent to Howe Sound (Howe Sort dry land sort) on Sunshine Coast, B.C (Paul van Poppelen, 2002). Storm water runoff from urban and agricultural areas, which may contain metals, pesticides, and petroleum products, has received considerable attention by municipal, environmental and industrial regulators (deHoop etal., 1998, Grout etal., 1999). In contrast, liquid discharges from dry land sorts have, until recently, eluded the attention of regulators and the environmental and scientific community. First and foremost, runoff from any industrial operation is a nebulous entity to characterize, as it is often precipitation-dependent and may contain a vast mixture of compounds and which may vary significantly in concentration (Grout etal., 1999). Furthermore, seasonal and operational variations, and wood species sorted contribute to the variability in total flow rates and constituents of DLS runoff. For this reason, incongruity exists on whether conventional water quality standards apply (NCASI, 1992). However, recent studies on DLS runoff samples have detected both: a) concentrations of materials above British 6 Columbia Approved Water Quality Guidelines (BC Approved Water Quality Guidelines, 2001) and b) acute trout toxicity (Zenaitis etal. 1999, Bailey etal., 1999). In light of these results, Department of Fisheries and Oceans (DFO) regulators have made attempts to motivate DLS operations into voluntary treatment of runoff, in most cases of little to no avail. This context changed when Weyerhaeuser Company Limited (previously MacMillan Bloedel Limited) was charged with depositing deleterious substances into fish habitat and failing to comply with an Inspector's Direction (Fisheries Act, 1985). According to line 35 of The Fisheries Act: (1) 'No person shall carry on any work or undertaking that results in the harmful alteration, disruption or destruction of fish habitat'. Furthermore line 36 states: (3) 'Subject to subsection (4), no person shall deposit or permit the deposit of a deleterious substance of any type in water frequented by fish or in any place under any conditions' (Fisheries Act, 1985). The company was also charged with three counts under the provincial Waste Management Act for introducing waste into the environment. The maximum penalty for this violation is a one million dollar fine (Fisheries and Oceans Canada New Release, March 29, 2000). Companies may also be charged up to $300, 000 for their first offence in contravention of the Pulp and Paper Effluent Regulations of the Fisheries Act. This act regulates the release of Biochemical Oxygen Demand, Suspended Solids and Trout toxicity (Environmental Acts and Regulations, 2002). 2.1.2 Characterization and Toxicity of DLS Runoff The most apparent and immediate water quality concern of DLS runoff is its appearance (Orban etal., 2002, NCASI, 1992). Wood waste materials such as bark, chips and fibres, suspended particles, and tannins/lignins darken the runoff water and can form a thick, wood waste mat on the shoreline of receiving waters. These settleable materials may destroy or force sensitive filter fauna to move. Fish may find these areas unattractive and relocate as well. As these materials decompose, oxygen becomes limited and anaerobic activity may release hydrogen sulfide, which is toxic to some fish and pelagic invertebrate (for example zooplankton) species (NCASI, 1992). Foam is another visible entity of concern. Foam is the extensive dispersion of gas bubbles in a liquid and is thought to arise from the resinous materials in conifers, such 7 as resin acids, fatty acids, and lignins. It is unclear if foam itself poses a toxicity threat (NCASI, 1992). The wood waste contingent of DLS runoff is a complex mixture of (mostly organic) compounds. These include carbohydrates (approximately 60-65% by weight, for example: cellulose and hemicelluloses), lignins (approximately 26-32%), and phenolic substances (for example tannins and lignans), terpenes (for example resin acids), alcohols, proteins and inorganic compounds (McDougall, 1996). The presence and concentration of each compound is dependent on amount of precipitation and the type of wood species sorted. For example, leachates from Douglas fir, spruce, pine, and larch typically produce concentrations of resin acids greater than other tree species (Samis etal., 1999). The most common tree species sorted on dry land sorts in British Columbia are conifers such as Douglas fir, hemlock and pine (Orban etal., 2002). In conifers, organic wood extractives such as tannins, phenolics, tropolones and resin acids have been reported as the major toxic constituents (Hoel etal., 1995, Samis et al., 1999, Bailey etal., 1999a, Magnus etal., 2000). Wood extractives are non-structural components and may comprise between 1% and 5% of the dry weight of a tree. They function as energy sources, catalysts or biological defence. For example, resin acids and phenolics are toxic to microbial pathogens and some insects and are used as agents of microbiological and insect defence by trees (Sjostrom, 1993). Thus, it is intuitive that resin acids and phenolics may show toxicity to other organisms. Accordingly, resin acids found in similar wood waste streams to DLS runoff such as thermomechanical pulping (TMP) and softwood debarking effluents have exhibited acute trout toxicity (Hoel etal., 1996, Samis etal., 1999). Fish exposed to resin acids in a laboratory study experienced red blood cell lysis, increase haemoglobin and bilirubin breakdown and conjunction path overload (Ard etal., 1996). In another study, tannins and lignins (TL) were correlated with acute toxicity in DLS runoff (Bailey etal., 1999a). Metals are also of considerable concern in DLS runoff. One study found zinc, at concentrations (0.1-0.9 mg/L) which are above BC Approved Water Quality Guidelines (0.01 mg/L), to be the primary cause of runoff acute trout toxicity from sawmills (Bailey etal., 1999a, 1999b, British Columbia (BC) Water Quality Report, 1998). Metals in DLS 8 runoff are believed to originate from galvanized buildings and machinery tires and brake linings located on dry land sorts (Orban etal., 2002). Although the causative agent of DLS runoff toxicity is still unclear, a recent study found 72% of 58 runoff samples collected from nine British Columbia sawmills to exhibit LC 5 o trout toxicity (Bailey et al., 1999a). In a separate study, 96% of 27 runoff samples taken from 3 Vancouver Island sawmills, exhibited LC50 trout toxicity (Bailey etal., 1999b). In an earlier study conducted in California, 17 log yards were monitored over a 3-year period. The toxicity testing yielded an average of 70% survival rate for trout exposed to runoff samples (NCASI, 1992). In addition to testing for proven pollutants such as metals and resin acids, researchers and regulators alike have used conventional water quality criteria to characterize and compare wood waste effluents from different streams. Measures such as chemical oxygen demand (COD), five-day biochemical oxygen demand (BOD5), total suspended solids (TSS), dissolved oxygen (DO) and pH have been tested and have shown dramatic variation from day to day and from site to site. In studies conducted through the 70's and 80's by the USEPA, the following water quality value-ranges were compiled for log storage and sort decks in Northern California: BOD 5 (11-840 mg/L), COD (35-11,000 mg/L), and TSS (76-9,100 mg/L) (NCASI, 1992). In a more recent study, runoff samples collected from one log storage deck in Louisiana during storm events between June and November reported the following data ranges: COD (0 -14,723 mg/L), BOD 5 (0 - 48.4 mg/L), TSS (6.7 -20,0078 mg/L) and pH (6.7-8.1) (deHoop etal., 1998). In 2 independent studies conducted in Alberta (McDougall, 1996) and British Columbia (Zenaitis and Duff, 2002), the following data ranges were determined: BOD 5 (4-920 mg/L), COD (75-3740 mg/L) and TSS (7-812 mg/L). The following BC Approved Water Quality criteria can be applied to DLS runoff for fresh, marine and estuarine waters: particulates less than 25 mg/L when background in less than 25 mg/L, total organic carbon (TOC) within 20% of median background concentration, pH between 6.5 and 8.5, and zinc less than 0.010 mg/L for protection of marine life. When compared to BC Approved Water Quality, DLS runoff has most likely exhibited concentrations of organic carbon (COD, BOD 5, TOC), particulates (TSS), and metals (zinc) that are in contravention (BC Approved Water Quality Guidelines, 1998). In addition DLS runoff poses a toxicity threat in contravention of the Fisheries Act (1995) and the Waste Management Act. 2.2 Treatment of DLS runoff 2.2.1 Current Practices on Dry Land Sorts In a recent study conducted in BC, only 35 of 64 (-55%) sites surveyed collected runoff, whereas 67% of these sites were labelled as high risk to receiving waters. Of 35 sites collecting runoff, 29 sites treated runoff, most using passive treatment methods such as sediment traps and horizontal flow wetlands (Orban et al., 2002). In a separate study conducted on 33 log yards in Alberta, two sites used conventional biological treatment systems, 12 sites used passive treatment methods such as infiltration riparian buffer strips, another 12 redirected runoff into ditches and 7 sites used detention ponds (McDougall, 1996). Many different treatment technologies have been suggested for DLS runoff. These include physical and chemical treatment methods such as: screening, sedimentation, aeration, adsorption, chemical oxidation, precipitation, flocculation and reverse osmosis and biological processes such as: constructed wetlands and activated sludge (Samis etal., 1999, Orban etal., 2002). Nevertheless, a review of the scientific and engineering literature, found only three articles referring to treatment processes tested for DLS and/or log yard runoff (Zenaitis etal., 1999, Zenaitis etal., 2002). An intensive internet search conducted on the subject of wood waste, sawmill, log yard and/or DLS runoff treatment technologies provided only fifteen references to runoff and only two private companies were evaluating wood waste runoff treatment technologies. No commercially available treatment methods were advertised on-line for log yard runoff or DLS runoff as of January 2005. In contrast, an internet search 10 conducted for treatment technologies for urban storm water runoff produced approximately 15,000 references and hundreds of commercially available technologies. Of the few treatment studies conducted, Microtox toxicity removal has been correlated with the removal of organic carbon measures such as BOD 5, COD and TOC (Zenaitis etal., 1999, Zenaitis etal., 2002). Most of the organic content is found in the suspended particle phase (colloids and particulates), since most of the organic molecules are of wood fibre origin and show low to moderate solubility in water (Magnus etal., 2000). One particular study identified a direct correlation of high COD with high TSS in DLS runoff and recommended a TSS removal system, such as sedimentation or filtration, to control COD (deHoop etal., 1998). Studies of similar effluents to DLS runoff, such as TMP wastewater, suggest that toxicity is attributable to particulate and colloidal-bound resin acids (Hoel etal., 1995, Magnus etal. 2000). These studies recommend a particle removal system as a likely candidate for a DLS runoff treatment technology. The following sections review mechanisms of particle removal such as sedimentation and flocculation, degradation, sorption and filtration. 2.2.2 Mechanisms of Particle Removal Every suspended particle in solution will settle under favourable conditions (Hammer etal., 2001, Cheremisinoff, 2002, Tchonanoglous etal., 2003). Stability is dependant upon many physical factors such as particle size, density and concentration and chemical factors such as surface functional groups and solution chemistry (pH and ions). Stoke's Law gives the terminal settling velocity for a spherical particle, at low Reynold's number. V s = q ( p g - p)d2 18 (2.1) Where p s is the density of the particle p is the density of the liquid g is the acceleration due to gravity u is the viscosity of the fluid Vs is the terminal settling velocity of the particle d is the particle diameter 11 Table 2.1 gives the minimum diameter of spherical particles that will sediment due to gravity or centrifugation. The densities of particles increase from the left to right columns and are be representative of more organic to less organic spheres, for example a humic acid particle compared to an inorganic iron oxide particle (Perret etal., 1994) Table 2.1: Minimum diameters of particles that will sediment under gravitational sedimentation or centrifugation (Perret etal., 1994) Minumum Diameter (urn) of Particles Sedimented p s = 1.1 g /cm 3 p s = 1 .5g/cm 3 p s = 2 . 0 g / c m 3 p s = 3 .0g / cm 3 Sedimentation > 9 >4 > 3 >2 Centrifugation: 1500 rpm (0.5 hr) >2 > 0.9 > 0.6 > 0.45 Centrifugation: 4000 rpm (5 hr) >0 .23 > 0.11 > 0.075 > 0.5 Centrifugation: 30 000 rpm (1 hr) >0.065 >0.03 > 0.02 > 0.013 Centrifugation: 30 000 rpm (14 hr) >0.020 > 0.008 > 0.006 > 0.0035 The primary treatment process for suspended particles in wastewater treatment is typically clarification by sedimentation and/or flocculation (Hammer, 2001). This process may entail gravity settling alone or rapid mixing to promote particle collision and flocculation (Tchonanoglous etal., 2003). Particles greater than 1 to 2 urn, such as clays, silts and floes are then sedimented in settling chambers or filtered through granular media. Figure 2.2 shows a typical water and wastewater treatment plant scheme. Wastewater influent Rapid mix Flocculation Settling Sand Filter Coagulant Auxiliary chemicals —•» Activated cartoon Figure 2.2: Common water and wastewater treatment sequence of mixing, flocculation and sedimentation (adapted from Hammer, 2001) 12 In a study characterizing COD, particle and resin acids removal from TMP wastewater using a similar primary treatment scheme as given in Figure 2, the primary clarifier (after settling) removed 34% COD (of 5000 mg/L), 17% total solids (of 4100 mg/L) and 53% of the extractables (of 463 mg/L) (Magnus etal., 2000). Thus sedimentation has been used effectively as a primary treatment process for wood waste effluents similar to DLS runoff. Suspended materials larger than the 1.2 pm, GF/C standard filter are operationally defined as suspended particulates or total suspended solids (TSS) (Cheremisnoff, 2002, Tchonanoglous etal., 2003). Colloids are suspended particles that pass through the standard 1.2 pm filter and theoretically will not settle under gravity alone. This definition of colloids is somewhat loose as colloids may form stable aggregates with a mean diameter of up to 10 pm (Perret etal., 1993). Furthermore, both particulates and colloids hold the capacity to interact physio-chemically by ligand exchange, surface-complexation, hydrophobic, and ion-exchange reactions, thus the dissertion between the two has largely been based on conventional and somewhat arbitrary sizing, rather than function (Buffle etal., 1998, Ran etal., 2000). Alternatively, suspended particulates may be cited as those colloidal aggregates larger than 1.2 pm. Thus particulates may be larger polymers (more aggregated particles) constructed from colloidal monomer aggregates. For the purpose of this thesis, no physiochemical dissertion between the surface-chemistry of suspended particulates and colloid aggregates, other than relative size, will be made. It is widely accepted that organic colloidal particles in natural and waste- waters have a negative surface charge in the circum-neutral pH range due to the deprotonation of surface carboxylic, phenolic, and phosphoric functional groups (Perret etal., 1993, Buffle etal., 1998). 13 The combined effects of Brownian motion and negative charge repulsion between particles cause the formation of stable suspensions. In colloidal systems, some ions of opposite charge, through electrostatic and van der Waals forces, will overcome thermal agitation and get fixed to the charged surface of the colloids. This forms the Stern layer and the subsequent potential drop from P0 to Pd as shown below in Figure 2.3. Surrounding the Stern layer is a diffuse layer of with a potential of P d to O. The Stern and diffuse layer combined form the electric double layer which is an important theoretical model to understand the origin of charge and the prospect of bringing particles of the same charge together. Figure 2.3: The potential energy diagram of the electric double layer at a negatively charged particle surface (adapted from Poortinga etal., 2002) In order to reduce repulsive effects of particle surface charge, potential determining ions (for example strong acids or bases) or electrolytes (for example aluminium, iron or calcium) can be added to lessen surface charge or reduce the thickness of the diffuse electric layer (Tchonanoglous etal., 2003). Similarily, wastewater treatment systems use chemical coagulants and flocculants to destabilize particles and encourage the flocculation process. Coagulants and flocculants include 14 metal salts such as alum or ferric sulfate and synthetic organic polyelectrolytes such as polyaluminum and polyiron chloride respectively. Metal salts function to adsorb negatively charged particle surfaces and neutralize surface charge. At high concentrations, metal floes will form. Cationic polyelectrolytes, typically organic, are employed as flocculants. The flocculation process involves sorption of the polyelectrolyte, charge neutralization and formation of interparticle bridging. Figure 2.4 illustrates the function of a polyelectrolyte in forming settleable floes with colloidal particles. particles in solution polyelectrolyte polyelectrolyte f l 0 C formation Figure 2.4: Application of a cationic polyelectrolyte to promote interparticle bridging and floe formation of negtively charged colloids (drawn by Peter Doig). The use of flocculants as the sole treatment method for wood waste runoff was previously described (BC Research, 1973, Zenaitis etal., 1999). In both studies, chemical flocculation was not an effective treatment method. At best, only 12 % of Microtox toxicity and 27% of TOC removal was reported (Zenaitis etal., 1999). Adsorption is a ubiquitous mechanism in particle removal systems. It is an active component of sedimentation/flocculation, biological and filtration treatment processes. Similar physiochemical mechanisms occur in all systems: particles (adsorbates) are adsorbed to surfaces (adsorbents) by covalent, electrostatic (including hydrogen bonding), hydrophobic or van der Waals forces (Cheresminisinoff, 2002). The kinetics of adsorption can be estimated by batch, shake flask, adsorption experiments and by applying an adsorption isotherm, for example the Freundlich, Langmuir or Emmet Isotherm. After equilibrium is reached, the adsorbent phase concentration (qe) is determined using the following equation: m (2.2) qe = mg adsorbate/g adsorbent C0 = initial concentration of adsorbate, mg/L Ce = final concentration of adsorbate, mg/L V = volume of liquid in reactor, L m = mass of adsorbent, g A common sorbent used in wastewater treatment is activated carbon (Tchonanoglous etal., 2003). One study found activated carbon to be effective (100% actute toxicity removal) in treating wood waste effluents (Zenaitis etal., 1999). However due to its relatively high expense, the authors concluded that no further investigations into its application would be conducted. Alternatively, low cost, industrial and agricultural, waste streams as well as natural substances are gaining increased usage (Baily etal., 1999). These include peat moss, brewery waste, leaf compost, coffee grounds, and chitosan. Their exact adsorption capacity are not well defined, however their effectiveness in treating metals may be explained by a net negative surface charge (Bailey etal., 1999). Other adsorbents include tailored minerals such as oxide coated sands and zeolites. The use of tailored minerals for 2 wood waste samples showed 31% toxicity and 35% TOC removal (Zenaitis etal., 1999). Biological treatment is perhaps the most widely used process for wood waste effluents. More specifically, several studies have shown that biological treatment is effective in removing BOD 5, COD, TOC, and resin acids to acceptable water quality levels (Borga etal., 1996, Zenaitis etal., 1999, Magnus etal., 2000, Zenaitis etal., 2002). Biological treatment methods assessed or suggested for wood waste runoff include: constructed wetlands, aeration and stabilization lagoons, and suspended and fixed biofilm reactors (Zenaitis etal., 2002). 16 The advantages of constructed wetlands are the diversity of contaminant-uptake processes, such as soil adsorption, macrophyte uptake and microbial catabolism. These systems are inherently stable and productive due to the following macrophyte attributes: • Leaves and stems both provide heat insulation and shield the water column from sunlight which prevents algal blooms • Stems and roots trap colloid and release oxygen, thus encouraging microbial metabolism A laboratory-scale, constructed wetland removed 63 to 95% of BOD, COD and tannins/lignins (TL) and reduced toxicity by a factor of 25 from hog fuel leachate over a hydraulic retention time (HRT) of 14 days (Zenaitis etal., 1999). In the subsequent pilot-scale study, removal of BOD, COD, TL ranged from 20 to 45% and only 49% of trout toxicity was removed. Due to the long retention time, this method shows potential for small liquid loads (for example hog fuel leachate), however the liquid discharge produced during a storm event on a dry land sort would most likely exceed capacity. Aeration (mechanical aeration) and stabilization (solar oxygenation) lagoons are a low cost alternative treatment strategy, typically employed for sewage treatment in rural areas or industrial operations able to donate large land areas to waste management (Rittmann and McCarty, 2001). They function similar to industrial sedimentation basins in the removal of particulates, however differ in that they are open systems that encourage microbiological degradation of organic materials and nutrients. A lab-scale model of an aerated lagoon showed moderate to high treatment potential for a mechanical pulping effluent as it removed BOD reduction from 52-87% and medium to high toxicity (B.C. Research, 1973a). Nevertheless, the large land requirements and long retention time required, have dictated that lagoons will typically not be employed for treatment of wood waste effluents, especially DLS runoff. Suspended and fixed growth reactors have shown the greatest potential to treat woodwaste runoff. In suspended growth bioreators, a microbial seed is typically administered and the microorganisms are maintained in solution by stirring and forced aeration. The suspended biomass is responsible for degradation and/or adsorption and flocculation of suspended particles (Tchonanoglous etal., 2003). This type of reactor was previously shown to remove 99% BOD, 80% COD, 90% tannin and lignin, and lower Microtox toxicity from EC50 of 1.83% to 50.4% in one sample of wood waste runoff from a sawmill in British Columbia (Zenaitis etal., 2002). A full-scale HCR (High Efficiency Compact Reactor) activated sludge plant showed excellent treatment efficiency for effluent from an integrated newsprint mill. The HCR system removed 79% COD (of 5000 mg/L), 98% BOD (of 1490 mg/L), 100% toxicity, 85% of extractible materials (160 mg/L) and 99% (15.6 g/L) of the resin acids (Magnus etal., 2000). Fixed film bioreactors have also been assessed for treatment of woodwaste effluents. Microbial populations, biofilms, are attached to inert packing materials such as rock, gravel, sand, and wide range of synthetic materials (Tchonanoglous etal., 2003). Common systems include trickling filters, suspended carrier and rotating biological contactors. A rotating biological contactor removed an average of 70% BOD (ranging 2600 to 5400 mg/L) and 46% COD (ranging 4610 to 7960 mg/L) over loading rates of 0.007 to 0.033 kg BOD/m 3 and 0.012-0.055 kg COD/m 3 respectively in the treatment of chemical TMP wastewater. The authors concluded that the liquid loading rates tested exceeded the optimum capacity of the system (Mathys etal., 1997). In a recent study, a lab-scale, trickling filter biological contactor method was assessed for the treatment of log yard runoff. At reactor temperatures of 34 °C, 94-100% of BOD, 86-93% of COD, and 91 -97% of tannins and lignans (TL) were removed from solution. Treatment at 5 °C and 24 °C removed 76% and 97% BOD, 64% and 91% COD and 67% and 95% of TL respectively (Woodhouse, 2003). This treatment process shows great potential, however several challenges remain to prove its treatment performance. For example, the effect of temperature and intermittent loading upon biofilm health and performance. Consequently, relying on biological treatment alone may present some technical challenges. The success of biological treatment for treatment of wood waste effluents has been attributed to both consumption of organic carbon (lowering COD and BOD) and more specifically to the degradation of resin acids (Magnus etal., 2000). However, for pulp effluents, variations in pH and resin acid loading rates were shown to correspond 18 with resin acid and toxicity breakthrough. pH fluctuations alter the physiochemical state of resin acids. For example at low pHs (below 6) resin acids are more hydrophobic and bound to the particulate and colloidal phases and thus more bioavailable. Inversely, at pH's above 6, resin acids are more soluble and less bioavailable. pH and resin acid concentration changes also may change the active populations of microorganisms. While a lag stage in microbial growth occurs (metabolic dormancy), resin acids may break through reactor treatment (Zhang etal., 1997, Werker and Hall, 1999). Measures to increase hydraulic retention time (HRT), have to be taken when fluctuations in pH and resin acid loading rates occur. Filtration, similar to adsorption, is a widespread mechanism in particle removal processes. There are three types of filtration processes typically employed: surface filtration, membrane filtration and depth filtration. Surface filtration employs mechanical sieving similar to the function of a kitchen colander. Filters are made of woven metal, cloth or synthetic materials and effective pore size can range from 10 to 30 pm or larger (Tchonanoglous etal., 2003). These filters would be effective in filtering large wood waste particles, however sedimentation may be a more energy and cost effective means to achieve removal of large particulates. On the opposite extreme to surface filtration lies membrane filtration. There are several size ranges used in membrane filtration technologies. These include microfiltration (0.08-2.0 pm filters for TSS, turbidity, and protozoa), ultrafiltration (0.005-0.2 pm filters for macromolecules, colloids, and most bacteria), and nanofiltration (0.001-0.01 pm for viruses and micromolecules). Membrane filtration is typically a polishing step for wastewater recycling projects and potable water production. Drawbacks to membrance filtration are high capital costs and operational costs due to rapid membrane fouling (Tchonanoglous etal., 2003). Depth filtration is conventionally used for potable water. Its main advantages are low capital cost and simplicity of design and operation (Brink and Parks, 1996). For most processes, water is introduced onto a bed of granular media set on top of support media, typically gravel. Particles are strained reducing inorganic, organic and biological particulates and colloids. During successive batch filter runs, particles accumulate in the interstices of the granular media and the headloss builds to a level beyond the initial headloss. At a saturation level, headloss development will exceed a threshold and filter 19 performance becomes compromised. Filter regeneration is achieved by back washing, a reverse reaction. Filtration efficacy is dependent upon the physical features of column (depth, width), media characteristics (grain size), liquid loading rates, and the development of biological activity (Haarhoff and Cleasby, 1991). The first depth filtration method developed was slow sand filtration. Slow sand filtration has recently seen resurgence of use worldwide. There are an estimated 225 slow sand filtration plants in the US (Brink and Parks, 1996). Slow sand filters utilize low infiltration rates (0.04 to 0.2 m/h), small grain sizes (0.15-0.30 mm mean diameter) and shallow depths (0.8-0.9 m). Typical design periods are 24 h/d of continuous operation for 10-15 years. As slow sand filters run without pre-chlorination and with high retention times, biological activity is prevalent (Haarhoff and Cleasby, 1991). Rapid sand filters typically use larger grain sizes and higher filtration rates with chlorination. Rapid filters may require cleaning every 1 to 2 days when threshold headloss develops. Table 2 compares slow sand filtration with rapid sand filtration. Table 2.2: Typical characteristics of slow and rapid sand filters (Logsdon, 1991) Slow Filters Rapid Filters Filtration Rate 0.1 m/h 10 m/h Water above top of sand 1.5 m 1.5 m Sand depth 800 mm 800 mm retention time above sand 15 hour 9 min retention time in sand 3.2 hour 2 min cvcle lenath 1 -6 months 1 -4 davs Grain size is the primary characteristic, which affects filter performance. It affects both the initial and operational headloss development. A small grain size builds a filter media able to capture small particles, however increased friction and increased terminal headloss is directly correlated with small grain size. This, in turn, increases retention time and lowers the liquid loads that the filter bed can accommodate. Thus, design of slow sand filters must account for estimated liquid loads. The exact mechanisms of particle removal in sand filtration will be discussed in more detail in Chapter 4. 20 As a final introductory note, there are several inherent functional differences that one must consider when assessing the treatment systems used in conventional pulp and paper mills and those proposed for dry land sorts. First and foremost, because the milling process requires sophisticated machinery and technicians, there is ample infrastructure such as power and personal already established on site. In contrast dry land sorts are often far removed from municipal power supplies and their associated resources, thus treatment systems may be required to consume less power. Secondly, because effluents from pulp mills are potentially many orders of magnitude more toxic, strict regulations are enforced. As a result, wastewater treatment systems for pulping operations are integrated into the pulping process; whereas treatment systems for dry land sorts are largely add on systems, which presents significant technical limitations such as space, energy, cost, operation-simplicity and solids handling. In light of this, many conventional treatment systems such as chemical treatment, integrated bioreactors, aeration ponds and wetlands have limited potential on dry land sorts. 2.3 Preliminary Runoff Characterization and Treatment Trials In April of 2001, Microbial Technologies Inc., was contracted by two dry land sorts to evaluate vertical flow wetlands for treatment of runoff. These installations were estimated at reducing the immediate land requirements by 80%. In the first series of studies, samples of runoff were found to contain TSS (400-4000 mg/L), BOD 5 (up to 400 mg/L), COD (300-1200 mg/L), resin and fatty acids (0.7 mg/L), metals (Fe, Al, Zn), and pH (4.5-5.5) all exceeding BC water quality guidelines. DLS run-off was found to fail the LC 5o Rainbow Trout test, which is taken as evidence of deposition of a substance deletious to fish (Fisheries Act 1995). In preliminary trials, Microbial Technologies Inc. assessed a sand column filtration process. In lab-scale models, runoff samples were circulated through separate granular media columns as a sequencing batch reactor. All columns were observed to reduce BOD 5 and TSS, however one specific sand media, showed enhanced ability to improve water quality. In the latter column, zinc, aluminium and Microtox toxicity was removed to satisfy BC Approved Water Quality Quidelines. Furthermore, this specific sand was observed to treat successive batches of DLS runoff without a consequent decrease in treatment efficiency or the requirement of column regeneration (back-washing). Whereas, particles were trapped within the other columns and consequent headloss was observed, the preferred column was observed to produce settleable solids in the solution reservoir. Microbial Technologies suggested that this specific sand column treatment process, with the preferred sand, was not simple sand filtration, sensu strico, and that it may be self-sustaining: in contrast to conventional sand filtration processes. Their proposed sand filtration system was operating as a black box (FRBC Technical Report, 2001). Microbial Technologies findings precede this thesis work on characterization of DLS runoff and there proposed sand column treatment process. 22 References Journal articles, conference proceedings and books Ard, T.A. and T.J. McDonough, 1996. Toxicity assays in the pulp and paper industry- a review and analysis. Tappi International Environmental Conference and Exhibit Proceedings, 901-913. Bailey, H.C., J.R. Elphick, A. Potter, E. Chao, D. Konasewich and J.B. Zak, 1999a. Causes of toxicity in stormwater runoff from sawmills. Envrionmental Toxicoloty and Chemistry, 18:1485-1491. Bailey, H.C., J.R. Elphick, A. Potter, E. and J.B. Zak, 1999b. Zinc toxicity in stormwater runoff from sawmills in British Columbia. Water Research 33: 2721 -2725. Bailey, S.E., T.J. Olin, R. M. Bricka, and D.D. Adrian, 1999. A Review of potentially low-cost sorbents for heavy metals, Water Research, 33:2469-2479. Borga, P., T. Elowson and K. Liukko, 1996a. Environmental loads from water sprinkled softwood timber: 1. Characteristics of and open and a recycling watering system. Environmental Toxicology and Chemistry\5: 856-867. Borga, P., T. Elowson and K. Liukko, 1996b. Environmental loads from water sprinkled softwood timber: 2. Influence of tree species and water characteristics on wastewater discharges. Environmental Toxicology and Chemistry 15: 1445-1454. Buffle, J., K.J. Wilkenson, S. Stoll, M. Filella, and J. Zhang, 1998. A generalized description of aquatic colloidal interactions: the three-colloidal component approach. Environmental Science and Technology, 32: 2887-2899. Cheremisinoff, P., 2002. In Handbook of Water and Waste-water Treatment Technologies, Butteworth and Heinemann: Boston. •deHoop, C.F., D. A. Einsel, K. S. Ro, S. Chen, M. D. Gibson and G. A. Grozdits, 1998. Stormwater runoff quality of a Louisiana log storage and handling facility. Journal of Environmental Science and Health, A33:165-177. Graham, N. and R. Collins, 1996. In Advances in Slow Sand and Alternative Biological Filtration. Wiley and Sons, New York. Haarhoff, 1991. In Advances in Slow Sand and Alternative Biological Filtration, Wiley and Sons, New York. Hoel and Aarsand, 1995. Accute toxicity of colloidal and dissolved material in TMP effluents. Nordic Pulp and Paper Research Journal, 2:98-109. 23 Logsdon, G.S., 1991. In Slow Sand Filtration. American Society of Civil Engineers: New York. Magnus, E., G. E. Carlberg, and H. Hoel, 2000a. TMP wastewater treatment, including a biological high-efficiency compact reactor. Nordic Pulp and Paper Research Journal, 15:29-36. Magnus, E., G. E. Carlberg, and H. Hoel, 2000b. TMP wastewater treatment, including a biological high-efficiency compact reactor. Nordic Pulp and Paper Research Journal, 15:37-45. Mathys, R.G., R.M.R., Branion, K.V. Lo, K.B. Anderson, P. Leyen, and D. Louie, 1997. CTMP wastewater treatment using a rotating biological contactor, Water Quality Research Journal, 32:771-794. Orban, J. L, R. A. Kozak, R. C. Sidle, and S. J. B. Duff, 2002. Assessment of relative environmental risk from logyard run-off in British Columbia. The Forestry Chronicle, 78:145-151. Perret, D., M. E. Newman, J.C. Negre, Y. Chen and J. Buffle, 1993. Submicron particles in the Rhine river I: physiochemical characterization. Water Research, 28:91-106. Ran, Y., J. M. Fu., G.Y. Sheng, R. Beckett, and B.T Hart, 2000. Fractionation and composition of colloidal and suspended particulate materials in reivers. Chemosphere, 41: 33-43. Ryan, J.N., M. Elimelech, A.A. Ard, R.W. Harvey, and P.R. Johnson, 1999. Bacteriophage PRD1 and silica colloid transport and recovery in an iron oxide-coated sand aquifer. Environmental Science and Technology, 33:63-73. Spath, R., H. C. Flemming, and S. Wuertz, 1998. Sorption Properties of Biofilms, Water Science and Technology, 37:207-210. Tchonanoglous, G., F.L. Burton, and H.D. Stensel, 2003. In Wastewater engineering: treatment, disposal, and reuse, Metcalf and Eddy Fourth Edition, McGraw and Hill Co: New York. Tian, F., A. L. Wilkens and T. R. Healy, 1998. Accumulation of resin acids in sediments adjacent to log handling area, Tauranga Harbour, New Zealand. Bulletin of Environmental Contamination and Toxicology, 60:441-447. Werker, A. G. and E. R. Hall, 1999. Limitations for biological removal of resin acids from pulp mill effluent. Water Science and Technology, 40:281 -288. Winegardner, D.L., 1996. In An Introduction to Soils for Environmental Professionals. Lewis Publishers: New York. Woodhouse, C. A., 2003 Attached growth biological treatment of stormwater run-off from log yards, M.A.Sc. thesis, The University of British Columbia, Vancouver, B.C. 24 Zenaitis, M. and S. J. B. Duff, 2002. Ozone removal of acute toxicity from logyard runoff. Ozone Science and Engineering, 24:83-90. Zenaitis, M., K. Frankowski, K. Hall, and S. Duff, 1999. Treatment of runoff and leachate from wood processing operations. The Sustainable Forest Management Network Conference, Science and Practice: Sustaining the Boreal Forest, Edmonton, Alberta. Zenaitis, M., H. Sandhu, and S. J. B. Duff, 2002. Combined biological and ozone treatmen of log yard run-off. Water Research, 36: 2053-2061. Zhang, Y., P.A. Bicho, C. Breuil, J. N. Saddler and S. N. Liss, 1997. Resin acid degradation by bacterial strains grown on CTMP effluent. Water Science and Technology, 35:33-39. Government and industry reports Alberta Forest Products Association (AFPA), 1999. Characterization of surface water run-off from log yard sites in Alberta. British Columbia Approved Water Quality Guidelines, 1998. A compendium of working water quality guidelines in British Columbia: 1998 Edition, http://wlapwww.gov.bc.ca/wat/wq/BCguidelines/approved.html B.C. Research, 1973a. Identification and treatment of toxic materials in mechanical pulping effluents, Environment Canada Forestry Service, Project Report 149-1. B.C. Research, 1973b. Identification and treatment of toxic materials in pulp and paper woodroom effluents, Environment Canada Forestry Service, Project Report 148-1. Doig, P., and P. van Poppelen, 2002. Assessment of DLS runoff water quality and toxicity on the Sunshine Coast of British Columbia, Forest Renewal British Columbia Report. Environmental Acts and Regulations, 2002. Acts administered in part by the minister of the environment, Fisheries Act, Pulp and Paper Effluent Regulations, (SOR/92-269), http://www.ee.gc.ca/EnviroRegs/ENG/SearchDetail.cfm?intReg=83, Environment Canada. McDougall, S., 1996. Assessment of log yard runoff in Alberta, Preliminary Evaluation. Alberta Environmental Protection. Moore, K. 1992. Wood waste leachate characterization study, B.C. Environment Lower Mainland Region Industrial Section. National Council of the Paper Industry for Air and Stream Improvement (NCASI), 1992. Storm water from log storage sites: A literature review and case study. Technical Bulletin, 637. 25 Samis, S. C , S. D. Liu, B. G. Wernick and M. D. Nassichuck, 1999. Mitigation of fisheries impacts from the use and disposal of wood residue in British Columbia and the Yukon, Canadian Technical Report of Fisheries and Aquatic Sciences, 2296. 26 Chapter 3 Characterization of Dry Land Sort Runoff In this Chapter, the physiochemical and biological properties of 12 different DLS runoff samples are characterized. In the first series of experiments, DLS suspended particles, which have been identified previously as toxic, are characterized both chemically and morphologically. In a separate series of experiments, the propensity for particles to sediment and/or biodegrade over storage is tested in order to determine the shelf-life of samples for subsequent treatment experiments in Chapter 4 and also to determine if a primary clarifier could potentially be an effective treatment tool. 3.1 Introduction Colloidal (suspended) particles play an important role in the transport, physiochemical properties and fate of organic materials, metals, and pollutants in natural waters (Ran etal. 2000, Grout etal. 1999, Buffle etal. 1998, Perret etal., 1992). For example, the toxic constituents of TMP pulp mill effluents, resin acids, are found in the colloidal (20 nm<particles<1-2 urn) and particulate (particles>1-2 urn) particle fractions as defined in previous studies (Magnus etal., 2000, Hoel etal., 1995). Consequently, it is important for the purpose of this thesis to explore natural processes such as sedimentation and treatment process options that minimize suspended particles in the effluent of the treatment system, either by using primary sedimentation or filtration or a combination of both. Although very little is known empirically about the composition and the variability of colloidal structures, some general properties have been recognized. In natural waters, colloidal materials may be classified under three types: (i) compact, spherical, inorganic colloids: the oxy-hydroxides of Fe, Al, and Si (ii) small, humic substances: the multi-component products of nutrient cycling and microbial transformation (iii) large, rigid or flexible biopolymers: fibrous sugars, lignins, and proteins (Buffle etal., 1998). 27 In the circum-neutral (5-8) pH range, inorganic colloids such as Al and Fe oxy-hydroxides are neutral or positively charged, whereas the organic colloids (humics and biopolymers) are negatively charged (Pizarro etal., 1995). Though there are three main types of colloids, size fractionation of natural waters has revealed uniformity in size among all fractions, as these colloids tend to form heterogeneous aggregates (hetero-aggregates) with a net negative surface charge (Ran etal., 2000, Buffle etal., 1998). Studies have shown that the organic colloids (humics and bio-polymers) fix the surface charge of colloidal aggregates to negative due to surface coverage of their inorganic counterparts (Ran etal. 2000, Grout etal. 1999, Buffle etal. 1998, Perret etal., 1992). Thus, colloids in natural waters are often treated as uniform, negatively-charged suspended particles large enough to have supramolecular structure and properties, however, small enough not to sediment quickly in the absence of sedimentation-factors (Buffle et al. 1998). It is understood that the relative concentration of each colloid in suspension determines the stability. Generally the small humic substances stabilize the inorganic colloids, whereas the large biopolymers tend to cross-link and destabilize heterogeneous colloidal aggregates, leading to the formation of settleable hetero-aggregates (Buffle etal., 1998). Figure 3.1 shows theoretical colloid aggregation processes in natural waters. In the absence of cross-linking biopolymers, stable hetero-aggregates persist. 28 Figure 3.1: Colloidal aggregation processes in natural waters (adapted from Buffle etal., 1998) Although weak hydrophobic or van der Waals forces may lead to the formation of unstable (settleable) aggregates, generally, in the absence of large bio-polymers, negative charge repulsion and stearic-hindrance represses the formation of settleable structures. Such factors as ionic strength, for example a high calcium concentration, and pH variability may enhance aggregate sedimentation (Oste etal., 2002). In order to gain insight into colloidal structure/function reactions and aggregation processes, several studies have characterized colloids in natural waters both physio-chemically and morphologically. In contrast, few efforts have been made to characterize wastewater streams or storm water runoff by microscopy (Perret etal., 1993, Hoel etal., 1995, Grout etal., 1999, Magnus etal., 2000, Ran etal., 2000). This is somewhat surprising, as there is an immediate pollution potential with the release of wastewater or storm water runoff. The primary focus of this study was to characterize the DLS runoff particles, specifically with regards to particle physio-chemistry, morphology and stability in 29 solution over storage. In the first series of experiments, two DLS runoff samples which represented far spectrums of DLS storage age, particle concentration, and chemical oxygen demand (COD) were fractionated into particulate (>1 urn), colloidal, and dissolved phases and characterized chemically and morphologically. These experiments yielded valuable information about the nature and properties of DLS runoff which can be exploited for treatment purposes. In a separate series of experiments, the effect of storage time and temperature upon DLS runoff particle size distribution and COD was observed to determine to what extent settling and biodegradation would occur and to determine the shelf-life of fresh sample for treatment experiments in Chapter 4. Both of these series of experiments yielded information to first establish the treatability of DLS runoff and secondly to help elucidate treatment mechanisms within the sand filtration process. 3.2 Materials and Methods 3.2.1 Sampling of DLS run-off Twelve samples of DLS run-off, from three different dry land sorts located on the Sunshine Coast (Howe Sound Sort, Fleatwood Forest Products, and Terminal Forest Products), were collected between October 25 t h, 2002 and August 28 t h, 2003. Each of three sites was located adjacent to Howe Sound: the body of water which originates near Whistler, British Columbia, flows North-South, and separates West Vancouver from the Sunshine Coast. At each site, logs were unloaded from land (by truck) or from water (log booms) and stored on open, paved decks. Figure 3.2 shows log storage piles during a storm event at Fleatwood Forest Products. Figure 3.2: Storm event causing runoff from log storage area at Fleatwood Forest Products (van Poppelen, 2002) Rain or sprinkler water, which had come into contact with buildings, machinery, logs piles, chip piles, and leachate generated DLS runoff, most of which was channeled by incline into Howe Sound. Terminal Forest Products collected runoff into a catch basin and pumped it to a nearby holding lagoon. Figure 3.3: Chip pile producing leachate located at Terminal Forest Products (van Poppelen, 2002) Each sample, with the exception of sample 4, was collected directly from the deck on route to being pumped into the catch basin during storm or sprinkling events. Sample 4 was collected from the storage lagoon at Terminal Forest Products. The storage lagoon functioned in part as a flow equalization lagoon and as primary treatment for the sand filter. The dominant wood species sorted on each site were Douglas fir and Hemlock. Table 3.1 contains the sampling date, site, runoff type and area collected. Table 3.1 Forestry DLS runoff sample descriptions 32 sample date DLS runoff number site event 1 25-Oct-02 Howe Sound rain 2 25-Oct-02 Fleatwood rain 3 11-Mar-02 Fleatwood rain 4 25-Apr-02 Terminal dry/lagoon 5 16-Sep-02 Terminal rain 6 28-Oct-02 Terminal rain 7 17-Jan-03 Terminal sprinkler 8 25-Feb-03 Terminal sprinkler 9 20-May-03 Terminal sprinkler 10 23-Jul-03 Terminal sprinkler 11 17-Aug-03 Terminal sprinkler 12 28-Aua-03 Terminal sorinkler Samples were collected in acid-washed 2L or 4L plastic containers after rinsing with an aliquot of run-off sample. Samples were then covered with sterile cheesecloth and stored at 4 °C until analysis was performed. To reduce the physiochemical and/or microbial alteration of samples, all initial characterization analyses were performed within 4 days of sample collection. 3.2.2 Fractionation and characterization of DLS run-off Filtration is the most common fractionation technique employed, however it can be error-prone due to coagulation of colloids at membrane surfaces (Schafer etal., 2000). Samples that contain a significant particulate fraction (particles >1.2um) will quickly mat filters and block smaller particles from passing through, thus also counting colloidal particles as filterable particles. For this reason a successive centrifugation cascade was chosen to first remove larger particles, followed by ultracentrifugation to embed the remaining colloids onto a viewable microgrid separating the colloids from the dissolved phase. Samples 3 and 4 were taken through this cascade, as shown in Figure 3.4, in order to analyze the fractions designated as: particulate, colloidal and dissolved. 33 Sample (Centrifuge) Pellet 1 + Supernatant 1 (Ultracentrifuge) Pellet 2 onto TEM grid Pellet 2 -+ Supernatant 2 -+• Chemical Analysis Chemical Analysis Chemical Analysis Figure 3.4 Fractionation and characterization scheme for samples 3 and 4 For the fractionation cascade, 500 mL of crude sample was centrifuged in ten, 50 mL Falcon tubes at a relative centrifugal force of 290 g (1,500 rpm at 25 °C) for 30 min using an Eppendorf 5810 R centrifuge. The supernatant containing the dissolved and colloidal fraction was carefully pipetted off (450 mL) leaving the top 2 mL of liquid. The pellet containing the particulate fraction (>2pm) was discarded. Two hundred mL of the supernatant was set aside for chemical analysis. In parallel, 250 mL of supernatant 1 (containing the dissolved and colloidal phases) was ultra-centrifuged using a Storvall RC 80 (Beckman SW 40 Ti rotor) at a relative centrifugal force of 1.6 x 10 5 g (30,000 rpm at 4 °C) directly onto carbon-coated, micro-grids. The difference in COD and metals/cations between the fresh sample and supernatant's 1 and 2 represented the COD and metals/cations in the particulate (Pellet 1) and colloidal (Pellet 2) phases respectively. Table 2.2 contains the minimum diameter of Stoke's radii that is eliminated during centrifugation and ultra-centrifugation. It was estimated that the particulate fraction would be greater than 2 pm; the colloidal fraction would be <2 pm and > 0.002 pm, and the dissolved fraction would be less than 0.02 pm (20 nm). 34 Fresh samples were stored overnight at 4 °C, prior to analysis, to allow for the settling of large particulates and woody fibers under gravity. Aliquots were pipetted off the top of the sampling containers in order to exclude settleable materials from the analysis. Fresh sample and supernatant from each centrifugation step, were measured for organic carbon (COD), metals/cations (ICP-AES) and morphologically (TEM-EDX). The BOD 5 was determined for fresh (un-fractionated sample), only. Although most aromatics or straight-chain aliphatics are not oxidized during the COD test, it was chosen over BOD 5 and TOC for measurement of organic carbon throughout the fractionation process due its relative rapidity of measurement and ability to detect, on average, 95% of the organic carbon present in solution. COD values are generally higher then BOD, which are in turn higher then TOC, since the COD usually includes the BOD, the TOC as well as other chemical demands, such as oxidizable metals such as ferrous iron. 3.2.3 Aging Study Two separate aging studies were conducted upon samples 7 and 9, respectively. For the first analysis, sample 7 was covered with sterile cheesecloth and stored at 4 °C. Particle size analysis (PSA) and COD were measured on day 1,7, 11, 17, 28, and 36. All measures were taken from the sample supernatant. In the second analysis, 50 mL aliquots of sample 9 were stored in Falcon tubes covered with sterile cheesecloth. Two sample sets were stored separately in the dark at 4 °C and 25 °C respectively. Each measurement day, triplicate tubes from each sample set were forfeited for analysis by CODwnoie vs. CODsupematant- Measurements were made on days 1, 4, 7, 10 and 13. The heterotrophic bacteria population was assessed by heterotrophic plate count (HPC). Heterotrophic plate counts (HPC) were conducted by asceptically plating serial dilutions of runoff samples onto BBL® standard methods agar composed of 5.0 g pancreatic digest of casein, 2.5 g yeast extract, 1.0 g dextrose, and 15.0 g agar. In total, 1 mL of each sample was plated over the dilution range of 10° to 107. Plates containing between 30 and 300 colony-forming units per mL (CFU/mL) were counted as viable numbers and the true concentration of cells was estimated. 35 3.2.4 Analytical methods used for characterization of DLS runoff samples Samples were characterized by BOD 5, COD, TSS, pH, turbidity, metals/cations, particle size analysis (PSA), transmission electron microscopy with electron dispersive X rays (TEM-EDX) and HPC. The five-day BOD test was performed on triplicate samples according to standard methods (APHA, 1992). Three litres of dilution water was bubbled with an aeration stone for 1 hour prior to use. One nutrient buffer pillow and BOD seed, provide by VWR Canlab, were added to 500 mL dilution water. Runoff sample were diluted between 2x and 40x (v/v) based on estimated strength of sample. Samples were incubated at 20 °C for 5 days in 300 mL, standard BOD bottles. Dissolved oxygen was tested before and after incubation. COD analysis was performed using the closed reflux-tube method (APHA, 1992). COD solutions were made by combining 0.5 mL of sample, 0.5 mL dH 20, 1.2 mL of 0.04 M potassium dichromate and 2.8 mL 0.02 M silver sulfate dissolved in 25% H 2 S0 4 . Solutions were digested in closed tubes at 150°C in a heating block for 2 hours. Adsorbance relative to a Lamotte COD standard preparation was determined at 600 nm on a Lamotte Inc., spectrophotometer (PO Box 329, 802 Washington Avenue, Chestertown, MD 21620 - USA). TSS were determined by vacuum filtration through Whatman 1.2 pm glass filter (GF/C) standard filters. The mass of filter papers, dried at 105 °C, was measured before and after filtration. pH was measured using a VWR SB20 pH meter. Turbidity was measured using a Lamotte SMART colorimeter and expressed in formazine turbidity units (FTUs) over a wavelength range of 430 to 620 nm. The concentrations of metals/cations were determined after 100 mL of sample was digested in aqua regia: 2mL of 50% H N 0 3 and 1 mL of 50% HCI. This reaction was conducted at 85 °C for 16 hours or until the volume reached 10 mL. The solution was then covered and refluxed for 1 hour. Samples were vacuum-filtered through Whatman GF/C filters and analyzed by inductively coupled plasma, atomic electron spectroscopy (ICP-AES) in flame ionization mode by NorWest Laboratories, Inc. (203 -20771 Langley By-Pass, Langley, BC, V3A 5E3). 36 Particle size analysis (PSA) was performed using the Malvern Mastersizer 2000. Solutions were analyzed by light scattering over the size range of 0.02 urn to 2000 urn at a refractive index for wood particles in solution of 1.33. TEM-EDX analysis was performed using a Hitachi H800 Scanning Transmission Electron Microscope. Liquid samples were deposited and dried on carbon coated, Cu 200, mesh microgrids and viewed at 100 kV. Serial dilutions of (2, 10,100, 200x) were made in order to discern individual colloids entrapped by ultra-centrifugation onto microgrids. 3.3 Results and Discussion 3.3.1 Primary characterization of all DLS runoff samples All samples were tested for pH, colour and COD as given in Table 3.2. Samples 1-3 and 5 were collected during rainfall events and had CODs ranging from 346 to 1260 mg/L. Of these, sample 3 was particularly high in COD (1260 mg/L) as it was collected during the first hour of a rainfall event following an extended dry period. The higher relative COD may be attributed to its collection during the first flush of materials from the deck before considerable dilution occurred. Due to the remoteness of the sampling sites, it was difficult to repeat first flush sampling. Therefore, in an effort to collect strong and potentially toxic samples, samples 6 to 12 were collected during the first flush of the DLS deck runoff with sprinkler water (sprinkler runoff). These samples collected from sprinkler runoff had CODs ranging from 1115 mg/L to 3690 mg/L. 37 Table 3.2: Results of primary characterization of DLS runoff samples in which triplicate COD was measured and the standard deviation was determined (bracketed numbers). sample number PH colour runoff event COD (+/-) (ma/L) 1 5.5 brown/opaque rain 583 (22) 2 5.1 brown/translucent rain 714(16) 3 4.6 brown/translucent rain 1260 (37) 4 5.5 black/opaque dry/lagoon 378 (23) 5 6.3 black/opaque rain 346 (11) 6 5.5 brown/opaque sprinkler 1900(43) 7 5.6 brown/opaque sprinkler 2870 (102) 8 6.0 brown/opaque sprinkler 2645 (89) 9 6.1 brown/opaque sprinkler 1620 (80) 10 5.2 brown/opaque sprinkler 2905 (62) 11 6.1 brown/opaque sprinkler 1115 (47) 12 5.1 brown/oDaaue SDrinkler 3690(113) COD is not regulated directly in Canada or the US, however it is related as discussed previously with TOC (regulated under the BC Approved Water Quality Guidelines) and BOD 5 (regulated under the BC Waste Management Act)) for organic discharges. The BC Approved Water Quality Guidelines stipulate that TOC shall not exceed 20% of the median background levels in receiving waters. The median background TOC for, fish bearing, receiving waters in BC is usually less than 5 mg/L. In contrast, waters with high natural organic material content, such as bogs, may contain between 26 and 105 mg/L TOC and are subsequently devoid of fish and most invertebrates. Other waters with high measured TOC (up to 20 mg/L) are usually situated near anthropogenic sources (BC Approved Water Quality Guidelines, 1998). The Municipal Sewage Regulation releases permits based on the sensitivity of the receiving water. For the requirement of primary treatment the effluent shall not exceed 130 mg/L TSS and 130 mg/L BOD 5 and for secondary treatment the effluent shall not exceed 45 mg/L TSS or 45 mg/L BOD 5 (Waste Management Act, 1999). Although DLS runoff is not currently regulated, COD values ranging from 346-3690 mg/L are clearly in contravention of both the BC Approved Water Quality Guidelines and the Municipal Sewage Regulations and are thus an environmental concern to receiving waters. This was apparent to regulators who applied the Pulp and Paper Effluent Regulations of the Fisheries Act to charge a Weyerhaeuser-owned dry land sort on 3 counts of non-compliance in 2000 (Fisheries and Oceans Canada New Release, March 29, 2000). 3.3.2 Fractionation and Characterization of DLS run-off Two different samples, representing a fresh sample (sample 3) and a stored sample (sample 4) were taken through the fractionation scheme shown in Figure 3.4. The fresh sample was taken during the first hour of a storm event following a month long dry period. The stored sample was taken from the storage lagoon at Terminal Forest Products site. Lagoons, primary clarifiers or equalization tanks are a common primary treatment strategy for settling of larger particulates as discussed in Chapter 2. In addition to settling, phase partitioning and particle transformation may occur in lagoons. For example, large particles may be biologically degraded into smaller particles. This is of particular concern for treatment processes using filtration mechanisms. Primarily, microbial degradation of filterable molecules into non-filterable molecules, for example particulates into colloids, may encourage particle breakthrough, through the filtration media. Secondly, phase partitioning may cause the release of toxic substances, for example metals or resin acids into a more mobile phase, for example the colloidal or dissolved phase. The chemical and morphological characteristics of samples 3 {fresh) and 4 {stored) are compared in Table 3.3. 39 Table 3.3: Chemical and morphological characteristics of samples 3 and 4 sample 3 (+/-) sample 4 (+/-) colour brown/translucent black/opaque odour woody methane/sulfur gas bubbles no yes PH 4.5 5.1 COD 1260(37) 378 (23) BOD 336(16) 66(13) BOD/COD 0.27 0.17 biodegradability medium low aggregate size 5-10 pm 1-2 pm Observations from this study found that typical fresh rainfall-generated runoff bears a strong woody-smell, a brown/translucent color, moderate to high COD and relatively low pH from 4.5 to 5.0. Stored sample has a strong; often methane or sulfur smell (typical of anaerobic degradation), black/opaque color, low relative COD (378 mg/L) and pH somewhat higher at 5 to 6. The BOD/COD ratio of fresh sample (0.27) showed a medium potential for biodegradability (0.1 is low, 0.5 is relatively high according to Tchonanoglous, 2003). The BOD/COD ratio for stored sample was 0.17, exhibiting a relatively lower potential for biodegradability. TEM images of fresh and stored samples at 2.5K magnification are shown in Figures 3.5 and 3.6 respectively. 40 Figure 3.6: TEM image at 2.5K magnification of DLS run-off sample 4 (stored) Typical particle structures observed in the fresh sample were typically 5-10 pm in mean radius. Typical particle structures observed in the stored sample were 1-2 pm in mean diameter. The persistence of small aggregate-structures only, in the stored 41 sample, is likely due to microbial degradation, sedimentation and/or dissociation of large aggregates due to dilution of the sample by rainwater. Table 3 contains the results of elemental analysis by EDX of the un-fractionated particulates in both samples. Carbon is not accurately quantified by EDX analysis; however the carbon weight percentage can be estimated by low weight percentages of the other quantitative elements. Table 3.4: Weight percents as determined by TEM-EDX of a typical 10 urn and 2pm aggregate structure found in the fresh and stored samples, respectively. fresh wt % stored wt % element (10 L«m) (2 Mm) Carbon 44 14.1 Oxygen 39.3 68.8 Aluminum 2.4 2.6 Calcium 0.9 2.1 Iron 0.7 0.9 Magnesium 7.7 9.1 Silicon 4.8 2.9 The 10 pm fresh particulate contained 16.5 % bound ions or oxides of aluminum, calcium, iron, magnesium and silicon. The weight percent (wt %) of oxygen was 39.3 % , which may have been organic or inorganic such as aluminum, silicon or iron oxides. The other 44 % of particulates was likely organic carbon. The wt % distribution of inorganic ions in the 2 pm-stored particulate was 17.6 % . This was quite similar to that of the fresh particulate. In contrast, the wt % of oxygen in the stored sample was 68.8 % , with the remaining 14.1 % being carbon. The lower concentration of carbon observed in the stored sample particulates indicated that these particulates were mostly of inorganic. This result is logical, as the stored sample appeared to be highly degraded of organic carbon indicated by its relatively low BOD, COD and biodegradability. The colloids of both samples (fresh and stored) were viewed on microgrids after removing the particulate and dissolved phases. TEM images of the colloidal fractions of fresh and stored samples at 10K and 7K magnification respectively are shown in Figures 3.7 and 3.8, respectively. 42 1 urn Figure 3.7: TEM image at 10.0 K magnification of the colloidal fraction of DLS run-off sample 3 (fresh) Figure 3.8: TEM image at 10.0 K magnification of the colloidal fraction of DLS run-off sample 4 (stored) 43 Morphologically, colloids in the fresh sample were predominantly heterogeneously-shaped particles including some aggregated structures. In addition to sphere-shaped particles, long, string-like particles between 5 and 10 pm in length were abundant. The string-like particles were most likely woody polymers, biopolymers or microorganisms which did not sediment under the first round of centrifugation (which pelleted most particles with a mean diameter of > 2 pm) due to their low relative density. None of these particles were observed in the stored sample suggesting that they had either sedimented out of solution, been degraded into smaller particles or in the case of microorganisms were not a dominant part of the flora in the storage lagoon. In contrast, the stored sample colloids were predominantly homogenously-shaped particles, with a relatively low degree of aggregation. Both samples had particles in the size range of 0.005 pm (5 nm) and up to 1-2 pm. The EDX analysis for individual fresh and stored colloids are given in Table 3.5 and 3.6 respectively. The following sized particles were analyzed: 0.005-0.5 pm, 0.5pm-1 pm, and 1-2 pm. Table 3.5: EDX analysis of fresh sample 3 colloids fresh sample (wt %) element (0.005-0.5pm) (0.5-1 pm) (1-2 pm) Carbon 41.1 43.8 38.7 Oxygen 53.7 53.0 54.6 Aluminum 0.5 0.2 1.6 Calcium 0.2 0.0 0.3 Iron 0.1 0.4 2.1 Magnesium 1.1 0.0 0.9 Silicon 3.3 2.9 2.2 44 The EDX analysis of the fresh individual colloids in the 3 different size ranges generally showed chemical uniformity amongst all fractions. Particles were mostly organic with a small silicon contingent, most likely silica dioxide. The particle scanned in the 1-2 pm range also contained colloidal iron and aluminum. These colloids were representative of inorganic/organic hetero-aggregates. The long, thin particle was entirely organic (EDX analysis not shown). Table 3.6: EDX analysis of 3 stored sample colloids stored sample (wt %) element (0.005-0.5pm) (0.5-1 pm) (1-2 pm) Carbon ' 89.0 26.9 8.5 Oxygen 11.0 61.1 62.2 Aluminum 0.0 0.0 9.8 Calcium 0.0 0.1 0.0 , Iron 0.0 0.3 7.9 Magnesium 0.0 0.1 0.0 Silicon 0.0 11.5 8.9 The EDX analysis of stored colloids (Table 3.6) revealed 3 quite different colloid types. Particles in the size range 0.005-0.5 pm were the most abundant colloids observed. These colloids completely lacked an inorganic contingent and were especially low in oxygen. Colloids in the 0.5-1.0 pm sub-range contained mostly silicone, oxygen, and carbon. These colloids resembled the 2 pm particulates observed in the un-fractionated stored sample and were most likely inorganic/organic hetero-aggregates with bound Si02. The colloids analyzed in the sub-range of 1-2 pm were predominantly inorganic aggregates. As observed with particulates in the un-fractionated stored TEM-EDX analysis, the organic carbon concentration in aggregated particles was relatively low. The persistence of abundant organic particles around 0.005 pm suggests that most of the readily biodegradable organics, such as large biopolymers, were broken down into smaller components. These particles were most likely humic compounds, the end products of microbiological degradation which bear few available reactive groups and tend to persist in solution (Buffle etal., 1998). 45 Using EDX analysis as the sole characterization tool can give misleading information, as EDX is not a quantitative technique. For example, dissolved ions such as calcium and magnesium may form salt bridges with particles under desiccation of the TEM vacuum, and may consequently get scanned as particulate. Coupling chemical characterization with TEM-EDX alleviates this uncertainty and it provides quantitative information about the fractional distribution of chemical species, such as organic carbon, cations and metals. COD and ICP-AES were performed on all fractions. Figures 3.9 and 3.10 for fresh and stored sample, respectively, summarize the results of the chemical analysis in weight percents. Appendix A, Table A.1 contains the actual concentration data. ,c9 & <& & & <<? ^ ^ ^ * ^ s <* °° 1? a particulate a colloidal • dissolved Figure 3.9: Chemical characterization of particulate, colloidal, and dissolved fractions of fresh sample of DLS runoff 46 Figure 3.10: Chemical characterization of particulate, colloidal, and dissolved fractions in stored sample of DLS runoff The TEM-EDX analysis of particulates (un-fractionated particles) and colloids in both fresh and stored samples suggested that these elements were bound in organic/inorganic hetero-aggregates. Accordingly, the majority of the aluminum (85% of 28.7mg/L), iron (75% of 22.9 mg/L), and silicon (83% of 20.6 mg/L) were found in the particulate and colloidal fractions of the fresh sample. Similarly, the particulate and colloidal fractions of the stored sample contained the majority of aluminum (85% of 4.54), however a relatively smaller percentage of the iron (50% of 13.7 mg/L) and silicon (37% of 10.8 mg/L). Colloidal aluminum (Al203) has a higher pHpzc (-9.5) then colloidal iron (Fe 2 0 3 , FeOOH) and is subsequently more positive at the pH 5 (pH of stored sample) (Buffle etal., 1999). Thus, aluminum is expected to have more surface coverage with negatively charged organic colloids relative to iron and consequently elude reduction and dissolution into the dissolved phase. The relatively lower concentration of particle-bound silicon may simply be explained by the lower 47 concentration of particles in solution and the consequent tendency for particles to diffuse and disperse in solution if given the space (Harris, 1996). The dissolved phase of both samples contained concentrations of cations, such as Na, K, Ca, and Mg, typical of brackish waters (Manahan, 1994). All of these cations were located in the dissolved phase of the stored sample. A small percentage (<20%) of these cations were associated with the colloidal phase of the fresh sample only. The weight percents of particulate magnesium (7-9%) and calcium (1-2%) reported by EDX analysis for the un-fractionated samples were most likely due to formation of salt bridges under the TEM vacuum and not a true report of particulates. The particulate and colloidal fractions of the fresh sample contained 52% and 39% respectively of the total COD (1260 mg/L). In contrast, the particulate and colloidal fractions of the stored sample contained 8% and 84% respectively of the total COD (378 mg/L). The absence of long, thin biopolymers in the stored sample colloidal fraction may explain the relatively low distribution of COD in the particulate phase. As discussed previously, biopolymers tend to cross-link colloidal particles and consequently form large and settleable particulates. Consequently, these aggregates may have settled out of solution or may have alternatively been biodegraded into humic compounds during storage in the lagoon. Likewise the stored sample contained an abundance of small (0.005-0.50 pm), spherical, organic particles as shown in Figure 3.8. Considering the evidence of microbial activity in the stored sample (odor, color, gases, and low BOD 5) and absence of large biopolymers, these colloids were most likely humic substances. Humic substances are important in the collection, transportation and partitioning of organic pollutants and metals in natural waters as discussed previously (Perret etal. 1993). Accordingly, 93% of the zinc was found in the colloidal fraction (7% dissolved) of the stored sample, which incidentally was observed to have a high concentration of humic colloids. In contrast, only 18% of zinc was found in the colloidal fraction (82% dissolved) for the fresh sample which was less degraded, thus had a lower concentration of humic substances. The distribution of metals between the dissolved and colloidal phases of solution may have important implications for the treatment 48 strategies of DLS runoff. Accordingly, the concentrations of zinc were 1.18 mg/L and 0.91 mg/L for the fresh and stored samples, respectively. These concentrations are approximately two orders of magnitude above the BC water quality guidelines limit of 0.030 mg/L and thus must be considered as potentially toxic (BC Approved Water Quality Guidelines, 2001). 3.3.3 Aging study The COD of sample 7 stored at 4 °C decreased rapidly over the first 7 days (-33% reduction) and decreased at a steady rate over the 29 succeeding days (54% reduction) as shown in Table 3.8. The particle size analysis histograms are given in Figure 3.11 for the supernatant of the DLS runoff samples (settled materials were not disrupted). Table 3.7: Change to PSA and COD over 36 days of storage time COD COD/COD d a yi > 1pm wt% d a y 1 - wt% (davs) fma/L) % wt. % wt% 1.0 2870.0 100.0 75.8 0.0 7.0 1920.0 66.9 70.7 5.1 11.0 1625.0 56.6 72.0 3.8 17.0 1480.0 51.6 65.7 10.1 28.0 1370.0 47.7 42.8 33.0 36.0 1310.0 45.6 59.6 16.2 49 Particle Size Otetriinrtion 11 10 I 9 I 3 ! 7 ! Putl'Kh Si™ Patributipti Day7 Psrtida Size (pm) Partici* Siz« Dtetrftiutlen Day 11 10 100 Particle Size (Mm) Parties 8tt« Or«tfiButloo Day 17 Particie Size (pm) Particle Size Ogtributton Part ltd* Sfea DHtrtbution Day 28 10 100 Particle Size (pm) Day 36 Particle Size (urn) Figure 3.11: Particle size histograms constructed using Malvern Mastersizer (2000) software showing change in PSA over 36 storage days of storage at 4 °C Approximately 75% of the total particles of sample 7 measured by light scattering were above 1 pm (operationally defined as particulates). After 11 days of storage, the weight percentage of particulates had only decreased slightly from 75.8% to 72.0%, suggesting that storage of sample for approximately 11 days is suitable to achieve a reasonable degree of repeatability for sample composition during the successive batch treatment experiments conducted in Chapter 4. After storage for 36 days, the weight percentage of particulates had decreased by 16.2 % . Sediments that accumulated in 5 0 the solution reservoirs during the experiment would account for some of the COD and particulates decrease (not measured). Another cause of COD and particulates reduction may have been due to microbial degradation and consequent release of carbon ( C O 2 / C H 4 ) . Thus, the amount of settling and bacterial degradation was measured for another sample. Accordingly, a heterotrophic plate count of sample 9 revealed a cell concentration of 1 x 107 CFU/mL. This significant concentration of heterotrophic microorganisms in the sample supported the hypothesis that microbial degradation may have a significant role in the removal of COD from DLS runoff suspension. Thus, sample 9 was subjected to an aging study, similar to the first study, with an additional temperature variable (4 °C vs. 25 °C) and measures of COD d e g r a d e d and CODsedimented-Tables 3.9 and 3.10 contain the calculated COD d e g raded and COD s e d imented and Figure 3.12 summarizes the changes in COD measured for each sample over storage. Table 3.8: COD d e g r a d e d and COD s e d imented during storage of sample 9 for 13 days at 4°C. C O D s U p e r denotes the COD in the sample supernatant. day Cookie COD s u p e r COD d eg r a d e d C O D s e d i m e n t e d ( C O D ^ - C O D ^ ) (COD^e-CODs^) (+/-) (+/-) (+/-) (+/-) 1 1620(104) 1620(104) 0(146) 0(146) 4 1749(113) 1305(68) -129(153) 444(132) 7 1669(137) 1289(16) -49(171) 380(138) 10 1655 (86) 1342 (67) -35(135) 313(110) 13 1574(60) 1135(11) 46(120) 439 (61) Table 3.9: COD d e g r a d e d and COD s e d imented during storage of sample 9 for 13 days at 25°C day CODwrioie COD s u p e r C O D d e g r a d e d C O D s e d i m e n t e d ( C O D ^ - C O D ^ ) ( C O D ^ - C O D ^ ) (+/-) (+/-) (+/-) (+/-) 1 1620(104) 1620(104) 0(146) 0(146) 4 1141 (91) 994(14) 479(104) 147 (92) 7 1111 (139) 696 (22) 509(106) 415(141) 10 1136(147) 672 (35) 484(109) 464(151) 13 938 (86) 533(17) 682(105) 405 (87) 51 2200.0 2000 !> 1400.0 o o 1000.0 800.0 600.0 400.0 0 whole_4C — A — super_4C whole_25C super_25C 6 8 time (days) 10 12 14 Figure 3.12: Change in CODwhoie and COD s u p ernatant over storage of sample 9 at 4°C and 25°C for 13 days There was no difference between CODsedimentation for sample stored at 4 °C vs. sample stored at 25 °C. In both cases, supernatant C O D sedimented rapidly during the first seven days of storage. Over 13 days, approximately 25% of the supernatant COD (1620 mg/L) sedimented. The previous aging analysis conducted on sample 7 found a 44% (1263 of 2870 mg/L) decrease in the supernatant COD over 11 days of storage. The apparent higher CODsedimentation rate for sample 7 may have been attributed to a greater relative particle concentration and thus particle collision probability and consequent formation of larger, settleable particles. Samples stored at 4 °C appeared to be stable with respect to biodegradation up to 13 days of storage. In contrast, samples stored at 25 °C lost approximately 42% (682 +/-105 mg/L of 1620 mg/L) COD. These results suggest that stored DLS runoff is readily biodegraded by natural flora found in the wood waste or originating from the paved DLS decks. This property of DLS runoff may be integral to its treatment. 52 3.4 Conclusions DLS runoff varies in physiochemical properties based on sampling site and rainfall conditions. For example, COD strength varied from 346 to 3690 mg/L over 12 samples collected from 3 sites. CODs in this range are unacceptable by BC water quality guidelines and municipal sewage regulations for release into adjacent waterways. The high organic loading of DLS runoff thus poses an immediate threat to aquatic life. In this study, over 90% of this COD was found in the suspended particles of 2 samples tested. Since previous work has found toxicity of wood waste effluents to be associated with suspended particles, specifically resin acids (Hoel et al., 1995, Magnus et a)., 2000) or perhaps metals (Woodhouse, 2003), it was the purpose of this study to investigate the physiochemical properties of DLS runoff suspended particles in order to elucidate conditions that could be potentially exploited in a treatment process. 3.4.1 Characterization of fresh and stored runoff The two samples investigated, /ires/? and stored were extremely different in physiochemical properties suggesting that storage of DLS runoff readily changes its composition. For example, the fresh DLS runoff had COD of 1260 mg/L and was predominantly composed of particles up to 5 to 10 pm in mean diameter, including long, string-like biopolymers. In contrast, the stored DLS runoff had COD of 378 mg/L and appeared to be highly degraded, consisting of mostly smaller (1-2 pm) particulates and colloids. Size fractionation and chemical analysis showed that the fresh sample contained 52% particulate COD and 39% colloidal COD (9% dissolved), whereas the stored sample contained 8% particulate COD and 84% colloidal COD (8% dissolved). TEM-EDX analysis of the colloidal phase of the stored runoff indicated the presence of inorganic colloids and humic substances. Since both are involved in the phase partitioning and transport of pollutants, this was an important observation. The concentrations of zinc for the fresh (1.18 mg/L) and stored (0.91 mg/L) and samples were well above the BC water quality guidelines for discharges (0.030 mg/L) and thus pose an environmental concern toxic (BC Approved Water Quality Guidelines, 2001). Upon fractionation of the stored DLS runoff sample, 93% of the zinc was found in the colloidal phase (only 18% colloidal, 92% dissolved in the fresh sample), which gives evidence that the small organic particles seen in the stored colloids of the were humic substances as heavy metals such as Al, Cu, and Zn are known to readily adsorb to their surfaces (Benjamin et al., 1996). The question remains as to which phase partition of metals is more readily treatable: dissolved or colloidal. 3.4.2 Aging Study The aging study confirmed the hypothesis that storage of DLS runoff sample greatly altered its physiochemical properties. For example, the COD of DLS runoff sample 7 decreased from 2870 mg/L to 1310 mg/L (54% decrease) and the weight percent of particulates (particles > 1.2 pm) decreased from 75.8% to 59.6% (16.2 decrease) over 36 days of storage at 4 °C. In the second analysis, DLS runoff sample 9, stored at 4 °C and 25 °C for 13 days, lost 439 mg/L and 405 mg/L COD (of 1620 mg/L COD ~ 25%) respectively to sedimentation (CODsedimentation). In addition, storage at 25 °C caused the loss of an additional 533 mg/L COD (of 1629 mg/L COD ~ 33%) presumably to degradation (CODbi0degradation)- Sample 9 stored at 4 °C was relatively stable from degradation for 13 days. 54 3.5 References Journal articles, conference proceedings and books Aim, R.B, S. Vigenswaran, H. Prasanthi, and V. Jegatheesan, 1997. Influence of particle size distribution in granular bed filtration and dynamic microfiltration. Water Science and Technology, 36: 207-215. American Public Health Association (APHA) Greenberg, A.E., L.S. Clesceri and A.D. Eaton, 1992. in Standard Methods for the Examination of Water and Wastewater 18th edition, American Water Works Association. Water Environment Federation: Washington, D.C. ^ Benjamin, M., R.S. Sletten, R.P. Bailey, and T. Bennette, 1996. Sorption and filtration of metals using iron-oxide-coated sand. Water Research. 30: 2609-2620. Buffle, J., K.J. Wilkenson, S. Stoll, M. Filella, and J. Zhang, 1998. A generalized description of aquatic colloidal interactions: the three-colloidal component approach. Environmental Science and Technobgy, 32: 2887-2899. Chen, Y. and J. Buffle, 1996. Physiochemical and Microbial Preservation of colloid characteristics of natural water samples 1: Experimental Conditions, Water Research. 30:2178-2184. Hoel and Aarsand, 1995. Accute toxicity of colloidal and dissolved material in TMP effluents. Nordic Pulp and Paper Research Journal, 2:98-109. Grout, H., M. R. Wiesner, and J.Y. Bottero, 1999. Analysis of colloidal phases in urban stormwater runoff, Environmental Science and Technology, 33:831-839. Oste, L.A., E.J.M. Temminghoff and W.H. Van REisdijk, 2002. Solid-solution partitioning of organic matter in soils as influenced by an increase in pH or Ca concentration, Environmental Science and Technology, 36:208-214. Perret, D., M. E. Newman, J.C. Negre, Y. Chen and J. Buffle, 1993. Submicron particles in the Rhine river I: physiochemical characterization. Water Research, 28:91-106. Ran, Y., J. M. Fu., G.Y. Sheng, R. Beckett, and B.T Hart, 2000. Fractionation and composition of colloidal and suspended particulate materials in reivers. Chemosphere, 41: 33-43. Schafer, A.I., U. Schwicker, M.M. Fischer, A.G. Fane, and T.D. Waite, 2000. Microfiltration fo colloids and natural organic matter, Journal of Membrane Science 171:151-172. Tchonanoglous, G., F.L. Burton, and H.D. Stensel, 2003. In Wastewater engineering: treatment, disposal, and reuse, Metcalf and Eddy Fourth Edition, McGraw and Hill Co: New York. 55 Government and industry reports Alberta Forest Products Association (AFPA), 1999. Characterization of surface water run-off from log yard sites in Alberta. B.C. Approved Water Quality Guidelines, 1998. A compendium of working water quality guidelines in British Columbia: 1998 Edition, http://wlapwww.gov.bc.ca/wat/wq/BCguidelines/approved.html B.C. Waste Management Act, 1999. Municipal Sewage Regulation, http://www.qp.gov.bc.ca/statreg/reg/W/WasteMgmt/l 29_99.htm, B.C. Ministry of Water, Land and Air Protection. Environmental Acts and Regulations, 2002. Acts administered in part by the minister of the environment, Fisheries Act, Pulp and Paper Effluent Regulations, (SOR/92-269), http://www.ec.gc.ca/EnviroRegs/ENG/SearchDetail.cfm?intReg=83, Environment Canada. McDougall, S., 1996. Assessment of log yard runoff in Alberta, Preliminary Evaluation. Alberta Environmental Protection. Moore, K. 1992. Wood waste leachate characterization study, B.C. Environment Lower Mainland Region Industrial Section. Chapter 4 In this Chapter, results are reported for sand column treatment and batch adsorption experiments of dry land sort (DLS) run-off. First a literature review of the mechanisms responsible for particle removal in sand filtration processes is presented. This is followed by a description of the materials and methods used in the sand column treatment experiments. Finally, the sand column test results are presented and their significance discussed. 4.1 - Background The suspended particles of dry land sort (DLS) runoff were characterized previously as negatively charged hetero-aggregates of inorganic and organic colloids. These colloidal particles contained the majority of the DLS runoff BOD 5, COD, turbidity, resin acids and acute trout toxicity which all exceeded BC water quality guidelines (FRBC Technical Report, 2002). Therefore, DLS runoff requires treatment, specifically by particle removal, before it is discharged into its adjacent watersheds. Sand filtration is a common particle removal method employed in wastewater treatment systems. Its main advantage is its simple design characteristics and low capital cost (Logsdon, 1991). Its main disadvantage is column saturation causing head-loss development and/or particle breakthrough limiting sand filter to a finite amount of runoff events. Sand filters need to be continuously cleaned or replaced. For this reason, sand filtration is conventionally used as a tertiary polishing step to remove residual BOD and TSS (Logsdon, 1991). In contrast to convention, a sand filtration process developed by Microbial Technologies Inc. has shown potential as a primary treatment step, perhaps providing adequate overall treatment for DLS runoff before it is released into the receiving environment. This filter was observed to treat multiple successive batches of DLS runoff (without pre-treatment) and without any decrease in filter performance, when operated at low infiltration rates (0.1 m/h) and in a closed loop configuration. The preliminary results of these treatment trials were extremely surprising considering the 57 high turbidity and particle loads of DLS runoff as discussed in Chapters 2 and 3 of this thesis. There are several key mechanisms that may be responsible for this sand column's enhanced treatment capacity. The primary mechanism is filtration. Filtration is achieved by two processes: physical straining and surface deposition. Straining is the entrapment of particles in pore restrictions and is effective when the ratio of colloid size (d) to sand grain size (D) exceeds a threshold range, typically 0.05-0.2 (Wan et al., 1997). Surface deposition occurs by collision (interception, diffusion, sedimentation) and attachment (electrostatic, van der Waals, hydrophobic forces). Most theoretical modeling has been done on water saturated porous media, however many studies show that particle retention is more effective with unsaturated (partially dry) media (Haarhoff etal., 1991, Wan etal., 1997). An unsaturated medium is achieved by using low infiltration rates. This phenomenon was described previously as 'film straining', in which the replacement of water films bearing colloids with gas (during drying) causes the subsequent enhanced deposition of colloids (Wan etal., 1997). Conventional filter sands are composed of negatively charged quartz minerals (Si0 2 pHpzc ~ 2-4). Removal of negatively charged, colloids from a percolated solution is therefore, largely a filtration process. In contrast, sand grains containing oxide surfaces are host to a number of surface selective reactions. Colloid adsorption, desorption and transport along Al- and Fe-oxide surfaces has been well documented (Sposito, 1989, Schulthess etal., 1991, Benjamin etal., 1996, Lo etal., 1997, Ryan et al. 1999). Sand grains containing iron or aluminium oxides bear a high pHpzc (-8-10) and are thus positively charged in the circumneutral pH range. These oxide surfaces elicit pH-dependant adsorption and desorption reactions with negatively charged particles. Thus, at pHs below 8-10 Al and Fe oxides have the ability to adsorb negatively charged ligands such as those found on organic colloids (carboxylic acids -COO" groups for example). A third mechanism present in sand filtration may be flocculation. Previous work has shown that once particles are adsorbed to oxide surfaces, their unabsorbed sites are exposed for further surface reactions which include cation exchange, ligand 58 exchange, adsorption and hydrophobic interactions (Lenhar and Saiers, 2002, Weng et al., 2002, Lu and Pignatello, 2002). These reactions may lead to the formation of larger, settleable aggregates that may be eluted from the column as floes. Accordingly, in preliminary treatment trials conducted by Microbial Technologies Inc., settled floes were observed in the solution reservoir after a treatment run was conducted. It was hypothesized that floe formation, elution, and settling may be one of the mechanisms that enabled the column to treatment successive batches of DLS runoff without any apparent column saturation occurring. Perhaps the most important, yet least understood mechanisms in sand filtration are biological. Accordingly, collections of microorganisms on and in the interstices between filter grains, biofilms, are found in almost every municipal sand filter that is free of pre-chlorination (Haarhoff etal., 1991). Biofilms in these conventional slow sand filters are typically responsible for adsorption, filtration, degradation and flocculation processes (Logsdon etal. 1991, Grahm etal., 1996). As such, particulate materials intercepted by biofilms represent an important carbon source for microorganisms (Larson etal., 1994). The dominant population of microorganisms found in fresh DLS runoff (Heterotrophic plate count (HPC) up to 10 8 CFU/mL - Chapter 3) are expected to thrive on the DLS runoff particles as both the microorganisms and the particles originated from logs stored on site and are expected to be wood-degrading, potentially resin acid degrading bacteria. For example, two separate studies isolated five and eleven bacteria species respectively from paper mill effluents that were capable of growth on dehydroabietic acid (DHA), a common resin acid found in wood waste effluents (Mohn, 1995 and Zhang etal., 1997). Accordingly, several studies have used pulp mill sludge as an effective seed for biological contactors in the treatment of BOD 5, COD, turbidity and resin acid degradation in forestry effluents (Woodhouse, 2003, Zenaitis etal., 2002, Werkerand Hall, 1999, Zhang etal., 1997). Particle removal mechanisms within the sand column may be enhanced by the application of a closed loop configuration at low infiltration rates in which the sand column and liquid phase may reach a pseudo-steady state (Eienmann etal., 2001). Likewise, degradation and flocculation processes may act to prevent column saturation and consequent particle breakthrough or headloss. The following experiments were 59 conducted to test the efficacy of a sand filtration process as well as to determine the important treatment mechanisms. First, the physical and chemical properties of the sand were investigated. Secondly, a standard set of process conditions, such as flow-rate, batch to column volume, and sand preparation were devised. Time-course experiments were performed to monitor treatment performance to determine treatment run length for typical DLS run-off samples. The final set of experiments functioned to identify and discern the key mechanisms, such as filtration, adsorption, flocculation and biological degradation, in the treatment process. 4.2 Materials and Methods 4.2.1 Sand Preparation and Characterization The following section describes the physiochemical properties of the three different types of sand media used in the reactor treatment experiments: oxide-coated, oxide-stripped, and conditioned sand. The oxide-coated sand was commercially available Target Play S a n d ™ and was the sand that Microbial Technologies Inc. found to show enhanced efficacy when compared to common quartz filter sand for treatment of DLS runoff. The analysis given by Target Inc. and shown in Table 4.2.1 reports a 5% iron oxide coating. When compared with the scientific literature, a 5% by weight, iron oxide coated sand was up to twice as high as that reported for other iron oxide coated sands used in sand filtration experiments (Benjamin etal., 1996, Lo. etal., 1997). The oxide-stripped sand was prepared by acid digestion. The conditioned oxide-coated sand was removed from the top 40 cm of Microbial Technology's pilot scale treatment reactor that had been effectively treating successive batches of DLS runoff for approximately 1 year. 60 Table 4.1 Chemical analysis provided by Target Inc. for Play S a n d ™ (coated sand) Mineral % wt. Chemical Oxide % wt. Quartz 60 Silica 68 Feldspar 20 Calcium 4 Chlorite 10 Alumina 14 Iron Oxide 5 Iron (Fe 20 3) 5 Horneblende 5 Magnesium (MgO) 2 Determination of oxide coating Although the commercial supplier of the sand provided the above mineralogical analysis, its accuracy could not be assumed for the purpose of a scientific thesis. The true iron (Fe) and aluminum (Al) oxide coating was determined after measuring the Fe and Al content of aqua regia and citrate-bicarbonate-dithionite (CBD) extractions by ICP-AES . Aqua regia dissolves the oxide coating and the free-oxides from sand surfaces, whereas citrate-bicarbonate-dithionite dissolves the free-oxides only. Previous work has reported that this analysis correlates well with X-ray diffraction (XRD) analysis (Klute, 1986). For the aqua regia digest, 1g of sand was refluxed at 85°C in 4 mL 50% H N 0 3 and 10 mL 20% HCI for 1 hour. For the CBD extraction, 1 g of sand was placed in a dialysis membrane and dialyzed against 0.5 M sodium acetate buffer for 16 hours to remove soluble salts and carbonates. Following this, any bound organic materials were removed from the sand by washing with 30% H2O2 at 70°C. The CBD extraction functions to complex free Fe-oxides, buffer the reaction (sodium bicarbonate) and reduce the extract (sodium dithionite) for detection of ferrous Fe. One gram of pretreated sand was heated at 80°C while stirring for 10 minutes in 40 mL 0.4M sodium citrate and 5 mL 0.5M sodium bicarbonate. One gram of sodium dithionite was added in 5 minute increments for 15 minutes while stirring. After 15 minutes, 10mL NaCI was added to promote flocculation. Samples were centrifuged at a relative centrifugal force of 290 g (1,500 rpm at 25 °C) for 10 minutes and the supernatant was collected for detection of Fe and Al by ICP-AES. 61 Oxide stripped sand was prepared for column experiments. For this reaction, the coated sand was stripped of its oxide coating using a simplified, scaled-up version to the aqua regia digest described above. The aqua regia digest was conducted instead upon 500 mL (795 g) of sand in a 25 L Nalgene bottle fitted with a heating element and an Arrow 850 multiple speed biotech stirrer. To this, 1.59 L H N 0 3 and 1.99 L HCI was added to 795 g of sand and 7.55 L dH 2 0. The extraction solution was agitated rapidly for one hour at 85 °C. The extract was quantified for Fe and Al by ICP-AES. The resultant pH of the uncoated sand was adjusted to the pH of the coated sand using successive washes with 1 M NaOH. This sand preparation was used for the reactor experiments. Determination of minerological species of sand A n XRD-Rei tve ld analysis was conducted to give a complete mineralogical breakdown of the sand for the purpose of prediction and explanation of plausible surface chemistry reactions that the sand may host. The XRD-Rei tve ld analysis was performed by Dr. Matti Raudsepp (Department of Earth and Ocean Sciences). Oxide-stripped sand was used for this analysis as metal oxides interfere with X R D patterns. For this analysis, the particle size of the sample was reduced to the optimum grain-size range for X-ray analysis (<10 pm) by grinding under ethanol in a vibratory McCrone Micronising M i l l (McCrone Scientific Lts., London, U . K . ) for 7 minutes. Step-scan X-ray powder diffraction data were collected over a range 3-70° with C u K radiation on a standard Siemens (Bruker) D5000 Bragg-Brentano diffractometer . X-ray powder diffraction data were refined with Rietveld Topas 2.0 (Bruker A X S ) running on a Pentium III 10000 M h z personal computer. The X-ray diffractogram was analyzed using the International Centre for Diffraction Database PDF2 Data Sets, 1-49 plus 70-86, using Search-Match software by Siemens (Bruker). Particle size analysis, pH and cation exchange capacity Particle size distribution was performed on 100 g each of oxide-coated and oxide-stripped sand by sieve analysis. Sand grains were fractionated sequentially using sieve numbers 16, 30, 50, and 200 and thus separated into 2 -1 mm (very course 62 sand), 1 - 0.6 mm (course sand), 0.6-0.3 mm (medium sand) and 0.3 - 0.1 mm (fine sand) respectively based on standard methods of soil analysis (Sparks, 1996). The pH of the coated sand was measured with ananalytical pH probe after suspending and shaking the sand in distilled H 2 0 (1:5) for 30 min. The cation-exchange capacity (CEC) and percent base saturation were determined for the oxide-coated and oxide-stripped sand by the ammonium actetate method at pH 7.0 based on standard methods of soil analysis (Sparks, 1996). In order to determine the exchangeable cations, 10 g of sand was shaken with 40 mL 1M NH 4OAc for 15 minutes then washed/vacuum filtered through a 0.45 membrane with 4 x 30 mL washes. The extract was then analyzed for Na, Ca, Mg, and K using a Varian Spectra 220 FS atomic absorption spectrophotometer (AAS). For total CEC, the ammonium-saturated sand was washed with 3 x 40 mL portions of isopropanol (discarded), followed by 4 x 50 mL portions of 1M KCI that were collected for analysis of NH 4 +using a Lachat XYZ QuickchemAE auto-analyzer using method #10-107-06-2-A. The percent base saturation was calculated as the sum of all exchangeable cations per CEC. 4.2.2 DLS runoff Sample Preparations Table 4.2: Characteristics of the DLS Runoff Samples used in Sterilization, Batch Adsorption Experiments I and II and the Column Treatment Experiments A, B, and C Sample PH COD (+/-) (mg/L) Experiment(s) 3 4.6 1260 (37) Column A 7 5.6 2870(102) Sterilization, Adsorption I 9 6.1 1620 (80) Column B 11 6.1 1115 (47) Adsorption II, Column C 12 5.1 3690(113) Column C All samples were measured for pH, COD, turbidity, and heterotrophic plate counts (HPCs). Samples were collected, stored and tested according to methods outlined in Chapter 3. In addition Sample 3 was measured by TEM-EDX, TSS and for metals by ICP-AES. Samples 3 and 7 were measured for particle size distribution by Malvern Mastersizer analysis. 63 4.2.3 Sterilization of DLS runoff Sample 7 Mercuric Chloride (HgCI2) was tested as a broad spectrum antibiotic to achieve abiotic DLS runoff preparations as it has been previously used as a non-disruptive preservative for environmental colloidal systems (Chen and Buffle, 1996, Perret etal., 1993). Nevertheless, the effect of HgCI2 treatment on both sterilization and particle aggregation processes had to be determined for this unique system. A concentration range was tested based on the previous findings that 0.11 g/L HgCl2 sterilized river water systems with little effect on colloidal processes. Runoff Sample 7 was added to serial dilutions of HgCI2 stock solution resulting in 0.11, 0.055, 0.011, 0.0011 g/L HgCI2 solutions. The resultant solutions were agitated for 4 h, and then plated on standard heterotrophic media as described in Chapter 3. Sterilization was achieved in samples when 0 CFU/mL was obtained after triplicate plating. The lowest concentration of HgCI2 that was effective in sterilization yet showed no effects upon particle size distribution, pH and COD was chosen for sand column treatment experiments. One hundred and twenty-five mililitres of runoff samples used for treatment columns labeled 'abiotic' were agitated with HgCI2 at 100 rpm for 4 hours in 200 mL Erlenmeyer flasks covered with sterile cheesecloth. In parallel, all other samples were shaken for 4 hours in 200 mL Erlenmyer flasks covered with sterile cheesecloth. 4.2.4 Batch Adsorption Experiments The first series of treatment experiments functioned to test whether adsorption was a key mechanism in removal of DLS runoff particles. This was determined by measuring the COD removal capacity of the sand(s) as previously it was shown that COD concentration correlates with the suspended particles concentration. Batch Adsorption Experiment I For the first experiment, 10 g of oxide-coated and oxide-stripped sand were agitated at 100 rpm for 24 h in 100 mL of DLS runoff Sample 7 contained in 125 mL Erlenmeyer shake flasks covered with sterile cheesecloth. A shaken runoff control (containing no sand) was included in this analysis. The flasks were allowed to settle for an additional 24 h before analysis. All samples were prepared in triplicate and tested for COD removal and pH. Comparing the COD removal capacities for oxide-coated and oxide-stripped sand would illustrate the effect that the oxide coating had on adsorption. Batch Adsorption Experiment II For the second adsorption experiment 1, 5 and 10 g of oxide-coated sand were tested for adsorption of Sample 11 using the same experimental conditions as above. The following controls were included in the analysis: A) static runoff control, B) agitated runoff control, and C) distilled water control containing 10 g of the oxide-coated sand preparation. 4.2.5 Sand Column design Sand columns were constructed out of 500 mL graduated cylinders with a diameter of 4.5 cm and a cross sectional area of 15.9 cm 2 as shown in Figure 4.1 (in section 4.2.6). Fifty milliliters of course gravel (to a height of approximately 3.1 cm) was placed in the bottom of each reactor for the purpose of drainage and aeration enhancement. Polypropylene discs containing 2 mm diameter holes were wrapped in a chemically inert cloth with a mean pore diameter of 50 pm and used to separate the gravel and filter sand. Five hundred millilitres of filter sand was added atop of the filter cloth to a height of 31.5 cm. The bottoms of the cylinders were fitted with effluent drainage tubes. The sand was settled and compacted uniformly for all columns by sealing the effluent tube at the bottom of the reactor and then loading sterilized water until the reactor was saturated with water. The void volume of the reactors was then measured based on how much water was necessary to saturate the media. This volume was 125 mL for the oxide-coated and oxide-stripped sands, and 100 mL for the conditioned sand. For each column run, 125 mL of runoff was loaded from the influent reservoir (200 mL Erlenmeyer flasks covered in sterile cheesecloth) directly at a height of 5 cm above the sand onto the middle of the sand media using a Masterflex, variable-speed 65 (Cole-Parmer) pump and Masterflex PharMed tubing. The influent was added at the threshold rate in which headloss (puddling) occurred for the conditioned sand column. This was determined to be approximately 10 mL/min (1.5 m/h) for the fresh oxide-coated and oxide-stripped columns and 3.2 mL/min (0.6 m/h) for the conditioned sand column. The residence time for the fresh sand and the conditioned sand were 12.5 min and 31.25 min respectively. Each treatment batch was conducted at 20 °C for 24 h. 4.2.6 Column Treatment Experiments Column Experiment A: Treatment of sample 3 with conditioned and fresh oxide-coated sand columns For this experiment, DLS sample 3 was treated using two columns: conditioned oxide-coated and fresh oxide-coated and tested for COD, turbidity, and pH. The effluent was tested after every cycle for a total of 10 column cycles and again for the above at the end of 24 hours. To achieve this, 125 mL of sample was loaded from a 200 mL Erlenmeyer flask as described above and allowed to drain into a clean 200 mL Erlenmeyer flask. This effluent was then used as the influent for the next column cycle thus simulating a closed loop recirculating filter. The conditioned oxide-coated sand column effluent of sample 3 was also tested by TSS and Malvern PSA after the first cycle and TSS, Malvern PSA, and TEM-EDX after 24 hours. Column Experiment B: Treatment of Sample 9 in a) fresh oxide-coated, b) conditioned oxide-coated sand c) fresh oxide-stripped sand, abiotic runoff. For this analysis, Sample 9 was subjected to treatments a), b), and c) in duplicate columns (except c) and tested for COD removal after the first cycle using the simulated closed loop as described for Column Experiment A and again at 2 h, 5 h, 10 h, and 24 h when operated as a real recirculating filter. Heterotrophic plate counts (HPCs) were tested after samples were prepared, at the end of the first cycle and again at the end of 24 h of treatment. The runoff Sample 9 used for the oxide-coated, 'abiotic' run was prepared according to Section 4.2.3. One of the purposes of this experiment 66 was to test if treatment c) ('abiotic') would indeed remain sterilized during a treatment run. Column Experiment C: Successive batch treatment experiments The final series of sand column experiments tested the efficacy of variable sand preparations: Treatment 1 - oxide-coated sand, biotic runoff (2 column replicates) Treatment 2 - oxide-coated sand, abiotic runoff (2 column replicates) Treatment 3 - oxide-stripped sand, biotic runoff (2 column replicates) Treatment 4 - oxide-stripped sand, abiotic runoff (2 column replicates) Treatment 5 - conditioned oxide-coated sand, biotic runoff (1 column replicate) In total 8 successive batches of DLS runoff sample 11 (6 runs) and sample 12 (2 runs) were treated and tested for COD, turbidity, HPC and pH over 24 hours. Figure 4.1 shows the experimental setup. Figure 4.1: Sand columns used in successive batch treatment experiments: from left to right columns 1 and 2 (treatment 1), columns 3,4 (treatment 2), columns 5,6 (treatment 3), columns 7,8 (treatment^), column 9 (treatment 5) 67 The overall objective of these experiments was to determine the physiochemical attributes of the oxide-coated sand and sand/runoff interaction, which makes this specific sand filtration process a potential commercially viable on-site treatment method for D L S runoff. Within this design, there were 4 main mechanisms of sand column treatment that were investigated: filtration, adsorption, flocculation, and the biological activity. In order to quantify the percent of C O D removal that each mechanism contributed to the overall process, the following set of assertions were devised. 1) Filtration - C O D removed by treatment 4 (the oxide-stripped abiotic columns) will be represented as the filtration mechanism alone ( C O Deration). 2) Adsorption - The difference between the C O D removed by treatment 2 (oxide-coated abiotic) and treatment 4 (oxide-stripped abiotic) will be represented as the mechanism associated with the oxide coating or adsorption mechanism ( C O D adsorption). 3) Biological - The difference between the C O D removed by treatment 1 (oxide-coated biotic) and treatment 2 (oxide-coated abiotic) as well as treatment 3 (oxide-stripped biotic) and treatment 4 (oxide-stripped abiotic will be represented as the mechanism associated with the biological component ( C O D biological). 4) Flocculation - The C O D which settles (CODsettied) in the static control during the 24 hour treatment run will be compared with the C O D which settles and/or was caused to flocculate (CODsettied+fioccuiated) during the treatment runs. The following equations will be applied: Static control: i COD(w) - C O D ( S ) = CODsettied All Columns: C0D(W) - C O D ( S ) = CODsettied+fioccuiated CODSettled+flocculated " CODs ettled = CODfiocculation 68 C O D 0 = untreated Sample C O D ( w ) = C O D of whole effluent COD( S ) = C O D of supernatant effluent CODsettied = C O D settled in static control over 24 hours CODsettied+fioccuiated = C O D settled and/or flocculated during column treatment CODfioccuiation = C O D flocculated out of solution attributed to column treatment 4.3 Results and Discussion 4.3.1 Sand Characterization The physical and chemical properties of the oxide-coated and oxide-stripped sand are presented in Table 4.3. Acid digestion of the oxide-coated sand resulted in a color change from brown/grey to white/grey and a pH change from 6.2 (coated) to 3.5. Since Fe-oxides are brown in color, the change from brown to white corresponded well with the removal of iron oxides (Brady and Weil, 1996). The pH was adjusted to 6.2 after rinsing in several batches of NaOH. There was no difference in particle size analysis between the oxide-coated and resultant oxide-stripped sands, which meant that the oxide-coated sand required no further size fractionation to standardize with the oxide-coated sand for the treatment experiments. 69 Table 4.3: Properties of the Oxide-coated and Oxide-stripped Sands used in the Column Treatment Experiments oxide-coated sand (+/-) oxide-stripped sand (+/-) colour brown/grey white/grey PH 6.2 6.2 PSA (wt. %) (2-1 mm) 36.5 (2.2) 37.6(1.8) (1-0.6 mm) 32.5 (1.7) 30.1 (2.1) (0.6-0.3 mm) 30.3 (0.8) 31.1 (0.3-0.1 mm) 0.9 (0.1) 1.1 (0.2) total oxide (wt. %) 2.44 (0.03) 0 Fe - oxide 1.53 (0.5) 0 Al - oxide 0.91 (0.2) 0 C E C (meq/100 g) 0.62 1.01 % base saturation 204.6 349.5 The amount of oxide coating (2.4 wt %) was similar to a previous study that constructed Fe-oxide sands from quartz sand (Benjamin etal., 1996). The coated sand had a 2.4 to 3.2 wt.% Fe-oxide coating, which in turn dramatically increased the sand's surface area and p H p z c (pH p z c from 0.7 to 9.3) and subsequent ability to filter/adsorb particulates and both cationic and anionic metals (Benjamin etal., 1996). Oxide coated surfaces are extremely flexible ion exchange mediums as their charge is pH-variable and can contain both negatively and positively charged sites dependent on solution pH (Brady and Weil, 1996). Accordingly, a p H p z c of 9.3 for the Fe oxide-coated sand produced a net positive charge in the circum-neutral pH range and since most organic colloids and particulate are negatively charged in this pH range, the sand readily adsorbed these particles (Benjamin etal., 1996, Buffle etal., 1998). 70 Table 4.4: Cation Exchange Capacities (CEC) for Selected Soils and Minerals (adapted from Winegardner, 1996) Soil C E C (meq/100g) Agricultural Soils (Netherlands) 38.3 Agricultural Soils (California) 20.3 Mollisol (Russia) 56.1 Sodic Desert Soil (California) 18.9 Clay Subsoil (Alabama) 18 Montmorollionite (mineral) 92 Kaolinite (mineral) 3.3 Biotite 38.7 Weathered Biotite 1.9 Mica 25 Vermiculite (mineral) 125 The cation exchange capacity is an important measure, as given in Table 4.4 for several soils, for the classification of soils. It represents the sum of the total exchangeable cations (soil surface and interlayer) that a soil can adsorb (Sposito, 1989). The clay mineral and organic contingents of soils contain most of the C E C due to the high level of available negative charges spread over a large surface area. For example, biotite (a clay-mineral) has a C E C of 38.7 meq/100g, whereas weathered biotite (dissolved clay minerals) has a C E C of 1.9 meq/100g. The oxide-coated and oxide-stripped sands had relatively low CECs suggesting a low percentage of clay minerals. Interestingly, the oxide-stripped sand had a 39% greater C E C (1.01 meq/100 mL) then the oxide-coated sand (0.62 me q/100 mL). This is consistent with previous work that has shown that free oxides of Fe and Al bound to mineral surfaces inhibit the exchange of interlayer cations between the influent liquid phase and the soil matrix (Ghabru etal., 1989). As a result, the free oxides of Fe are Al (upon the oxide-coated sand) were expected to be a major contributor to the cation exchange process. 71 Table 4.5: Mineralogical Content and Characteristics of the Sand used in Column Treatment Experiments Mineral wt. % Characteristic Quartz 45.5 Si02 (silicate) Plagioclase 35.4 Ca, Na, Al silicate (feldspar) Chlorite 7.3 Fe, Al, Mg silicate hydroxide (clay mineral) Hornblende 3.5 Ca, Mg, Fe, Al silicate hydroxide Muscovite 3.2 K, Al silicate hydroxide fluoride Potassium Feldspar 2.5 K, Al silicate Titanite 0.9 Ca, Ti silicate Ankerit 0.7 carbonate The XRD-Reitveld analysis showed that the sand used in the treatment experiments was composed predominantly of quartz and feldspar, primary minerals. Primary minerals such as quartz are negatively charged in the circum-neutral pH range and weather slowly in water (Winegardner, 1996). For this reason, these minerals can be considered relatively inert to surface reactions with DLS runoff particles and will most likely function as filtration and support media in the sand column treatment experiments. In contrast, Chlorite is a source of exchangeable cations such as Na and Ca and may have a significant effect on surface chemistry (Winegardner, 1996). 4.3.2 Sterilization of DLS runoff Sample 7 Sample 7, which had a heterotrophic plate count of 3 x 10 6 CFU/mL, was used to determine if and at what concentration HgCI2 could be added to DLS runoff to achieve effective sterilization. Table 4.5 summarizes the results for three concentrations of HgCb tested. At HgCI2 concentrations of 0.055 g/L and 0.11 g/L, effective sterilization was achieved with no difference in COD, pH or particle size analysis. These results correlate well with previous experiments that tested HgCI2 as a broad-spectrum antibiotic for colloidal samples extracted from river water (Chen et al., 1994). A concentration of 0.11 g/L HgCI2 was chosen for sterilization of DLS runoff samples for the subsequent abiotic sand column runs. Table 4.6: Effect of HgCI2 on sterilization, particle size analysis and COD 72 Concentration HgCI2 Bacteria PSA (wt %) COD (g/y (CFU/mL) >1(pm) mg/L(+/-) 0 3.6 x 10 6 82.8 2870(37) 0.011 1 x10 1 69.1 2851(31) OJJ 0 71.3 2818(65) 4.3.3 Adsorption shake flask experiments Two important mechanisms of particle removal in sand filtration are adsorption and filtration. Both mechanisms are dependent upon the sands physical and chemical characteristics. For example, filtration will be dependent upon the sand grain size (in mm), channel-pore size and the linear liquid velocity (in m/h). A decrease in the minimum m/h required to cause head-loss indicates that the filtration capacity of the sand is decreasing due to decreasing pore-size. Total filtration saturation results in the development of headloss at which the sand filter is no longer effective (Logsdon, 1991). The adsorption mechanism is dependent more on the chemical surface groups of the sand grains-the free sorptive sites such as Al- and Fe-oxides. Adsorptive saturation results in particle breakthrough. Particle breakthrough can be measured for DLS runoff reactor treatment runs, by a decrease in the COD and Turbidity removal rate. In order to determine whether the oxide-coated and oxide-stripped sands had different adsorption capacity for DLS runoff, two separate shake flask adsorption experiments (Batch Adsorption I and Batch Adsorption II) were conducted. Batch Adsorption I As shown in Figure 4.2, the oxide-coated sand removed on average 17% of the DLS runoff Sample 7 COD. This adsorption capacity was statistically significant as determined by Excel ANOVA when compared with the oxide-stripped (p-value = 6.41 x10"5, F(28. 7) > Fcrit(4.5)) and shaken runoff control (p-value = 1.76 x10"6, F ( 9 7 .7) > Fcrit(5.o))-The oxide-stripped sand had no statistical difference from the shaken runoff (no sand) 73 control (p-value 0.65, F(0.2)<Fcrit(5.o))- As a result, the oxide-coated sand's enhanced adsorptive property may explain any increase in the treatment performance of the oxide-coated sand columns over the oxide-stripped sand columns. 3500.0 3000.0 runoff, no sand coated 1 coated 2 coated 3 stripped 1 stripped 2 stripped 3 Figure 4.2: Batch Adsorption Experiment I comparing the adsorption capacity, for DLS runoff Sample 7 COD, of the oxide-coated sand (coated replicates 1, 2, and 3), the oxide-stripped sand (stripped replicates 1,2, and 3), and the shake runoff control (runoff, no sand). Error bars represent the standard deviation of triplicate measurements. Batch Adsorption Experiment II For the second analysis, 1, 5 and 10 grams of oxide-coated sand were tested for absorption of Sample 11 COD as shown in Figure 4.3. Three runoff controls were tested: 1) static runoff, no sand 2) shake runoff, no sand, and 3) shake dH 20, sand. 74 1200.0 1000.0 staticrunoff.no shake runoff, no dH20, 10g sand skake runoff, 1 g shake runoff, 5 g shake runoff, 10 g sand sand sand sand sand Figure 4.3: Batch Adsorption Experiment II comparing the adsorption capacity for DLS runoff Sample 11 COD, of 1, 5, and 10 g of oxide-coated sand with that of the shake runoff control, no sand. Error bars represent the standard deviation of triplicate measurements. There was no statistical difference in the shake runoff control and the static runoff control (P-value = 0.07, F(5.9)<Fcrit(7.7)). This result implies that the removal of COD from DLS runoff will be due to collision with, and adsorption to, sand particles and not as a result of agitation-induced aggregation processes. Furthermore, it shows that DLS runoff particles are extremely stable in agitated solution and further suggests that a conventional primary agitation/settling scheme may not be effective, unless perhaps in combination with chemical flocculants. The resultant COD of the shaken dH 20, 10g sand control was 0 (the same as dH20) and illustrated that the sand did not contribute COD to the analysis. There was no statistical difference in the adsorption of DLS runoff COD for 1 g of oxide-coated when compared with the shake runoff control (P-value = 0.07, F(5.9)<Fcrit(7.7). In contrast, the 5 g treatment removed on average 26% of DLS runoff 75 COD which was statistically significant when compared with the shake runoff control (P-value = 0.016, F(15.6)>Fcrit(7.7). Likewise, the 10g treatment removed 23% of DLS runoff which was statistically significant when compared with the shake runoff control (P-value 0 0.011, F(19.7)>Fcrit(7.7). There was no difference between the 5 g treatment and the 10 g treatment (p-value = 0.71, F(0.16)<Fcrit(7.7). The value of the adsorption threshold was not investigated any further. 4.3.4 Column Treatment Experiments Column Experiment A: Treatment of sample 3 with conditioned oxide-coated sand and fresh oxide-coated sand columns As shown in Figures 4.4 and 4.5, sand column treatment of sample 3 using both the conditioned oxide-coated and oxide-coated fresh sands had decreased the COD and turbidity significantly after one column cycle. Accordingly, the TSS was reduced from 330 mg/L to 0 mg/L (+/-5 mg/L) and the color changed to a pale, translucent yellow after one column cycle for both columns. The pH also increased from 4.5 to approximately 6.5 (+/-0.2) for both columns. After the second cycle, the COD and turbidity had increased sharply. This was followed by a gradual decrease in the COD, turbidity and color for the remaining 22 hours and 40 minutes of treatment for both columns. In total, the conditioned oxide-coated sand column removed approximately 90% COD and 92% turbidity and the fresh oxide-coated sand column removed approximately 80% COD and 92% turbidity. 76 Figure 4.4: Comparison of COD removal efficacy of oxide-coated conditioned sand and oxide-coated fresh sand during 24 treatment runs Figure 4.5: Comparison of turbidity removal efficacy of conditioned oxide-coated sand and fresh oxide-coated sand during 24 treatment runs 7 7 The slightly enhanced treatment efficacy of the conditioned sand column is not surprising as the buildup of organic materials and biological flora is expected to both decrease the pore size of the filter media and enhance biological COD removal rates (Eisenmann etal., 2001). Accordingly, the establishment of biofilms for sorption, filtration, and degradation is standard pretreatment for most biological contactors and has been exploited previously for treatment of woodwaste effluents (Mathys etal., 1997, Woodhouse, 2003). Likewise, the conditioning effect attributed to the buildup of filtered organic particles, not necessarily biological, has been also shown to enhance the adsorptive properties of filter medias such as sand and soil (Lu and Pignatello, 2002). As a second component to this study, the conditioned oxide-coated sand treated effluent for Sample 3 was characterized for the same physiochemical characteristics of untreated Sample 3 as was investigated in Chapter 3: appearance, particle size, morphology, and metals. Figure 4.6 shows the visual difference and Figure 4.7 shows the particle size difference between untreated Sample 3 and conditioned oxide-coated 24 hour effluent. Figure 4.6: Untreated fresh DLS runoff Sample 3 (left) and conditioned, oxide-coated treated 24 hour effluent (right) using sand filtration process. 78 Figure 4.7: Malvern Mastersizer pictograms of untreated fresh Sample 3 (left) and conditioned oxide-coated sand 24 hour effluent (right) The PSA analysis showed that the total weight percent of particles above 1.2 pm decreased from 79.8 wt.% to 11.2 wt.% after 24 hours of treatment. This analysis showed a similar trend to the TSS analysis (removal of 330 mg/L TSS -see page Chapter 4 text page 19), however complete removal of particles above 1.2 pm was not shown by the Malvern Mastersizer histogram. One explanation for this difference could be due to artifact introduction during the particle size analysis. The physiochemical properties of fresh DLS runoff Sample 3, such as particle morphology and chemical constituency, were investigated in Chapter 3. The un-fractionated sample was determined to be composed of predominantly large (5-10 pm) hetero-aggregates of inorganic and organic colloids with a total COD of 1260 mg/L. Accordingly, this sand column treatment method was chosen to target the suspended particles of DLS runoff. The particles remaining in the 24 hour Sample 3 effluent are shown in Figure 4.8. Table 4.7 summarizes the EDX spectra for the TEM-EDX analysis. 79 1 pm Figure 4.8: TEM image (10,000 x) of Sample 3 effluent from the conditioned oxide-coated sand column Table 4.7: TEM-EDX analysis of Sample 3 effluent from the conditioned oxide-coated sand column element (<0.1 pm) treated effluent (wt%) (0.1-0.5 pm) (1 pm rods) Carbon 68.7 10.2 29.3 Oxygen 29.7 51.6 53.2 Aluminum 0.0 8.1 1.0 Calcium 1.6 0.7 0.8 Iron 0.0 0.3 0.0 Magnesium 0.0 1.2 1.2 Silicon 0.0 17.3 4.8 The conditioned sand column treated Sample 3 effluent contained 3 main particle-types: organic spheroids less than 0.1 pm, inorganic spheres from 0.1-0.5 pm, and rod-like particles with a length of 1 pm. The smallest particles (< 0.1 pm) were almost entirely organic, similar to particles seen in the colloidal fractions of the untreated samples, and were most likely humic substances. These particles were the most 80 abundant in the column effluent, and most likely comprised the majority of the COD. The 0.1-0.5 pm spheres were mostly O, Si, and Al. These particles may have been the inorganic contingent of the hetero-aggregate colloids (as seen in the untreated sample 3 - Chapter 3, Figure 3.5), however stripped of their organic constituents during sand column treatment. Alternatively, these particles may have been stripped from the column sand as free Al-oxides and/or alumino-silicates suggesting that some of the sand surfaces are readily leachable. The effect of continuous sand leaching on sustained treatment efficacy is unknown. Table 4.8: Comparison of the COD, pH and metals analysis of untreated DLS runoff Sample 3 and conditioned oxide-coated sand column treated Sample 3 untreated treated removal (mg/L) (mg/L) (%) COD 1260 146 88 PH 4.5 6.5 increase 2 units Al 28.7 0.1 100 Ca 97.1 246 increase 153 Cu 0.1 0 100 Fe 22.9 0.1 100 Mn 209 174 17 Mg 18.5 1.6 91 Na 1780 2150 increase 21 Si 20.6 13.3 35 Zn 1.2 0 100 Coupling chemical analysis with the TEM-EDX analysis, verified that conditioned oxide-coated sand column treatment removed the entire particulate phase and most of the colloidal particles. As shown in Chapter 3, Sample 3's COD fractionation profile was as follows: 655 mg/L particulate (> 1.2 pm), 491 mg/L colloidal (1.2 pm>colloids >0.02 pm) and 114 mg/L dissolved (>0.02 pm). Accordingly, conditioned oxide-coated sand column treatment removed 88% (1114 mg/L) of the COD. Treatment also removed close to 100% of the Al and Fe from solution, which was 85% and 75%, respectively, bound to the particle phases. Furthermore, treatment also removed 100% of the Cu and Zn, which had both previously, exceeded the approved BC water quality guidelines. As 81 determined in Chapter 3, approximately 30% and 15% of the Cu and Zn respectively were colloidal (0% particulate). This result was somewhat surprising as treatment also removed dissolved Cu and Zn, however consistent with other studies which have used oxide coated sands for the treatment of dissolved and colloidal metals (Benjamin etal., 1996, Lo etal., 1997, Khaodhiar etal., 2000). Therefore, the oxide-coated sand is a potential treatment method for both particles and metals. Another interesting outcome of the treatment was the pH increase effect from 4.5-6.5. This may be explained by two different mechanisms. First, by acid-base equilibria, in which the exposed hydroxyl groups on the silica or oxide surfaces of the sand may have accepted protons from the DLS run-off solution, thus effectively removing acidity from solution. Secondly, this phenomenon may be explained by cation-exhange reactions and the consequent release of the base cations (Na+, Ca 2 +) into the runoff solution. The Chlorite mineral detected by XRD analysis at a total of 7.1 wt. % may have been the source of the Na and Ca. The pH increase effect may be the key mechanism that explains the adsorption, aggregation and desorption of particles upon the sand surfaces. Figure 4.9 shows the first 10 cycles for the conditioned, oxide-coated treatment of Sample 3. 82 1400 column cycle Figure 4.9: Change in pH, COD and Turbidity during time-course, conditioned oxide-coated sand column treatment of Sample 3 After the first column pass, the pH of effluent had increased from 4.5 to 6.5 and the COD of the effluent had decreased from 1260 to 360 mg/L. After the second column pass, with the influent pH of 6.5, COD was added back into column effluent. One hypothesis to explain this increase in COD is that these negatively charged particles (hetero-aggregates representing most of the COD) react within the sand matrix in three successive steps: (1) Adsorption on the first column pass to the positively charged oxide surfaces of the sand renders these particles as substrate for subsequent reactions with the re-circulated run-off. Re-circulating run-off solution passed through the sand increases collision efficiency of particles. (2) Protonation of the exposed carboxylic and phenolic residues of the runoff particles removes the exposed negative surface charge, thus allowing more hetero-aggregates to bind through hydrophobic interactions. 83 {3) Desorption occurs after the first cycle as the pH increases from 4.5 to 6.5 and the positively charged oxide surfaces become less positive (pK 8-10 for Al and Fe-oxides), and consequently release some of the absorbed particles (representing an increase in the COD and turbidity). Column Experiment B: Treatment of Sample 9 by a) fresh oxide-coated sand , b) conditioned oxide-coated sand c) fresh oxide-coated, abiotic runoff The COD treatment profiles for Sample 9 were similar to the results of Column Experiment A (Sample 3) for all columns in that the COD decreased on the first cycle increased on the second cycle, and then decreased gradually over the remaining treatment run. The treatments removed on average: 86% COD for treatment a), 86% COD for treatment b) and 88% COD for treatment c) as shown in Figures 4.10, 4.11, and 4.12 respectively. There was no statistical difference in the column effluent COD for the 3 treatments (P-value = 0.1, F(5.8) < Fcrit (19). 1800 200 - • 0 -I 1 1 1 1 0 5 10 15 20 25 time (hrs) Figure 4.10: 24 hour treatment of Sample 9 COD using treatment a) with two identical oxide-coated sand columns in parallel (replicate 1 and 2) 84 Figure 4.11: 24 hour treatment of Sample 9 COD using treatment b) with two identical oxide-coated, abiotic sand columns in parallel (replicate 1 and 2) Figure 4.12: 24 hour treatment of Sample 9 COD using treatment c) conditioned, oxide-coated sand column (no column replicates) 85 Sample 9 preparations used for the oxide-coated and conditioned oxide-coated sand had a heterotrophic plate count (HPC) of 5 x 106 CFU/mL. The Sample 9 preparation for the oxide-coated, abiotic sand column was sterilized with 0.11 g/L HgCl2 and thus had an HPC of 0 CFU/mL. After one column cycle, the oxide-coated sand column effluents and the conditioned oxide-coated sand column contained 2 x 105 CFU/mL (average of 2 replicates) and 1 x 104(no replicates) respectively, whereas the oxide-coated, abiotic (sterilized) effluent had a CFU/mL of 0 and effectively remained sterilized. The decrease in CFU/mL in sand column effluents of both columns (relative to untreated sample) suggests that a population of heterotrophic bacteria were absorbed to the sand matrix. At the end of the 24 hour treatment runs, the biomass in the effluent was assumed to have reached a quasi-steady state with the bacteria sorbed to the column matrix. The oxide-coated sand and conditioned sand column effluent contained 2 x 105 and 10 x 105 respectively, whereas the oxide coated, abiotic sand column effluent remained sterilized. The implications of the results of this experiment were two-fold. First, it was assumed that using the HgCl2 sterilization method for abiotic treatment runs was effective. Secondly, since there was no difference between the treatment performance of the oxide-coated sand column runs and the oxide-coated abiotic sand column runs, it was assumed that the concentration of HgCI2 used had no residual effects on treatment performance. Column Experiment C: Successive batch treatment experiments of Samples 11 and 12. Figure 4.13 shows the treated effluent of treatments 1, 3, and the static control. The oxide-coated and conditioned oxide-coated sand columns (Treatments 1, 2 and 5) produced translucent effluent, whereas the oxide-stripped sand columns' effluents (treatments 3 and 4) remained dark brown and turbid at the end of the 24-hour treatment run. In fact, there here was no discernable visual difference between the static control and the oxide-stripped sand treated effluent. Accordingly, the oxide-coated columns performed better in both COD and turbidity removal for the first treatment run as shown in Figures 4.13 and 4.14. Appendix A.3 through A. 10 contains the comprehensive experimental results. Figure 4.13: Treated DLS runoff sample 11 from right to left respectively: treatment 1 (oxide-coated, biotic), treatment 2 (oxide-stripped, abiotic) and the static control untreated static control oxide-coated oxide-coated oxide-abiotic stripped oxide- conditioned stripped abiotic Figure 4.14: COD after the first 24 hour batch treatment run for Sample 11 for treatments: 1 - oxide-coated, biotic, 2 - oxide-coated abiotic, 3 - oxide-stripped, biotic, 4 - oxide-stripped abiotic and 5 - conditioned oxide-coated. Error bars represent the 87 standard deviations of triplicate measurements of COD. Treatments 1-4 are the combined results for duplicate columns. 1200 1000 800 «? I— LL ^ 600 •a 400 200 untreated static control oxide-coated oxide-coated oxide-abiotic stripped oxide-stripped abiotic conditioned Figure 4.15: Turbidity after first 24 hour batch treatment run for Sample 11 for treatments 1-5. Error bars represent the standard deviations of triplicate measurements of turbidity. Treatments 1 -4 are the combined results for duplicate columns. The oxide-coated sand columns (treatments 1 and 2) treated DLS runoff significantly better, on average 86% COD and 90% turbidity removal, when compared with the oxide-stripped sand columns (treatments 3 and 4), average 52% COD and 60% turbidity removal (P-value = 1 x10"8, F (288)>Fcrit (5.0)). The conditioned oxide-coated sand column (treatment 5) showed no statistical treatment difference when compared to the fresh oxide-coated sand column (P-value = 0.3, F (1.38)<Fcrrt(5.1). The static control showed a 17% COD and 13% turbidity decrease over 24 hours of storage, possibly due to settling and biodegradation. 88 Untreated Sample 11 had a heterotrophic plate count of 6 x 105 CFU/mL. The abiotic column effluents (treatments 2 and 4) contained 0 CFU/mL, thus maintaining sterilized column runs. Nevertheless, there was no statistical difference between the treatments 1 and 2 (biotic vs. abiotic) for the first batch run (P-value = 0.82, F (0.05)<FCrit (5.0)). Likewise, there was no statistical difference treatments 3 and 4 (biotic vs. abiotic for the first batch column run (P-value = 0.5, F(0.6)<Fcrit (5.0)). As a result, the biological component did not appear to have an effect on treatment performance for the first column run. The treatment performance of the 5 sand column treatments over 8 successive batch treatment runs is summarized in Figure 4.16 for COD removal and Figure 4.17 for turbidity removal. Appendix A, Tables A.2 through A.9 contains the full experimental data. Q o o c o o TJ OJ 100 90 80 70 60 50 40 -I 30 -•— oxide coated oxide-stripped -•— conditioned -k—oxide-coated abiotic • — - - oxide-stripped abiotic 0 run Figure 4.16: Average percent COD reductions for the 5 different sand column treatments: oxide-coated, oxide-coated/abiotic; oxide-stripped; and oxide-stripped/abiotic duplicates and oxide-coated/conditioned. Averages and standard deviations are based on two column replicates for each treatment) and 3 replicate COD measurements (except conditioned treatment - one column, 3 replicate COD measurements). 89 100 90 -I 80 70 c g t> 60 T 3 ^ 50 40 30 • oxide-coated abiotic -conditioned - oxide-stripped - oxide-coated oxide-stripped abiotic 4 5 run Figure 4.17: Average percent turbidity reductions for the 5 different sand column treatments: oxide-coated, oxide-coated/abiotic; oxide-stripped; and oxide-stripped/abiotic duplicates and oxide-coated/conditioned. Averages and standard deviations are based on two column replicates for each treatment) and 3 replicate turbidity measurements (except conditioned treatment - one column, 3 replicate turbidity measurements). The treatment performance of the conditioned oxide-coated, the fresh oxide-coated and the abiotic oxide-coated sand (oxide-coated sands) appeared to decrease in gradually over successive treatment experiments, especially on the seventh and eighth successive treatment runs which had extremely high COD (> 3000 mg/L) as shown in Figure 4.16 and 4.17. The overall decrease in treatment performance (approximately 10% for COD removal and 5% for turbidity removal) on the seventh and eighth treatment runs may be explained by particle breakthrough caused by the high COD loading of Sample 12 (3690 mg/L), which was a large increase from Sample 11 (1115 mg/L). This result may suggest that the proposed sand column treatment method may have some limitations with regards to wastewater strength, specifically COD loading 90 rates. A primary clarifier/equalization tank or a polishing stage may be required for full-scale treatment. In contrast to the oxide-coated sand columns, the oxide-stripped sand columns' treatment performance appeared to increase (approximate increase of 15% C O D and turbidity removal) over successive treatment runs. This result may be explained by the conditioning effect wherein the buildup of organic materials enhances filtration. Another interesting result, was that the oxide-coated abiotic column preparations did not remain abiotic throughout the analysis as the HPC was determined to be 5 x 101 CFU/mL and 1 x 102 CFU/mL on the seventh and eight runs respectively. The source or cause of contamination is unclear. The results for the analysis of sand column treatment mechanisms are summarized below for the set of assertions described in Section 4.2.6 (p. 12): 1) Filtration - COD removed by the oxide-stripped abiotic columns was represented as the filtration mechanism alone (CODfntratkm)- The CODfatration was an average of 56.3% (+/- 5.6%) for the eight successive column runs. On the eighth and final run, the CODfiitration increased to 62.5%. 2) Adsorption - The difference between the COD removed by the oxide-coated abiotic and oxide-stripped abiotic was represented as the mechanism associated with the oxide coating or adsorption mechanism (COD adsorption). COD adsorption was an average of 21.5% (+/- 6.0%) for the eight successive columns. 3) Biological - There was no CODbi0iogicai apparent for the first 8 batch treatment runs as there was no statistical difference in the treatment efficacy of the for the oxide-coated vs. oxide-coated abiotic columns (P-value = 0.2, F(2.0)< Fcrit(4.6)) or the oxide-stripped vs. oxide-stripped abiotic columns (P-value = 0.2, F(2.0)<Fcrit(4.6)). This suggests that the biological component may not be of importance in treatment performance of the columns. Furthermore, no headloss was observed for the abiotic columns showing that the biological component may not be the factor, as previously hypothesized, which prevents the sand columns from saturating during treatment and forming considerable 91 headloss. Alternatively, perhaps the eight successive column runs did not provide enough time for the biological component to establish itself as a key mechanism and thus have an effect on treatment performance. If this is the case and the assumption is made that the conditioned sand it an established biologically-active system and the fresh oxide-coated sand was not, then by comparing the treatment efficacy of both, the effect of the biological component upon sand column treatment can be elucidated. Consequently, there was no average difference between the fresh oxide-coated sand and the conditioned oxide-coated sand (P-value = 0.2, F(1.8) < Fcrit(4.6)). 4) Flocculation - CODsettied was determined to be170 mg/L for the static control for the first column run. In contrast CODSettied+fioccuiated was determined to be on average 0 mg/L for all treatments, including the conditioned oxide-coated treatment. Therefore, CODfioccuiation did not appear to be a mechanism of COD removal from D L S runoff solution for the first column run. And since, the conditioned oxide-coated treatment did not show any CODfiO C C Ui ation it was assumed that column conditioning of the remaining treatments would not cause CODfioccuiation to occur. Consequently, CODfioccuiation was not calculated for the remaining treatment runs. 4.4 Conclusions Chapter 4 functioned to characterize the physiochemical properties of the filter sand (Target Play Sand™) used in treatment experiments, to assess the performance of the proposed sand column treatment process at a lab-scale and to determine the primary mechanisms which contribute to the overall treatment process. The rational for this research was: a) to recommend the vital features of the sand column treatment process necessary for commercialization of this technology and b) to advance the overall understanding of mechanisms in sand filtration processes. 4.4.1 Sand Characterization and Sand Column Performance It was hypothesized very early in the design of this thesis that some property(s) specific to the Target Play Sand™ itself, were vital to the performance of the chosen 92 sand column treatment process. For this reason, the sand was characterized by pH, CEC, grain size, mineralogy and its surface properties, specifically oxide-coating. Consequently, the 2.4% oxide-coating was determined to be the distinguishing factor for successful sand column treatment of DLS runoff. The oxide-coated sand absorbed 17% and 26% of the COD (from 2870 mg/L and 1115 mg/L COD respectively) over two separate shake flask experiments (Batch Adsorption I and II), whereas the oxide-stripped sand showed no adsorption capacity (Batch Adsorption I). Furthermore, the oxide-coated sand removed up to 86% of the COD and 92% of the turbidity over three series of treatment experiments (Column Experiments A, B and C), whereas the oxide-stripped sand removed only 52% of the COD and 60% of the turbidity (Column Experiment C). The discrepancy in treatment performance between the oxide-coated and the oxide-stipped sands was most obvious by the visual appearance: the oxide-coated sand effluents were clear and the oxide-stripped effluents were dark brown to black. 4.4.2 Determining Key Sand Column Treatment Mechanisms In Chapter 3, it was determined that the main constituents of DLS runoff were organic/inorganic (hetero-aggregated) colloidal particles, which most likely bear a negative-charge due to the high weight percentage of organic molecules (Buffle et al., 1998). In addition to particles, individual microbes were observed by TEM and enumerated by HPC at a very high concentration (up to 108 CFU/mL), similar to the mixed liquor of activated sludge (Cheremisinoff, 2002). Accordingly, the aging study conducted in Chapter 3 showed that the COD of DLS runoff was readily biodegradable. These two properties of the runoff suggested that surface chemistry reactions, especially adsorption with the oxide-coating of the sand surfaces, flocculation processes and biological reactions would be of importance to the sand column treatment process. The role of the oxide-coating was quantitatively valued at an average of 21.5% COD removal for the oxide-coated sands over 8 successive treatment runs. More qualitatively, the oxide-coating also may have caused the pH increase effect as discussed in the proposed Adsorption, Protonation, and Desorption hypothesis as outlined. There was no flocculation of COD measured suggesting that organic 93 materials are trapped within the sand column and/or biologically degraded over time. Similarly, the biological component of DLS runoff did not appear to have an effect on treatment performance. Unfortunately, the importance of the biological role to the DLS runoff sand column treatment process may not have been realized in the 8 successive batch treatment experiments as there was no increase in the COD removal and/or alleviation of the sand columns from saturation and headloss. The biological component may very well fulfill these tasks; however it is most likely that 8 successive treatment experiments were not enough to saturate the abiotic sand columns. Albeit potentially challenging to maintain this experimental design by keeping abiotic treatment runs, an improvement to this study will be an increase in successive batch runs. 94 References Journal articles, conference proceedings and books Aim, R.B, S. Vigenswaran, H. Prasanthi, and V. Jegatheesan, 1997. Influence of particle size distribution in granular bed filtration and dynamic microfiltration. Water Science and Technology, 36: 207-215. American Public Health Association (APHA) Greenberg, A.E., L.S. Clesceri and A.D. Eaton, 1992. in Standard Methods for the Examination of Water and Wastewater 1&h edition, American Water Works Association. Water Environment Federation: Washington, D.C. Bailey, S.E., T.J. Olin, R. M. Bricka, and D.D. Adrian, 1999. A Review of potentially low-cost sorbents for heavy metals, Water Research, 33:2469-2479. Boiler, M., D. Kobler, and G. Koch, 1997. Particle separation, solids budgets and headloss development in different biofilters. Water Science and Technology, 36: 239-247. Benjamin, M., R.S. Sletten, R.P. Bailey, and T. Bennette, 1996. Sorption and filtration of metals using iron-oxide-coated sand. Water Research. 30: 2609-2620. Buffle, J., K.J. Wilkenson, S. Stoll, M. Filella, and J. Zhang, 1998. A generalized description of aquatic colloidal interactions: the three-colloidal component approach. Environmental Science and Technobgy, 32: 2887-2899. Butterfield, P.W., A. K. Camper, J. A. Biederman and A. M. Bargmeyer, 2002. Minimizing biofilm in the presence of iron oxides and humic substances, Water Research. 36:3898-3910. Chen, Y. and J. Buffle, 1996. Physiochemical and Microbial Preservation of colloid characteristics of natural water samples 1: Experimental Conditions, Water Research. 30:2178-2184. Cheremisinoff, P., 2002. In Handbook of Water and Waste-water Treatment Technologies, Butteworth and Heinemann: Boston. Eisenmann, H., I. Letsiou, A. Feuchtinger, W. Beisker, E. Mannweiler, P. Hutzler, and P. Arnz, 2001. Interception of small particles by flocculent structures, sessile ciliates, and the basic layer of a wastewater biofilm. Applied and Environmental Microbiology, 67:4286-4292. Ghabru, S.K., A.R. Mermut, and R. J. St. Arnaud, 1989. Layer-charge and cation-exchange characteristics of vermiculite (weathered biotite) isolated from a gray luvisol in NorthEastern Saskatchewan, Clays and Clay Minerals, 37:164-172. 95 Graham, N. and R. Collins, 1996. In Advances in Slow Sand and Alternative Biological Filtration. Wiley and Sons, New York. Haarhoff et al., 1991. In Advances in Slow Sand and Alternative Biological Filtration, Wiley and Sons, New York. Harris, D.C., 1995. In Quantitative Chemical Analysis. Fourth Edition, W.H. Freeeman and Company: New York. Hoel and Aarsand, 1995. Accute toxicity of colloidal and dissolved material in TMP effluents. Nordic Pulp and Paper Research Journal, 2:98-109. Khaodhiar, S., M.F. Azizian, K.O., and P.O. Nelson, 2000. Copper, chromium, and arsenic adsorption and equilibrium modeling in an iron-oxide-coated sand, background electrolyte system. Water, Air, and Soil Pollution, 119:105-120. Kim, S. and M.Y. Corapcioglu, 1997. The role of biofilm growth in bacteria-facilitated contaminant transport in porous media. Transport in Porous Media, 26:161 -181, 1997. Lenhart, J. J. and J. E. Saiers, 2002. Transport of silica colloids through unsaturated porous media: Experimental results and model comparisons, Environmental Science and Technology, 36:769-777. Lo, S.L., H.T. Jemg, and C H . Lai, 1997. Characteristics and adsorption properties of iron-coated sand. Water Science and Technology, 35:63-70. Logsdon, G.S., 1991. In Slow Sand Filtration. American Society of Civil Engineers: New York. Lu, Y. and J.J Pignatello, 2002. Demonstrating the conditioning effect in soil organic matter in support of a pore deformation mechanism for sorption hysteresis. Environmental Science and Technology, 36:4553-4561. Mohn, W.W., 1995. Bacteria obtained from a sequencing batch reactor that are capable of growth on dehydroabietic acid, Applied and Environmental Microbiology, 61: 2145-2150. Orban, J. L, R. A. Kozak, R. C. Sidle, and S. J. B. Duff, 2002. Assessment of relative environmental risk from logyard run-off in British Columbia. The Forestry Chronicle, 78:145-151. Oste, LA., E.J.M. Temminghoff and W.H. Van REisdijk, 2002. Solid-solution partitioning of organic matter in soils as influenced by an increase in pH or Ca concentration, Environmental Science and Technology, 36:208-214. Paulsen, J. E., E. Oppen, and R. Bakke,1997. Biofilm morphology in porous media, a study with microscopic and image techniques, Water Science and Technology, 36:1-9. 96 Ryan, J.N., M. Elimelech, A.A. Ard, R.W. Harvey, and P.R. Johnson, 1999. Bacteriophage PRD1 and silica colloid transport and recovery in an iron oxide-coated sand aquifer. Environmental Science and Technology, 33:63-73. Schulthess, C. P. and C. P. Huang, 1991. Humic and fulvic acid adsorption by silicon and aluminum oxide surfaces on clay minerals, Soil Science Society of America Journal, 55:34-42. Spath, R., H. C. Flemming, and S. Wuertz, 1998. Sorption Properties of Biofilms, Water Science and Technology, 37:207-210. Tchonanoglous, G., F.L. Burton, and H.D. Stensel, 2003. In Wastewater engineering: treatment, disposal, and reuse, Fourth Edition, McGraw and Hill Co: New York. Wan, J. and T.K. Tokunaga, 1997. Film straining in unsaturated porous media: conceptual model and experimental testing. Environmental Science and Technology, 31: 2413-2420. Wen, L, E. P. Fest, J. Fillus, E. J. Temminghoff, and W.H. van Riemsdijk, 2002. Transport of humic and fulvic acis in relation to metal mobility in a copper-contaminated acid sandy soil, Environmental Science and Technology, 36:1699-1704. Werker, A. G. and E. R. Hall, 1999. Limitations for biological removal of resin acids from pulp mill effluent. Water Science and Technology, 40:281-288. Winegardner, D.L., 1996. In An Introduction to Soils for Environmental Professionals. Lewis Publishers: New York. Woodhouse, C. A., 2003 Attached growth biological treatment of stormwater run-off from log yards, M.A.Sc. thesis, The University of British Columbia, Vancouver, B.C. Zenaitis, M., H. Sandhu, and S. J. B. Duff, 2002. Combined biological and ozone treatmen of log yard run-off. Water Research, 36: 2053-2061. Zhang, Y., P.A. Bicho, C. Breuil, J. N. Saddler and S. N. Liss, 1997. Resin acid degradation by bacterial strains grown on CTMP effluent. Water Science and Technology, 35:33-39. Government and industry reports B.C. Approved Water Quality Guidelines, 1998. A compendium of working water quality guidelines in British Columbia: 1998 Edition, http://wlapwww.gov.bc.ca/wat/wq/BCguidelines/approved.html B.C. Waste Management Act, 1999. Municipal Sewage Regulation, http://www.qp.gov.bc.ca/statreg/regAA//WasteMgmt/129_99.htm, B.C. Ministry of Water, Land and Air Protection. 97 Environmental Acts and Regulations, 2002. Acts administered in part by the minister of the environment, Fisheries Act, Pulp and Paper Effluent Regulations, (SOR/92-269), http://www.ec.gc.ca/EnviroRegs/ENG/SearchDetail.cfm?intReg=83, Environment Canada. 98 Chapter 5 Conclusions and Recommendations for Future Work Chapter 3 focused on characterization of DLS runoff with regards to particle size, morphology, physiochemistry, and the propensity for particle to sediment and biodegrade over storage. Treatment experiments conducted in Chapter 4 built upon the knowledge gained of the physiochemical and biological properties of DLS runoff. In Chapter 5, the main conclusions reported thus far are briefly revisited in point form and recommendations for future work on DLS runoff characterization and treatment are made. 5.1 Runoff Characterization Several important conclusions can be drawn from the characterization work conducted upon DLS runoff in Chapter 3. • DLS runoff has organic carbon loads and potentially zinc concentrations above the discharge limits for BC Approved Water Quality Guidelines. • Fresh DLS runoff is composed of hetero-aggregates of inorganic colloids, organic colloids and long, thin biopolymers • Over storage DLS runoff is considerably altered in physiochemical properties, however maintains a high concentration of suspending particles and zinc in the dissolved and colloidal phase, thus settling lagoon treatment does not provide complete treatment • DLS runoff is biologically active and will degrade COD when stored at 25 °C, but is relatively stable from degradation when stored at 4 °C 99 • DLS runoff readily sediments at both 25 °C and 4 °C making a primary clarifier a potential pre-treatment option for DLS runoff COD, if a suitable retention time is chosen within the constraints of space allowable on dry land sorts While this study showed that DLS runoff poses an environmental threat due to high particle loading and potentially metals (zinc), more work can be done to identify other toxic components such as xenobiotics, phenols, resin acids and metals as other studies have identified these species as toxic to fish in other wood waste effluents as discussed in Chapter 2. A similar fractionation and characterization study as conducted in Chapter 3 could be expanded to test for the phase partitioning (particulate, colloidal, dissolved) of other suspect toxic components, for example resin acids. This study could be used to quantitatively determine if indeed these substances are particle-bound and the acutely toxic contingent of DLS runoff. 5.2 Lab-scale DLS runoff treatment trials The following conclusions can be drawn with respect to laboratory-scale sand batch adsorption and batch column treatment experiments conducted in this thesis: • The oxide-coated sand removed DLS runoff COD, turbidity, zinc and raised the pH to a level which may be acceptable to regulators when compared with BC Approved Water Quality Guidelines • The oxide-coated sand removed an additional 34% COD and 32% turbidity when compared with the oxide-stripped sand most likely due to enhanced absorption (as proven shown in the batch absorption experiments) and filtration due to increase surface area (as discussed in Chapter 2) • Flocculation was not a key mechanism in sand column treatment over eight successive treatment experiments • The role of the biological component of sand column treatment was not realized over eight successive treatment experiments 100 Future lab scale treatment trials should test a primary clarifier/flow equalization tank and/or polishing stage as DLS runoff had highly variable COD strength and volume wastewater (as in a storm event) may require flow equalization. In addition a primary clarifier would decrease loading on a sand filter. In addition to a primary clarifier a polishing stage may be required when wastewater strength exceeds ah undetermined threshold. The successive batch treatment experiments conducted in Chapter 4 could be repeated and extended for more than eight successive treatment runs to test two important hypotheses. Firstly, it would be useful to determine if the biotic columns show enhanced treatment performance over the abiotic columns over time, for example to determine if the abiotic columns saturate. Secondly, to determine if the oxide-stripped columns continue to increase their treatment performance over time as the results of this study showed that perhaps the COD and turbidity removal performance was increasing over successive batches. In addition, the sand column's performance should be tested for treatment performance at low temperatures, for example 0 - 4 °C. Furthermore, the fate of other compounds of interest such as metals and resin acids could be tested more comprehensively during the treatment process. 5.3 Full-scale DLS runoff treatment trials The immediate appeal of this proposed sand filtration process for dry land sort operators is its low-cost and ease of design and implementation. However, as discussed for lab-scale studies, commercialization of this technology will require serious consideration and perhaps design of primary treatment and/or possibly polishing technology which in turn will be dictated by on-site hydrological conditions, liquid loading rates, and the potential for solids handling. Due to the intermittent flow, volume and strength of DLS runoff, a real-time control activated sludge or sequencing batch reactor (SBR) may be considered as a treatment option, in order to optimize treatment performance and space on dry land sorts. 101 Appendix A - Supplementary experimental results Tables A.1 and A.2 contain the experimental data for Chapter 3, chemical characterization of particulate, colloidal, and dissolved fractions in fresh and stored for DLS runoff. Tables A.3 through A. 10 present the experimental data from Chapter 4 Column Experiment C. Figures A.1 through A.7 illustrate the treatment of COD for run 2 through run 8 for Column Experiment C. Coated refers to the oxide-coated sand and uncoated refers to the oxide-stripped sand for Tables A.2 through A.9 and Figures A.1 through A.7. Table A.1: Results of Chapter 3 chemical characterization of particulate, colloidal and dissolved fractions in fresh DLS runoff. Values are reported as averages of 3 replicates as determined by ICP-AES. Crude Particulate Colloidal Dissolved (mg/L) (mg/L) (mg/L) (mg/L) COD 1260 660 486 74 Al 28.7 20.35 4.06 4.29 Ba 0.34 0.044 0.064 0.232 Ca 97.1 0 13.7 84 Cu 0.079 0 0.028 0.06 Fe 22.9 13.51 3.58 5.81 K 215 0 41 178 Mg 209 0 44 168 Mn 18.5 0 3.5 15.4 Na 1780 0 353 1487 P 6.41 1.63 1.95 2.83 Si 20.6 13.2 4 3.4 Sr 0.7 0 0.04 0.66 s 82.8 0 6.75 78.25 Zn 1.18 0 0.215 0.995 102 Table A.2: Results of Chapter 3 chemical characterization of particulate, colloidal and dissolved fractions in stored DLS runoff. Values are reported as averages of 3 replicates as determined by ICP-AES. Crude Particulate Colloidal Dissolved (mg/L) (mg/L) (mg/L) (mg/L) COD 378 32 319 27 Al 4.54 2.59 1.25 0.7 Ba 0.99 0 0.936 0.054 Ca 52.8 0.3 0 55.6 Cu 0.094 0.008 0.007 0.079 Fe 13.7 3.1 3.73 6.86 K 69.3 3.9 0 73.7 Mg 63.9 0 0 91.5 Mn 1.16 0 0 1.27 Na 868 17 0 992 P 2.63 0.64 1.26 0.73 Si 10.97 3.14 0.96 6.87 Sr 0.41 0.01 0 0.46 S 40.2 0 0 44.53 Zn 0.91 0 0.847 0.063 103 Table A.3: Results of Chapter 4 column experiment C run 1 column description COD std. dev. Turbidity Turbidity a v e std. dev. pH (mg/L) (mg/L) (mg/L) (mg/L) (FTUs) (FTUs) (FTUs) (FTUs) (FTUs) untreated untreated 1109 1074 1162 1115 44 980 940 1000 973 31 6.1 untreated static control (whole) 1109 1126 1183 1139 39 untreated static control (super) 975 935 866 925 55 850 850 840 847 6 6.1 1 coated (whole) 172 114 156 147 30 1 coated (super) 133 167 148 149 17 76 78 80 78 2 6.7 2 coated (whole) 163 222 184 190 30 2 coated/(super) 123 201 137 154 42 78 78 80 79 1 6.9 3 coated/abiotic (whole) 163 116 161 147 27 3 coated/abiotic (super) 111 123 244 159 74 70 68 68 69 1 6.8 4 coated/abiotic (whole) 143 259 218 207 59 4 coated/abiotic (super) 119 124 226 156 60 67 68 67 67 1 • 6.1 5 uncoated (whole) 460 411 484 452 37 5 uncoated (super) 480 506 512 499 17 420 410 440 423 15 6.1 6 uncoated (whole) 580 516 556 551 32 6 uncoated (super) 554 602 492 549 55 380 410 400 397 15 6.3 7 uncoated/abiotic (whole) 544 624 582 583 40 7 unocated/abiotic (super) 510 624 562 565 57 440 420 400 420 20 6.2 8 unocated/abiotic (whole) 562 545 597 568 27 8 uncoated/abiotic (super) 532 510 526 523 11 380 380 390 383 6 6.2 9 conditioned (whole) 118 133 106 119 14 9 conditioned (super) 112 104 122 113 9 72 72 72 72 0 6.9 Table A.4: Results of Chapter 4 column experiment C run 2 column description (mg/L) COD (mg/L) (mg/L) (mg/L) std. dev. (FTUs) Turbidity (FTUS) (FTUS) Turbidity™, (FTUs) std. dev. pH (FTUs) untreated untreated 1160 1235 1228 1208 41 940 980 1000 973 31 6.1 untreated static control (whole) 1192 1147 1264 1201 59 untreated static control (super) 1091 990 1091 1057 58 840 850 850 847 6 6.1 1 coated (whole) 226 218 208 217 9 1 coated (super) 191 146 221 186 38 82 80 80 81 1 i 6.7 2 coated (whole) 182 280 174 212 59 2 coated/fsuper) 206 105 151 154 51 74 76 76 75 1 6.9 3 coated/abiotic (whole) 129 244 164 179 59 3 coated/abiotic (super) 214 192 126 177 46 68 70 70 69 1 6.8 4 coated/abiotic (whole) 182 163 221 189 30 4 coated/abiotic (super) 221 164 108 164 57 66 66 67 66 1 6.1 5 uncoated (whole) 559 569 562 563 5 5 uncoated (super) 581 662 489 577 87 320 310 340 323 6.1 6 uncoated (whole) 353 411 425 396 38 15 6 uncoated (super) 410 440 441 430 18 280 300 310 297 6.3 7 uncoated/abiotic (whole) 545 610 606 587 36 15 7 unocated/abiotic (super) 615 595 520 577 50 440 420 420 427 6.2 e unocated/abiotic (whole) 545 623 536 568 48 12 8 uncoated/abiotic (super) 609 516 576 567 47 410 400 380 397 6.2 9 conditioned (whole) 208 242 227 226 17 15 9 conditioned (super) 155 208 168 177 28 72 72 72 72 0 6.9 Table A.5: Results of Chapter 4 column experiment C run 3 104 column description • (mg/L) C O D <mo/L) (mg/L) C O D „ „ . (mg/L) std. dev. (FTUs) Turbidity (FTUs) (FTUs) Turbidity^ (FTUs) std. dev. (FTUS) pH untreated untreated 1190 1220 1308 1239 61 900 980 920 933 42 6.1 untreated static control (whole) 1335 1260 1281 1292 39 untreated static control (super) 1279 1245 1202 1242 39 860 880 880 873 12 6.1 1 coated (whole) 235 208 216 220 14 1 coated (super) 190 184 156 177 18 82 80 80 81 1 6.7 2 coated (whole) 272 186 206 221 45 2 coateoV(super) 178 156 190 175 17 78 76 74 76 2 6.9 3 coated/abiotic (whole) 282 264 318 288 27 3 coated/abiotic (super) 265 334 386 328 61 90 88 90 89 1 6.8 4 coated/abiotic (whole) 392 441 415 416 25 4 coated/abiotic (super) 219 340 322 294 65 88 91 86 88 3 6.1 5 uncoated (whole) 533 562 632 576 51 5 uncoated (super) 508 498 662 556 92 320 310 340 323 15 6.1 6 uncoated (whole) 589 582 614 595 17 6 uncoated (super) 498 523 541 521 22 300 310 310 307 6 6.3 7 uncoated/abiotic (whole) 533 551 549 544 10 7 unocated/abiotic (super) • 557 631 536 575 5 0 . 340 380 380 367 23 6.2 e unocated/abiotic (whole) 601 589 623 604 17 e uncoated/abiotic (super) 576 548 616 580 34 310 300 280 297 15 6.2 9 conditioned (whole) 208 216 248 224 21 9 conditioned (super) 242 204 184 210 29 72 72 72 72 0 6.9 Table A.6: Results of Chapter 4 column experiment C run 4 column description (mg/L) C O D (mg/L) (mg/L) C 0 D „ „ (mg/L) std. dev. (FTUs) Turbidity (FTUs) (FTUs) Turbid i ty M (FTUs) std. dev. (FTUs) pH untreated untreated 1284 1226 1308 1273 42 910 920 940 923 15 6.1 untreated static control (whole) 1302 1294 1236 1277 36 untreated static control (super) 1178 1215 1285 1226 54 870 880 870 873 6 6.1 1 coated (whole) 259 267 312 279 29 1 coated (super) 191 282 280 251 52 120 130 130 127 6 6.7 2 coated (whole) 272 243 218 244 27 2 coated/(super) 260 223 227 237 20 120 110 110 113 6 6.9 3 coated/abiotic (whole) 322 395 371 363 37 3 coated/abiotic (super) 296 392 291 326 57 120 120 120 120 0 6.8 4 coated/abiotic (whole) 316 359 343 339 22 4 coated/abiotic (super) 324 308 296 309 14 110 100 110 107 6 6.1 5 uncoated (whole) 585 550 592 576 23 5 uncoated (super) 536 580 519 545 31 340 330 340 337 6 6.1 6 uncoated (whole) 566 528 625 573 49 6 uncoated (super) 589 621 560 590 31 330 350 350 343 12 6.3 7 uncoated/abiotic (whole) 691 720 682 698 20 7 unocated/abiotic (super) 668 681 666 672 8 350 350 340 347 6 6.2 8 unocated/abiotic (whole) 689 692 626 669 ' 37 8 uncoated/abiotic (super) 660 608 661 643 30 330 330 320 327 6 6.2 9 conditioned (whole) 263 208 263 245 32 9 conditioned (super) 184 224 216 208 21 88 90 90 89 1 6.9 Table A.7: Results of Chapter 4 column experiment C run 5 105 column description (mg/L) COD (mg/L) (mg/L) C 0 D T O T O (mgn.) std. dev. (FTUs) Turbidity (FTUs) (FTUs) Turbidity (FTUs) std. dev. (FTUs) pH untreated untreated 1160 1186 1208 1185 24 900 920 920 913 12 6.1 untreated static control (whole) 1124 1120 1143 1129 12 untreated static control (super) 922 846 912 893 41 860 880 870 870 10 6.1 1 coated (whole) 150 118 178 149 30 1 coated (super) 120 165 189 158 35 82 86 80 83 3 6.7 2 coated (whole) 203 186 218 202 16 2 coated/(super) 186 208 162 185 23 70 70 72 71 1 6.9 3 coated/abiotic (whole) 202 219 295 239 50 3 coated/abiotic (super) 218 241 235 231 12 80 82 82 81 1 6.8 4 coated/abiotic (whole) 265 287 222 258 33 4 coated/abiotic (super) 205 218 268 230 33 96 90 92 93 3 6.1 5 uncoated (whole) 439 402 436 426 21 5 uncoated (super) 362 418 409 396 30 320 320 340 327 12 6.1 6 uncoated (whole) 442 486 502 477 31 6 uncoated (super) 414 462 421 432 26 330 330 340 333 6 6.3 7 uncoated/abiotic (whole) 496 421 396 438 52 7 unocated/abiotic (super) 452 392 419 421 30 320 310 320 317 6 6.2 8 unocated/abiotic (whole) 456 502 478 479 23 8 uncoated/abiotic (super) 416 478 436 443 32 330 330 320 327 6 6.2 9 conditioned (whole) 120 182 148 160 31 9 conditioned (super) 88 108 116 104 14 62 70 72 68 5 6.9 Table A.8: Results of Chapter 4 column experiment C run 6 column description (mg/L) COD (mg/L) (mg/L) C O D „ „ . (mg/L) std. dev. (FTUs) Turbidity (FTUs) (FTUs) Turbidity „ . (FTUs) std. dev. (FTUs) pH untreated untreated 1178 1213 1169 1187 23 880 920 910 903 21 6.1 untreated static control (whole) 960 922 942 941 19 untreated static control (super) 858 812 886 852 37 850 860 860 857 6 6.1 1 coated (whole) 166 246 208 207 40 1 coated (super) 207 258 282 249 38 110 100 120 110 10 6.7 2 coated (whole) 203 257 226 229 27 2 coatedV(super) 191 226 218 212 18 120 120 120 120 0 6.9 3 • coated/abiotic (whole) 258 269 254 260 8 3 coated/abiotic (super) 236 179 256 224 40 130 130 140 133 6 6.8 4 coated/abiotic (whole) 242 218 225 228 12 4 coated/abiotic (super) 256 262 177 232 47 120 130 130 127 6 6.1 5 uncoated (whole) 438 489 372 433 59 5 uncoated (super) 436 441 398 425 24 350 340 340 343 6 6.1 6 uncoated (whole) 418 451 478 449 30 6 uncoated (super) 414 402 471 429 37 340 360 340 347 12 6.3 7 uncoated/abiotic (whole) 439 478 456 458 20 7 unocated/abiotic (super) 501 418 443 454 43 330 340 340 337 6 6.2 8 unocated/abiotic (whole) 495 421 454 457 37 i 8 uncoated/abiotic (super) 514 477 421 471 47 350 340 330 340 10 6.2 9 conditioned (whole) 218 264 189 224 38 9 conditioned (super) 167 254 202 208 44 92 90 95 92 3 6.9 Table A.9: Results of Chapter 4 column experiment C run 7 106 column description (mpA) COD (mpA) (mg/L) C 0 D „ „ (mg/L) std. dev. (FTUs) Turbidity (FTUs) (FTUs) Turbidity„. (FTUs) std. dev. (FTUS) pH untreated untreated 3700 3548 3824 3691 138 4400 4200 4300 4300 100 6.1 untreated static control (whole) 3666 3618 3510 3598 80 untreated static control (super) 2954 2847 3012 2938 84 3200 3600 3300 3367 208 6.1 1 coated (whole) 1099 1151 1086 1112 34 1 coated (super) 947 919 975 947 28 540 520 520 527 12 6.7 2 coated (whole) 1108 1056 1063 1076 28 2 coated/(super) 903 956 925 928 27 630 600 610 613 15 6.9 3 coated/abiotic (whole) 1249 1174 1207 1210 38 3 coated/abtotic (super) 1098 1036 1015 1050 43 700 680 670 683 15 6.8 4 coated/abiotic (whole) 1156 1108 1147 1137 26 4 coated/abiotic (super) 1024 1021 1002 1016 12 680 680 670 677 6 6.1 5 uncoated (whole) 1452 1523 1587 1521 68 5 uncoated (super) 1484 1502 1554 1513 36 1200 1300 1200 1233 58 6.1 6 uncoated (whole) 1556 1585 1607 1583 26 6 uncoated (super) 1545 1523 1558 1542 18 1200 1200 1300 1233 58 6.3 7 uncoated/abiotic (whole) 1656 1587 1555 1599 52 7 unocated/abiotic (super) 1542 1565 1578 1562 18 1100 1200 1100 1133 58 6.2 8 unocated/abiotic (whole) 1605 1592 1560 1586 23 8 uncoated/abiotic (super) 1547 1563 1618 1576 37 1200 1300 1300 1267 58 6.2 9 conditioned (whole) 708 746 722 725 19 9 conditioned (super) 723 712 717 717 6 480 520 500 500 20 6.9 Table A. 10: Results of Chapter 4 column experiment C run 8 column description <mg/L> COD (mg/L) (mg/L) COD„„„ (mg/L) std. dev. (FTUs) Turbidity (FTUs) (FTUs) Turbidity*, (FTUs) std. dev. (FTUs) pH untreated untreated 3189 3063 3014 3089 90 3700 3700 3800 3733 58 6.1 untreated static control (whole) 2842 2789 2808 2813 27 untreated static control (super) 2601 2686 2642 2643 43 2900 3000 2800 2900 100 6.1 1 coated (whole) 862 889 855 869 18 1 coated (super) 883 877 891 884 7 470 460 470 467 6 6.7 2 coated (whole) 823 852 801 825 26 2 coated/(super) 806 875 896 859 47 560 560 580 567 12 6.9 3 coated/abiotic (whole) 992 1009 983 995 13 3 coated/abiotic (super) 956 953 986 965 18 620 600 600 607 12 6.8 4 coated/abiotic (whole) 1098 1142 1188 1143 45 4 coated/abiotic (super) 989 957 961 969 17 600 610 600 603 6 6.1 5 uncoated (whole) 1053 1071 1040 1055 16 5 uncoated (super) 1002 1028 990 1007 19 1000 980 1020 1000 20 6.1 6 uncoated (whole) 1061 1018 1024 1034 23 6 uncoated (super) 1061 1045 1008 1038 27 1100 990 990 1027 64 6.3 7 uncoated/abiotic (whole) 1142 1038 1136 1105 58 7 unocated/abiotic (super) 1171 1167 1048 1129 70 1100 1000 1100 1067 58 6.2 8 unocated/abiotic (whole) 1141 1189 1256 1195 58 8 uncoated/abiotic (super) 1162 1144 1256 1187 60 900 1200 1000 1033 153 6.2 9 conditioned (whole) 637 651 585 624 35 9 conditioned (super) 654 625 615 631 20 480 520 500 500 20 6.9 107 'mm coated (1) coated (2) coated/abiotic (1) coated/abiotic (2) uncoated (1) uncoated (2) uncoated/abiotic uncoated/abiotic conditio (D (2) Figure A.1 - Run 2 COD treatment performance for column experiment C - i -mm I coated (1) coated (2) coated/abiotic (1) coated/abiotic (2) uncoated (1) uncoated (2) uncoated/abiotic uncoated/abiotic conditio (D (2) Figure A.2 - Run 3 COD treatment performance for column experiment C 108 coated (1) coated (2) coated/abiotic (1) coated/abiotic (2) uncoated (1) uncoated (2) uncoated/abiotic uncoated/abiotic conditio (1) (2) Figure A.3 - Run 4 COD treatment performance for column experiment C coated (1) coated (2) coated/abiotic (1) coated/abiotic (2) uncoated (1) uncoated (2) uncoated/abiotic uncoated/abiotic conditio (1) (2) Figure A.4 - Run 5 COD treatment performance for column experiment C 109 Q 60.0 1 q a* 40.0 H coated (1) coated (2) coated/abiotic (1) coated/abiotic (2) uncoated (1) uncoated (2) uncoated/abiotic uncoated/abiotic conditio (1) (2) Figure A.5 - Run 6 COD treatment performance for column experiment C n Q O O 50.0 -§ g 40.0 -10.0 -M M coated (1) coated (2) coated/abiotic (1) coated/abiotic (2) uncoated (1) uncoated (2) uncoated/abiotic uncoated/abiotic conditioi (1) (2) Figure A.6 - Run 7 COD treatment performance for column experiment C 1 1 0 r - i - | — ± - i ] | 1 I coated (l) coated (2) coated/abiotic (1) coated/abiotic (2) uncoated (1) uncoated (2) uncoated/abiotic uncoated/abiotic c u J d) (2) Figure A.7 - Run 8 COD treatment performance for column experiment C I l l Appendix B - Statistical Analyses for Treatment Experiments Table B.1 presents the statistical data for Chapter 3 comparison of CODd egraded and CODsedimented during storage of sample 9 for 13 days at 4 °C and 25°C. Tables B.2 and B.3 present A N O V A statistical analysis for Chapter 4 experiments Batch Adsorption I (comparison of COD adsorption of oxide-coated sand and oxide-stripped sand) and Batch Adsorption II (comparison of different masses sand for adsorption capacity). Tables B.4 through B.7 present ANOVA statistical analysis for comparison of treatments 1 through 5 for run 1 of Column Experiment C. Tables B.8 through B.10 present ANOVA statistical analysis for comparison of the average COD treatment performance for sand column treatments 1 through 5 over 8 successive columns runs. Table B.1: Statistical analysis for Chapter 3 batch adsorption experiment 1 - adsorption of oxide-coated vs. oxide-stripped sand SUMMARY Groups Count Sum Average Variance oxide-coated 9 21650 2405.556 6150.278 oxide-stripped 9 25560 2840 53030 ANOVA ource of Variatio SS df MS F P-value F crit Between Groupi 849338.9 1 849338.9 28.70344 6.41 E-05 4.493998 Within Groups 473442.2 16 29590.14 Total 1322781 17 Table B.2: Statistical analysis for Chapter 3 batch adsorption experiment I - adsorption of oxide-coated vs. shake runoff control SUMMARY Groups Count Sum Average Variance shake control 3 8714 2904.667 4069.333 oxide-coated 9 21650 2405.556 6150.278 ANOVA ource of Variatio SS df MS F P-value F crit Between Group! 560501.8 1 560501.8 97.74906 1.76E-06 4.964591 Within Groups 57340.89 10 5734.089 Total 617842.7 11 112 Table B.2: Statistical analysis for Chapter 3 batch adsorption experiment I - adsorption of oxide-stripped vs. shake runoff control SUMMARY Groups Count Sum Average Variance oxide-stripped 9 25560 2840 53030 shake control 3 8714 2904.667 4069.333 ANOVA ource of Variatio SS df MS F P-value F crit Between Group; 9409 1 9409 0.21761 0.650863 4.964591 Within Groups 432378.7 10 43237.87 Total 441787.7 11 Table B.2: Statistical analysis for Chapter 3 batch adsorption experiment II - shake control vs. static control SUMMARY Groups Count Sum Average Variance static runoff control 3 3202 1067.333 2297.333 shake runoff control 3 3049 1016.333 4134.333 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 3901.5 1 3901.5 1.213216 0.332518 7.70865 Within Groups 12863.33 4 3215.833 Total 16764.83 5 113 Table B.3: Statistical analysis for Chapter 3 batch adsorption experiment II - 1 g sand/runoff vs. shake control SUMMARY Groups Count Sum Average Variance shake runoff control 3 3049 1016.333 4134.333 1 g sand/runoff 3 2434 811.3333 17169.33 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 63037.5 1 63037.5 5.917995 0.071772 7.70865 Within Groups 42607.33 4 10651.83 Total 105644.8 5 Table B4: Statistical analysis for Chapter 3 batch adsorption experiment II - 5 g sand/runoff vs. shake control SUMMARY Groups Count Sum Average Variance shake runoff control 5 g sand/runoff 3 3 3049 2254 1016.333 751.3333 4134.333 9354.333 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 105337.5 1 105337.5 15.61867 0.016789 7.70865 Within Groups 26977.33 4 6744.333 Total 132314.8 5 Table B5: Statistical analysis for Chapter 3 batch adsorption experiment 11-10 sand/runoff vs. shake control SUMMARY Groups Count Sum Average Variance shake runoff control 3 3049 1016.333 4134.333 10 g sand/runoff 3 2336 778.6667 4474.333 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 84728.17 1 84728.17 19.68439 0.011363 7.70865 Within Groups 17217.33 4 4304.333 Total 101945.5 5 Table B6: Statistical analysis for Chapter 3 batch adsorption experiment II - 1g, 5g sand/runoff SUMMARY Groups Count Sum Average Variance 1 g sand/runoff 3 2434 811.3333 17169.33 5 g sand/runoff 3 2254 751.3333 9354.333 10 g sand/runoff 3 2336 778.6667 4474.333 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 5414.222 2 2707.111 0.261995 0.777882 5.143249 Within Groups 61996 6 10332.67 Total 67410.22 8 115 Table B.4: Statistical analysis for comparison of treatment 1 and 2 average COD treatment performance over eight successive batch runs SUMMARY Treatment Count Sum Average Variance 1 - oxide-coated 6 909 151.5 813.5 2 - oxide-coated, abiotic 6 947 157.8333 3626.167 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 120.3333 1 120.3333 0.054208 0.820592 4.964591 Within Groups 22198.33 10 2219.833 Total 22318.67 11 Table B.4: Statistical Analysis for comparison of treatment 3 and 4 COD treatment performance over eight successive batch runs SUMMARY Treatment Count Sum Average Variance 3 - oxide-stripped 6 3146 524.3333 2082.267 4 - oxide-stripped, abiotic 6 3264 544 1900.8 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 1160.333 1 1160.333 0.582633 0.462909 4.964591 Within Groups 19915.33 10 1991.533 Total 21075.67 11 116 Table B.5: Statistical Analysis for comparison of treatments 1 and 3 COD treatment performance over eight successive batch runs SUMMARY Treatment Count Sum Average Variance 1 - oxide-coated 6 909 151.5 813.5 3 - oxide-stripped 6 3146 524.3333 2082.267 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 417014.1 1 417014.1 288.0164 1.06E-08 4.964591 Within Groups 14478.83 10 1447.883 Total 431492.9 11 Table B.6: Statistical Analysis for comparison of treatments 1 and 5 COD treatment performance over eight successive batch runs SUMMARY Treatment Count Sum Average Variance 1 - oxide coated 7 910 130 3913.667 5 - conditioned 4 343 85.75 2952.25 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 4984.159 1 4984.159 1.387111 0.269098 5.117357 Within Groups 32338.75 9 3593.194 Total 37322.91 10 

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.831.1-0058943/manifest

Comment

Related Items