Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Hydrogen sulphide removal kinetics in a trickling biofilter Symons, Stephen Albert 2003

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

Item Metadata

Download

Media
831-ubc_2003-0096.pdf [ 4.94MB ]
Metadata
JSON: 831-1.0058964.json
JSON-LD: 831-1.0058964-ld.json
RDF/XML (Pretty): 831-1.0058964-rdf.xml
RDF/JSON: 831-1.0058964-rdf.json
Turtle: 831-1.0058964-turtle.txt
N-Triples: 831-1.0058964-rdf-ntriples.txt
Original Record: 831-1.0058964-source.json
Full Text
831-1.0058964-fulltext.txt
Citation
831-1.0058964.ris

Full Text

H Y D R O G E N S U L P H I D E R E M O V A L K I N E T I C S I N A T R I C K L I N G BIOFILTER by Stephen Albert Symons B. Sc., Queen's University, 1994. A thesis submitted in partial fulfilment of the requirements for the degree of Master of Applied Science The Faculty of Graduate Studies Department of Chemical and Biological Engineering University of British Columbia January 2003 We accept this thesis as conforming to the required standard © Stephen A . Symons, 2003. In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of CtteHV6dn_ A o f ^ (SigLOtVctVi €^0>M<3ax.ti4(^ . The University of British Columbia Vancouver, Canada Date lo6S-ct.ll A B S T R A C T The first-order kinetics o f hydrogen sulphide oxidation was examined in two trickling biofilter reactors using residence time distribution analysis and the Tanks in Series model for packed bed reactors. The reactor's liquid phase was maintained at p H 5.0 in Reactor 1 and p H 2.5 in Reactor 2. Carbon dioxide was added as a supplemental carbon source in varying concentrations. Hydrogen sulphide concentrations to 300 ppm were investigated. The Tanks in Series parameter N was found to be approximately 7 for both reactors, and was insensitive to the air flow rate. The hydrogen sulphide removal rate constant decreased with increasing hydrogen sulphide inlet concentration. Substrate inhibition was suspected due to the toxic nature o f hydrogen sulphide on the trickling biofilter microbes, but was shown to be insignificant at the hydrogen sulphide concentrations under investigation. A t equal hydrogen sulphide inlet concentrations, the oxidation rate constant at p H 5.0 was approximately twice that at p H 2.5. The greatest removal rate constant found was k=0.130 s 1 at a hydrogen sulphide inlet concentration o f 50 ppm. The Tanks in Series model was shown to be a viable means o f estimating the l s t -order rate constant for the removal o f gaseous hydrogen sulphide from air streams in a trickling biofilter. A scale-up approach using the Tanks i n Series model to determine the kinetic rate constant and literature data for determining the parameter N in a production-scale trickling biofilter is briefly discussed. This wil l aid in the application o f economical and environmentally preferable trickling biofilters to the emissions control systems for Kraft pulp mills and other H , S producing industries. T A B L E O F C O N T E N T S Abstract - ii Table of Contents iii List of Tables v List of Figures vi List of Equations ix Acknowledgments x Dedication xi Chapter 1 - Introduction • 1 1.1 Objectives 2 Chapter 2 - Literature Review 3 2.1 Health Risks •. 3 2.2 Kraft Pulp Odour Control 5 2.3 Biological Waste Gas Treatment 7 2.4 Gas Residence Time Distribution 12 2.5 Tanks in Series Model 13 2.6 Hydrogen Sulphide Removal 15 2.7 Biological Considerations 16 2.8 Modelling Biological Gas Treatment 18 Chapter 3 - Materials & Methods 30 3.1 Reactor Design 30 3.2 Residence Time Distribution 32 3.3 Trickling Biofilter Operation 34 3.4 Experimental Program 36 3.5 Gas Concentration Measurement 37 3.6 Data Analysis 38 Chapter 4 - Residence Time Distribution Results 41 4.1 Pressure Drop Results 41 4.2 R T D Results 42 Chapter 5 - Hydrogen Sulphide Removal Results 49 5.1 50 ppm H,S + 0 ppm C O , Results 49 iii 5.2 50 ppm H 2 S +150 ppm C O , Results : 54 5.3 50 ppm H 2 S + 300 ppm C O , Results 59 5.4 50 ppm H 2 S + 1500 ppm C O , Results 65 5.5 100 ppm H 2 S + 0 ppm C O , Results : 70 5.6 100 ppm H 2 S + 300 ppm C O , Results 75 5.7 100 ppm H 2 S + 650 ppm C O , Results : 80 5.8 100 ppm H,S + 1500 ppm C O , Results : 91 5.9 200 ppm H,S + 0 ppm C O , Results 96 5.10 300 ppm H,S + 0 ppm C O , Results 101 Chapter 6 - Discussion of Results : 106 6.1 Summary of Results 106 6.2 Residence Time Distribution 107 6.3 First Order Assumption 108 6.4 Substrate Inhibition 110 6.5 Acid Yield and Energy Storage 112 6.6 Effect o f p H 114 6.7 Effect of Carbon Dioxide 116 6.8 R T D Scale-Up Approach 117 Chapter 7 - Conclusions • 119 Chapter 8 - Recommendations 121 Chapter 9 - Nomenclature • 122 Chapter 10 - References 124 Chapter 11 - Appendices • 131 11.1 Nutrient Media :.. 131 11.2 Reaction Order Verification ; 132 iv LIST O F T A B L E S Number Page Table 2-1 TRS Gas Odour Threshold Limits 3 Table 2-2 Risks & Effects of Acute H 2 S Exposure 4 Table 2-3 Material Phase Comparisons of Biological Waste-Gas Treatment Technologies. 8 Table 2-4 Comparison of Selected Treatment Technologies for H V L C TRS Gases 12 Table 2-5 Selected Literature Results for H,S Removal by Laboratory and Pilot-Scale Biofilters... 15 Table 2-6 Various Kinetic Expressions Used To Model Substrate Biodegradation 25 Table 2-7 Comparison of Quasi-Steady-State Models for Biofilters 28 Table 2-8 Comparison of Dynamic Models for Biofilters 29 Table 3-1 Air Flowrates Used in R T D Experiments 32 Table 3-2 Significant Events and Changes During Trickling Biofilter Reactor Start-Up '. 35 Table 3-3 Experimental Program Sequence 37 Table 4-1 Summary of R T D Parameter Results 48 Table 6-1 Summary of Experimental Results: Maximum H 2 S Removal Rate & lst-Order Rate Constant 107 Table 6-2 Summary of Substrate Inhibition Model Parameters with Various Initial Value Sets ... I l l Table 6-3 Summary of Acid Yield Results 113 Table 9-1 Table of Nomenclature 122 Table 11-1 Nutrient Media Formulations 131 v L I S T O F F I G U R E S Number Page Figure 3-1 Schematic o f Two Parallel Trickling Biofilters 31 Figure 4-1 Reactor pressure drop for Reactors 1 and 2 showing A P differences between reactors and hysteresis 41 Figure 4-2 Reactor 1 R T D tracer response curves with peak asymmetry and tails due to back-mixing 42 Figure 4-3 Reactor 2 R T D tracer response curves with peak asymmetry and tails due to back-mixing : 43 Figure 4-4 Reactor 1 mean residence time x; E B R T calculation is shown for reference 44 Figure 4-5 Reactor 2 mean residence time x; E B R T calculation is shown for reference 45 Figure 4-6 Reactor 1 number of CSTR's N with mean value o f 7.374 46 Figure 4-7 Reactor 2 number of CSTR's N with mean value o f 6.446 47 Figure 5-1 H 2 S removal efficiency at 50 p p m H 2 S and 0 ppm C O , 50 "Figure 5-2 H 2 S removal rate at 50 ppm H 2 S and 0 ppm C O , 51 Figure 5-3 H 2 S rate constant at 50 ppm H , S and 0 ppm C O , 52 Figure 5-4 H ,S acid yield at 50 ppm H 2 S and 0 p p m C O , 53 Figure 5-5 H ,S removal efficiency at 50 ppm H , S and 150 ppm C O , 55 Figure 5-6 H ,S removal rate at 50 ppm H , S and 150 p p m C O , ; 56 Figure 5-7 H , S rate constant at 50 ppm H , S and 150 p p m C O , 57 Figure 5-8 H ,S acid yield at 50 ppm H , S and 150 p p m C O , 58 Figure 5-9 H ,S removal efficiency at 50 ppm H , S and 300 ppm C O , 61 Figure 5-10 H ,S removal rate at 50 ppm H , S and 300 p p m C O , 62 Figure 5-11 H , S rate constant at 50 ppm H , S and 300 ppm C O , 63 Figure 5-12 H ,S acid yield at 50 ppm H,S and 300 ppm C O , 64 Figure 5-13 H , S removal efficiency at 50 ppm H , S and 1500 ppm C O , 66 Figure 5-14 H 2 S removal rate at 50 ppm H , S and 1500 ppm C O , 67 Figure 5-15 H ,S rate constant at 50 ppm H , S and 1500 ppm C O , 68 Figure 5-16 H , S acid yield at 50 ppm H , S and 1500 p p m C O , 69 Figure 5-17 H , S removal efficiency at 100 ppm H , S and 0 ppm C O , : 71 Figure 5-18 H ,S removal rate at 100 ppm H , S and 0 ppm C 0 2 72 v i Figure 5-19 H 2 S rate constant at 100 ppm H , S and 0 ppm C O , 73 Figure 5-20 H ,S acid yield at 100 ppm H , S and 0 ppm C O , 74 Figure 5-21 H ,S removal efficiency at 100 ppm H,S and 300 ppm C O , 76 Figure 5-22 H ,S removal rate at 100 p p m H , S and 300 ppm C O , 77 Figure 5-23 H ,S rate constant at 100 ppm H , S and 300 ppm C O , 78 Figure 5-24 H 2 S acid yield at 100 ppm H , S and 300 ppm C O , 79 Figure 5-25 H ,S removal efficiency at 100 ppm H , S and 650 ppm C O , 82 Figure 5-26 H ,S removal rate at 100 ppm H , S and 650 ppm C O , 83 Figure 5-27 H 2 S rate constant at 100 ppm H 2 S and 650 ppm C O , 84 Figure 5-28 H ,S acid yield at 100 ppm H , S and 650 ppm C O , 85 Figure 5-29 H,S'removal efficiency at 100 p p m H,S and 650 ppm C 0 2 87 Figure 5-30 H 2 S removal rate at 100 ppm H , S and 650 ppm C O , 88 Figure 5-31 H ,S rate constant at 100 ppm H 2 S and 650 ppm C 0 2 : 89 Figure 5-32 H , S acid yield at 100 ppm H , S and 650 ppm C O , 90 Figure 5-33 H ,S removal efficiency at 100 ppm H , S and 1500 ppm C O , 92 Figure 5-34 H ,S removal rate at 100 p p m H , S and 1500 ppm C O , 93 Figure 5-35 H , S rate constant at 100 ppm H , S and 1500 ppm C O , 94 Figure 5-36 H ,S acid yield at 100 ppm H , S and 1500 ppm C O , 95 Figure 5-37 H ,S removal efficiency at 200 ppm H , S and 0 ppm C O , 97 Figure 5-38 H 2 S removal rate at 200 ppm H , S and 0 ppm C O , 98 Figure 5-39 H ,S rate constant at 200 ppm H , S and 0 ppm C 0 2 : 99 Figure 5-40 H , S acid yield at 200 ppm H 2 S and 0 ppm C O , 100 Figure 5-41 H , S removal efficiency at 300 p p m H 2 S and 0 ppm C 0 2 102 Figure 5-42 H 2 S removal rate at 300 ppm H , S and 0 ppm C 0 2 103 Figure 5-43 H ,S rate constant at 300 ppm H , S and 0 ppm C O , 104 Figure 5-44 H , S acid yield at 300 ppm H , S and 0 ppm C O , 105 Figure 6-1 Reaction order comparison showing validity o f I s -order model assumption, compared to 0-order and 2 n d-order, at 100 p p m H , S and 1500 ppm C O , 110 Figure 6-2 H , S Substrate Inhibition M o d e l 112 Figure 11-1 Reaction order comparison for Reactor 1 at 50 ppm H , S + 0 ppm C O , 132 Figure 11-2 Reaction order comparison for Reactor 2 at 50 ppm H , S + 0 ppm C O , 132 Figure 11-3 Reaction order comparison for Reactor 1 at 50 ppm H , S + 150 ppm C O , 133 Figure 11-4 Reaction order comparison for Reactor 2 at 50 ppm H , S + 150 ppm C O , 133 vii Figure 11-5 Reaction order comparison for Reactor 1 at 50 ppm H,S + 300 ppm C 0 2 134 Figure 11-6 Reaction order comparison for Reactor 2 at 50 ppm H 2 S + 300 ppm C 0 2 134 Figure 11-7 Reaction order comparison for Reactor 1 at 50 ppm H 2 S + 1500 ppm G O , 135 Figure 11-8 Reaction order comparison for Reactor 2 at 50 ppm H 2 S + 1500 ppm C 0 2 . 135 Figure 11-9 Reaction order comparison for Reactor 1 at 100 ppm H , S + 0 ppm C O , 136 Figure 11-10 Reaction order comparison for Reactor 2 at 100 ppm H 2 S + 0 ppm C 0 2 136 Figure 11-11 Reaction order comparison for Reactor 1 at 100 ppm H , S + 300 ppm C O , 137 Figure 11-12 Reaction order comparison for Reactor 2 at 100 ppm H 2 S + 300 ppm C O , 137 Figure 11-13 Reaction order comparison for Reactor 1 at 100 ppm H , S + 650 ppm C O , . . ; . 138 Figure 11-14 Reaction order comparison for Reactor 2 at 100 ppm H 2 S + 650 ppm C O , 138 Figure 11-15 Reaction order comparison for Reactor 1 at 100 ppm H , S + 650 ppm C O , 139 Figure 11-16 Reaction order comparison for Reactor 2 at 100 ppm H 2 S + 650 ppm C 0 2 139 Figure 11-17 Reaction order comparison for Reactor 1 at 100 ppm H 2 S + 1500 ppm C O , 140 Figure 11-18 Reaction order comparison for Reactor 2 at 100 ppm H 2 S + 1500 ppm C O , 140 Figure 11-19 Reaction order comparison for Reactor 1 at 200 ppm H 2 S + 0 p p m C 0 2 141 Figure 11-20 Reaction order comparison for Reactor 2 at 200 ppm H,S + 0 ppm C 0 2 141 Figure 11-21 Reaction order comparison for Reactor 1 at 300 ppm H , S + 0 ppm C 0 2 142 Figure 11-22 Reaction order comparison for Reactor 2 at 300 ppm H 2 S + 0 ppm C O , 142 vi i i L I S T O F E Q U A T I O N S Number Page Equation 2-1 Mean Residence Time Equation 13 Equation 2-2 Distribution Variance Equation 13 Equation 2-3 Dimensionless Time & Dimensionless Variance 13 Equation 2-4 Number o f Tanks in Series Parameter 14 Equation 2-5 Kinetic Rate Constant Estimate 14 Equation 2-6 Calvin Cycle Reaction Stoichiometry 17 Equation 2-7 Sulphur Compound Oxidation Reactions 17 Equation 2-8 Henry's Law & Constant for H 2 S 20 Equation 2-9 Typical Saturation Kinetics 23 Equation 2-10 Substrate Inhibition Kinetics 24 Equation 3-1 H ,S Solubility in Water 30 Equation 3-2 H 2 S Removal Efficiency 38 Equation 3-3 H 2 S Loading Rate 38 Equation 3-4 H 2 S Removal Rate 39 Equation 3-5 Complete hydrogen sulphide oxidation & subsequent neutralisation with sodium hydroxide 40 Equation 3-6 H 2 S A c i d Yie ld 40 Equation 6-1 Performance equation for a 0 t h order reaction in N tank reactors in series 109 Equation 6-2 Performance equation for a 2 n d order reaction in N tank reactors in series; the 2 n d -order rate constant was evaluated using a spreadsheet Solver function 109 Equation 6-3 H 2 S Dissociation Reactions; pH-dependent Solubility 115 Equation 6-4 Dispersion Number for Small Extents o f Dispersion 118 Equation 6-5 Dispersion Number for Large Extents o f Dispersion for Closed Vessels 118 ix A C K N O W L E D G M E N T S The author wishes to thank D r . K e n Pinder for his patience, understanding and generosity as the supervisor o f this research project. H e provided the opportunity for this research project and the advice & understanding to ensure its completion. The National Science & Engineering Research Council ( N S E R C ) and the Pulp and Paper Research Institute of Canada ( P A P R I C A N ) provided financial support. Their support is gratefully acknowledged. Special thanks are due to the staff o f U B C ' s Pulp & Paper Centre. Their assistance in solving the day-to-day research and administrative problems was invaluable. Friends and colleagues at the P P C • helped "find the scotch tape". D E D I C A T I O N This work is dedicated to my wife Heather. Y o u gave me the courage to succeed where before I would not have tried. X I Chapter 1 - Introduction The odour o f Kraft pulping was once considered "the smell o f money". It is now considered a nuisance to be contained, controlled and eliminated. Due to increasing community pressures and increasingly stringent environmental regulations, Kraft pulp mills are aiming to reduce their overall emissions o f total reduced sulphur (TRS) gases and volatile organic compounds (VOCs) . Current odour treatment practice is mainly limited to incineration o f the T R S gases in operational mil l furnaces such as the lime kiln or recovery boiler. Biological treatment o f these T R S gases offers a cost-effective alternative with high removal efficiency. Biofilters and trickling biofilters employ a fixed microbial biofilm supported on organic or inert packing such as compost or ceramic rings. The T R S gases diffuse into the biofilm and are oxidized by the microbes. They are converted into odourless gases, biomass and sulphate salts. Biological treatment o f industrial exhaust gas pollutants has been gaining in popularity due to its lower capital and operating costs compared to conventional (physical and thermal) air pollution control techniques (Bibeau, L . , K . Kiared, et al. 1997). This is particularly true for high-volume low-concentration waste gas streams (Hodge, D . S. and J . S. Devinny 1995). These biological methods provide high pollutant removal efficiencies at low operating temperatures and pressures (Hwang, S.-J., H . - M . Tang, et al. 1997). A further advantage over some conventional methods is that the pollutants are eliminated by biological conversion and degradation, not simply shifted to another waste stream (Kirchner, K . , C. A . Gossen, et al. 1991). These T R S compounds are by-products o f the Kraft pulping process and are emitted at a number o f points in a typical Kraft mill. The T R S gases contain hydrogen sulphide (H 2S), methanethiol ( C H 3 S H ) , dimethyl sulphide ( C H 3 S C H 3 ) and dimethyldisulphide ( C H 3 S 2 C H 3 ) , which are responsible for a large portion o f the offensive odours normally associated with Kraft pulping. The main V O C emissions are methanol ( C H 3 O H ) and formaldehyde ( C H 2 0 ) . Efforts to reduce Kraft mi l l water consumption and close the mil l water cycle wi l l increase the demand for T R S and V O C removal technologies since these components must be removed before the water can be reused in the pulping process. Existing emissions treatment methods for removing these air contaminants have focussed on either incinerating the gases until all contaminants are completely oxidized, or 1 removing the gas phase pollutants to a liquid phase and treating them in the mil l wastewater treatment facility. Biological oxidation is a relatively novel approach to removing pollutants from the air emissions. It offers Kraft pulp mills a potentially low-energy, low-cost method for oxidizing the foul air pollutants to non-odorous and non-hazardous end products. Further details on the operating principles o f trickling biofilters are provided in section 2.3 Biological Waste Gas Treatment. Previous work in this research group (Lee, D . - H . 1999) has shown that the effects o f substrate inhibition can be reduced and the pollutant removal rates can be increased by increasing the degree of back-mixing in the packed bed trickling biofilter reactor. This back-mixing would manifest itself in a lower value o f the Tanks in Series parameter N . 1.1 Objectives This study was the first phase in a research program for analysing H 2 S removal in trickling biofilters. As such the objectives were limited in scope. The objectives o f this project were: 1. Design & build two laboratory scale trickling biofilters in parallel, 2. Demonstrate that a trickling biofilter system can be applied to treating Kraft pulp high-volume, low-concentration ( H V L C ) T R S emissions, 3. Evaluate the use of residence time distribution (RTD) analysis and the Tanks in Series model for determining removal kinetics, 4. Determine the kinetic parameters that describe H , S biodegradation in the trickling biofilter, 5. Examine the effect o f liquid phase p H on the H 2 S removal rate, 6. Evaluate the use o f C O , as a microbial carbon source to increase H , S removal rates, 7. Identify and characterize any inhibitory kinetics. The study was limited to using H , S as a model T R S pollutant, and limited to using H ,S concentrations typical o f I T V L C emission streams. 2 Chapter 2 - Literature Review 2.1 Health Risks 2.1.1 Odour Problems Tremendous effort has been put forth by the industry in the past decade to combat odours, and advances in odour control technology have reduced T R S emissions considerably (Jarvensivu, M . , R. Lammi, et al. 1997; Pinkerton, J . E . 1999). However, these T R S gases are detectable at extremely low concentrations and even small releases may trigger odour complaints from the surrounding community. Modern Kraft mills that incorporate state-of-the-art odour control equipment are still subject to these complaints. The Peace River Pulp Divis ion (Peace River, A B . ) bleached market Kraft pulp mill was commissioned in 1990 with a low odour boiler, non-condensable gas ( N C G ) collection and incineration system, and a steam stripper. Despite these advanced odour control systems, the mil l received over 300 odour complaints from the surrounding community i n their first year o f operation alone. They estimated that these complaints were triggered by ambient T R S concentrations as low as 2 ppb (Tarpey, T. 1995). Odour threshold limits for the T R S gases are shown in Table 2-1 (Jarvensivu, M . , R. Lammi, et al. 1997). A l l are detectable at very low concentrations and public pressure to reduce emissions even further is certain to continue. Table 2-1 TRS Gas Odour Threshold Limits TRS Compound Formula Threshold (ppb) Hydrogen Sulphide H 2 S 0.5 - 5 Methanethiol (MT) C H 3 S H 0.3 - 3 Dimethyl Sulphide (DMS) C H 3 S C H 3 1 - 15 Dimethyl Disulphide ( D M D S ) C H 3 S S C H 3 1 - 20 A l l the reduced sulphur compounds are toxic and can create health hazards under certain conditions. Acute effects are those created by short-term exposure to high concentrations while chronic effects are due to long-term exposure at low concentrations. 3 2.1.2 Acute Effects Acute T R S exposure risks are well documented, particularly for H 2 S , and can occur following accidental vent gas release or confined space entry by mil l personnel. Hydrogen sulphide inhibits cellular respiration and the lethal effects o f high concentration exposure are due to its effect on the nervous system. Methanethiol and D M D S are considered to be slighdy less toxic than H , S while D M S is roughly 100 times less toxic. Acute responses to various H 2 S exposure levels are summarised in Table 2-2. Less is known about the acute health hazards o f other T R S gases, but M T and D M D S produce similar effects at somewhat higher levels (Tatum, V . L . 1995). T a b l e 2-2 R i s k s & Effects o f A c u t e H 2 S E x p o s u r e Concentra t ion (ppm) Acu te H e a l t h Effect 0.5 - 5 ppb Odour threshold 5 - 1 0 Possible eye irritation 50 - 1 0 0 Respiratory tract & eye irritation 250 - 500 Severe pulmonary edema (swelling due to collected fluid in lungs) > 500 Anxiety, headache, dizziness, brain's respiratory centre depressed; death within ~1 hour 900 - 1 0 0 0 Immediate respiratory arrest & death 2.1.3 Chronic Effects Chronic health effects o f T R S gases are much more difficult to study as they occur over long periods o f time. Ambient T R S levels in Kraft mills and the surrounding communities are generally well below those required to produce acute effects and thus the main concerns about potential T R S health risks are those associated with long-term, low-level chronic exposure. There is no clear evidence that long-term exposure to low levels o f T R S gases cause increased incidence o f serious diseases such as cancer (Tatum, V . L . 1995). Studies reporting adverse health effects were not reproducible. But the uncertainty surrounding the issue will almost certainly remain a source of controversy and pulp mills will be required to control and reduce their T R S emissions. 4 2.2 Kraft Pulp Odour Control 2.2.1 TRS Emissions T R S gases are generated during pulp digesting and contribute to foul odour problems in the communities surrounding the mills. They can pose significant acute and chronic health risks to mill workers and are released from a wide variety o f sources, generally at low concentrations. The T R S gas emission points in a Kraft mil l are widespread and varied. They are typically classified as: a) low volume, high concentration ( L V H C ) or b) high volume, low concentration ( H V L C ) sources. Actual T R S composition and concentration varies between systems and with time within any given system (Jarvensivu, M . , R. Lammi, et al. 1997). Current odour treatment practice is limited to incineration for the L V H C streams while incineration, wet scrubbers, and occasionally chemical scavengers are used to treat the H V L C streams. Incineration in existing boilers and furnaces is currendy the most common method used to treat T R S gas emissions but can lead to process upsets and operational problems (Banks, D . 1997). The dilute T R S gases are oxidized to S O x and C O , in the boiler. Since the T R S gases are dilute, this requires additional energy inputs (natural gas or hog fuel) to the boiler. There is also a movement to start using dedicated emissions incinerators to prevent process upsets in the boilers caused by variable emissions loads. This may prevent process disturbances but can be expensive to operate. 2.2.2 LVHC Streams Typical L V H C streams will have T R S concentrations above 40,000 ppm (Spizzirri, P. 1995). These streams can account for 80 — 85% of the total T R S releases despite their small volumes. Sources generally include turpentine recovery systems, digester relief condensers, evaporator hotwells, foul condensate storage tanks, and strong black liquor storage tanks (Tarpey, T . 1995; Jarvensivu, M . , R. Lammi, et al. 1997). These gas streams are collected at concentrations above their upper explosion limit (UEL) to prevent safety hazards; that is, there is not enough oxygen in the stream to support combustion. Incineration is the most efficient way to treat these T R S gas streams and the mill's lime kiln is generally the first choice for this operation. These L V H C streams can generally support combustion with little or no added fuel and can be a valuable energy input to the mill . Recovery boilers and dedicated incinerators have also been used (Banks, D . 1997; Jarvensivu, M . , R. Lammi, et al. 1997) to incinerate these streams. 5 2.2.3 HVLC Streams High volume, low concentration T R S streams originate anywhere the brownstock or black liquor comes in contact with air and contribute the remaining 15 — 20% of T R S emissions. These sites include the brownstock washer hoods & storage tanks, black liquor storage tanks, chip bin vents, the smelt dissolving tank, and mud filters (Tarpey, T. 1995; Jarvensivu, M . , R. Lammi, et al. 1997). H V L C streams typically have T R S levels o f 200 - 400 ppm with flowrates o f 10,000 - 30,000 a c f m / A D T . The recovery boiler is currently the most common incineration point for the collected H V L C streams while the use of dedicated incinerators is apparently growing to reduce process upsets (Banks, D . 1997; Jarvensivu, M . , R. Lammi, et al. 1997). Because o f their low concentration o f combustible gases, H V L C streams require substantial fuel inputs to support combustion, making incineration uneconomical as a waste gas treatment strategy. 2.2.4 Incineration A s noted above, incineration is the most efficient means to treat concentrated T R S gases. The chosen incinerator must have on operating temperature above 650 °C, minimum 0.5 sec residence time and an excess oxygen level over 3% to ensure complete T R S oxidation (Jarvensivu, M . , R. Lammi, et al. 1997). However, these conditions may be difficult or impossible to maintain in an existing furnace due to process demands and constraints. Dedicated incinerators generally require a separate wet scrubber to reduce their S O x emissions and wil l require an emission permit (Banks, D . 1997). 2.2.5 Scrubber Wet scrubber systems have also been used to control T R S odours. Venturi scrubbers (Frederick, W . J. , J . P. Danko, et al. 1996) and packed tower scrubbers (Bowman, R. 1997) have been used for various H V L C T R S treatment applications. In both cases, the T R S gases are absorbed into the scrubbing liquid and chemically oxidised to less volatile and less odorous compounds. Alkaline scrubbing solutions with p H 9 — 11 are generally used to increase the T R S gas solubility. Various existing mil l streams such as green liquor and weak wash (smelt dissolving tank feed) have been used (Frederick, W . J . , J . P. Danko, et al. 1996) as well as fresh chemical additions such as sodium hypochlorite and sodium hydroxide. Hypochlorite chemical costs were estimated at $120/d in one application (Bowman, R. 1997). Corrosion, handling hazards, and possible generation o f chlorinated organic compounds were cited as potential problems. 6 2.2.6 Scavenger Organic T R S scavenger chemicals have been used to selectively complex T R S compounds rendering them much less volatile and odorous. These scavengers are injected into scrubber vent stacks and into mil l sewer or wastewater treatment systems to counter T R S odours generated by liquor spills (Spizzirri, P. 1995; Hagen, C. E . and R. W . Hartung 1997). Final T R S concentrations o f less than 1 ppm have been reported following these scavenger treatments. Operating costs using these proprietary formulations is unknown and scavengers are probably best reserved for intermittent use during regular odour control unit outages. 2.3 Biological Waste Gas Treatment Biological oxidation o f T R S gases is a novel alternative method to remove these odorous gases from Kraft mil l emissions. Microbes fixed in a biofilm oxidize the offensive gases to innocuous end products, eliminating their odours. Trickling biofilters offer other advantages over more traditional biofilters since they can be scaled vertically rather than horizontally. Additionally, they are not subject to media consumption and ageing. There are several biological waste-gas treatment technologies available to remove and treat air emissions. For all technologies, it is generally held that the pollutants o f interest must be soluble in water, and they must be easily biodegradable. But current research is expanding the boundaries o f this field and increasing the number o f pollutants that have been successfully treated with these technologies. The three main biological waste-gas treatment technologies are a) biofilters, b) trickling biofilters, and c) bioscrubbers. These are described in detail below. But the operational principles for all three technologies are similar. The pollutant laden waste gases enter the reactor where the pollutants are transferred to the liquid phase, diffuse to and are absorbed by microorganisms, which oxidize the pollutants to non-toxic and odourless products (biomass, C O , , H , 0 , S 0 4 " ) . The microbes use the pollutant gases as food and energy sources. The differences between the three are in the structure of the liquid phase, the location o f the biological phase, and the nature o f the reactor's packing phase. The main features are summarised in Table 2-3 (Devinny, J . S., M . A . Deshusses, et al. 1999). 7 Table 2-3 Material Phase Comparisons of Biological Waste-Gas Treatment Technologies Technology Liquid Phase Biological Phase Packing Phase Biofilter Discontinuous Fixed Organic Trickling Biofilter Discontinuous Fixed Inert Bioscrubber Continuous Suspended Inert Biological treatment o f exhaust gas pollutants has been gaining in popularity due to its lower capital and operating costs compared to conventional (physical and thermal) air pollution control methods (Bibeau, L . , K . Kiared, et al. 1997). This is particularly true for H V L C waste gas streams (Hodge, D . S. and J . S. Devinny 1995) and these biological methods provide high pollutant removal efficiencies at low operating temperatures and pressures (Hwang, S.-J., H . - M . Tang, et al. 1997). Biofilters and trickling biofilters are the two most commonly used methods o f biological gas treatment. In the case o f T R S gases, the compounds are oxidized to sulphuric acid & sulphate salts by the microbes, and are also incorporated into the microbial biomass (Yang, Y . and E . R. Al len 1994a). Other odourless gases, free C O , , and H , 0 may also be produced. 2.3.1 Biofilter Biofilters are generally considered the simplest to operate o f the three biological waste-gas treatment methods. Contaminated air enters and passes through a packed bed where microbes attached to the packing material absorb and consume the pollutants before the cleaned air is vented. Biofilters use an organic packing material such as peat, compost, wood bark, or soil. The organic packing serves both as the physical support as well as an additional nutrient source for the biofilm microbes. This packing is often amended with bulking agents (wood chips, perlite) to reduce compaction, p H buffers (limestone, carbonates), and additional nutrients (commercial N P K fertilisers) to assist microbial growth. Spray jets or an upstream bubble column humidifies the incoming contaminant stream and the packing can be irrigated periodically using water spray jets at the top o f the biofilter. Biofilters may be the simplest to operate o f the three biological waste-gas treatment methods but suffer from a number o f drawbacks. Control o f the bed moisture content and p H are common operating problems found in biofilters (Yang, Y . and E . R. Al len 1994a; Yang, Y . and E . R. Al len 8 1994b). These are two critical parameters for efficient contaminant removal and control is difficult. Organic packing materials require minimum moisture content in order to operate effectively. Pre-humidification o f the incoming air and occasional spraying help maintain the biofilter's moisture content at 60 - 70% (Mohseni, M . and D . G . Al l en 1997). Pollutant removal efficiency generally drops dramatically when the packing bed moisture content falls below 50 — 55%. Anaerobic conditions can be produced by excess moisture that also contributes to bed compaction and high pressure drops. A s a result the packing may have to be replaced prematurely. Control o f packing p H is a problem when treating ammonia, sulphur- or cUorine-containing compounds. The sulphur or chlorine compounds are oxidized to sulphuric or hydrochloric acid. This media acidification can physically damage the packing and reduce the effective lifetime. It may also create a hostile environment for the microorganisms. Ageing o f the packing, increasing pressure drop, channelling, and compaction of the bed material are also frequendy cited problems in organic-packed biofilters (Pisotti, D . A . 1997). Increased pressure drop is caused by growth of biomass into the packing void spaces. This leads to channelling and the bypassing o f pollutant gases through the packing with a reduced and erratic removal efficiency (Sorial, G . A . , F . L . Smith, et al. 1995). As the packing ages, channels can be created whereby the waste gas bypasses sections o f the packing. This short-circuiting reduces the removal efficiency. The microbes can consume the packing material, causing compaction that further reduces the biofilter's effectiveness. A s a result o f these drawbacks, biofilter packings are replaced frequently. Because o f packing compaction and pressure drop, organic-packed biofilters usually have a larger footprint than an equivalent trickling biofilter. This may be a considerable problem, particularly in coastal pulp mills where space is at a premium. 2.3.2 Trickling Biofilter A trickling biofilter consists o f a packed bed reactor column with a recirculating liquid nutrient stream. The inert packing supports a fixed biofilm o f microbes that are capable o f consuming and metabolising the unwanted pollutant gases. A s the pollutant gases move through the column, they are absorbed into the trickling liquid phase and diffuse through the biofilm. The pollutants are then metabolised and oxidised by the biofilm microbes. The nutrient solution provides essential minerals to the microbes and serves to keep the biofilm moist. This liquid drains to a sump where its p H can be controlled and additional nutrients can be added. The cleaned gaseous emissions are then vented. 9 These reactors may be operated in both co- and countercurrent flow modes. Cocurrent flow is generally preferred for sparingly soluble contaminants to prevent the back-transfer o f contaminants from the liquid to the gas phase (Ockeloen, H . F. , T. J . Overcamp, et al. 1996). The target pollutant is generally the primary carbon and energy source, as in the case o f volatile organic compounds (VOCs) (Bibeau, L . , K . Kiared, et al. 1997), or may be some other limiting nutrient (Yang, Y . and E . R. Al len 1994a). Trickling biofilters have been applied successfully for the treatment of H V L C V O C s such as toluene (Sorial, G . A . , F . L . Smith, et al. 1995), mixed B T E X (Sorial, G . A . , F. L . Smith, et al. 1997), propanal (Kirchner, K . , C. A . Gossen, et al. 1991), ethanol (Hodge, D . S. and J . S. Devinny 1994), methanol (O'Connor, B . 1996), acetone (Rho, D . , N . Matte, et al. 1998) and dichloromethane (Diks, R. M . M . , S. P . P. Ottengraf, et al. 1994). T o date, applied research into using tricMing biofilters for the treatment o f T R S gases has been limited. Trickling biofilters are particularly beneficial when treating low concentration gas pollutants due to the uneconomical aspects o f incinerating low energy content air streams. Pollutants producing acidic metabolites, such as the T R S gases (metabolized to sulphuric acid) and chlorinated organics (hydrochloric acid), are exceptionally good candidates for trickling biofilter treatment since the liquid p H can be easily controlled. Trickling biofilters overcome many o f the problems commonly encountered with traditional biofilters. A s noted above, the moisture content o f a biofilter bed is a critical and difficult parameter to control. The trickling biofilter has a continuous liquid phase that removes the need for moisture control and pre-humidification o f the inlet air stream. The oxidation of T R S gases creates p H control problems in a biofilter as the sulphur is oxidized to sulphuric acid which causes the bed p H to drop below levels at which the sulphur consuming microbes can thrive (Park, S.-J., K . - S . Cho, et al. 1993; Yang, Y . and E . R. Al len 1994b). Conversely, controlling the p H o f the trickling biofilter's liquid phase is straightforward: the recirculating liquid phase can be placed under automatic p H control using any basic solution to neutralize the acid generated (Kirchner, K . , G . Hauk, et al. 1987). The inert support packing used in trickling biofilters does not suffer from channelling, compaction and ageing problems. A n y bed plugging due to excess microbial biofUm growth can be solved by fluidization and washing the packing (Alonso, C , M . T. Suidan, et al. 1997). 10 2.3.3 Bioscrubber Bioscrubbers consist o f two separate reactor components: a liquid scrubber column and a mixed tank biooxidation reactor. The gas absorption is separated from the biological oxidation o f the contaminants. The gas absorption takes place in a classical packed tower scrubber and the scrubbed air exits the column. The pollutant-laden water is sent to the oxidation tank where air is added and the liquid agitated. Suspended microbes in the biooxidation reactor oxidize the contaminants and rejuvenate the scrubber water. Nutrients and p H control chemicals may be added easily. Bioscrubbers are most often used to treat exhaust streams with very dilute contaminants or difficult to degrade V O C ' s (Hecht, V . , D . Brebbermann, et al. 1995; Barton, J . W. , K . T. Klasson, et al. 1997; DeHollander, G . R., T . J . Overcamp, et al. 1998). Their use is not widespread due to the their additional capital costs and operational complexity compared to biofilters and trickling biofilters. 2.3.4 Other Pulp & Paper Applications Biofilters have found use in several pulp & paper applications, primarily to control volatile organic compound (VOC) emissions. Terpenes such as OC-pinene have been successfully removed by biofilters with greater than 99% removal efficiency (Mohseni, M . and D . G . Al len 1997; Pisotti, D . A . 1997). Other regulated V O C s such as phenol and formaldehyde have been treated to 99+% removal in biofilters (Pisotti, D . A . 1997). Mos t o f the available literature on biofilters and trickling biofilters examines their use for V O C removal. They have proven to be very effective in these applications, but examining V O C removal is beyond the scope o f this work. 2.3.5 Advantages oj"Biological Treatment Biological treatment o f T R S emissions offers a number o f advantages over conventional methods for H V L C T R S streams. A s noted above, incineration is the most efficient method for controlling. L V H C T R S emissions due to their relatively high heating value. Capital costs for H V L C T R S treatment systems wil l be comparable for most o f the currently available methods. The exception wil l be a dedicated T R S incinerator, which entails additional capital and operating costs. L o w operating costs are expected for biological treatment and wet scrubbers. The expected removal efficiency o f the biological methods is quite high. A comparison o f these important factors is presented in Table 2-4. 11 Table 2-4 Comparison of Selected Treatment Technologies for HVLC TRS Gases Biological Incineration Wet Scrubber Scavenger Capital Cost L o w High (dedicated) L o w (existing) L o w L o w Operating Cost L o w M i d L o w - M i d Variable T R S Fate Biomass, Sulphate S O x Sulphate, Dissolved T R S TRS-organic complex (sequestered) T R S Emissions < 0.1 ppm Unknown ~1.3 ppm < 1 - 3 ppm T R S Removal 99.9+% Unknown - 9 2 % 97 - 99% Comments Acclimation required Process upsets in existing boilers Several operating hazards Reserved for intermittent use Reference (Yang, Y . and E . R. Al len 1994a) (Banks, D . 1997) (Bowman, R. 1997) (Hagen, C. E . and R. W. Hartung 1997) 2.4 Gas Residence Time Distribution A s a fluid moves through a reactor, some fluid elements may follow a more indirect route and hence take a longer time to exit the reactor. Knowing the extent o f this residence time distribution (RTD) is an important diagnostic tool for identifying and rectifying channelling and other non-ideal behaviour in any packed bed reactor. This R T D and Tanks in Series reference material was taken from Levenspiel's book (Levenspiel, O . 1972). The R T D is usually evaluated by creating a pulse or step-change in the concentration o f a tracer fluid and then measuring the tracer concentration in the reactor outlet over time. The R T D is then characterized by two values: the mean residence time X, and the variance CT2. The mean is an average value o f all the recorded data and the variance measures the spread o f the data. When the reactor oudet tracer concentration C is monitored continuously following a pulse-input, the resulting C vs. time curve allows the reactor's R T D to be evaluated. The mean residence time X can be evaluated for both continuous and discrete time values and is given by: 12 T = 0 0 Equation 2-1 Mean Residence Time Equation The variance can also be described in continuous and discrete form and is defined as: j{t-tfCdt jt2Cdt 0 _ 0 - 2 jcdt JCdt 0 0 Equation 2-2 Distribution Variance Equation For ease o f automatic data logging on a computer, the discrete forms o f the mean residence time and variance equations were used throughout the experiments. It is often convenient to express time in units o f the mean residence time. The dimensionless time and dimensionless variance are denoted as: Equation 2-3 Dimensionless Time & Dimensionless Variance 2.5 Tanks in Series Model The Tanks in Series Mode l is one o f two methods routinely used to characterize non-ideal fluid flow through packed bed reactors. This is a one-parameter model that treats the packed column as a series o f identical continuous stirred tank reactors (CSTRs). This model characterises large amounts o f axial dispersion in the packed column better than the Dispersion Mode l (Levenspiel, O . 1972). The column's performance is evaluated based on the number N o f theoretical CSTRs required to give the observed tracer response. The larger the observed N , the more the reactor 13 behaves as a pure plug flow reactor. A t intermediate N values (N=2-20), the R T D tracer response curve is asymmetrical with a tail extending over time. The value o f N for any reactor system can be estimated directly from the dimensionless variance as: Equation 2-4 Number of Tanks in Series Parameter This relation was used to determine the number o f tanks in series N for each trickling biofilter reactor over a range of air flow rates. For l s t -order reactions in a series o f identical mixed reactors, the kinetic rate constant k can be evaluated direcdy i f the inlet and outlet concentrations (C 0 & C N respectively), the mean residence time X, and the number of tanks in series N , are known: T - 1 Equation 2-5 Kinetic Rate Constant Estimate With Equation 2-5, the T'-order removal kinetics o f hydrogen sulphide in the trickling biofilter can be evaluated. Levenspiel also provides graphs for estimating reaction kinetics for other reaction orders (Levenspiel, O . 1972). The Tanks in Series model was chosen for this study for two reasons. This model allows analysis o f a larger range o f conditions than the Dispersion model, and allows simpler calculations for determining the kinetic parameters (Levenspiel, O . 1972). In addition, the terminology commonly used in this field to quantify the air residence time in the packing can be misleading. Other measurement methods such as empty bed residence time (EBRT) and superficial velocity specifically ignore the effect o f column packing volume on the residence time. 14 2.6 Hydrogen Sulphide Removal Biological oxidation and removal o f hydrogen sulphide from waste gases has been studied in the past. However most research here has been performed using biofilters and not trickling biofilters as in this study. A s noted above, biological oxidation o f hydrogen sulphide produces acidic metabolites, mainly sulphuric acid. I f not controlled, this lowers the reactor system's p H and may cause an upset in the reactor performance, thereby reducing the gas removal efficiency. A t higher concentrations, hydrogen sulphide is toxic and inhibits the growth and metabolism o f the microorganisms responsible for the oxidation. These two factors impose control problems and constraints on biofilter & trickling biofilter design for hydrogen sulphide removal. T R S removal using biofilters has been shown to be effective for a number o f industries including wastewater treatment, petroleum refining, food processing, and pulp & paper manufacturing. One group o f researchers found their H 2 S removal was greater than 99.9+% in a compost biofilter with a gas retention time o f 20 s and H 2 S inlet concentration o f 5 — 2650 ppm (Yang, Y . and E . R. Al len 1994a; Yang, Y . and E . R. Al len 1994b). Their maximum observed gas treatment rate was 130 g-S / m 3 packing/h. The tested hydrogen sulphide concentrations are well above those noted above for H V L C T R S streams. This indicates that these streams could be readily treated using biofilters. The Peace River Pulp Division was, in 1995, proceeding with a pilot-scale biofilter to treat miscellaneous T R S emission sources (Tarpey, T. 1995). A variety o f conditions have been used to study H 2 S removal in biological reactors with varying results. These are summarised below in Table 2-5. N o literature reports were found for H 2 S removal in trickling biofilters. Table 2-5 Selected Literature Results for H2S Removal by Laboratory and Pilot-Scale Biofilters Maximum Removal ~ Maximum Cone. Packing & Notes Reference (g-H^/nr'h) (ppm H2S) 21.5 150 Immobilized Ca-alginate beads (Chung, Y . - C , C. Huang, et al. 1996) 4.3 70 Peat; mixed T R S gases (Park, S.-J., K . - S . Cho, et al. 1993) 15 M aximurri i - : s Remova l M a x i m u m .', Cone . ^ ; P a c k i n g & N o t e s Reference ( g - H 2 S / m 3 h ) (ppm H 2 S ) 26.7 200 Compost (Ward, A . H . , R. M . R. Branion, et al. 1998) 138.1 2000 Yard waste compost; maximum removal highly dependant on compost properties. (Yang, Y . and E . R. Al l en 1994a) . 25.3 150 Peat; mixed T R S gases (Hirai, M . , M . Ohtake, et al. 1990) 800 Selective H 2 S wet scrubber & catalytic oxidation. 1 (Oloman, C , F. E . Murray, et al. 1969) 300 30,000 Intermittent & discontinuous H 2 S spikes. (Cook, L . L . , P. A . Gostomski, et al. 1999) 24.4 200 Peat (Cho, K . - S . , M . Hirai, et al. 1991) 3.3 45 Peat (Cho, K . - S . , M . Hirai , et al. 1992) 10.0 180 D r y wastewater sludge; low loading, 187 d at 100% removal. (Degorce-Dumas, J . R., S. Kowal , et al. 1997) 129.2 2344 D r y wastewater sludge; high loading, only 10 d at 100% removal. (Degorce-Dumas, J . R., S. Kowal , et al. 1997) 26.6 60 Immobilized Ca-alginate beads. (Chung, Y . - C , C. P. Huang, et al. 1997) 2.7 B i o l o g i c a l Considerat ions The fundamental distinction between biological waste-gas treatment technologies and physico-chemical treatment methods is o f course the use o f microorganisms. It is the microorganisms that metabolize and oxidize, and therefore remove, the pollutants from the gas stream. I f the microbial Wet scrubber for H?S shown only for comparison to biological waste gas treatment. 16 population, usually found as a fixed biofilm on support packing, is unable to survive and thrive, the entire treatment reactor will cease to function. In the biological removal of hydrogen sulphide, the H ,S functions as the main energy source for the microbes. Carbon required for cellular growth is generally obtained from atmospheric carbon dioxide. The Calvin cycle is the biological mechanism that regulates the autotrophic growth o f microorganisms involved in hydrogen sulphide oxidation (Brock, T. D . 1997). The Calvin cycle reactions convert six C O , molecules to one fructose-6-phosphate molecule using 12 nicotinamide-adenine dinucleotide phosphate molecules ( N A D P H ) and 18 adenosine triphosphate (ATP) molecules. N A D P H provides the reducing power required to chemically reduce the carbon dioxide to carbohydrate, and the A T P provides the energy necessary for the phosphorylation reactions. The microorganisms produce the N A D P H and A T P from the energy released through hydrogen sulphide oxidation. Magnesium is an important required ion in the proper operation o f the Calvin cycle. (Britton, A . 2001). The overall Calvin cycle reaction is: RubisCO 6 C02 +12 NADPH + ISATP -» C6Hl206(P03H) + 12 NADP+ +18 ADP + 17 Pt PRK Equation 2-6 Calvin Cycle Reaction Stoichiometry A s noted above, the energy required to drive the Calvin cycle is derived from the oxidation o f hydrogen sulphide and other sulphur compounds as follows (Brock, T. D . 1997): H2S + 202 ->5042" +2/T; - 798.2 Ul mol S (a) HS~ + \02 + H+ -^S° +H20; -209AkJ/molS (b) S° + H20 + \-2-02 ^S024~ +2H+; - 587.1*7/mol S (c) S203 +H20 + 202 -^2S024~ +2H+; - 441.3 kJI mol S (d) Equation 2-7 Sulphur Compound Oxidation Reactions The biofilm microorganisms have the ability to oxidize the hydrogen sulphide to elemental sulphur (S°), which can be stored internally during periods o f high H , S exposure for future use. When the H,S concentration later decreases, the stored elemental sulphur is further oxidized to sulphate. This has been previously demonstrated (Buisman, C . J . N . , P. IJspeert, et al. 1991; Chung, Y . - C , C. Huang, et al. 1996). The microbes can thus survive for extended periods with little or no external 17 energy source (Brock, T. D . 1997). This is very similar to mammals storing excess food as fat for later use. A direct result o f this storage mechanism is that, under high H 2 S loads, the trickling biofilter reactor may not be acidified, or may be acidified much less than expected (Chung, Y . - C , C . Huang, et al. 2001). Later while under subsequent low H 2 S loads, the reactor may be acidified more than expected as the microbes consume thek stored elemental sulphur reserves. The carbon dioxide required for microbial growth is obtained from the free C O , in the air added to the trickling biofilter. The standard C O , concentration in air is approximately 300 ppm. This may be supplemented by additional C 0 2 i f desired. Supplementary C 0 2 levels o f up to 5% have been shown to increase the microbial growth rate o f Thiobadllus type bacteria (Kargi, F . 1982). The economics of supplementary C O , on an industrial scale have not been evaluated. 2.8 Modelling Biological Gas Treatment Trickling biofilters are complex units involving a number o f biological and physical principles. Each o f these principles must be modelled accurately i f the overall model is to provide a realistic prediction o f the actual process. Mass transport from the gas to liquid phase, material balances, diffusion through the biofilm, biodegradation o f the target pollutant, and microbial growth are normally included in published models. Early models were based on restrictive assumptions but were still used to design operational trickling biofilter units (Zarook, S. M . and A . A . Shaikh 1997). It should be noted that no modelling work was performed during the course i f this research project. The modelling literature is provided here for reference to the most important operating parameters and issues in biological waste-gas treatment. While most research efforts have been focussed on biofilters, the modelling issues and equations hold equally for trickling biofilters. Modell ing efforts to date have focussed on steady state and quasi-steady state models (Yang, Y . and E . R. Al len 1994b; Yang, Y . and E . R. Al len 1994a; Ockeloen, H . F. , T . J . Overcamp, et al. 1996; Alonso, C , X . Zhu, et al. 1998). The main objective of these research efforts is to facilitate improved design and operation compared to the gross assumptions and rule-of-thumb guidelines often used. Flowever, no published rule-of-thumb design guidelines have been found to date. Dynamic models in the literature are limited due to the complex nature o f trickling biofilter processes. Pollutant emission levels from industrial processes are unlikely to be constant and response to substrate concentration and flowrate changes can be better understood through 18 dynamic models (Deshusses, M . A. , G . Hamer, et al. 1995a; Deshusses, M . A. , G . Hamer, et al. 1995b; Zarook, S. M . , A. A . Shaikh, et al. 1997). Published biological gas treatment models employ a combined diffusion-reaction scheme where diffusion of contaminant through the biofilm and reaction in the biofilm are treated separately. The diffusion rate and the reaction rate are considered to be equal. The trickling biofilter's overall substrate removal capacity may be limited either by the diffusion of substrate through the biofilm or by the substrate oxidation rate. These different limitations may even be found witbin different zones of the same trickling biofilter and as such substrate diffusion through the biofilm is an important factor that should not be neglected (Zarook, S. M . and A . A . Shaikh 1997). Target pollutants must diffuse through the biofilm before they are metabolised. The relative diffusion and oxidation rates of the substrates in the biofilm determine an effective biofilm thickness beyond which no substrate is available to the microbes (Yu, J. and K . L. Pinder 1994). A n effectiveness factor indicating the relative importance of diffusion and reaction rates has been employed (Hwang, S.-J., H . -M. Tang, et al. 1997). Some models assume the effective biofilm thickness remains constant throughout the entire trickling biofilter length despite the fact that the equilibrium liquid phase substrate concentration, and thus the diffusional driving force, changes (Zarook, S. M . and A . A . Shaikh 1997). Assumed effective biofilm thickness values range from 20 — 100 u.m in various models (Deshusses, M . A. , G . Hamer, et al. 1995b; Hwang, S.-J., H . - M . Tang, et al. 1997). 2.8.1 G as ¥ low Patterns Nearly all published biofilter and trickling biofilter models implicitiy or explicidy assume ideal plug flow behaviour in the gas phase. Depending on the reactor construction, the gas flow rate, and the type of reactor packing, this may or may not be a valid assumption (Barton, J. W., X . S. Zhang, et al. 1998; Zarook, S. M . , A . A . Shaikh, et al. 1998). It is certainly an understandable assumption since the data analysis may be gready simplified. One notable exception is work with ammonia biofilters (Lee, D . - H . 1999). It was shown that increasing the degree of back-mixing of the gas phase in the reactor actually increased the removal rates of ammonia by the biofilter, since a more uniform distribution of microbial growth is formed. 19 In an ideal plug flow reactor, the reactant concentration is highest at the inlet and decreases towards the outlet according to the flow rates and the reaction kinetics. Reaction rate is highest near the inlet due to the high reactant concentration, and reaction rate is low near the outlet. In a continuous mixed reactor, the reactant concentration is uniform throughout the reactor, and is equal to the outlet concentration. For a given reactant conversion, the mixed reactor is smaller in size than the plug flow reactor (Levenspiel, O. 1972). This leads to lower capital costs. In the case of hydrogen sulphide removal, where the reactant is toxic and inhibits the oxidation reaction above a certain concentration, the plug flow reactor may become poisoned at the inlet if the reactant concentration is too high. Since the trickling biofilter is biologically based, this is equivalent to killing the biofilm microbes, the oxidation catalyst. A front of poisoned biofilm would move axially through the reactor until all the biofilm microbes were dead. The trickling biofilter would have to be shut down and re-seeded with a new biological inoculum. Using a continuous mixed reactor, on the other hand, would prevent the reactant concentration from ever reaching toxic levels since the toxic reactant is effectively diluted. This would allow higher concentrations of hydrogen sulphide to be treated with a trickling biofilter than would be possible in an ideal plug flow reactor. 2.8.2 Mass Transfer & Diffusion Henry's Law provides the fundamental relationship between equilibrium gas and liquid concentrations of a chemical compound. Here, C a is the concentration of a species in the aqueous phase and p g is the partial pressure of that species in the gas phase. Units for k H vary, but one common set is (mol/m3Pa). The Henry's Law constant for hydrogen sulphide is 9.869x10"4 mol/m 3Pa (Sander, R. 1999). ; k"lS =9.869x1(T4^-H „ ' H nfPa " S Equation 2-8 Henry's Law & Constant for H 2S The Henry's Law constant k H can vary by several orders of magnitude and determines the extent to which the target pollutant can be absorbed into the liquid stream prior to degradation (Ockeloen, H . F., T. J. Overcamp, et al. 1996). This intrinsic parameter can have a profound impact 20 on the design and operation o f a trickling biofilter. For example, compounds with high k H (high solubility) can be removed to a high degree with either counter-current or co-current gas — liquid flow. But as k H decreases (low solubility), high removal efficiency can only be achieved with co-current operation (Ockeloen, H . F. , T . J . Overcamp, et al. 1996). Counter-current mass transfer operations are well known to be generally more efficient than co-current. However, the liquid recirculation with counter-current gas — liquid flow creates a situation where pollutants in the liquid phase can be stripped back into the gas phase immediately prior to the gas exit, thus reducing the mass transfer efficiency. Substrate mass transfer rates from the gas to liquid phase are generally described by the liquid film mass transfer coefficient k La. Several groups have shown that the liquid film resistance dominates the gas absorption limitations in biofilters and trickling biofilters (Ockeloen, H . F. , T. J . Overcamp, et al. 1996; Hwang, S.-J., H . - M . Tang, et al. 1997). Others simply assume liquid film resistance dominates (Kirchner, K . , C . A . Gossen, et al. 1991). Oxygen microsensors have been used to experimentally measure the dissolved oxygen profiles within biofilms (Cunningham, A . B . , E . Visser, et al. 1995). Microsensors for other target substrates may also available but no data on this topic has been found to date. Other studies have found the oxygen diffusion rate was a significant limiting step in the pollutant degradation process (Hwang, S.-J., H . - M . Tang, et al. 1997). Oxygen can be depleted at relatively shallow biofilm depths compared to other substrates leading to drastically reduced substrate conversion (Zarook, S. M . , A . A . Shaikh, et al. 1997). This is due to the relative insolubility o f oxygen in water. A wide variety o f packing materials have been used in trickling biofilters. The packing acts as a support surface for biofilm growth and as a gas — liquid contact enhancing surface. Column packing properties manifest themselves in model development by their contributions to the mass transfer coefficient k L a, available surface area and the bed void volume. There are also operational considerations such as microbial attachment and bed cleaning. A suitable packing material should have the following qualities (Tchobanoglous, G . and F . L . Burtop 1991): • H igh surface to volume ratio, 21 • High bulk void volume, • Structural strength, • Provide secure microbial attachment surface, • A l l o w biofilm removal by fluidization, • Chemically inert, • Provide high liquid-side mass transfer. Direct comparisons o f various packing media have validated the above criteria. Hydrophilic support media provide superior microbial attachment compared to hydrophobic surfaces (Yu, J . and K . L . Pinder 1992). The relatively low liquid turbulence in random packed tubes compared to Raschig rings resulted in approximately 20% lower substrate removal due to relatively poor mass transfer (Kirchner, K . , G . Hauk, et al. 1987). The ability to clean the packing and remove surplus biomass is also relevant. One group found their glass packing disintegrated during washing attempts (Kirchner, K . , G . Hauk, et al. 1987). Others evaluating a monolithic channel support medium encountered severe plugging problems that could not be solved by washing (Sorial, G . A . , F. L . Smith, etal. 1995). The judicious choice o f trickling biofilter column packing should allow stable long-term operations over a wide operating range. 2.8.3 Reaction Kinetics Many different reaction kinetics models have been used in biological waste-gas treatment modelling. These range from simplified, discontinuous saturation kinetics equations (1 s t order at low concentration, zero order at high concentration) to equations accounting for interaction and inhibition between multiple substrates. 2.8.3.1 Saturation Kinetics In this field, saturation kinetics equations are often referred to as " M o n o d " kinetics or "Michaelis-Menten" kinetics, but neither term is entirely correct. Saturation kinetics refers to kinetics equations where the reaction rate is 1 s t order at low reactant concentrations and gradually changing to zero order as the reactant concentration increases. Saturation kinetics are typical o f biological 22 processes and are used extensively in modelling biofilter and trickling biofilter operations. A typical saturation kinetics expression is shown in Equation 2-9 (Shuler, M . L. and F. Kargi 1992), where r M A x is m e maximum reaction rate, and Kg is the half-saturation constant: r C _ R _ 'MAX'-Ks+C Equation 2-9 Typical Saturation Kinetics 2.8.3.2 Inhibitory Kinetics Inhibitor)' kinetics expressions have also been used in modelling these processes to account for several phenomena: the possible effect of one easily degraded substrate in a mixed contaminant stream, and the possible effect of highly toxic substrate concentrations in the contaminant stream. In the case of one easily degraded substrate, the trickling biofilter microbes preferentially consume one pollutant at the expense of consuming others. The presence, and concentration, of the first substrate inhibits the removal of the other substrates. This will be important in the design of industrial trickling biofilters, particularly in the pulp & paper and wood products industries since the contaminant streams are expected to contain mixed VOC's and/or TRS gases. Deshusses showed one method of accounting for this (Deshusses, M . A. , G . Hamer, et al. 1995b). Variations of inhibitory kinetics expressions have been called Andrews kinetics or competitive inhibition kinetics. These have been used to provide a better explanation of experimental data (Zarook, S. M . . A . A . Shaikh, et al. 1997). In the case of toxic substrates, the substrate of interest is both a food source and poison for the microbes in the trickling biofilter biofilm. At higher concentrations, increasing the substrate concentration actually reduces the reaction rate. This is particularly important in designing trickling biofilters for TRS and H,S emissions since the pollutants are known to be toxic above certain concentrations. Substrate inhibition kinetics are generally expressed as shown in Equation 2-10 (Shuler, M . L. and F. Kargi 1992), where K , = inhibition constant. 23 Equation 2-10 Substrate Inhibition Kinetics A t low substrate concentration, C 2 / K , << 1 and the inhibition effect is not observed. But as C increases, C 2 / K , tends to dominate the relation and the reaction rate r is reduced. Substrate inhibition is expected to be a significant concern in this work. However the extent o f inhibition, and the H 2 S substrate concentration at which the inhibition becomes significant are unknown. The inhibition constant K , is unique for each substrate-microbe system and may be determined through kinetic analysis. Analysis o f K , for this system is presented in section 6.4 Substrate Inhibition. 2.8.4 Biofilm Growth Many biofilm growth models assume no net biofilm growth without validating the assumption (Diks, R. M . M . , S. P. P. Ottengraf, et al. 1994). Several researchers assumed no growth in order to simplify the model and this allowed them to assume a constant biofilm thickness (Ockeloen, H . F. , T . J . Overcamp, et al. 1996; Hwang, S.-J., H . - M . Tang, et al. 1997). Methods for measuring growth vary from periodically scraping the biofilm from test plates (Yu, J . and K . L . Pinder 1993) to measuring the relative rates o f oxygen and substrate consumption i n the trickling biofilter (Escot, A . , C. Chavarie, et al. 1996). Table 2-6 reviews various selected kinetic expressions used to model substrate utilisation and microbial growth in biofilms. Those models actively accounting for growth have provided additional explanation for experimental observations, particularly those related to bed plugging (Alonso, C , M . T. Suidan, et al. 1997). 24 Table 2-6 Various Kinetic Expressions Used To Model Substrate Biodegradation Substrates . Kinetics Interaction Comments Reference 3 Andrews Yes Inhibition and interaction between substrates (Zarook, S. M . , A . A . Shaikh, et al. 1997) 2 Competitive Yes Provided better model fit to experimental data (Deshusses, M . A . , G . Hamer, et al. 1995a) 2 M o n o d N o (Hwang, S.-J., H . -M . Tang, et al. 1997) 1 M o n o d N o Desired simple biofilm model (Ockeloen, H . F. , T. J . Over camp, et al. 1996) 1 M o n o d N o Implicitly assumed 0 2 was in excess (Alonso, C , M . T. Suidan, et al. 1997) 2.8.4.1 Single Substrate A single limiting substrate has been assumed when a simple substrate utilisation expression was desired (Ockeloen, H . F. , T . J . Overcamp, et al. 1996). Few recent models account for increasing biofilm thickness with time. One study incorporated biofilm growth as a means to evaluate the available microbial surface area during their experiments (Alonso, C , M . T. Suidan, et al. 1997). They found that available surface area initially increased and then decreased as the packing void volume became filled with biomass. This resulted in decreased substrate removal efficiency and the model predictions were confirmed experimentally. This model also incorporated terms to describe the loss o f biomass to shear and microbial decay. 2.8.4.2 Multiple Substrates Having multiple pollutant substrates in the waste gas stream is a much more realistic scenario than a single substrate. These multiple substrate systems have been modelled in several ways. Oxygen and the V O C substrate as multiple non-interacting limiting substrates in the M o n o d equation have been evaluated (Hwang, S.-J., H . - M . Tang, et al. 1997). One model (Zarook, S. M . , A . A . Shaikh, et 25 al. 1997) expressed V O C degradation rates with interacting (i.e.: competitive removal) M o n o d and Andrews inhibitory kinetics but did not provide adequate validation for doing so. Others have evaluated competitive, non-competitive and uncompetitive Michaelis - Menten kinetics for their model with no net biofilm growth (Deshusses, M . A . , G . Hamer, et al. 1995b; Deshusses, M . A . , C. T.Johnson, et al. 1999). They found competitive kinetics gave superior results for the substrates being studied and incorporated this kinetic scheme into their model. Many models explicitly assume oxygen is available in excess (Ockeloen, H . F . , T . J . Overcamp, et al. 1996) while others make this assumption implicitly (Alonso, C , M . T. Suidan, et al. 1997). Oxygen's importance to the overall bio degradation process is determined primarily by the substrate loading to the trickling biofilter. A t high pollutant concentrations, increasing the oxygen partial pressure in the trickling biofilter has been shown to cause significant increases in the trickling biofilter's maximum pollutant removal capacity (Kirchner, K . , S. Wagner, et al. 1992). The potential for having multiple limiting substrates in a trickling biofilter has been shown to be a serious consideration and cannot be ignored. 2.8.4.3 Plugging Plugging o f trickling biofilters is a well-known phenomenon caused by excess biofilm growth (Diks, R. M . M . , S. P. P. Ottengraf, et al. 1994). Many researchers have acknowledged plugging without addressing the problem directly (Kirchner, K , C. A . Gossen, et al. 1991). This problem results in reduced and erratic removal efficiency behaviour and is usually revealed by an increased pressure drop across the packing (Sorial, G . A . , F . L . Smith, et al. 1995). One study has shown that a biological equilibrium can be reached where there is no net biomass growth and stable long-term operation without plugging can be achieved (Diks, R. M . M . , S. P. P. Ottengraf, et al. 1994). However, this has only been observed with one particular V O C substrate (dichloromethane) and is not a general result. The researchers suggest that endogenous decay and predation by secondary microbial populations contribute to this biological equilibrium. The recalcitrant nature of this substrate, and the associated slow biological growth rate, probably also contributes to this observation. 26 Different washing strategies have been tested to compare their effectiveness in removing excess biomass from the bed packing in order to combat plugging (Sorial, G. A. , F. L. Smith, et al, 1995; Alonso, C , M . T. Suidan, et al. 1997). These backwash treatments were part of a model validation in the form of washing frequency and duration. They found that full medium fluidization for 1 h in every 48 h resulted in the greatest overall substrate removal. 2.8.5 Model Results 2.8.5.1 Steady State Early trickling biofilter models only considered limiting case scenarios (zero order reaction & diffusion limited, zero order reaction & reaction rate limited, first order reaction) under very high or very low substrate concentration conditions. This approach has been superseded by a more general approach providing superior modelling results (Zarook, S. M . and A . A . Shaikh 1997). These quasi-steady state models assume that the biofilm characteristics do not change appreciably during the short measurement time periods. Lack of consistent numerical evaluation for model error has prevented a direct comparison of various models' accuracy. One method used was the mean square error (MSE) between experimental data and model predicted substrate concentrations (Alonso, C , M . T. Suidan, et al. 1997). This group found their highest error at high gas flowrates due to the increasing importance of factors neglected in the model such as oxygen limitations and mass transfer resistance. Another method was to provide the percent error (difference over model) in substrate removal rates (Zarook, S. M . and A . A . Shaikh 1997). A more rigorous and standardised statistical treatment of model error would enable the direct comparison of various models. Table 2-7 summarises the number of explicit assumptions, parameters and equations required by various models. Those with fewer assumptions are generally more complex but yield improved predictions of the experimental data. 27 Table 2-7 Comparison of Quasi-Steady-State Models for Biofilters Assumptions Parameters Equations Model Comments Reference »'K'.'v.-1""-:', Prediction 5 12 4 N / A Simulation (Ockeloen, H . only F . . T . J . Overcamp, et al. 1996) 9 17 6 Poor Assumes 1 s t (Hodge, D . S. order and J . S. reaction rate Devinny 1995) 5 15 11 Satisfactory (Hwang, S.-J., H . - M . Tang, et al. 1997) 6 11 11 M S E 2.63 - (Alonso, C , M . 28.21 T. Suidan, et al. 1997) N o t Given 15 12 0 . 0 - 1 5 . 4 % General (Zarook, S. M . and A . A . A v g . 6.7 % Shaikh 1997) 1.6-18.1 % Zero order, diffusion A v g . 8.4 % limited 3.1 - 67.7 % Zero order, reaction A v g . 10.8 % limited 3.6 - 36.0 % 1 s t order reaction A v g . 12.6 % 2.8.5.2 Dynamic T o compete with conventional waste gas treatment methods, trickling biofilters must be able to provide long-term operational stability and consistent regulatory compliance (Diks, R. M . M . , S. P. P . Ottengraf, et al. 1994). Dynamic models should have the ability to predict the surge capacity o f trickling biofilter designs and validate their continuous compliance with air emission regulations. Again there is a lack o f consistent numerical evaluation o f model error. 28 One novel approach to dynamic modelling has been to consider the trickling biofilter packed bed as finite layers with biofilm subdivisions (Deshusses, M . A . , G . Hamer, et al. 1995b; Deshusses, M . A . , G . Hamer, et al. 1995a). Multiple substrate concentrations were assumed to be uniform within each layer and subdivision. Dynamic mass balances over each layer and subdivision provide for transient response to step changes in inlet concentration o f the two substrates. A s with the quasi steady-state models, dynamic model evaluations have found that increasing gas flowrate leads to increased model error (Zarook, S. M . , A . A . Shaikh, et al. 1997). These dynamic models are much more complex than their quasi-steady state counterparts. They incorporate a large number o f equations and parameters but may not provide increased model accuracy. In one case, a small number o f realistic assumptions gready reduced the model complexity without sacrificing much predictive power (Zarook, S. M . , A . A . Shaikh, et al. 1997). Table 2-8 summarises the complexity o f selected dynamic models. Table 2-8 Comparison of Dynamic Models for Biofilters Assumptions Parameters Equations Model Prediction Comments Reference 10 14 10 Satisfactory (Deshusses, M . A . , G . Hamer, et al. 1995b) N o t given 29 50 0.3 - 29 % A v g . 6.7 % Approximate model (Zarook, S. M . , A . A . Shaikh, et al. 1997) N o t given 29 230 0.0 - 22.6 % General model (but fewer) A v g . 5.2 % 29 Chapter 3 - Materials & Methods 3.1 Reactor Design T o alleviate any safety concerns related to the release o f hydrogen sulphide to the working environment, the trickling biofilter reactors were designed and built in our shops to fit inside the laboratory fume hood. This resulted in a relatively small and compact reactor design. The reactors incorporated co-current gas-liquid flow to minimise the back-diffusion o f H 2 S pollutant to the air stream prior to its exit from the column. H 2 S is sparingly soluble in water with mole fraction solubility in water o f 1.85xl0"3 at 298K and an H 2 S partial pressure o f 101.325 kPa. The mole fraction solubility (X,) o f H 2 S in water at temperature T (K) is given by the following equation (Gevantman, L . H . ): 3477 In JT. =-24.912 + + 0.3993 ln7* + 0.01577* i T Equation 3-1 H2S Solubility in Water T w o identical trickling biofilter reactors were built. See Figure 3-1 Schematic o f T w o Parallel Trickling Biofilters. They shared common compressed air and contaminant gas supplies. A l l tubing was 0.64 cm QA") or 1.27 cm (V2") O D polyethylene. Reactor columns were 15.2 cm O D / 1 4 . 0 cm I D (6" O D / 5 Y 2 " ID) and 45.7 cm (18") tall made from 0.64 cm (Vi") Plexiglas. They had a 10.1 cm (4") headspace at the top and a 10.1 cm (4") gas-liquid disengagement zone at the bottom. Packing space was 30.5 cm (12") high. Both reactors were piped in the same manner. A final packed scrubber was used to remove any remaining H 2 S from the reactor outlet air. The scrubber consisted o f a 1 L cylinder filled approximately 3A full with 0.64 cm ('A") Raschig rings and 0.1 M N a O H scrubbing solution. The reactor outlet gases were piped to the bottom o f the scrubber and the scrubbing solution was refilled as required. The reactors were not designed for complete hydrogen sulphide removal, as would be expected for an industrial design. Instead, they were built to allow a measurable H 2 S concentration at the outlet so that the kinetic rate constant could be calculated. 30 HOUSE AIR REACTOR 1 REACTOR 2 NaOH CO 6 I o vJ PACKING a: UJ 111 m Z) o CO CO X CD O PACKING LU m 00 => a o CO © Figure 3-1 Schematic of Two Parallel Trickling Biofilters 3. /. / Flow Control Compressed house air was controlled using a 0.64 cm QA") needle valve from Parker Fluid Connectors (Ravenna, OH) and measured with an Omega FL4512 rotameter (Laval, QC). The compressed air pressure was controlled with a Speedaire filter-regulator at 20 psig supply pressure to the control valves. Hydrogen sulphide gas (10%, balance Nj) was supplied by Praxair (Vancouver, BC) and controlled with an Omega FL-3GP-41ST-05ST Gas Proportioning rotameter. These gases were mixed at a T-junction prior to entering the reactor column. Carbon dioxide gas (Praxair, Vancouver) was metered using a micro-adjust needle valve from Parker and pre-diluted with air before mixing with the main air stream at a separate T-junction. 3.1.2 Column Packing The column packing used was Celite R-635, a sintered diatomaceous earth product provided by World Minerals Inc (Lompoc, CA). This material is made of rigid inorganic materials and provides an excellent surface for biofilm adhesion with a predictable pressure drop over a wide airflow range. Since this is a biologically inert material, it is not degraded over time by the microbes in the column. 31 3.1.3 Pressure Drop Pressure drop across the reactor packing was measured using manometers made from Plexiglas tubing and 0.1" manometer scales. The pressure drop across reactor packing for a range of airflow rates was recorded with dry packing prior to any biomass inoculation to serve as a baseline for later comparison. This provides an indication of biofilm growth as well as imminent plugging and channelling in the packing. Initial pressure drop results are reported in section 4.1 Pressure Drop Results. 3.2 Residence Time Distribution The residence time distribution (RTD) measurements were made with dry packing to prevent the C 0 2 from being absorbed by and then desorbed from the liquid media. This would have the effect of incorrectly increasing the apparent R T D variance. It was expected that the low nutrient media flowrates to be used during the kinetics experiments would have a negligible hold-up volume and thus would not significantly affect the mean residence time. The carbon dioxide gas was injected direcdy into the inlet air stream. The C O z in the outlet air stream was measured with a Horiba APBA-210 non-dispersive infrared (NDIR) C O , analyser. The 0 - 1 0 0 mV output signal from the N D I R analyser was recorded at 500 ms intervals on a computer using a CIO-EXP16 data acquisition board (Computer Boards) and custom software. Conversion of the output signal to concentration units (parts per million) was performed in the software. The following procedure was used to measure the RTD in the two trickling biofilter reactors. 3.2.1 Preparation The air flowrate in the reactor was set at the desired value; Table 3-1 below shows the air flowrates used. The air rotameters were calibrated for direct reading of S C F M values. The metric equivalents are shown here for reference. Table 3-1 Air Flowrates Used in RTD Experiments Standard ft3/min (SCFM) Standard m3/h 0.7 1.19 0.8 1.36 32 Standard ft-'/min (SCFM) Standard m 3 /h •1.0 1.70 1.2 2.04 1.4 2.38 1.6 2.72 1-8 3.06 , The C O , analyser output was set at approximately 50 ppm using the Zero calibration knob to establish a baseline reading, since calibration gases were not available. The data acquisition computer was readied with the data file name and data logging interval. The 5.0 mL gas syringe (Precision Sampling Corp., Baton Rouge, LA) was flushed three times with C O , (10%, balance N 2 ; Praxiar, Vancouver, BC) before filling with 5.0 mL for injecting into the reactor. The C 0 2 was held at a positive pressure of approximately 5 psig and the syringe was filled through a rubber septum. 3.2.2 Injection The C O , was injected through a rubber septum into the inlet air stream immediately upstream of the upper headspace. Data logging on the computer was started simultaneously with the injection. The C O , analyser's readout was monitored and the data logging was stopped when the readout returned to its baseline value for approximately three seconds. In rare cases, the readout did not return to its baseline value and the experiment was discarded. The R T D measurement was repeated six times at each air flowrate in both reactors. 3.2.3 Data Analysis A l l data analysis calculations were performed in a spreadsheet using Microsoft Excel. The C O , baseline value of approximately 50 ppm was corrected by subtracting the minimum reading in the data set from all readings. The baseline-adjusted concentration data were multiplied by the Att (500 ms) time interval and summed according to Equation 2-1 (page 13) to calculate the mean residence time X. The data were multiplied by At} and summed according to Equation 2-2 (page 13) to calculate the distribution variance a2. Equation 2-3 and Equation 2-4 (page 14) were used to calculate the Tanks in Series parameter N for each R T D experiment. Results, for X and N are reported in section 4.2 R T D Results. 33 3 . 3 Trickling Biofilter Operation 3.3.1 Nutrient Media The nutrient media used was based on A T C C Culture Medium 238 Thiobacillus Medium B. The original A T C C media formulation and the trickling biofilter media used are listed in Appendix 11.1 Nutrient Media. The media used to initially establish the biofilm was denoted S+C+ media (contains sulphur and carbon). Media used during the kinetic experiments was denoted S-C- media (does not contain sulphur and carbon). During all experiments, media was circulated at 500 mL/min using a Cole-Parmer 77300-00 peristaltic pump and controller system (Chicago, IL). This flow rate was deemed to provide adequate wetting and coverage of the packing for biofilm growth and maintenance. Media volume in the reservoir tanks was kept at approximately 2 L by removing excess liquid as required; usually 1.0-1.5 L / d . The reservoir tanks were located between the reactor liquid outlet and the recirculation pumps. Excess liquid was due to sodium hydroxide addition for pH control. Approximately half of the media was replaced once per day as one litre was removed (in addition to media removed for level control) and replaced with fresh media. 3.3.2pH Control Nutrient media pH was controlled by the addition of 0.2 M sodium hydroxide using two Cole-Parmer 7142 p H Control/Pump systems (Chicago, IL). The N a O H solution was made from N a O H pellets (Fisher Scientific, Nepean, ON). The Reactor 1 media was controlled at pH 5.0 and the Reactor 2 media was controlled at pH 2.5. 3.3.3 Biofilm Inoculum & Ke actor Start-Up The objective during trickling biofilter start-up was to establish a stable biofilm capable of consuming and oxidizing reduced sulphur compounds. A compost sample from a biofilter used for hydrogen sulphide and mixed TRS removal experiments (Wani, A . H . 1999) was taken and used as the initial microbial inoculum. This compost (4 x 25 g) was incubated in 1-L shake flasks with 250 mL S+C+ media for 18 h. The medium was decanted and filtered to remove compost particles, yielding approximately 1 L of gold coloured hazy compost filtrate. The liquid medium reservoir for each reactor was filled with 2 L fresh S+C+ media, 1 L distilled water, and 250 mL compost filtrate. 34 Table 3-2 below shows significant events and operating changes for both reactors during the 66-day start-up period. During the entire start-up, media recirculation rate was 500 mL/min, and 1 L media was replaced once per day. The media volume was supplemented with distilled water and/or additional media as required to replace evaporative losses. Table 3-2 Significant Events and Changes During Trickling Biofilter Reactor Start-Up Day , Air Flow (scfm) H2S Cone, (ppm) ' Notes 1 0.8 0 Begin reactor start-up; p H S P = 4.2 (both reactors) with 0.1 M NaOH. 3 0.8 0 Reactor pH rising; change to pH control with 0.1 M HCI, p H S P = 4.2. 10 0.8 50 Begin H 2 S feed. 12 0.8 50 Observe patches of white film and bumps on packing in both reactors; bumps resemble microbial colonies. 17 0.8 0 End H 2 S feed. 24 0.0 0 Shut down both reactors for cleaning and maintenance. 27 1.0 0 Continue reactor start-up. 30 1.0 0 Reactor p H dropping; change to p H control with 0.2 M N a O H , P H S P = 5.0. 36 0.0 0 Shut down both reactors for cleaning and maintenance. 51 1.0 0 Continue reactor start-up. 56 1.0 50 Begin acclimatizing reactors to H 2 S. 58 1.0 50 Begin using S-C+ media (contains no sulphur). 59 1.0 50 Observe white film forming on reactor packing. 62 1.0 50 H,S removal efficiency approx. 90% in both reactors. 63 1.0 50 Observe white film covering most reactor packing. 66 0.0 0 End reactor start-up. 35 3.4 Experimental Program The experimental program was designed to examine the removal kinetics of hydrogen sulphide at levels up to those found in typical Kraft pulp mill H V L C TRS emissions streams. The upper H,S limit of 300 ppm was selected since it is the midpoint of a typical 200 - 400 ppm TRS H V L C emissions stream. Carbon dioxide was added during the experiments at several different levels relative to the average C O , concentration in the atmosphere, approximately 300 ppm. The objective of the C O , addition was to increase the H,S removal rate by increasing the microbial growth rate. Without the supplementary carbon dioxide, atmospheric C O , was the only source of carbon available to the biofilm microbes for their metabolic functions: growth, reproduction, and cellular repair. The same air flowrates were used for all experiments as were used in the R T D experiments, with the exception of 1.19 m 3 / h (0.7 scfm) and 3.06 m 3 / h (1.8 scfm), which were not used. The low air flowrate corresponded to the minimum flow rate that could be reproduced reliably on the rotameters used, and the high air flowrate corresponded to the pressure drop limit of the manometers used. Experiments were performed in a random order to prevent the entry of systematic error into the removal kinetics analysis. Potential systematic effects include long-term biofilm growth (changing biofilm density) and microbial adaptation & mutation. During all experiments, the p H in Reactor 1 was controlled with a setpoint of 5.0 and the p H in Reactor 2 was controlled with a setpoint of 2.5. The experimental parameters are listed below in Table 3-3 with the actual experimental run order noted. The acclimation phase was deemed to be complete when the first of the two reactors reached a steady-state H,S outlet concentration. Steady-state was defined as no significant change in outlet concentration over the previous 12 hours. The acclimation phase was observed to be between 0 -72 h long. The H,S loading rate was increased when the first of the two reactors reached a new steady-state. Recall the H,S loading rate was increased by increasing the air flow rate at constant inlet concentration. 36 Table 3-3 Experimental Program Sequence H2S Cone, (ppm) C 0 2 Cone, (ppm) Experiment Results Section 50 0 1 5.1 100 0 2 5.5 200 0 3 5.9 100 650 4,5' 5.7 100 1500 6 5.8 50 300 7 5.3 50 150 8 5.2 50 1500 9 5.4 300 0 10 5.10 100 300 11 5.6 3.5 Gas Concentration Measurement Hydrogen sulphide concentrations at the inlet and outlet of each trickling biofilter reactor were measured with a Gastec gas-sampling pump (Gastec, Japan) and tubes. The disposable tubes were purchased from Levitt Safety (Vancouver, BC). The tubes are available in several measurement ranges. The 4L (range), 4M (range), and 4 H M (range) tubes were used in this study. A predetermined gas volume is drawn into the tubes and the tube packing changes colour. For the hydrogen sulphide tubes this colour change is white to brown as lead acetate (PbCH 2 COOH) is converted to lead sulphide (PbS) in the presence of hydrogen sulphide. The gas sampling tubes were used according to the manufacturer's instructions with the following exception: tubes were used for multiple samples as described here. To reduce experimental costs, the hydrogen sulphide measurement tubes were used for several successive readings by noting the reading from the 1 Experiments 4 & 5 were performed at the same conditions; #4 was terminated prematurely when the CO2 supply was depleted unexpectedly. 37 previous sample and subtracting this result from the current sample. The measured concentration was then divided by the number of 100 mL pump strokes used; generally V2, 1 or 2. Tubes were not used again after being open for more than 24 hours. With proper range selection, a single tube could be used for three or four gas samples. No significant error was expected using this method. The validity of using one tube for several readings was verified on several occasions by performing duplicate gas concentration measurements with both new tubes and previously used tubes. In all instances, the concentration measurements were equal providing the previously used tubes had not been opened for more than 24 h. Hydrogen sulphide concentration measurements were made by placing the open tip of the gas sampling tube directly inside the end of a 0.64 cm (%") O D permanently connected valved sample lines leading from the reactor inlet or outlet tube. Sample lines were purged for 1-2 minutes prior to sampling. 3.6 Data Analysis 3.6.1 Removal Efficiency Removal efficiency as defined in this work is the percentage of initial H,S that is removed by the trickling biofilter. It is often used as a rapid method to express the relative amount of contaminant gas removed by a trickling biofilter. It is calculated as: 77: f C ^ 1--^ v c o y x l 0 0 % Equation 3-2 H2S Removal Efficiency 3.6.2 H2S Loading Rate The H,S loading rate is the mass feedrate of H,S into the trickling biofilter per unit volume of packed-bed reactor. It was calculated as: C xF ^ _ ' w AIR V ' PACKING Equation 3-3 H2S Loading Rate 38 3.6.3 H2S Removal Rate The H,S removal rate is mass rate at which H,S is removed by the trickling biofilter per unit volume of packed-bed reactor. It is a more useful measure of the amount of H 2 S removed by the tricHing biofilter. It was calculated as: R_(C0-CN)xFMR v v PACKING Equation 3-4 H 2S Removal Rate 3.6.4 H2S Rate Constant The l s t-order kinetic rate constant for H 2 S removal was evaluated direcdy using Equation 2-5 Kinetic Rate Constant Estimate (page 14). The previously estimated mean gas residence time X and the number of tanks in series N from the R T D experiments at the same flow conditions and the H 2 S concentration data for each set of experimental conditions were used in the analysis. In examining this equation it is apparent that it is not applicable as the outlet concentration C N approaches zero, since the equation becomes undefined. In this case the actual fraction of reactor bed used to completely remove the H 2 S was unknown. This limited the analysis during some experiments as the H 2 S was completely removed by the trickling biofilters. The l s t-order reaction model was chosen as the reaction scheme based on the uncertainty of the limiting reaction step. If diffusion were limiting, the reaction order would be ls t-order. If the biological oxidation reaction were limiting, the reaction order would be either l s t-order or possibly 0-order at concentrations above the biological saturation concentration. Reaction orders are compared for 0-order, T'-order and 2nd-order reactions in section 6.3 First Order Assumption (below). It is shown there that the T'-order reaction model is the most reasonable of the three orders tested. 3.63 H2S Add Yield The acid yield as defined in this work is the molar quantity of sodium hydroxide required to neutralize the sulphuric acid produced by the biofilm microbes. This is synonymous with maintaining a constant pH in the reactor's liquid media. The following simplified reaction scheme describes this neutralisation: 39 H2S + 202 > 2H + + S0A 2 " (aq) 20H~ +2H+ >2H20 Equation 3-5 Complete hydrogen sulphide oxidation & subsequent neutralisation with sodium hydroxide The quantity of sulphuric acid produced by the biofilm microbes was calculated by measuring the quantity of N a O H solution required to maintain a constant p H in the reactor's liquid media. From Equation 3-5 we see that two moles of hydroxide are required to neutralize one mole of hydrogen sulphide. The acid yield was calculated as: cumulative NaOH added (mol) cumulative H2 S removed (mol) 1-1 C NaOH ' NaOH ~*~ 2^  ' NaOH 1 FAIR-{C0-CN)-At , ^ 1- > m„ Equation 3-6 H2S Acid Yield In dynamic or non-ideal situations, the acid yield Y may be greater or less than the stoichiometric value of two. A n acid yield Y < 2 indicates that H 2 S is being removed from the contaminated air stream but the oxidized sulphur (as sulphuric acid) is not being neutralized; i.e.: the sulphur is being stored by the microbes or deposited on the packing as elemental sulphur S°. A n acid yield Y > 2 indicates that more sodium hydroxide is being used to neutralize sulphuric acid than is required given the H 2 S removal rate; i.e. there is an additional acid load on the reactors from elemental sulphur that was previously stored by the biofilm microbes is oxidized. See Equation 2-7 Sulphur Compound Oxidation Reactions (page 17). The cumulative N a O H usage was used in this evaluation to average out any potential transient yield spikes that may result from the pH controllers' actions. 40 Chapter 4 - Residence Time Distribution Results 4.1 Pressure Drop Results Air flowrates in the reactors were varied from 1.2 — 3.4 m 3 /h (0.7 - 2.0 scfm) and the corresponding pressure drops were recorded. Some hysteresis was observed and this was attributed to measurement error and float sticking in the air rotameters. The manufacturer's stated accuracy for these rotameters is ±2.5% of full-scale. The observed hysteresis is well within the rotameters' accuracy as shown in Figure 4-1. Error in the pressure drop readings is ± 0.025 kPa (± 0.1 "WC). Reactor Pressure Drop Dry Packing 1.5 2.0 2.5 Air Flow Rate (m'/h) 3.0 4.0 • Reactor 1 • Reactor 2 Figure 4-1 Reactor pressure drop for Reactors 1 and 2 showing AP differences between reactors and hysteresis The pressure drop results show differences between the two reactor columns and this was attributed to differences in how the packing settled as it was dumped into the columns. This 41 difference in itself is not significant since the effect will be accounted for by the Tanks in Series parameter N , shown in section 4.2 R T D Results. 4.2 R T D Results Air flowrates in the reactors were varied from 1.2-3.1 m 3 / h (0.7 — 1.8 scfm) and the R T D was evaluated as described above. Typical residence time distribution curve plots for each reactor are shown in Figure 4-2 and Figure 4-3. These show the pronounced C O , concentration peak asymmetry that indicates significant deviation from plug flow in the gas phase. As expected, it also shows the curve peak location at progressively shorter residence times as the air flowrate increases. A l l R T D curves were repeated six times for each air flowrate. The mean residence time X and the variance CT were calculated for each curve according to Equation 2-1 and Equation 2-2. RTD Tracer Response Reactor 1 60 i — 0 20 40 60 80 100 120 Time (s) Figure 4-2 Reactor 1 R T D tracer response curves with peak asymmetry and tails due to back-mixing 42 RTD Tracer Response Reactor 2 80 70 0 1 0 20 30 4 0 50 60 Time (s) Figure 4-3 Reactor 2 R T D tracer response curves with peak asymmetry and tails due to back-mixing 43 Tanks in Series Model Analysis Reactor 1 Mean Residence Time Figure 4-4 Reactor 1 mean residence time x; EBRT calculation is shown for reference 44 Tanks in Series Model Analysis Reactor 2 Mean Residence Time Figu re 4-5 Reactor 2 m e a n residence t ime i; E B R T ca lcu la t ion is shown for reference 45 Tanks in Series Model Analysis Reactor 1 Number CSTR's 1.5 2.0 Air Flow Rate (m3/h) 3.5 Figure 4-6 Reactor 1 number of CSTR's N with mean value of 7.374 46 Tanks in Series Model Analysis Reactor 2 Number CSTR's 10 4 • • s D o • • • B B • B o • a |N=6.446±1.116| • • • • • o B • • • • • • • • n o • 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Air Flow Rate (m3/h) F i g u r e 4-7 Reactor 2 number o f C S T R ' s N w i t h m e a n value o f 6.446 The calculated mean residence time X and the Tanks in Series parameter N for the range of air flow rates studied are shown for Reactor 1 and Reactor 2 in Figure 4-4 and Figure 4-5, respectively. As expected the reactors had lower X at increasing air flowrates. This decrease is not linear due to the increasing pressure drop at higher flow rates. One X and one N value were calculated from each R T D tracer response curve. The X values were correlated empirically and the correlation was used in subsequent kinetic analysis. The N values appeared to be represented by a horizontal line and as such the mean values of N were taken as constants in the kinetic analysis. Results are shown in Figure 4-6 and Figure 4-7 for Reactor 1 and 2 respectively. The reported errors for N are + one standard deviation. These results are summarised in Table 4-1, where F is the air flowrate (m3/h). 47 Table 4-1 Summary of RTD Parameter Results Reactor • Empirical Mean RT Mean Tanks in Relation (s) Series (N) Reactor 1 28.350-F1 0 9 3 6 7.374±1.101 Reactor 2 29.851-F'1 1 0 7 7 6.446±1.116 The empty bed residence time (EBRT) is a parameter commonly used in the literature to describe the gas residence time in biofilters and trickling biofilters. It is calculated simply as the reactor volume divided by volumetric gas flow rate and is shown on the previous figures for comparison only. For these reactors the E B R T does not differ significantly from the experimental X since the packed portion of the reactors is relatively small in comparison to the reactor volume. The E B R T makes no allowance for reactor packing and is expected to differ significantly from X for larger reactors. 48 Chapter 5 - Hydrogen Sulphide Removal Results Several experiments required an acclimation period to allow the H 2 S outlet concentration to stabilize. If this acclimation period was required, it was generally between 24 h and 48 h long. Concentration data taken during the acclimation periods were removed from the kinetic analysis. In general it was noted that the acclimation period was shorter when experiments were performed in rapid succession. Each experiment was performed at a set target H 2 S inlet concentration. Increasing the air flowrate through the reactor columns, and thus reducing the mean residence time of the H 2 S in the reactors, increased the H 2S loading rate. As noted above Reactor 1 was controlled at pH 5.0 and Reactor 2 at pH 2.5. The differences in H,S rate constant attributed to this p H difference are explained below. Carbon dioxide was added to the reactors during selected experiments at varying concentrations as an additional source of carbon for the biofilm microbes. The experiments were performed in a random order to reduce or eliminate systematic errors. The experiments are presented here in a logical, rather than chronological, order to facilitate presentation and discussion. Table 3-3 (above) shows the actual chronological experiment order. 5.1 50 p p m H 2 S + 0 p p m C 0 2 Results The previous experiment was the Reactor Start-Up at 50 ppm H 2 S; see section 3.3.3 Biofilm Inoculum & Reactor Start-Up and Table 3-2. The acclimation period for this experiment was approximately 36 h. Following the acclimation, the H 2 S outlet concentration in Reactor 1 was 0 ppm thus giving 100% H 2 S removal and a maximum H 2 S removal rate of 34.0 g/m 3h; see Figure 5-1. The H 2 S rate constant could not be evaluated using the Tanks in Series model due to the C N value of 0; the model becomes undefined at this point. Reactor 2 had an H,S removal efficiency of 76.8%. The maximum H 2 S removal rate was 21.1 g/m 3h as shown in Figure 5-2 and the H 2 S rate constant was 0.136 ± 0.017 s"1 as shown in Figure 5-3. 49 Figure 5-4 shows the acid yield results from this first experiment and provides a good example of the expected results from this data. The acid yield surge and subsequent levelling after 36 h corresponds to the acclimation phase noted above. The acid yield for Reactor 1 following the acclimation phase is 2 mol/mol as expected. Reactor 1 at this point was operating at 100% H 2 S removal and therefore some portion of the biofilm microbes would have been under starvation conditions. No S° storage would have been possible at this stage. Reactor 2 was operating at 76.8% removal efficiency but the acid yield was approximately 1 mol/mol. This indicates that some fraction of the H 2 S was removed from the air stream but was not oxidized to sulphuric acid. Therefore this suggests that the biofilm microbes in Reactor 2 were storing this unaccounted sulphur as S° or some other sulphur species. Results for this experimental series are shown below. H2S Removal Efficiency 40.0 35.0 25.0 " E 20.0 15.0 10.0 • Reactor 1 o Reactor 2 Target Load Figure 5-1 H2S removal efficiency at 50 ppm H2S and 0 ppm c o 9 50 H2S Removal Rate 40.0 -, 35.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 H2S Load (g/m3h) • Reactor 1 • Reactor 2 Complete Removal | Figure 5-2 H2S removal rate at 50 ppm H2S and 0 ppm C0 2 51 H2S Rate Constant 15.0 20.0 25.0 H2S Load (g/m'h) 40.0 • Reactor 1 • Reactor 2 Figure 5-3 H2S rate constant at 50 ppm H2S and 0 ppm C0 2 52 H2S Acid Yield CD D dp 40 Time (h) 50 • Reactor 1 D Reactor 2 Figure 5-4 H2S acid yield at 50 ppm H2S and 0 ppm C0 2 53 5.2 50 ppm H 2 S +150 ppm C 0 2 Results The previous experiment was conducted at 50 ppm H 2 S and 300 ppm CO, ; see section 5.3. The acclimation period for this experiment was negligible. A t these gas inlet concentrations, Reactor 1 maintained 100% removal efficiency to an H 2 S loading rate of approximately 23 g/m 3h; see Figure 5-5. Above this load, the H 2 S removal rate was constant at 23.1 + 2.5 g/m 3h as shown in Figure 5-6. The mean H 2 S rate constant was 0.122 + 0.026 s"1 as shown in Figure 5-7. Reactor 2 never achieved 100% H 2 S removal at these conditions; see Figure 5-5. The H,S removal rate was 12.7 + 3.6 g/m 3h for the entire experiment except at the highest loading, when the removal rate increased to approximately 20 g/m 3h as shown in Figure 5-6. The H 2 S rate constant, was 0.046 + 0.013 s"1 as shown in Figure 5-7. The acid yield data in Figure 5-8 shows transient behaviour in both reactors up to approximately 60 h. The acid yield stabilizes at approximately 3.8 mol/mol in Reactor 1 and approximately 3.2 mol/mol in Reactor 2. These acid yield values are significantly greater than any recorded from the earlier experiments. Recall this experiment was performed near the end of the experimental program; see Table 3-3 Experimental Program Sequence. The only plausible explanation that exists for such high acid yield values is the continued and increasing reliance of the biofilm microbes on elemental sulphur that had been stored during earlier experiments. Acid yield values of this magnitude were not expected. In this experiment Reactor 1 shows a classic example of the expected biofilter behaviour as the pollutant load increased above the removal capacity of the biofilter. The removal rate increases with the load at 100% removal efficiency until a certain maximum removal rate is achieved. At this point the removal efficiency decreases but the removal rate remains constant. This is generally considered to be a limitation of the biofilm microbes. The microbes can only consume and oxidize the pollutant at a certain rate. To increase this rate would require a greater overall quantity of microbes, or different environmental conditions under which the existing microbes could operate more efficiently or effectively. 54 H2S Removal Efficiency 90% 70% 60% 0 • 50% 40% $ 30% 20% 10% 0% • m • di • o • • a • • a • • • • • • • • 4- 5.0 20 40 60 80 100 120 140 160 Time (h) • Reactor 1 • Reactor 2 Target Load | 200 Figure 5-5 H2S removal efficiency at 50 ppm H2S and 150 ppm c o 2 55 H2S Removal Rate 45.0 45.0 • Reactor 1 • Reactor 2 Complete Removal | Figure 5-6 H 2S removal rate at 50 ppm H 2S and 150 ppm C O z 56 H2S Rate Constant 20.0 25.0 H2S Load (g/m3h) • Reactor 1 • Reactor 2 I 40.0 45.0 F i g u r e 5-7 H 2 S rate constant at 50 p p m H 2 S and 150 p p m C O z 57 H2S Acid Yield 6.0 • • Q • • of a t>aaa 100 Time (h) 120 • Reactor 1 o Reactor 2 140 160 180 200 Figure 5-8 H2S acid yield at 50 ppm H2S and 150 ppm C0 2 58 5.3 50 ppm H 2 S + 300 ppm C 0 2 Results The previous experiment was conducted at 100 ppm H 2 S and 1500 ppm CO, ; see section 5.8. In Reactor 1, the acclimation period for this experiment was approximately 70 hours. The H 2 S removal efficiency increased from 60% to 98% during this time while the H 2 S loading rate was held constant at 19.4 g/m 3h. At a loading rate of 24.3 g/m 3h, the removal efficiency remained constant at approximately 95%. At progressively higher loading, the removal efficiency gradually decreased to approximately 65% at an H 2 S loading rate of 38.8 g/m 3h; see Figure 5-9. As with the experiment 5.2, the maximum H 2 S removal rate achieved was approximately 23 g/m 3h; in this experiment it was 22.8 + 2.1 g/m 3h as shown in Figure 5-10. The average H,S rate constant for loading greater than 25 g/m h was 0.143 ± 0.027 s"1 as shown in Figure 5-11. At loading less than this level, the H 2 S removal efficiency was too high to allow reliable rate constant calculation using Equation 2-5 Kinetic Rate Constant Estimate. Recall this equation becomes undefined as C N approaches zero. Reactor 2 had a relatively constant H 2 S removal efficiency of approximately 50% throughout the experiment; see Figure 5-9. The H 2 S removal rate tended to increase with the H 2 S loading rate with an average of 15.0 ± 4.3 g/m 3h. Figure 5-10 shows the H 2 S removal rate in Reactor 2 increasing with H 2 S loading rate but without reaching an apparent maximum removal rate. The H 2 S rate constant was 0.058 + 0.019 s"1 as shown in Figure 5-11. Acid yield data from this experiment show the acclimation period to be even longer than shown by the H,S removal efficiency data. It is apparent that the reactors required approximately 100 hours to stabilize metabolically, rather than the 70 hours determined from the H 2 S removal efficiency in Figure 5-9. The acclimation transients in this case are similar for both reactors and show evidence of dynamic overshoot as the biofilm comes to equilibrium. See Figure 5-12. Since the biofilm microbes were exposed to a higher H,S concentration in the previous experiment (section 5.8), their metabolism was operating at a level that could not be sustained by the now-lower H 2 S concentration. The microbes required additional energy, in the form of elemental sulphur, as they became acclimatized to their new H 2 S inlet concentration. This would explain the initial spike and then drop in acid yield up to approximately 20 hours. During the acclimation 59 period observed between 20 - 70 h., the acid yield was stable at approximately 2.3 mol/mol in Reactor 1 and approximately 2.0 mol/mol in Reactor 2. Following the acclimation period the H,S loading rate was increased from approximately 20 g/m 3h to 25 g/m 3h. Subsequent H 2 S loading rate increases did not translate to increased H,S removal rates, as exhibited by the maximum H 2 S removal rate of approximately 23 g/m 3h. This helps explain the increase in acid yield in both reactors after 80 h to approximately 2.8 mol/mol in Reactor 1 and approximately 2.5 mol/mol in Reactor 2. The biofilm microbes were already oxidizing the incoming H,S at their maximum capacity, and that was not enough to satisfy their energy requirements as set in the previous experiment at 100 ppm H,S. Therefore the microbes oxidize S° from their storage reserves to satisfy their requirements. The long acclimation period of approximately 70 hours can be explained by the chronological order in which the experiments were performed. This experiment was performed following one with H 2 S inlet concentration of 100 ppm and C O , inlet concentration of 1500 ppm. At the higher H,S concentration, the biofilm microbes would be converting the H,S to an intermediate sulphur product such as elemental sulphur and storing this as an energy source for future requirements. During the next experiment at a lower H,S concentration, the microbes could preferentially consume the elemental sulphur rather than the incoming H,S. Only when the easily accessible S° was consumed would the microbes begin to consume the incoming H,S. 60 H2S Removal Efficiency 100% - 45.0 90% 50% -J 40% 30% 20% $ 10% 0% • • • • 20 • • • • • 80 100 Time (h) • • • • • • • • Reactor 1 • Reactor 2 Target Load r • • • • a • 40.0 • 35.0 25.0 120 140 160 180 200 Figure 5-9 H2S removal efficiency at 50 ppm H2S and 300 ppm c o 2 61 H2S Removal Rate 45.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 H2S Load (g/m3h) | • Reactor 1 • Reactor 2 Complete Removal | Figure 5-10 H2S removal rate at 50 ppm H2S and 300 ppm C0 2 62 H2S Rate Constant 0.30 0.25 -0.10 4 0.05 A • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • D • • • • • • B • • • o 0° • • 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 HZS Load (g/m3h) | • Reactor 1 • Reactor 2] Figure 5-11 H2S rate constant at 50 ppm H2S and 300 ppm C0 2 63 H2S Acid Yield • c P ncP • cPrxi • QUID • • rn qmD 100 Time (h) 140 160 180 200 • Reactor 1 • Reactor 2 Figure 5-12 H2S acid yield at 50 ppm H2S and 300 ppm C0 2 64 5.4 50 p p m H 2 S + 1500 p p m C 0 2 Results The previous experiment was conducted at 50 ppm H 2 S and 150 ppm C 0 2 ; see section 5.2. The acclimation period for Reactor 1 in this experiment was approximately 50 hours and the H 2 S removal efficiency reached a maximum of approximately 90% at this point. As before, the removal efficiency decreased gradually to 70% as the H 2 S loading rate was increased; see Figure 5-13. The H 2 S removal rate increased as the loading increased. This was different from the other experiments at lower C O , concentrations. Figure 5-14 shows the removal rate increased from approximately 16 g/m 3h at a loading of about 20 g/m 3h to approximately 32 g/m 3h at a loading of 43 g/m 3h. The maximum H 2 S removal rate achieved here was greater than the other experiments at 50 ppm H,S. However, the mean H,S removal rate was in the same range as the other 50 ppm H,S experiments at 22.8 ± 3.8 g/m 3h. The average H 2 S rate constant following the acclimation period was 0.130 ± 0.015 s"1 as shown in Figure 5-15. This was consistent with the other H,S experiments at 50 ppm. In Reactor 2, the removal efficiency was between 40%-60% for the entire experiment with a mean of 46.3 ± 10.6%; see Figure 5-13. The H,S removal rate increased steadily from 7 to 27 g/m 3h as the loading rate was increased from 19.4 to 43 g/m 3h. The mean H,S removal rate was 13.5 ± 5.7 g/m 3h as shown in Figure 5-14. The resulting H,S rate constant also showed a gradual increase from 0.023 s"1 to 0.111 s"1 as the loading rate was increased over this same range. The mean H,S rate constant was 0.051 + 0.026 s"1 as shown in Figure 5-15. The acid yield profiles in Figure 5-16 show a gradual increase over 70 h to stabilize at approximately 3.8 mol/mol in Reactor 1 and approximately 3.0 mol/mol in Reactor 2. Again the acclimation phase indicated by the acid yield is longer than that indicated by the removal efficiency. Chronologically, this was the second consecutive experiment with acid yield results of this magnitude. As in the previous experiment in section 5.2, the only plausible explanation for such high acid yield results was the continued and increasing reliance of the biofilm microbes on elemental sulphur that had been stored during earlier experiments. Recall this experiment was near the end of the experimental program; see Table 3-3 Experimental Program Sequence. 65 The increasing contaminant removal rate result is somewhat different from the classical result of complete removal followed by a constant maximum removal rate as the loading rate increases. In this experiment we observe the removal rate increasing, but at less than complete removal (<100%). One plausible explanation for the increasing H,S removal rate in both reactors during this experiment is increased microbial growth, both microbial population density and microbial biofilm surface area, due to the greater C O , addition. The C O , was added to provide a carbon source for autotrophic microbial growth. A n implicit assumption during these experiments was a constant microbial population. But the microbes require a carbon source for maintenance purposes. This must be provided by atmospheric C O , or by additional CO, . With the increased C 0 2 level in this experiment, the microbial growth may have been no longer carbon limited. The increasing population would consume more H,S and thus increase the H,S removal rate. H2S Removal Efficiency 80 , 100 120 Time (h) 180 | 25.0 -a 4- 15.0 H 200 • Reactor 1 • Reactor 2 Target Load [ Figu re 5-13 H 2 S removal efficiency at 50 p p m H 2 S and 1500 p p m C 0 2 66 H2S Removal Rate 45.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 H 2 S Load (g/m3h) | • Reactor 1 • Reactor 2 Complete Removal | Figure 5-14 H2S removal rate at 50 ppm H2S and 1500 ppm C0 2 67 H2S Rate Constant 20.0 25.0 H2S Load (g/m3h) j • Reactor 1 • Reactor 2 I 45.0 Figure 5-15 H2S rate constant at 50 ppm H2S and 1500 ppm CO, 68 H2S Acid Yield 6.0 5.0 4 4.0 in i p o 2.0 -I ^ D am a mtJm, 0 20 40 60 80 100 120 140 160 180 200 Time (h) • Reactor 1 • Reactor 2 Figure 5-16 H2S acid yield at 50 ppm H2S and 1500 ppm C0 2 69 5.5 100 ppm H 2 S + 0 ppm C O z Results The previous experiment was conducted at 50 ppm H 2 S and 0 ppm CO,; see section 5.1. There was no acclimation period for either reactor in this experiment. The H,S removal efficiency in Reactor 1 started at 100% and gradually decreased to approximately 64% as the H,S loading rate was increased from 38.8 g/m 3h to 87.4 g/m 3h over 5 days; see Figure 5-17. Below 95% removal efficiency, the H,S removal rate was fairly constant at 50.1 ± 6.7 g/m 3h as shown in Figure 5-18. This maximum removal rate was higher than that observed in the 50 ppm H,S experiments. The H 2 S rate constant was 0.129 ± 0.027 s'1 as shown in Figure 5-19. The outlying H,S rate constant points for Reactor 1 in Figure 5-19 (i.e.: > 0.25 s"1) were due to near-complete H,S removal, and were removed from the rate constant mean calculation. Recall the calculated rate constant becomes undefined as C N approached 0; see Equation 2-5 Kinetic Rate Constant Estimate. The initial H,S removal efficiency in Reactor 2 was 87% and this decreased to approximately 30% as the H,S loading rate was increased; see Figure 5-17. The maximum H,S removal rate was constant at 33.4 ± 7.8 g/m 3h as shown in Figure 5-18. The H,S rate constant was also stable at 0.079 ± 0.028 s"1 as shown in Figure 5-19. Acid yield results from this experiment show nearly constant acid yield in both reactors; see Figure 5-20. The lack of acclimation period is also seen in the acid yield plot as there was no apparent spike or valley. The acid yield in Reactor 1 is constant at approximately 1.6 mol/mol. The removal efficiency plot in Figure 5-17 indicates that the removal efficiency was less than 100% after the H,S loading rate was first increased and therefore the microbes were not subjected to starvation conditions. Since the acid yield is less than the stoichiometric value of 2 mol/mol, we can conclude that biofilm microbes are removing a portion of the incoming H,S but not oxidizing it to sulphuric acid. The missing sulphur is presumably being stored as S°. The acid yield in Reactor 2 is also constant but at a lower value of approximately 1.1 mol/mol. In this case roughly half of the H,S that is being removed from the contaminated air stream is not oxidized to sulphuric acid and subsequently neutralized. The H,S removal rate at 100 ppm H,S is roughly double that of the 50-ppm experiments while the H 2 S rate constant is roughly the same. This suggests that the contaminant removal rate is not fixed, 70 given a certain microbial density and set of treatment conditions, as is often assumed for biofilters. and trickling biofilters. This observation supports the T'-order kinetics assumption required to use the Tanks in Series model chosen. It may be that the H,S inlet concentration is not the factor limiting H,S removal as is usually assumed given the poisonous qualities of H,S. This apparent inlet concentration dependency suggests that the H 2 S removal rate is primarily mass-transfer limited instead of reaction rate limited. If this were indeed the case then the rate constant would be equal to the diffusion rate of hydrogen sulphide through the liquid film, which can be calculated. It is also possible that the microbes enter a different metabolic state at higher H 2 S concentrations; one where they may shift metabolic pathways from a survival mode to a mode where they consume and store energy for future needs. H2S Removal Efficiency 100% 50% 4 0 % 20% 0 % 60 80 Time (h) 100 120 • Reactor 1 • Reactor 2 Target Load [ Figure 5-17 H2S removal efficiency at 100 ppm H2S and 0 ppm CO, 100.0 90.0 + 80.0 70.0 60.0 | T3 CO 50.0 fj in I 40.0 2, c? CO H - 30.0 - - 20.0 10.0 0.0 140 71 H2S Removal Rate 120.0 T 100.0 0.0 20.0 40.0 60.0 80.0 100:0 120.0 H2S Load (g/m3h) • Reactor 1 • Reactor 2 Complete Removal Figure 5-18 H2S removal rate at 100 ppm H2S and 0 ppm C0 2 72 H2S Rate Constant 60.0 H 2S Load (g/m3h) I • Reactor 1 • Reactor 2 I 80.0 100.0 Figure 5-19 H2S rate constant at 100 ppm H2S and 0 ppm C0 2 73 H2S Acid Yield 3.5 H ft] • oo rjp 60 80 Time (h) I • Reactor 1 • Reactor 2 I 100 120 140 Figure 5-20 H2S acid yield at 100 ppm H2S and 0 ppm CO z 74 5.6 100 ppm H2S + 300 ppm C0 2 Results The previous experiment was conducted at 300 ppm H 2 S and 0 ppm CO, ; see section 5.10. The acclimation period for this experiment was approximately 5 h. The H,S removal efficiency in Reactor 1 was uniform throughout the experiment at 43.3 ± 11.0% as shown in Figure 5-21. This resulted in the H,S removal rate increasing from 14.2 to 44.4 g/m 3h as the H,S loading rate increased from 38.6 to 82.7 g/m 3h respectively; see Figure 5-22. The mean H,S removal rate was 27.1 + 13.8 g/m 3h. As a result the H 2 S rate constant exhibited a steady increase and was not uniform over the H 2 S loading rate range examined. The H,S rate constant varied from 0.024 to 0.085 s"1 as the H,S loading rate increased from 38.6 to 82.7 g/m 3h. The mean H,S rate constant was 0.048 ± 0.027 s"1 as shown in Figure 5-23. Reactor 2 exhibited a similar pattern of results during this experiment. The H,S removal efficiency was uniform at 23.9 + 8.8%; see Figure 5-21. The H,S removal rate increased from 7.6 to 26.1 g/m 3h as the H,S loading rate increased from 36.0 to 78.8 g/m 3h; the mean was 14.5 ± 9.1 g/m 3h as shown in Figure 5-22. The H,S rate constant varied from 0.011 to 0.042 s"1 over the same H,S loading rate range. The mean H,S rate constant was 0.022 ± 0.015 s"1 as shown in Figure 5-23. This experiment immediately followed one at a greater H,S inlet concentration; see section 5.10. Once again it is plausible that the biofilm microbes had been accustomed to a higher metabolic rate and required a higher rate of energy consumption to survive. In this case they would again consume S° to make up the energy difference between the H,S that was available to them and the H,S loading rate they were accustomed to. This would result in an acid yield Y > 2 mol/mol. Such behaviour is supported by the acid production yield data. Figure 5-24 shows that this was exactly the case for both reactors in the first 3A of the experiment. The gradual decrease in acid yield indicates the metabolic rate of the biofilm microbes is slowly adjusting to the lower H,S loading rate compared to the previous experiment at a much greater inlet concentration. The initial acid yield in Reactor 1 was 5.5 mol/mol and it gradually decreased to approximately 3.5 mol/mol. In Reactor 2 the initial acid yield was 4.0 mol/mol gradually decreasing to about 1.0 mol/mol. These initial acid yield values were the largest found during this work. 75 This pattern of uniformly increasing H 2 S removal rate was unusual and different from other experiments reported above. It is possible that the trickling biofilter microbial biofilm was still in the process of adapting & acclimatizing to the H,S loading rate. However, this experiment followed immediately after one with an H,S inlet concentration of 300 ppm, significandy higher than this experiment at 100 ppm. If anything, the reactors should have had an initial H 2 S removal efficiency of 100%. Another possible explanation for this unusual set of results is that during the previous experiment at 300 ppm H 2 S, the biofilm microbes partially oxidized the H 2 S and stored the intermediate sulphur compounds for future energy needs. Then when exposed to the lower H,S concentration of this experiment, the microbes preferentially used the stored sulphur to supply the energy requirements of the microbial population. This would result in lower observed H 2 S removal rates and rate constants since the H 2 S would pass through the reactors as the microbes consume the partially oxidized sulphur compounds first. H2S Removal Efficiency 40% 30% 20% 10% 0% 10 • • • o • D • • 20 30 Time (h) 40 • Reactor! o Reactor 2 Target Load • • • • D I • • ° • 80.0 70.0 60.0 — "E 50.0 -a CO o _1 + 40.0 £ to 30.0 H 20.0 10.0 0.0 60 Figure 5-21 H2S removal efficiency at 100 ppm H2S and 300 ppm C0 2 76 H2S Removal Rate 120.0 100.0 0.0 20.0 40.0 60.0 80.0 100.0 120.0 H 2 S Load (g/m3h) | ~ » Reactor 1 • Reactor 2 Complete Removal | Figure 5-22 H2S removal rate at 100 ppm H2S and 300 ppm c o 2 77 H2S Rate Constant 60.0 H2S Load (g/m3h) 100.0 120.0 • Reactor 1 • Reactor 2 Figure 5-23 H2S rate constant at 100 ppm H2S and 300 ppm c o 2 78 H2S Acid Yield 18.0 16.0 14.0 12.0 co - 5 1 0 0 o o 'I P 8.0 6.0 4.0 2.0 4 0.0 • • • • D o D • • • o o 0 o • ° O D • • D D 10 20 30 Time (h) 40 50 • Reactor 1 o Reactor 2 Figure 5-24 H 2S acid yield at 100 ppm H 2S and 300 ppm C0 2 79 5.7 100 ppm H2S + 650 ppm C0 2 Results This experiment was performed twice; the first was ended prematurely due to a lack of CO, , and the second was also ended prematurely for personal reasons. The two experiments have been reported separately below. The previous experiment was conducted at 200 ppm H,S and 0 ppm CO, ; see section 5.9. The first experiment at 100 ppm H,S and 650 ppm C O , showed no acclimation period in either reactor. The H,S removal efficiency in Reactor 1 remained relatively constant throughout the experiment at 71.7 + 8.4%. There was no apparent change in H,S removal efficiency due to the lack of C O , that occurred at approximately 50 h; see Figure 5-25. The H,S removal rate showed a steady increase from approximately 25 to 39 g/m 3h as the H,S loading rate increased from 31 to 52 g/m 3h. The mean H,S removal rate was 28.8 + 6.1 g/m 3h as shown in Figure 5-26. The Ff,S rate constant showed no apparent trend and was scattered about a mean of 0.078 ± 0.021 s"1 as shown in Figure 5-27. In Reactor 2 the H 2 S removal efficiency also remained relatively constant at 70.1 ± 10.6% as shown in Figure 5-25. The H,S removal rate increased uniformly from 24.1 to 36.4 g/m 3h as the H,S loading rate increased from 31.1 to 48.5 g/m 3h. The mean H,S removal rate was 28.8 ± 5.0 g/m 3h; see Figure 5-26. The mean H,S rate constant was 0.072 + 0.020 s"1 as shown in Figure 5-27. The acid yield data in Figure 5-28 show transients in both reactors up to about 48 h followed by a stable acid yield of approximately 2.1 mol/mol in Reactor 1 and approximately 2.0 in Reactor 2. Both the upward-transients and the stable acid yield Y ~ 2 support the S° storage hypothesis. Assuming the microbes were fully acclimatized to the H,S inlet concentration of the chronologically previous experiment (section 5.9) and then were subjected to a lower H,S concentration in this experiment, they would initially consume stored S° to compensate for the lower energy supply before stabilizing at an acid yield of Y ~ 2 where all removed H,S is oxidized to sulphuric acid. In this experiment the initial acid yield in Reactor 1 was approximately 2.3 mol/mol before peaking at approximately 2.7 mol/mol and subsequently stabilizing at approximately 2.1 mol/mol. The initial acid yield in Reactor 2 was approximately 1.5 mol/mol 80 before rising to approximately 2.0 mol/mol. In both reactors, the initial acid yield trend was upward as the microbes consumed stored S° to compensate for the lower H 2 S concentration. This was the only experiment where the H 2 S removal rate was approximately equal for both Reactor 1 and Reactor 2. The range of H 2 S loading rate studied in this experiment was lower than experiments reported above, and this was due to the early termination of the experiment. It seems evident that the reactors had not yet reached their maximum H,S removal rate levels. Had the experiment been taken to completion, we might have expected to see results patterns closer to those reported above in most other experiments. The H,S removal rate and H 2 S rate constant values are below those seen in previously reported experiments. This suggests the reactors have not yet reached their maximum operating removal rate. However, we would therefore expect the H,S removal efficiency to be close to 100%. Another possible explanation for these unusual results is that the C O , was somehow aiding the biofilm microbes to consume the H,S by providing a readily available carbon source. This carbon source would be required by the microbes to enter a growth mode instead of a survival-storage mode. 81 H2S Removal Efficiency 60 80 Time (h) • Reactor 1 • Reactor 2 Target Load | Figure 5-25 H 2 S removal efficiency at 100 ppm H 2 S and 650 ppm C 0 2 82 H2S Removal Rate 60.0 -, 0.0 10.0 20.0 30.0 40.0 50.0 60.0 H2S Load (g/m3h) | • Reactor 1 • Reactor 2 Complete Removal ] Figure 5-26 H2S removal rate at 100 ppm H2S and 650 ppm c o 2 83 H2S Rate Constant 30.0 H 2S Load (g/m3h) • Reactor 1 o Reactor 2 I 40.0 50.0 60.0 Figure 5-27 H2S rate constant at 100 ppm H2S and 650 ppm CO, 84 H2S Acid Yield ^ n • • CT • D • • 0.5 A 0.0 -I , , , , , , 1 0 20 40 60 80 100 120 140 Time (h) ["• Reactor 1 • Reactor 2~| Figure 5-28 H2S acid yield at 100 ppm H2S and 650 ppm C0 2 The second experiment at 100 ppm H 2 S and 650 ppm C O , again had no acclimation period. This experiment was also run with no C O , during the second half in order to test the C 0 2 effect, or rather the lack of effect, as seen in the previous experiment. In this experiment Reactor 1 again showed a relatively constant H,S removal efficiency at 54.4 + 15.2% as shown in Figure 5-29. The Ff,S removal rate varied between 9.4 and 41.8 g/m 3h as the H,S loading rate varied from 24.3 to 63.1 g/m 3h. The target H,S loading rate was 48.5 g/m 3h for the entire experiment; the loading rate variations were due to inlet concentration variations. The mean H,S removal rate was 24.5 + 10.2 g/m 3h as shown in Figure 5-30. There was no discernible trend in the H,S rate constant. It was found to be 0.056 ± 0.023 s"1 as shown in Figure 5-31. Reactor 2 had an H,S removal efficiency of 32.8 + 11.5% as shown in Figure 5-29. The H,S removal rate varied between 6.3 and 34.5 g/m 3h as the H,S loading rate varied from 36.4 to 60.7 g/m 3h. The target H,S loading rate remained 48.5 g/m 3h. The mean H,S removal rate was 15.3 ± 85 8.1 g/m 3h as shown in Figure 5-30. The H,S rate constant was found to be 0.026 ± 0.012 s"1 as shown in Figure 5-31. Figure 5-32 shows the acid yield in both reactors was relatively stable throughout the experiment following an initial transient period. Reactor 1 had an acid yield of approximately 2.2 mol/mol while Reactor 2 had an acid yield of approximately 1.2 mol/mol. At this point Reactor 1 may have still been consuming stored S° to compensate for the lower H 2 S inlet concentration compared to that reported in section 5.9. By the same token Reactor 2 should have had an acid yield of Y ~ 2, but that was not the case. The dramatic drop to an acid yield Y = 1.2 mol/mol may indicate that the Reactor 2 microbes were forced, by their harsher environment at pH 2.5, to more quickly return to a survival mode and begin again to store S°. It was possible the. microbes in Reactor 1 remained in a growth mode and were not required to store elemental sulphur. We observe that the performance level of Reactor 2 has returned to roughly half that of Reactor 1. As in the first experiment at 100 ppm H,S and 650 ppm C 0 2 , the range of H 2 S loading rate studied was limited. As above, it's possible that the reactors had not reached their maximum H,S removal capacity due to the H 2 S loading rate limitations. In the first of these two experiments, the C O , was not added during the second half because of an empty gas cylinder at approximately 50 h. No apparent change in H,S removal rate or H,S removal efficiency was observed as a result of this lack of CO, . The second of these two experiments was an attempt to confirm that the additional C O , had no effect on the H,S removal at these gas concentration levels. In the second experiment, at a constant H,S loading rate, there was no significant difference seen between the H,S removal with supplementary C O , and the H,S removal without supplementary CO, . 86 H2S Removal Efficiency • Reactor 1 • Reactor 2 Target Load Figure 5-29 H 2S removal efficiency at 100 ppm H 2S and 650 ppm C0 2 87 H2S Removal Rate 70.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 H 2S Load (g/m3 h) | • Reactor 1 o Reactor 2 Complete Removal | Figure 5-30 H 2S removal rate at 100 ppm H 2S and 650 ppm co2 88 H,S Rate Constant 0.10 30.0 40.0 H2S Load (g/m3h) I • Reactor 1 o Reactor 2 I 50.0 60.0 70.0 Figure 5-31 H2S rate constant at 100 ppm H2S and 650 ppm CO, 89 H2S Acid Yield 6.0 • • • • o arj • dfi 0.0 -I , , , , , 1 , 1 0 10 20 30 40 50 60 70 80 Time (h) • Reactor 1 o Reactor 2 I Figure 5-32 H2S acid yield at 100 ppm H2S and 650 ppm C0 2 90 5.8 100 ppm H 2 S + 1500 ppm C0 2 Results The previous experiment was conducted at 100 ppm H 2 S and 650 ppm CO, ; see section 5.7. This experiment had an acclimation period of approximately 24 h in both Reactor 1 and Reactor 2. The H,S removal efficiency in Reactor 1 decreased from approximately 76% to 32% as the H 2S loading rate increased; see Figure 5-33. With the H 2 S loading rate between 45 — 105 g/m 3h, Reactor 1 reached a mean maximum H 2 S removal rate of 34.9 ± 8.6 g/m 3h as shown in Figure 5-34. There was no distinct trend in the H 2 S rate constant data. Following the acclimation period, the mean H,S rate constant was 0.077 ± 0.027 s"1 as shown in Figure 5-35. In Reactor 2 the H 2 S removal efficiency decreased from 70% to 34% as the H 2 S loading rate increased; see Figure 5-33. Reactor 2 reached a mean maximum H,S removal rate of 28.9 ± 7.5 g/m 3h over the same H 2 S loading rate range as shown in Figure 5-34. There was no distinct trend in the H 2 S rate constant data. Following the acclimation period, the mean H,S rate constant 0.054 ± 0.018 s"1 in Reactor 2 as shown in Figure 5-35. The acid yield data showed the same trend in both reactors: transient for about 24 h followed by a gradual increase and stabilisation by 100 hours; see Figure 5-36. The acid yield in Reactor 1 stabilized at approximately 2.0 mol/mol while Reactor 2 stabilized at approximately 1.5 mol/mol. Similar to the previous experiment (section 5.7), Reactor 1 was still exhibiting approximately complete oxidation and neutralisation of all H 2 S that was removed. It was presumed that the biofilm microbes in Reactor 1 were still acclimatized to the higher H 2 S concentration as reported in section 5.9. And as in the second part of the previous experiment, Reactor 2 had reverted to storing a portion of the H 2 S it removed as S°. Recall it was presumed that the microbes in Reactor 2 were forced to switch back to a survival mode, while those in Reactor 1 continued to exist in a growth mode. 91 H2S Removal Efficiency 100% 9 0 % 7 0 % 6 0 % E 5 0 % a. 4 0 % 3 0 % 2 0 % 10% A Time (h) + 90.0 + 40.0 £. • Reactor 1 • Reactor 2 Target Load | Figure 5-33 H2S removal efficiency at 100 ppm H2S and 1500 ppm C0 2 92 H2S Removal Rate 120.0 120.0 [ • Reactor 1 D Reactor 2 Complete Removal | Figure 5-34 H 2 S removal rate at 100 ppm H 2 S and 1500 ppm co2 93 H2S Rate Constant • • • 8 • „ • • ft ft • • • • • • • • D ft a • • • • • • • • • • • D r, • ° • n • o.O 20.0 40.0 60.0 H2S Load (g/m3h) I • Reactor 1 a Reactor 2 I 80.0 100.0 Figure 5-35 H2S rate constant at 100 ppm H2S and 1500 ppm CO, 94 H2S Acid Yield 6.0 5.0 • 08m • P • t a n D o • attn • 150 250 Time (h) • Reactor 1 • Reactor 2 Figure 5-36 H2S acid yield at 100 ppm H2S and 1500 ppm C0 2 95 5.9 200 ppm H 2S + 0 ppm C 0 2 Results Due to a number of logistical and safety considerations, only one experiment was performed at 200 ppm H 2 S. No supplementary C O , was added. The previous experiment was conducted at 100 ppm H 2 S and 0 ppm CO, ; see section 5.5. There was no appreciable acclimation period in either reactor at this H,S level as shown in the H,S removal efficiency graph Figure 5-37. Reactor 1 had an initial removal efficiency of 76%, which decreased to approximately 25% as the H,S loading rate increased. Reactor 2 had an initial removal efficiency of approximately 65% decreasing to about 25%. The maximum H,S removal rate in Reactor 1 was scattered about a mean of 49.6 + 13.9 g/m 3h during the course of this experiment. It did not increase with the H,S loading rate as in other experiments at lower H,S levels. The maximum H,S removal rate in Reactor 2 was again lower at 39.0 ± 13.5 g/m 3h and was also constant with some data scatter. It was concluded that the initial H,S loading rate was already greater than the maximum removal capacity of the biofilm. See Figure 5-38. Reactor 1 had a mean H,S rate constant of 0.054 + 0.024 s"1 and there was no discernible trend in the data. One outlying data point was removed from the mean calculation. Reactor 2 had an H,S rate constant of 0.032 ± 0.015 s"1 as shown in Figure 5-39. The acid yield data for this experiment show transients up to approximately 36 h followed by fairly constant acid yield values of approximately 2.4 mol/mol in Reactor 1 and 2.2 mol/mol in Reactor 2; see Figure 5-40. The acid yield transients may have been due to the increase in H,S inlet concentration from the previous experiment (section 5.5). With the acid yield Y > 2 for both reactors, it is evident that the biofilm microbes were oxidizing additional sulphur sources; presumably to supplement their energy requirements. But this cannot be the case since the H,S concentration is higher than was provided in the previous experiment (100 ppm H,S). There is presumably sufficient energy available to satisfy the metabolic requirements since the H,S inlet concentration is double that of the previous experiment. The only reasonable explanation is that at this high H,S concentration, the microbes were shifted to a different metabolic pathway where they had higher-than-previous energy requirements. 96 The lack of an appreciable acclimation period, as seen in the H,S removal efficiency graph Figure 5-37 for both reactors, bears further discussion. Some acclimation period was expected as the microbes adjusted to the increased inlet concentration in going from the previous experiment at 100 ppm H,S to this experiment at 200 ppm H 2 S. It had been presumed that some time would be required for the microbial population to grow sufficiently to oxidize the increased H 2 S concentration. The lack of this expected acclimation period was taken as an indication of steady state in the reactors before increasing the loading rate to the next step. The removal efficiency trend was used in all experiments to provide an indication that the reactors had achieved a steady-state condition before increasing the H 2 S loading rate. If we can conclude that the acid yield data provides better insight into the underlying biological state of the biofilm microbes, then perhaps future work should rely on acid yield data to indicate that the reactors have achieved a steady-state condition. H2S Removal Efficiency 90% A 80% 70% A 50% 40% 30% 20% 10% 0% • 20 40 60 80 100 Time (h) 120 • Reactor 1 • Reactor 2 Target Load F i g u r e 5-37 H 2 S removal efficiency at 200 p p m H 2 S and 0 p p m C O , 180.0 160.0 + 140.0 + 120.0 — 100.0 v in 80.0 £ + 60.0 40.0 20.0 0.0 140 97 H2S Removal Rate 180.0 160.0 -0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0 200. H2S Load (g/m3h) | • Reactor 1 • Reactor 2 Complete Removal | Figure 5-38 H2S removal rate at 200 ppm H2S and 0 ppm C0 2 98 H2S Rate Constant 20.0 40.0 60.0 80.0 100.0 120.0 H2S Load (g/m3h) 140.0 160.0 180.0 200. • Reactor 1 o Reactor 2 Figure 5-39 H2S rate constant at 200 ppm H2S and 0 ppm C0 2 99 H2S Acid Yield 6.0 5.0 ] o o S i 3-0-"5 O < — 2.0 1.0 • o o W o a • • • ID 0.0 20 40 60 80 Time (h) 100 120 140 • Reactor 1 • Reactor 2 Figure 5-40 H2S acid yield at 200 ppm H2S and 0 ppm C0 2 100 5.10 300 ppm H 2 S + 0 ppm C 0 2 Results The previous experiment was conducted at 50 ppm H 2 S and 1500 ppm CO, ; see section 5.4. Neither reactor showed an acclimation phase when exposed to an H,S inlet concentration of 300 ppm. Reactor 1 had a relatively constant H 2 S removal efficiency at 42.8 + 7.8% while Reactor 2 had an H 2 S removal efficiency of 23.4 + 9.4%; see Figure 5-41. Both reactors showed a gradual upward trend in the H 2 S removal rate. This is contrary to conventional expectations for trickling biofilters but has been observed in other experiments. The H 2 S removal rate in Reactor 1 increased from 50.0 to 88.0 g/m 3h as the H,S loading rate increased from 112.1 to 202.3 g/m 3h (target H 2 S loading rate 116.5 to 203.9 g/m3h). Overall the H 2 S removal rate was 66.4 ±19.1 g/m 3h in Reactor 1. See Figure 5-42. The Reactor 2 H,S removal rate increased from 22.7 to 43.6 g/m 3h as the H,S loading rate increased from 113.7 to 194.2 g/m 3h (target H,S loading rate 116.5 to 203.9 g/m3h). The overall H 2 S removal rate in Reactor 2 was 35.8 + 15.6 g/m 3h. The H 2 S rate constant in Reactor 1 was 0.041 + 0.013 s"1 and showed no discernible trend vs. H 2 S loading rate. In Reactor 2 the H 2 S rate constant was 0.018 + 0.009 s"1 and showed no discernible trend. See Figure 5-43. The acid yield data in Figure 5-44 show a series of unusual sharp increases and gradual decreases. Neither this pattern nor the high acid yield values can be explained at this point. In Reactor 1, the acid yield decreased from 3.2 mol/mol to 2.9 mol/mol between 45 — 54 h, and from 3.4 mol/mol to 2.6 between 75 — 108 h. In Reactor 2, the acid yield decreased from 2.7 mol/mol to 2.5 mol/mol between 45 - 54 h, and from 3.1 mol/mol to 2.6 mol/mol between 75 - 108 h. Chronologically this was the third consecutive experiment with unusually high acid yield results. The gradual decreases seen in both reactors between 45 — 54 h and 75 — 108 h may possibly be attributed to the poisonous effects of hydrogen sulphide at this high concentration. It is possible that these periods of decreasing acid yield are a direct result of gradual biofilm microbe death. During the period between 54 - 75 h while the trickling biofilters were running but without any hydrogen sulphide (see explanation below), it is presumed that the biofilm microbes were able to 101 recover from the poisonous effects of the H 2 S. When the hydrogen sulphide was reintroduced at 75 h, the gradual biofilm death began once again. The lack of an appreciable acclimation period was unusual given that this experiment followed immediately after one at 50 ppm H,S; see section 5.4. In transitioning to the highest experimental H 2 S concentration, it was expected that the acclimation period, as observed in the H 2 S removal efficiency graph, would be prolonged. However, this may have been masked by the generally low removal efficiency recorded at this high H,S concentration. The Target H,S Loading Rate of 0 g/m 3h between 54 — 75 h in the H 2 S removal efficiency graph (Figure 5-41) was due to the rapid usage and depletion of the hydrogen sulphide gas supply at this extremely high loading rate. The replacement H 2 S cylinder was installed as soon as it was available. It was decided to continue running the trickling biofilters without any H 2 S load while awaiting delivery of the replacement gas cylinder. H2S Removal Efficiency a. CO i 90% \ 70% 60% \ 50% 40% 30% \ 10% 0% • D • • • • • • B ° 20 40 60 Time (h) • • • • o a • • 80 100 250.0 200.0 150.0 * CO I 100.0 H 50.0 0.0 120 • Reactor 1 • Reactor 2 • -Target Load | Figure 5-41 H2S removal efficiency at 300 ppm H2S and 0 ppm co9 102 H2S Removal Rate 250.0 H2S Load (g/m3h) | • Reactor 1 • Reactor 2 Complete Removal Figure 5-42 H 2 S removal rate at 300 ppm H 2 S and 0 ppm C0 2 103 H2S Rate Constant 100.0 150.0 H2S Load (g/m3h) I • Reactor 1 • Reactor 2 I 200.0 250. Figure 5-43 H2S rate constant at 300 ppm H2S and 0 ppm C0 2 104 H2S Acid Yield 120 • Reactor 1 • Reactor 2 Figure 5-44 H 2S acid yield at 300 ppm H 2 S and 0 ppm C0 2 105 Chapter 6 - Discussion of Results 6.1 Summary of Results Refer to Table 4-1 Summary of R T D Parameter Results (page 48) for a summary of residence time distribution results on which the kinetic analysis was based. Recall that for each reactor, the mean residence time X was empirically correlated with air flowrate, and that the mean Tanks in Series parameter N was used. Table 6-1 below shows a summary of the experimental results for hydrogen sulphide removal in two parallel trickling biofilter reactors. The maximum H 2 S removal rate and the H 2 S rate constant are reported for Reactor 1 (pH 5.0) and Reactor 2 (pH 2.5). 106 Table 6-1 Summary of Experimental Results: Maximum H2S Removal Rate & lst-Order Rate Constant H9S Cone, (ppm) . C0 2 Cone, (ppm) Expt. Sequence Reactor l 1 Max H2S H2S Rate Rate Constant (g/m3h) (s 1) Reactor 2 Max H2S H2S Rate Rate Constant (g/m3h) (s 1) 50 0 1 34.0 N / A 21.1 0.136 150 8 23.1 0.122 12.7 0.046 300 7 22.8 0.143 • 15.0 0.058 1500 9 22.8* 0.130* 13.5* 0.051* 100 0 2 50.1 0.129 33.4 0.079 300 11 27.1* 0.048* 14.5* 0.022* 650 4 28.8 0.078 28.8 0.072 650 5 24.5* 0.056* 15.3* 0.026* 1500 6 34.9 0.077 28.9 0.054 200 0 3 49.6 0.054 39.0 0.032 300 0 10 66.4* 0.041* 35.8* 0.018* 6.2 Residence Time Distribution The relatively low values of N observed indicate a large degree of back-mixing is taking place as the air moves through the trickling biofilter reactor columns. Experimental results showed that the N values were insensitive to the air flowrate. This was unexpected. The anticipated result was to observe N increasing with the air flowrate. The reactor headspace and liquid disengagement zones were included in the experiments & analysis and may have contributed to the observed back-mixing without affecting the actual flow pattern in the reactor packing itself. This would result in a lower value of N that would be found if the R T D experiments were performed on a reactor Notation * indicates the values unexpectedly increased with H 2S loading rate. Value reported is the mean value following acclimation. 107 without headspace and disengagement zones. The headspace and liquid disengagement zones combined account for 40% of the total reactor volume in these lab-scale reactors. It is expected that increasing the column length and/or reducing the air flowrate will allow the reactors to behave more like plug flow reactors. However, this is not necessarily desired. For pollutants such as hydrogen sulphide that are, at high concentrations, potentially toxic to the microbes in the biofilm, a plug flow reactor will cause the biofilm near the gas inlet to become poisoned. A trickling biofilter with a lower number of tanks N will behave more like a single stirred tank reactor. In the axial direction, the biofilm will be exposed to a lower and more uniform pollutant concentration that will help prevent substrate inhibition and biofilm poisoning. In addition, near the end of a plug flow reactor, the reactant conversion rate can be quite low due to the low reactant concentration. A larger reactor vessel will be required to achieve a given conversion. A thoroughly mixed reactor would reduce the required reactor size to achieve the desired conversion since the entire reactor volume would be working at its maximum removal rate. One researcher (Lee, D . - H . 1999) showed that a novel biofilter design with increased back-mixing effectively increased the biofilter's elimination capacity and reduced the reactor size required to achieve a given pollutant removal efficiency. 6.3 First Order Assumption The Tanks in Series model used in this work to evaluate the H,S rate constant assumes that the overall reaction process, or at least the rate-controlling step, is ls t-order. The oxidation process in a trickling biofilter is complex; it involves mass transfer from the gas to liquid phase, microbial uptake, and biological oxidation. Physical factors (pH, temperature), chemical factors (pollutant concentration, concentration of other chemical species), and biological factors (microbial concentration, microbial population) may influence the biological processes. The biological oxidation rate will probably be described by a saturation kinetics equation. Due to this complexity, an analysis of all H 2 S oxidation rate-controlling factors to determine the rate-controlling step was not performed. Since diffusion is a l s t-order process, these results may indicate that H,S removal was diffusion controlled or that the removal reaction step was ls t-order. Regardless of the exact rate controlling process, the l s t-order assumption can be shown to be a reasonable one by comparing the rate constant found using the 1 "-order Tanks in Series model to 108 other kinetic orders. A comparison was made between the rate constant equation,.Equation 2-5 Kinetic Rate Constant Estimate (page 14), to a zero-order equation and a 2nd-order equation also using the Tanks in Series model (Levenspiel, O. 1972). These two equations are shown below. N Equation 6-1 Performance equation for a 0th order reaction in N tank reactors in series f 4kT, 2 +2^-1... + 2^-1+ 2.^ 1 +4C0*T, J Equation 6-2 Performance equation for a 2" order reaction in N tank reactors in series; the 2nd-order rate constant was evaluated using a spreadsheet Solver function Plotting these three equations against the mean residence time showed that the I s-order model rate constant was the most consistent. The rate constant did not change appreciably as the mean residence time increased. In general, the R 2 correlation coefficient was low for the Tr-order model rate constant in this analysis, indicating that rate constant was not correlated with mean residence time. On this basis, the l s t-order assumption was deemed to be reasonable and valid for modelling purposes. A sample of these graphs is shown below in Figure 6-1 as a typical example. The remaining graphs are shown in Appendix 11.2 Reaction Order Verification. 109 Reaction Order Reactor 1 0.25 - i r 0.0005 FT = 0.937 4- 0.0004 R' = 0.0687 5.0 10.0 .15.0 Mean Residence Time (s) 20.0 k(n=1)(s-1) - Linear (k(n=1) (s-1)) k(n=2) (m3/mol.s) • k(n=0) (mol/m3.s) - Linear (k(n=2) (m3/mol.s)) Linear (k(n=0) (mol/m3.s)) 25.0 Figure 6-1 Reaction order comparison showing validity of re-order model assumption, compared to 0-order and 2nd-order, at 100 ppm H2S and 1500 ppm C0 2 6.4 Substrate Inhibition Since H 2 S is known to be toxic, substrate inhibition analysis was performed to determine the extent of substrate inhibition on the H,S removal kinetics. The experimental H 2 S rate constant data was fitted to Equation 2-10 Substrate Inhibition Kinetics (page 24) and the model parameters were determined by non-linear curve fitting using a spreadsheet program. The objective function was to minimise the sum of squared difference between the measured and model-predicted H,S removal rate values; e.g. minimise X(x-y)2. Results are shown below in Figure 6-2. The relatively large values of the inhibition constant K , compared to the H 2 S concentration suggest that the microbial inhibition caused by the H 2 S toxicity is insignificant at these concentrations. This is likely due to the fact that these microbes have evolved and adapted to be able to consume hydrogen sulphide as a primary energy source. 110 Although the final model parameters depended somewhat on the initial value used in the rninimisation function, these differences were deemed insignificant. Selected parameter sets are presented below in Table 6-2. Further justification for the conclusion that H 2 S toxicity is insignificant at these concentrations is the wide disparity in the final values of the K , parameter. This indicates that this parameter is not at all significant in the model analysis. Final parameter sets indicated in bold were selected for the model based on their lower ndnimisation function value. Table 6-2 Summary of Substrate Inhibition Model Parameters with Various Initial Value Sets Reactor v ^.Parameter Set • r M A X Ks 5>-y)2 1 Initial (1) 200 1 1 1957.508276 Final (1) 94.766 0.218 375.896 497.2209596 Initial (2) 100 1 10 6701.521868 Final (2) 95.600 0.222 39.166 498.6260632 2 Initial (1) 200 1 1000 883.1326927 Final (1) 54.433 0.165 999.986 334.34544 Initial (2) 100 1 1000 1762.677602 Final (2) 54.433 0.165 999.995 334.3454401 Initial (3) 100 1 10000 1762.570032 Final (3) 54.411 0.165 9999.999 334.3628639 111 H2S Substrate Inhibition 0.45 | • Reactor 1 (Data) Reactor 1 (Model) • Reactor 2 (Data) Reactor 2 (Model) | Figure 6-2 H2S Substrate Inhibition Model 6.5 Acid Yield and Energy Storage The concept of energy being stored as elemental sulphur S° and later consumed by the biofilm microbes during periods where the hydrogen sulphide supply is insufficient is the most intriguing aspect of this work. Table 6-3 summarises the acid yield results from all experiments. The different acclimation period durations (based on H 2 S oudet concentration stabilization and acid yield stabilization) are noted for comparison. 112 Table 6-3 Summary of Acid Yield Results Experiment Acid Yield ( mol/mol) Acclimation Period (h) Order Section H2S Cone, (ppm) Reactor 1 Reactor 2 By H2S Cone. By Acid Yield 1 5.1 50 2.0 1.0 36 36 2 5.5 100 1.6 1.1 0 0 3 5.9 200 2.4 2.2 0 36 4/5 1 5.7 100 2.1/2.2 2.0/1.2 0/0 48/15 6 5.8 100 2.0 1.5 24 100 7 5.3 50 2.8 2.5 70 100 8 5.2 50 3.8 3.2 0 60 9 5.4 50 3.8 3.0 50 70 10 5.10 300 3.2 2.0 0 22 11 5.6 100 5.5 4.0 5 5 Viewing the results in a chronological summary supports the hypothesis that the biofilm microbes are in fact storing energy during periods of abundance and consuming the stores during starvation periods. During the first six experiments, the biofilm microbe populations would have been growing as they were exposed to hydrogen sulphide concentrations in excess of their metabolic requirements. This is evidenced by the fact that the H 2 S contaminant gas was not completely removed throughout each experiment. With this population growth comes a concomitant growth in energy requirements to maintain the population. During the next three experiments with the H 2 S inlet concentration reduced to 50 ppm, the biofilm microbe populations would have been required to meet their energy requirements at the same levels as before but with reduced energy inputs from Experiments 4 & 5 were performed at the same conditions; #4 was terminated prematurely when the COa supply was depleted unexpectedly. 113 to stay alive. A more dramatic example of this phenomenon can be seen in experiments 10 (section 5.10) and 11 (section 5.6). As the H 2 S inlet concentration is reduced from 300 ppm to 100 ppm, we observe the acid yield rising from 3.2 mol/mol to 5.5 mol/mol in Reactor 1, and from 2.0 mol/mol to 4.0 mol/mol in Reactor 2. These results show that when the H,S inlet concentration is reduced, the acid yield generally increases to compensate for the reduced energy input. The microbes are consuming stored energy, probably elemental sulphur S°, in an attempt to maintain the existing biofilm population. The different acclimation period as shown by the acid yield stabilization bears further discussion. In all cases where a difference was noted in the acclimation period, the steady-state point as determined by the H 2 S outlet concentration was earlier than the steady-state point as determined by the acid yield. This indicates that stabilization of the contaminant outlet concentration may not be the best indication that steady-state has been achieved in a biofilter or trickling biofilter. It is possible that in addition to simply becoming acclimatized to a new inlet concentration, the biofilm microbes must also come to a new population steady-state before it can be said that the trickling biofilter has completely stabilized. 6.6 Effect of pH It was observed that in all experiments the H,S removal rate and H 2 S rate constant were greater in Reactor 1, which was running at a liquid p H setpoint of 5.0. The p H setpoint for Reactor 2 was 2.5. It is well established that various sulphur-oxidizing bacteria are capable of surviving over a wide pH range. It is reasonable to expect that running the biofilters at lower p H levels would reduce operating costs since less neutralizing solution would be required for pH control to counteract the acidification from H,S. However the experimental results indicate that the lower operating costs would be offset by increased capital costs since larger biofilters would be required to oxidize an equivalent amount of H 2 S at pH 2.5 compared to pH 5. The importance of reactor pH on the biooxidation of hydrogen sulphide in biofilters and trickling biofilters has not been conclusively decided. Some research groups found that maintaining a neutral reactor pH helped maintain H 2 S removal efficiencies (Yang, Y . and E . R. Allen 1994b), (Degorce-Dumas, J. R., S. Kowal, et al. 1997). Others have found that biofilter bed acidification from pH 6.6 to pH 3.1 during H,S biofiltration did not reduce the removal efficiency (Cook, L. L. , 114 P. A . Gostomski, et al. 1999). This work showed that the reactor liquid medium pH in a trickling biofilter did affect the hydrogen sulphide removal rate and rate constant. There are two possible explanations for the reduced H,S removal capacity at the lower p H level. The first is the microbial population in the trickling biofilter. Most microbial organisms will thrive over a pH range of about 3 units (Brock, T. D . 1997) and we can thus reasonably conclude that different species were dominating the microbial populations in the two reactors. The dominant species in Reactor 1 may have simply had a greater inherent H 2 S removal capacity than the dominant species in Reactor 2. This is unlikely since the seed inoculum for both reactors was obtained from the same source and prepared in the same manner, but over time the microbe populations in the two reactors may have diverged. The second explanation is the pH-dependant solubility of H 2 S in water. H,S is sparingly soluble in water with mole fraction solubility in water of 1.85xl0"3 at 298K and an H 2 S partial pressure of 101.325 kPa (Gevantman, L. H . ). In addition, H 2 S is a weak acid with pK^ = 6.96 and p K ^ = 12.87 (Jenkins, F., H . v. Kessel, et al. 1996). As the solution pH decreases, the H 2 S solubility is reduced. This is evident from the H 2 S dissociation reactions shown below in Equation 6-3. H2S(aq)< )HS~+H+ ( ^ >S2~ +2H+ Equation 6-3 H2S Dissociation Reactions; pH-dependent Solubility The incoming H,S in the gas phase must first dissolve into the liquid phase before it can be taken up for oxidation by the microbes. At higher pH levels, Equation 6-3 shifts to the right thereby increasing the H 2 S solubility. With a lower liquid solubility at lower p H levels, the mass transfer driving forces are reduced and thus the oxidation rate will also be reduced. The acid yield was uniformly lower in Reactor 2 than in Reactor 1. This is also presumably due to the pH difference in the reactors. This is likely due to reaction thermodynamics. From the sulphur oxidation reactions in Equation 2-7 Sulphur Compound Oxidation Reactions (page 17), it is apparent that reaction (b) is driven to the right at low pH and thus S° formation is favoured. Reactions (a), (c), and (d) are thermodynamically unfavourable at low pH. These will both limit the 115 oxidation of H 2 S and preferentially convert the hydrogen sulphide that is oxidized to elemental sulphur. Additional experiments over a wider pH range would help distinguish between which of these two mechanisms are more reasonable. Measuring the hydrogen sulphide ion concentration [HS] in solution would provide additional insight into the gas-liquid and liquid-microbe mass transfer rates. 6.7 Effect of Carbon Dioxide Carbon dioxide was added to the reactors in varying quantities in an attempt to increase the H 2 S oxidation rate by increasing the maximum growth rate of sulphur oxidizing microbes in the biofilter's biofilm layer. Al l C O , concentrations reported above are relative to atmospheric C O , levels; that is, no additional C O , is reported as 0 ppm. Atmospheric C O , levels were assumed to be approximately 300 ppm. In autotrophic organisms such as the sulphur oxidizing microbes in the biofilters, carbon dioxide acts as the source of carbon for biological processes. Hydrogen sulphide acts as the energy source. Carbon dioxide is reduced and converted to biological building blocks in the Calvin cycle (Brock, T. D . 1997), and the energy required for this process is provided by the oxidation of H,S. In the above experiments with added C O , levels ranging from 0 to 1500 ppm, the observed differences in maximum H,S oxidation rate and H,S rate constant cannot be completely attributed to the H,S inlet concentration alone. The experiment in section 5.4 showed a steadily increasing H,S removal rate (Figure 5-14) and H,S rate constant (Figure 5-15) as the H,S loading rate was increased. This was an unusual result and can be explained by a slow but continuing microbial growth rate in the reactor biofilm. The elevated C O , level may have permitted the microbial biofilm to continue growing where it would have otherwise come to equilibrium. It appears that 1500 ppm C O , was a minimum level for this phenomenon to occur. Carbon dioxide addition at lower levels did not appear to increase the H,S removal rate or H,S rate constant. Britton's review (Britton, A . 2001) indicated that approximately 0.5% (=5000ppm) was a minimum level of additional carbon dioxide required to provide increased microbial growth and sulphur 116 oxidation rates. This study has shown that lower levels of additional C O , at 1500 ppm (=0.15%) may have increased the H,S removal rates. 6.8 R T D Scale-Up Approach Section 2.8 Modelling Biological Gas Treatment provides a detailed discussion of some of the literature models published for modelling waste gas treatment in biofilters. It seems evident that the models reviewed therein are complex and detailed. However there is little evidence that they provide superior predictive capabilities compared to simpler models. Simplifying assumptions reduce the model complexity but have not always been adequately justified. In one example for a dynamic model (Zarook, S. M . , A . A . Shaikh, et al. 1997), assumptions that reduced the number of simultaneous differential equations from 230 to 50 did little to improve the predictive capabilities of the model; see Table 2-8 Comparison of Dynamic Models for Biofilters (page 29). The residence time distribution approach for design and scale-up that was used in this work provides a simple means to determine the l s t-order rate constant k for contaminant removal from a laboratory scale biofilter or trickling biofilter. At the laboratory or pilot scale, the mean residence time X can be determined using tracer analysis. From this the Tanks in Series parameter N may be calculated. With X, N and experimental values for the contaminant's inlet and outlet concentration C 0 and C N , the ls t-order rate constant for contaminant removal can be estimated. This was the process used in this work. A large body of literature data is available for estimating gas flow through dry packed beds (Hirnmelblau, D. M . and K . B. Bischoff 1968, Fig A10; Levenspiel, O. 1972, Ch 9 Fig 19). This can be used directly for scale-up to larger operational trickling biofilters. The Peclet number (Pe) or vessel dispersion number (D/uL) found from this literature is convertible to the Tanks in Series parameter N through the dimensionless variance as shown in Equation 6-4 and Equation 6-5 (Levenspiel, O. 1972). Recall Equation 2-4 Number of Tanks in Series Parameter (page 14) shows thatN = 1/Oe2. 117 \uL) Equation 6 -4 Dispersion Number for Small Extents of Dispersion 2 ( "L\ -2 l-e D \uL) K J Equation 6 - 5 Dispersion Number for Large Extents of Dispersion for Closed Vessels This work has assumed that the extent of dispersion would be large, given the size and construction of the trickling biofilters used. Figure 4-2 and Figure 4-3 (page 43) show this to be a valid assumption. As such, Equation 6-5 should be used to convert between the Tanks in Series parameter N found for the reactors in this work and the vessel dispersion number D / u L predicted from the literature. A simple and accurate scale-up to an operational trickling biofilter may now be made given: 1. The expected contaminant inlet concentration (design parameter), 2. The expected air flow rate (design parameter), 3. The desired contaminant removal (design parameter), 4. The bed packing properties (design parameter), 5. The estimated vessel dispersion number, converted to N (literature value), 6. The kinetic rate constant found from laboratory or pilot testing (experimental). 118 Chapter 7 - Conclusions The residence time distribution (RTD) results show that the two lab-scale reactors behaved slighdy differently in terms of their mean residence time T and the equivalent number of CSTR's N that would provide the same RTD. This is to be expected from dumped random packing as was used here. The observed number of tanks in series N for each reactor was constant over the air flowrate range studied. This was unexpected and probably due to the relatively large dead-space fraction of the reactors (40%). A number of conclusions may be drawn from the hydrogen sulphide results and discussions above. 1. The first is that for any experiment, both the H 2 S removal rate and the H,S rate constant were roughly double in Reactor 1 compared to Reactor 2. This is an effect of the reactor's liquid medium pH. It shows that the hypothesis of the lower p H being more favourable to these microbes, which produce sulphuric acid as a metabolic by-product, is false. In fact, higher pH levels may further enhance the H 2 S removal rates by providing a more favourable environment for the biofilm microbes. However there will be additional operating costs to keep the pH constant at a higher level. 2. The second conclusion to be made from these results is that at a given H 2 S inlet concentration, the H 2 S removal rate and H,S rate constant do not change significantly with increasing C O , concentration. While other researchers have found additional C O , increased H,S removal and/or microbial growth, it was not apparent at these C O , levels (Shiwers, D . W. and T. D . Brock 1973; Kargi, F. 1982). As an end-of-pipe pollution control system, a trickling biofilter must operate with bare minimum operating costs. The added expense of directly injecting C O , into the contaminant stream may not be justified unless a readily available C 0 2 source can be used from another mill emissions source such as a boiler or incinerator. 3. The third conclusion that may be drawn from these results is that the H,S is not appreciably toxic to the biofilm microbes at concentrations below 300 ppm. This is supported by the substrate 119 inhibition model analysis. The relatively large values found for K , in Table 6-2 and Figure 6-2 show that the microbes are not significantly poisoned or inhibited by the hydrogen sulphide substrate. If the substrate inhibition had been significant, the model results for the H,S removal rates shown in Figure 6-2 would have exhibited maximum values and a definite downward trend at increasing H,S concentrations. This is likely due to the fact that these microbes have evolved and adapted to be able to consume hydrogen sulphide as a primary energy source. This indicates that hydrogen sulphide poisoning of the biofilm is not a serious factor in using trickling biofilters to control H V L C TRS emissions from Kraft pulp mills. 4. The fourth is that monitoring the acid yield of the trickling biofilter provides an excellent means of indicating that steady-state hydrogen sulphide removal has been achieved. The indication from the acid yield was shown to generally lag the H 2 S removal efficiency as an indication of steady state. As they pass through 'feast and famine' cycles, the biofilm microbes have the ability to store elemental sulphur for future use when the H 2 S loading rate is greater than the biofilm's ability to completely oxidize the contaminant. When the H,S loading rate is then reduced below the level that provides sufficient energy, the elemental sulphur may be further oxidized to sulphuric acid to provide the remaining energy requirements. This will enable production-scale trickling biofilters to quickly adapt to changing operating conditions, particularly when they have been operating at less-than-peak capacity for some time. 5. The fifth, and most important, result from this work is that the Tanks in Series model analysis provides a viable means of estimating the l s t-order rate constant for the removal of gaseous hydrogen sulphide from air streams in a trickling biofilter. This will allow for a simple and rapid design and scale-up of trickling biofilters compared to the mechanistic modelling approach where the large number of model parameters does not necessarily increase the model's reliability. This will aid in the application of economical and environmentally preferable trickling biofilters to the emissions control systems for Kraft pulp mills and other H 2 S producing industries. 120 Chapter 8 - Recommendations Based on the results of this project and the operating experience gained a number of recommendations were developed for further investigation. 1. It is recommended that additional R T D experiments be performed using a reactor column without headspace and liquid disengagement zones. Due to the small size of the reactors, and the relatively large portion allotted to headspace and liquid disengagement zones, it was suspected that significant gas mixing may have occurred in these zones during the R T D tests. This mixing may have distorted the x and N values, and thus the kinetic rate constant calculations. The degree of gas mixing in the reactor packing only is desired for the rate constant calculations. 2. It is recommended that reactors be constructed that allow for independent changes in both the mean residence time x and the degree of back-mixing as represented by the Tanks in Series parameter N . Such a reactor may have removable baffles or other means to modify the air flow pattern inside the reactor. Increasing the back-mixing (N) should allow more of the trickling biofilter to be used, increasing the hydrogen sulphide removal capacity for a given reactor volume. 3. It is recommended that a wider range of reactor p H be explored to help optimize the H 2 S removal rate. Media formulation and other biological factors that would enhance the biofilm microbes' ability to thrive and consume hydrogen sulphide could also be investigated. 4. It is recommended that the production of elemental sulphur S° be recorded as a reaction product in addition to sulphuric acid H 2 S 0 4 . Closing the sulphur mass balance around the trickling biofilter should provide valuable insight into the metabolic processes involved, including storage and subsequent utilisation of S° during 'feast and famine' cycles. 5. It is recommended that larger trickling biofilter reactors be built to evaluate the whether this R T D approach of scale-up allows for a reasonable comparison of rate constants at similar process conditions. 121 Chapter 9 - Nomenclature Table 9-1 Table of Nomenclature Symbol Description Units C C 0 r Concentration Inlet concentration Outlet concentration g/m 3 ; also ppm D / u L Dispersion number -F Flowrate m 3 / h k Reaction rate constant -i s mol/m3-s (Equation 6-1) m3/mol-s (Equation 6-2) V Henry's Law constant mol/m 3-Pa K , Inhibition constant g /m 3 Ks Half-saturation constant g /m3 L Loading rate g/m 3-h N Number of tanks in series -R Removal rate g/m 3-h r Reaction rate g/m 3-h r M A X Maximum reaction rate g/m3-h t Time s V Volume m 3 Y Acid yield mol NaOH/mol H 2 S i\ e Removal efficiency Dimensionless time % a 2 Distribution variance s2 122 Symbol" Description Units Dimensionless variance -T Mean residence time s 123 Chapter 10 - References Alonso, C , M . T. Suidan, G. A . Serial, F. L. Smith, P. Biswas, P. J. Smith and R. C. Brenner (1997). "Gas Treatment in Trickle-Bed Biofilters: Biomass, How Much is Enough?" Biotechnology and Bioengineering 54(6): 583-594. Alonso, C , X . Zhu, M . T. Suidan, B. R. K i m and B. J. K i m (1998). Modeling of the Biodegradation Process in a Gas Phase Bioreactor - Estimation of Intrinsic Parameters: USC-T R G Conference on Bio filtration, Los Angeles, University of Southern California. Banks, D . (1997). "In Anticipation of Regulatory Impacts: Dedicated N C G Incineration with S02 Removal." Pulp & Paper Canada 98(5): 42-44. Barton, J. W., K . T. Klasson, J. Laurence J. Koran and B. H . Davison (1997). "Microbial Removal of Alkanes From Dilute Gaseous Waste Streams: Kinetics and Mass Transfer Considerations." Biotechnology Progress 13: 814-821. Barton, J. W., X . S. Zhang, B. H . Davison and K . T. Klasson (1998). Predictive Mathematical Modeling of Trickling Bed Biofilters. USC-TRG Conference on Biofiltration, Los Angeles, University of Southern California. Bibeau, L., K . Kiared, A . Leroux, R. Brezezinski, G. Viel and M . Heitz (1997). "Biological Purification of Exhaust Air Containing Toluene Vapor in a Filter-bed Reactor." Canadian Journal of Chemical Engineering 75: 921-929. Bowman, R. (1997). Wet Scrubber For TRS Control From A Brown Stock Washer. International Environmental Conference & Exhibit, Minneapolis, M N , TAPPI Proceedings. Britton, A . (2001). Maximizing Growth and Substrate Utilization Rates of Sulfur Oxidizing Bacteria in Biofilters Treating Hydrogen Sulfide. Personal Communication. University of British Columbia, CFIBR 559S Topics in Chemical Engineering. Brock, T. D . (1997). Biology of Microoganisms. Upper Saddle River, NJ , Prentice Hall, Inc. 124 Buisman, C. J. N . , P. IJspeert, A . Hof, A . J. H . Janssen, R. Hagen and G . Lettinga (1991). "Kinetic Parameters of a Mixed Culture Oxidizing Sulfide and Sulfur with Oxygen." Biotechnology and Bioengineering38(8): 813-820. Cho, K.-S., M . Hirai and M . Shoda (1991). "Degradation Characteristics of Hydrogen Sulfide, Methanethiol, Dimethyl Sulfide and Dimethyl Disulfide by Thiobarillus thioparus DW44 Isolated from Peat Biofilter." Journal of Fermentation and Bioengineering 71(6): 384-389. Cho, K - S . , M . Hirai and M . Shoda (1992). "Enhanced Removal Efficiency of Malodorous Gases in a Pilot-Scale Peat Biofilter Inoculated with Thiobarillus thioparus DW44." Journal of Fermentation and Bioengineering 73(1): 46-50. Chung, Y . - C , C. Huang, C. H . Liu and H . Bai (2001). "Biotreatment of Hydrogen Sulfide and Ammonia Containing Waste Gases by Fluidized Bed Bioreactor." Journal of the Air & Waste Management Association 51: 163-172. Chung, Y . - C , C. Huang and C.-P. Tseng (1996). "Biodegradation of Hydrogen Sulfide by a Laboratory-Scale Immobilized Pseudomonas putida CH11 Biofilter." Biotechnology Progress 12: 773-778. Chung, Y . - C , C. P. Huang and C. F. L i (1997). "Removal Characteristics of H2S by Thiobarillus novellus C H 3 Biofilter in Autotrophic and Mixotrophic Environments." Journal of Environmental Science and Health Part a- Environmental Science and Engineering & Toxic and Hazardous Substance Control 32(5): 1435-1450. Cook, L. L , P. A . Gostomski and W. A. Apel (1999). "Biofiltration of Asphalt Emissions: Full-Scale Operation Treating Off-Gases from Polymer-Modified Asphalt Production." Environmental Progress 18(3): 178-187. Cunningham, A . B., E . Visser, Z. Lewandowski and M . Abrahamson (1995). "Evaluation of a Coupled Mass Transport - Biofilm Process Model Using Dissolved Oxygen Microsensors." Water Science and Technology 32(8): 107-114. Degorce-Dumas, J. R., S. Kowal and P. Le Cloirec (1997). "Microbiological Oxidation of Hydrogen Sulphide in a Biofilter." Canadian Journal of Microbiology 43(3): 264-271. 125 DeHollander, G. R., T .J . Overcamp and C. P. L. Grady, Jr. (1998). "Performance of a Suspended-Growth Bioscrubber for the Control of Methanol." Journal of the Air & Waste Management Association 48: 872-876. Deshusses, M . A. , G. Hamer and I. J. Dunn (1995a). "Behavior of Biofilters for Waste Air Biotreatment. 2. Experimental Evaluation of a Dynamic Model." Environmental Science & Technology 29(4): 1059-1068. Deshusses, M . A. , G. Hamer and I. J. Dunn (1995b). "Behavior of Biofilters for Waste Air Treatment. 1. Dynamic Model Development." Environmental Science & Technology 29(4): 1048-1058. Deshusses, M . A. , C. T. Johnson and G. Leson (1999). "Biofiltration of High Loads of Ethyl Acetate in the Presence of Toluene." Journal of the Air & Waste Management Association 49: 973-979. Devinny, J. S., M . A . Deshusses and T. S. Webster (1999). Biofiltration for Air Pollution Control. Boca Raton, FL , Lewis Publishers. Diks, R. M . M . , S. P. P. Ottengraf and S. Vrijland (1994). "The Existence of a Biological Equilibrium in a Trickling Filter for Waste Gas Purification." Biotechnology and Bioengineering 44(11): 1279-1287. Escot, A. , C. Chavarie, R. Mayer and J. Ramsay (1996). " A Method for the Evaluation of Microbial Growth in a Trickling Bed Reactor." Canadian Journal of Chemical Engineering 74: 547-550. Frederick, W. J., J. P. Danko and R. J. Ayers (1996). "Controlling TRS Emissions From Dissolving-Tank Vent Stacks." TAPPI Journal 79(6): 144-148. Gevantman, L. H . (1996). Solubility of Selected Gases in Water. CRC Handbook of Chemistry & Physics. D . R. Lide, CRC Press. Hagen, C. E . and R. W. Hartung (1997). "New Chemical Treatment Method Controls Wastewater System Odor." Pulp & Paper: 81-89. 126 Hecht, V. , D . Brebbermann, P. Bremer and W.-D. Deckwer (1995). "Cometabolic Degradation of Trichloroethylene in a Bubble Column Bioscrubber." Biotechnology and Bioengineering 47: 461-469. Himmelblau, D . M . and K . B. Bischoff (1968). Process Analysis and Simulation. New York, John Wiley & Sons. Hirai, M . , M . Ohtake and M . Shoda (1990). "Removal Kinetics of Hydrogen Sulfide, Methanethiol and Dimethyl Sulfide by Peat Biofilters." Journal of Fermentation and Bioengineering 70(5): 334-339. Hodge, D . S. and J. S. Devinny (1994). "Biofilter Treatment of Ethanol Vapors." Environmental Progress 13(3): 167-173. Hodge, D . S. and J. S. Devinny (1995). "Modeling Removal of Air Contaminants by Biofiltration." Journal of Environmental Engineering 121(1): 21-32. Hwang, S.-J., H . -M. Tang and W.-C. Wang (1997). "Modeling of Acetone Biofiltration Process." Environmental Progress 16(3): 187-192. Jarvensivu, M . , R. Lammi and J. Kivivasara (1997). Odor Abatement Systems of the Modern Pulp Mill and Relative Importance of the Prevailing TRS Emission Sources. International Environmental Conference & Exhibit, TAPPI Press. Jenkins, F., H . v. Kessel, D . Tompkins and O. Lantz (1996). Chemistry. Toronto, Nelson Canada. Kargi, F. (1982). "Enhancement of Microbial Removal of Pyritic Sulfur from Coal Using Concentrated Cell Suspension of T. ferrooxidans and an External Carbon Dioxide Supply." Biotechnology and Bioengineering 24(3): 749-752. Kirchner, K . , C. A . Gossen and H . J. Rehm (1991). "Purification of Exhaust Air Containing Organic Pollutants in a Trickle-Bed Bioreactor." Applied Microbiology and Biotechnology 3 5 : 396-400. Kirchner, K . , G. Hauk and H . J. Rehm (1987). "Exhaust Gas Purification Using Immobilised. Monocultures (Biocatalysts)." Applied Microbiology and Biotechnology 26: 579-587. 127 Kirchner, K . , S. Wagner and H . J. Rehm (1992). "Exhaust Gas Purification Using Biocatalysts (Fixed Bacteria Monocultures) - the Influence of Biofilm Diffusion Rate (Ooj on the Overall Reaction Rate." Applied Microbiology and Biotechnology 37: 277-279. Lee, D . -H . (1999). Development of an Alternative Biofilter System for Odor Treatment. Ph. D . Thesis, Chemical & Bio-Resource Engineering, University of British Columbia, Vancouver. Levenspiel, O. (1972). Chemical Reaction Engineering. New York, John Wiley & Sons. Mohseni, M . and D . G . Allen (1997). Biofiltation of OC-Pinene and its Application to the Treatment of Pulp and Paper Air Emissions. International Environmental Conference & Exhibit, TAPPI Press. Ockeloen, H . F., T. J. Overcamp and C. P. L. Grady, Jr. (1996). "Engineering Model for Fixed-Film Bioscmbbers." Joumal^fFinvironmem 191-197. O'Connor, B. (1996). Methanol Emissions from Kraft Mil l Stationary Sources, Pulp and Paper Research Institute of Canada: 14. Oloman, C , F. E . Murray and J. B. Risk (1969). The Selective Absorption of Hydrogen Sulfide from Stack Gas. 55th Annual Meeting of the Technical Section, Canadian Pulp and Paper Association, Montreal, Canada, Canadian Pulp and Paper Association. Park, S.-J., K.-S. Cho and M . Shoda (1993). "Removability of Malodorous Gases from a Night Soil Treatment Plant by a Pilot-Scale Peat Biofilter Inoculated with Thiobarillus thioparus DW44." Journal of Fermentation and Bioengineering 76(1): 55-59. Pinkerton, J. E. (1999). "Trends in U.S. Kraft Mil l TRS Emissions." TAPPI Journal 82(4): 166-169. Pisotti, D . A . (1997). Biofilter Eliminates More Than Just VOCs From A Press Exhaust. International Environmental Conference & Exhibit. Rho, D. , N . Matte and M . Utiger (1998). Biodegradation Dynamics of Styrene and Acetone from Waste Gases in a Biotrickling Filter. USC-TRG Conference on Biofiltration, Los Angeles, University of Southern California. 128 Sander, R. (1999). Compilation of Henry's Law Constants for Inorganic and Organic Species of Potential Importance in Environmental Chernistry (Version 3). Internet: http://www.mpch-main^-tnpg.del —sander/ res/ henry.html Shiwers, D. W. and T. D . Brock (1973). "Oxidation of Elemental Sulfur by Sulfolobus acidocaldarius." Journal of Bacteriology 114: 706-710. Shuler, M . L. and F. Kargi (1992). Bioprocess Engineering: Basic Concepts. Englewood Cliffs, NJ , Prentice Hall. Sorial, G. A. , F. L . Smith, M . T. Suidan, P. Biswas and R. C. Brenner (1995). "Evaluation of Trickle Bed Biofilter Media for Toluene Removal." Journal of the Air & Waste Management Association 45: 801-810. Sorial, G . A. , F. L. Smith, M . T. Suidan, A . Pandit, P. Biswas and R. C. Brenner (1997). "Evaluation of Trickle Bed Air Biofilter Performance for B T E X Removal." Journal of Environmental Engineering 123(6): 530-537. Spizzirri, P. (1995). The Reduction fo TRS Gases During Non-Condensible Gas Venting and Liquor Spills to Mill Sewer. International Environmental Conference & Exhibit, Atlanta, Georgia, TAPPI Proceedings. Tarpey, T. (1995). Odour Reduction at Peace River Pulp. International Environmental Conference & Exhibit, TAPPI Proceedings. Tatum, V . L. (1995). Health Effects of TRS Compounds. International Environmental Conference & Exhibit. Tchobanoglous, G . and F. L. Burtop (1991). Wastewater Engineering: Treatment. Disposal, and Reuse. Toronto, McGraw-Hill Inc. Wani, A . H . (1999). Biofiltration for the Removal of Reduced Sulfur Gases from Low Concentration Air Streams. Ph .D. Thesis, Department of Chemical & Bio-Resource Engineering, University of British Columbia, Vancouver, BC. 129 Wani, A . H . , R. M . R. Branion and A . K . Lau (1998). Dynamic Behavior of Biofilters Degrading Pulping Odors and VOCs: Hydrogen Sulfide. International Environmental Conference & Exhibit, Vancouver, BC, TAPPI Press. Yang, Y . and E . R. Allen (1994a). "Biofiltration Control of Hydrogen Sulfide 1. Design and Operational Parameters." Journal of the Air & Waste Management Association 44: 863-868. Yang, Y . and E . R. Allen (1994b). "Biofiltration Control of Hydrogen Sulfide. 2. Kinetics, Biofilter Performance, and Maintenance." Journal of the Air & Waste Management Association 44: 1315-1321. ! Yu , J. and K . L. Pinder (1992). "Build-Up of Symbiotic Methanogenic Biofilms on Solid Supports." Biotechnology Letters 14(10): 989-994. Yu, J. and K . L. Pinder (1993). "Intrinsic Fermentation. Kinetics of Lactose in Acidogenic . i ' Biofilms." Biotechnology and Bioengineering 41: 479-488. Yu , J. and K . L. Pinder (1994). "Effective Diffusivities of Volatile Fatty Acids in Methanogenic Biofilms." Bioresource Technology 48: 155-161. i Zarook, S. M . and A . A . Shaikh (1997). "Analysis and Comparison of Biofilter Models." Chemical Engineering Journal 65: 55-61. 1 Zarook, S. M . , A . A . Shaikh, Z. Ansar and B. C. Baltzis (1997). "Biofiltration of Volatile Organic Compound (VOC) Mixtures Under Transient Conditions." Chemical Engineering Science 52(21/22): 4135-4142. Zarook, S. M . , A . A . Shaikh and S. M . Azam (1998). "Axial Dispersion in Biofilters." Biochemical Engineering Journal 1: 77-84. 130 Chapter 11 - Appendices 11.1 Nutrient Media Table 11-1 Nutrient Media Formulations Component ,,c • ATCC Medium S+C+ Medium S-C- Medium Ammonium Chloride NH 4 C1 (g) 0.10 0.10 0.10 Calcium Chloride CaCl 2 -2H 2 0 (g) 0.13 0.13 0.13 Magnesium Chloride MgCl 2 -6H 2 0 (g) 0.21 0.21 0.21 Potassium Phosphate Monobasic K H 2 P 0 4 ( g ) 3.00 3.00 3.00 Sodium Thiosulfate Na 2 S 2 0 3 -5H 2 0 (g ) 5.00 i 5.00 0.00 Methanol C H 3 O H (g) 0.00 5.00 0.00 Distilled Water (L) 1.0 1.0 1.0 Medium pH (-) (Reactor 1) (Reactor 2) 4.2 5.0 2.5 5.0 2.5 I I 131 11.2 Reaction Order Verification /1.2.1 50 ppm H2S + 0 ppm C02 Reaction Order Reactor 1 8.0 10,0 12.0 Mean Residence Time (s) | M k(n=1) (s-1) k(n=2) (m3/mol.s) • k(n=0> (mol/m3.s) Linear (k(n=2) (m3/mol.s» Linear (k(n=0) (mol/m3.s)) j Figure 11-1 Reaction order comparison for Reactor 1 at 50 ppm H 2S + 0 ppm C0 2 Reaction Order Reactor 2 0.0004 Mean Residence Time (a) B k(n=1}(s-1) & k(n=2)(m3/mol,s) • k(n=0) (mol/m3.s) Linear (k(n=1) (s-1)) Linear (k(n=2Km3/mol.s)) - Linear (k(n=0) (mol/m3.s)) Figure 11-2 Reaction order comparison for Reactor 2 at 50 ppm H 2S + 0 ppm C0 2 132 /1.2.2 50ppm H2S +150ppm C02 Reaction Order Reactor 1 10.0 15.0 Mean Residence Time (s) k(n=1) (s-1) -Linear (k(n=1) (s-1)) 0.0003 £ S f= o o II o c k(n=2) (m3/mol.s) • k(n=0) (mo1/m3.s) - Linear (k(n=2) (m3/mol.s)) Linear (k(n=0) (mo^ m3.s)) Figure 11-3 Reaction order comparison for Reactor 1 at 50 ppm H2S + 150 ppm C0 2 Reaction Order Reactor 2 10.0 15.0 Mean Residence Time (s) a k(n=1)(s-1) Linear (k(n=1) (s-1)) k(n=2) (m3/mol.s) • k(n=0) (mol/m3.s) - Linear (k(n=2) (m3/mol.s)) Linear (k(n=0) (moym3.s)) Figure 11-4 Reaction order comparison for Reactor 2 at 50 ppm H2S + 150 ppm C0 2 133 I I /1.2.3 50ppm H2S + 300ppm C02 , Reaction Order Reactor 1 » II 5.0 10.0 15.0 Mean Residence Time (s) • 20.0 m k(n=1)(s-1) Linear (k(n=1) (s-1)) • k(n=2) (m3/mol.s) • Linear (k(n=2) (m3/mol.s)) k(n=0) (mol/m3.s) -Linear (k(n=0) (mol/m3.s)) Figure 11-5 Reaction order comparison for Reactor 1 at 50 ppm H2S + 300 ppm C0 2 Reaction Order Reactor 2 10.0 15.0 Mean Residence Time (s) a k(n=1) (s-1) Linear (k(n=1) (s-1)) k(n=2) (m3/mol.s) • k(n=0) (mol/m3.s) - Linear (k(n=2) (m3/mol.s)) Linear (k(n=0) (mol/m3.s)) Figure 11-6 Reaction order comparison for Reactor 2 at 50 ppm H2S + 300 ppm CQ 2 134 /1.2.4 50ppm H2S + 1500ppm C02 Reaction Order Reactor 1 R2 = 0.7829 *• . _ f |-10.0 15.0 Mean Residence Time (s) k(n=1)(s-1) -Linear (k(n=1) (s-1)) k(n=2) (m3/mol.s) • k(n=0) (mol/m3.s) - Linear (k(n=2) (m3/mol.s)) Linear (k(n=0) (mol/m3.s)) Figure 11-7 Reaction order comparison for Reactor 1 at 50 ppm H 2 S + 1500 ppm C0 2 i Reaction Order Reactor 2 Figure 11-8 Reaction order comparison for Reactor 2 at 50 ppm H 2 S + 1500 ppm C0 2 135 1.2.5 100 ppm H-J + Oppm C02 Reaction Order Reactor 1 Figure 11-9 Reaction order comparison for Reactor 1 at 100 ppm H 2 S + 0 ppm CO z Reaction Order Reactor 2 Figure 11-10 Reaction order comparison for Reactor 2 at 100 ppm H 2 S + 0 ppm CO z 136 11.2.6 100ppm + 300ppm C02 Reaction Order Reactor 1 Figure 11-11 Reaction order comparison for Reactor 1 at 100 ppm H 2 S + 300 ppm C0 2 Reaction Order Reactor 2 Figure 11-12 Reaction order comparison for Reactor 2 at 100 ppm H 2 S + 300 ppm CO z 137 /1.2.7 100ppm H2S + 650ppm C02 Reaction Order Reactor 1 5.0 10.0 15.0 20.0 Mean Residence Time (s) • k(n=1) (s-1) k(n=2) (m3/mol.s) • k(n=0) (moi/m3.s) Linear (k(n=1) (s-1)) Linear (k(n=2) (m3/mol.s)) -Linear (k(n=0) (mol/m3.s)) Figure 11-13 Reaction order comparison for Reactor 1 at 100 ppm H2S + 650 ppm C0 2 Reaction Order Reactor 2 t. i 5.0 10.0 15.0 20.0 Mean Residence Time (s) a k(n=1) (s-1) k(n=2) <m3/mol.s) • k(n=0) (mol/m3.s) Linear (k(n=1) (s-1)) Linear (k(n=2) (m3/mol.s)) -Linear (k(n=0) (mol/m3.s)) Figure 11-14 Reaction order comparison for Reactor 2 at 100 ppm H2S + 650 ppm C0 2 138 Reaction Order Reactor 1 6.0 8.0 10.0 12.0 14.0 Mean Residence Time (s) |ak(n=1) (s-1) •:: k(n=2) (m3/mol.s) »k(n=0) (moi/m3.s) | Figure 11-15 Reaction order comparison for Reactor 1 at 100 ppm H 2 S + 650 ppm C0 2 Reaction Order Reactor 2 3.0 8.0 10.0 12.0 - Mean Residence Time (s) »k(n=1) (s-1) fti k(n=2) (m3/mol.s) *k(n=0) (mol/m3.s) I c o O II Figure 11-16 Reaction order comparison for Reactor 2 at 100 ppm H 2 S + 650 ppm C0 2 139 /1.2.8 100ppm H^S + 1500ppm C02 Reaction Order Reactor 1 10.0 15.0 Mean Residence Time (s) k(n=1)(s-1) - Linear (k(n=1) (s-1)) k(n=2) (m3/mol.s) - Linear (k(n=2) (m3/mol.s)) -k(n=0) (mol/m3.s) -Linear (k(n=0) (mol/m3.s)) Figure 11-17 Reaction order comparison for Reactor 1 at 100 ppm H 2 S + 1500 ppm C0 2 Reaction Order Reactor 2 10.0 15.0 Mean Residence Time (s) o k(n=1) (s-1) — — Linear (k(n=1) (s-1)) * k(n=2) (m3/mol.s) • k(n=0) (moi/m3.s) - - Linear (k(n=2) (m3/mol.s)) Linear (k(n=0) (mo[/m3.s)) Figure 11-18 Reaction order comparison for Reactor 2 at 100 ppm H 2 S + 1500 ppm C0 2 140 /1.2.9 200ppm + 0ppm C02 Reaction Order Reactor 1 SI H2 = 0.3474 m c o O II O c 5.0 10.0 15.0 20.0 Mean Residence Time (s) n k(n=1) (s-1) k(n=2) (m3/mol.s) • k(n=0) (mol/m3.s) Linear (k(n=1) (s-1)) Linear (k(n=2) (m3/mol.s)) -Linear (k(n=0) (mol/m3.s)) Figure 11-19 Reaction order comparison for Reactor 1 at 200 ppm H 2S + 0 ppm C0 2 Reaction Order Reactor 2 1- 0.0000 25.0 10.0 15.0 Mean Residence Time (s) k(n=1) (s-1) -Linear (k(n=1) (s-1)) ;. k(n=2) (m3/mol.s) — Linear (k(n=2) (m3/mol.s)) -k(n=0) (mol/m3.s) - Linear (k(n=0) (mol/m3.s)) Figure 11-20 Reaction order comparison for Reactor 2 at 200 ppm H 2S + 0 ppm C0 2 141 11.2.10 300ppm + 0 ppm C02 Reaction Order Reactor 1 : 10.0 15.0 Mean Residence Time (s) k(n=1) (s-1) - Linear (k(n=1) (s-1)) k(n=2) (m3(mol.s) - Linear (k(n=2) (m3/mol.s)) -k(n=0) (mo^m3.s) -Linear (k(n=0) (mol/m3.s)) Figure 11-21 Reaction order comparison for Reactor 1 at 300 ppm H 2S + 0 ppm C0 2 Reaction Order Reactor 2 0.05 0.00 4-0.0 • R2 = 0.9706 • R2 = 0.171 • 1 • ""- .^ " ~- - - . _ • • • B * • R2 = 0.1451 I 1 1 a 1 "8T 8 • B 9 B 5.0 10.0 15.0 Mean Residence Time (s) 20.0 B k(n=1) (s-1) •:, k(n=2) (m3/mol.s) • k(n=0) (mol/m3.s) Linear (k(n=1) (s-1)) Linear (k(n=2) (m3/mol.s)) - Linear (k(n=0) (mol/m3.s)) 0.0001 —I- 0.0000 25.0 Figure 11-22 Reaction order comparison for Reactor 2 at 300 ppm H 2S + 0 ppm C0 2 142 

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.831.1-0058964/manifest

Comment

Related Items