Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Making membranes more efficient: mapping surface shear in a pilot-scale submerged hollow-fibre membrane… Fulton, Blair G. 2009

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2009_fall_fulton_blair.pdf [ 17.63MB ]
Metadata
JSON: 24-1.0063129.json
JSON-LD: 24-1.0063129-ld.json
RDF/XML (Pretty): 24-1.0063129-rdf.xml
RDF/JSON: 24-1.0063129-rdf.json
Turtle: 24-1.0063129-turtle.txt
N-Triples: 24-1.0063129-rdf-ntriples.txt
Original Record: 24-1.0063129-source.json
Full Text
24-1.0063129-fulltext.txt
Citation
24-1.0063129.ris

Full Text

MAKING MEMBRANES MORE EFFICIENT: MAPPING SURFACE SHEAR IN A PILOT-SCALE SUBMERGED HOLLOW-FIBRE MEMBRANE CASSETTE USING ELECTROCHEMICAL SHEAR PROBES  by  Blair G. Fulton B.Sc., University of British Columbia, 2004  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE  in The Faculty of Graduate Studies (Civil Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) June, 2009  © Blair G. Fulton, 2009  ABSTRACT Optimization of gas sparging (ie. gas flow rate used for gas scouring) and module design has great potential to decrease fouling and improve energy efficiency in submerged membrane systems. Shear stress has been widely recognized as a controlling factor in the fouling of most types of membrane systems, but despite its relevance few researchers have attempted to quantify this variable in submerged hollow fiber systems, forcing membrane designers to infer processes at the membrane surface indirectly from flux and transmembrane pressure data. The present study utilized electrochemical probes to map the shear stresses on full-scale hollow fiber membrane modules in a pilot-scale air sparged submerged membrane system (GE-Zenon Zeeweed-500c). The effects of sparging rate, membrane module spacing, fiber tension, and sparging pattern were investigated, and all were determined to significantly affect shear stress profiles. The results of this study, presented largely as ‘shear maps’, provide insight into the roles of gas sparging and module configuration on shear stress profiles in submerged membrane systems.  ii  TABLE OF CONTENTS ABSTRACT........................................................................................................................ ii TABLE OF CONTENTS................................................................................................... iii LIST OF TABLES.............................................................................................................. v LIST OF FIGURES ........................................................................................................... vi ACKNOWLEDGEMENTS.............................................................................................. vii CHAPTER 1 INTRODUCTION AND RESEARCH OBJECTIVES ................................ 1 CHAPTER 2 LITERATURE REVIEW ............................................................................. 3 2.1 Fouling in membrane systems ............................................................................ 3 2.2 Approaches to fouling control ............................................................................ 4 2.3 Measuring shear stress and hydrodynamics........................................................ 6 2.4 Electrochemical shear probes ............................................................................. 7 2.4.1 The concept of limiting current................................................................... 8 2.4.2 The electrochemical method ....................................................................... 9 2.4.3 Previous uses of electrochemical probes in membrane systems............... 11 2.5 The knowledge gap: Scalable shear stress data for submerged hollow-fiber systems.............................................................................................................. 12 CHAPTER 3 MATERIALS AND METHODS ............................................................... 14 3.1 The structure of an electrochemical shear probe .............................................. 14 3.2 Reagents............................................................................................................ 16 3.3 Side loop membrane apparatus ......................................................................... 17 3.3.1 Side loop mechanical systems................................................................... 18 3.3.2 Side loop electrical system and data acquisition...................................... 20 3.3.3 Side loop experimental procedures........................................................... 22 3.3.4 QA/QC....................................................................................................... 23 3.4 Pilot-scale membrane system with full-scale membrane modules ................... 25 3.4.1 Mechanical systems .................................................................................. 25 3.4.2 Pilot electrical system and data acquisition ............................................. 31 3.4.3 Probe locations ......................................................................................... 35 3.4.4 Pilot apparatus experimental procedures................................................. 38 3.4.5 QA/QC....................................................................................................... 41 CHAPTER 4 EXPERIMENTAL RESULTS ................................................................... 42 4.1 Measurements taken within the pilot apparatus................................................ 42 4.2 QA/QC results................................................................................................... 46 4.2.1 Probe standardization using the side loop membrane apparatus ............ 46 4.2.2 QA/QC during pilot testing....................................................................... 48 4.3 Data limitations................................................................................................. 49 CHAPTER 5 DATA ANALYSIS AND DISCUSSION .................................................. 51 5.1 Experimental configurations with continuous sparging ................................... 55 5.1.1 Relationship between average shear stress and variation of the shear stress for individual probe measurements ................................................ 55 5.1.2 The effect of sparging rate ........................................................................ 57 5.1.3 The effect of module spacing..................................................................... 61 5.1.4 The effect of fiber tension.......................................................................... 66 5.1.5 The effect of vertical location within the membrane cassette ................... 71 iii  5.1.6 General observations ................................................................................ 72 5.2 Experimental configurations with discontinuous sparging............................... 73 5.2.1 Relationship between average shear stress and variation of the shear stress, individual probe measurements ..................................................... 73 5.2.2 The effect of sparging pattern ................................................................... 74 5.3 Assumptions regarding optimal shear stress..................................................... 82 CHAPTER 6 CONCLUSIONS ........................................................................................ 83 6.1 Suggestions for membrane design and configuration....................................... 84 6.2 Future research.................................................................................................. 85 REFERENCES ................................................................................................................. 86 APPENDIX 1: VOLTAGE vs. SHEAR STRESS............................................................ 92 APPENDIX 2: MANUFACTURING SHEAR PROBES ................................................ 93 APPENDIX 3: SHEAR STRESS IN ANNULAR FLOW ............................................... 94 APPENDIX 5: TABLE OF PROBE SURFACE AREA CORRECTION FACTORS .. 104 APPENDIX 6: COMPLICATIONS OF THE ELECTROCHEMICAL METHOD ...... 105 APPENDIX 7: PHOTOS OF THE EXPERIMENTAL APPARATUS ......................... 108 APPENDIX 8: MATLAB CODES USED FOR DATA ANALYSIS............................ 111 APPENDIX 9: SHEAR STRESS MAPS ....................................................................... 117  iv  LIST OF TABLES Table 2.1 Techniques for measuring hydrodynamics in membrane systems ................... 6 Table 3.1 Fiber tensions, module spacings, and sparging settings for the pilot-scale membrane apparatus ....................................................................................... 28 Table 3.2 Configurations examining the effect of module spacing, fiber tension, and sparging rate on shear stress..................................................................... 39 Table 3.3 Configurations examining the effect of sparging pattern on shear stress ....... 40 Table 4.1 Table of probe voltage correction factors ....................................................... 46 Table 5.1 Ranking configurations by cassette-average voltage and standard deviation of cassette-average voltage ............................................................. 54 Table 5.2 Ranking sparging patterns based on cassette-average voltage and standard deviation of cassette-average voltage ............................................................. 78  v  LIST OF FIGURES  Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure 3.5 Figure 3.6 Figure 3.7 Figure 3.8 Figure 3.9 Figure 3.10 Figure 3.11 Figure 3.12 Figure 3.13 Figure 3.14 Figure 3.15 Figure 4.1 Figure 4.2 Figure 4.3 Figure 5.1 Figure 5.2  Top and side views of a shear probe on a test fiber ..................................... 15 A test fiber shown with a Zeeweed-500 membrane fiber ............................ 15 Side loop apparatus used for all probe standardization tests ....................... 18 Photo of the Plexiglas column used in the side loop apparatus ................... 19 Electrical circuit used for all probe standardization tests ............................ 21 Pilot apparatus mechanical systems............................................................. 25 Photo of pilot apparatus mechanical systems: Zeeweed-500c cassette ....... 26 Close up of a single pneumatic actuator cylinder ........................................ 27 Top view (looking down) of the diffusers and membrane modules ............ 30 Diagram of the electrical components of the pilot apparatus ...................... 32 Schematic of the electrical circuit within the pilot apparatus ...................... 34 Photo of the 60-channel switch box connected to the 60 test fibers ............ 35 Probes were placed within one corner of the cassette.................................. 36 Side view of probe locations within the cassette ......................................... 37 Top view of probe locations within the cassette.......................................... 37 Typical raw unsorted voltage data ............................................................... 43 Typical 45 seconds of voltage data from a single probe.............................. 44 High frequency noise in the raw voltage data, an example ......................... 50 Cassette-averaged shear stress for each tested configuration ...................... 53 Typical mean shear stresses vs. their standard deviations for configurations using continuous gas flow ................................................... 56  Figure 5.3 Figure 5.4 Figure 5.5 Figure 5.6 Figure 5.7 Figure 5.8 Figure 5.9 Figure 5.10 Figure 5.11 Figure 5.12  Effect of sparging rate on cassette average shear stress .............................. 58 Typical sheet-averages illustrating effect of increasing sparging rate......... 59 Typical shear maps illustrating the effect of increasing sparging rate......... 60 Effect of module spacing on cassette average shear stress .......................... 63 Typical sheet averages illustrating effect of increasing module spacing..... 64 Typical shear maps illustrating the effect of increasing module spacing .... 65 Effect of fiber tension on cassette average shear stress ............................... 68 Typical sheet-averages illustrating effect of increasing fiber tension ......... 69 Typical shear maps illustrating the effect of increasing fiber tension ......... 70 Typical mean shear stresses vs. their standard deviations for configurations using discontinuous sparging............................................... 74  Figure 5.13 Figure 5.14 Figure 5.15 Figure 5.16  Typical effect of sparging pattern on cassette average shear stress............. 77 Typical sheet-averages illustrating effect of sparging pattern ..................... 79 Typical shear maps illustrating the effect of sparging pattern..................... 80 Alternating sparging between diffusers vs. continuous sparging ................ 81  vi  ACKNOWLEDGEMENTS  Completion of this project required a skill set and knowledge base beyond those possessed by the author; only through the mingling of many remarkable minds could this project have been accomplished. Dr. Pierre Bérubé was a key mind, both the source of several of the project’s founding ideas, and of continuous intellectual support to the author throughout the project’s many stages. Engineering Technician Bill Leung constructed many of the components of the bench and pilot-scale apparatuses, and often conceived of novel component designs superior to those suggested by the author. Engineering Technician Scott Jackson built many of the electronic components of the experimental apparatuses, and was a tremendous resource during setup and troubleshooting of electrical equipment. Laboratory Technicians Susan Harper and Paula Parkinson helped the author to develop appropriate laboratory methods, and to satisfy the author’s seemingly endless need for various pipe fittings. Graduate students Doug Beaton and Alejandro Bohl generously helped the author become familiar with Matlab software, which proved to be a valuable tool for dealing with the large quantities of data produced. PhD candidate Colleen Chan and MASc student Wade Archambault provided many useful suggestions that led to significant improvements in project design. Many graduate students generously contributed their time as safety personnel, ensuring the author could survive to analyze the project results. The author would also like to acknowledge Jeff Redwood, an undergraduate engineering student who performed many months of detailed work constructing electrochemical probes. Finally, the author is grateful to Annie VerSteeg, who provided her hilltop Mexican casa as the ideal location to write a thesis, though using a laptop in a hammock can be tricky!  vii  CHAPTER 1 INTRODUCTION AND RESEARCH OBJECTIVES Strengthening regulations regarding drinking water quality and wastewater discharge limits are generating increasing demand for advanced water/wastewater treatment technologies. Membrane treatment systems, with their modular design, small footprint, and impressive filtration capabilities, are an attractive option for treating a wide variety of source waters. Continued improvements in design and manufacturing methods continue to decrease the cost of many membrane systems, making them increasingly competitive with more conventional methods of water treatment. Membranes can be classified based on their pore size (microfiltration, ultrafiltration, nanofiltration), their material composition (ceramic, organic), and their design (tubular, flat sheet, hollow fiber). However, all membranes used for water treatment operate using the same basic principle, using pressure or suction (transmembrane pressure) to propel water across the membrane surface, which filters out everything too large to fit through the pores in its surface. Submerged outside-in hollow fiber organic membranes (referred to hereafter as submerged HF membranes) can be used for either microfiltration (for filtering out particulates in the bacteria size range, 1-10µm) or ultrafiltration (for filtering out viruses and large macromolecules, 0.1-1µm), making them widely applicable for the treatment for municipal water and wastewater (US EPA, 2003). In drinking water treatment, submerged HF membranes can be used as an alternative to sand filtration. In wastewater, submerged HF membranes are used as an alternative to the clarifier component in activated sludge systems, an arrangement called a membrane bioreactor (MBR), as well as for final effluent polishing as a tertiary treatment step. Submerged HF systems have particular advantages over side-stream tubular membrane systems, the most significant being increased permeate flux per membrane system volume (ie. greater packing density, allowing for a smaller plant footprint), and decreased energy demand (Ueda et al., 1997). Numerous major suppliers  1  of submerged HF systems have arisen to meet the growing world market for MBR technology, such as GE-Zenon, Memcor, and Mitsubishi, among others (Cui et al., 2003). During membrane filtration, particulate material removed from the source water collects at the membrane surface, increasing the resistance that water must overcome as it is drawn through the membrane from one side to the other (Bérubé et al., 2006b). This increase in membrane resistance, called fouling, is a major hindrance for membrane systems, as it both reduces the total amount of permeate that can be recovered for a given filtration time, and drives up energy costs due to the greater transmembrane pressure that must be applied to keep water passing through the pores of the membrane. Due to its great importance, fouling control has been a relatively active area of research since the development of membranes in the early 1980’s (Hilal et al., 2005; Cui et al., 2003). Presently, designers of submerged HF membrane systems must rely on permeate flux and transmembrane pressure data when comparing the effectiveness of various fouling reduction strategies. However, while flux and transmembrane pressure data are of critical importance to membrane performance, they fail to provide a clear understanding of the underlying physical processes affecting fouling. Shear stress (N/m2, Pa) is widely recognized as a fundamental variable controlling the mass transfer in most membrane systems (Bellara et al., 1996; Cabassud et al., 2003; Cui et al., 2003), but to date no quantitative information has been available above the bench-scale regarding shear stress in submerged HF systems. The objective of the present study, building upon the recent bench-scale research by Chan et al. (2007) and Bérubé et al. (2006a), was to map shear stresses on full-scale hollow fiber membrane modules in a pilot-scale air sparged submerged membrane system (GE-Zenon Zeeweed500c). Custom modifications to a pilot-scale Zeeweed-500c system allowed investigation of the affect of sparging rate, fiber tension, and module spacing on shear stress within full-scale membrane modules.  2  CHAPTER 2 LITERATURE REVIEW Submerged HF membrane systems, a concept initially developed in Japan in the early 1980’s (Cui et al., 2003), are an increasingly popular technology for both drinking water and wastewater treatment applications. Submerged HF systems have several advantages over tubular and flat sheet membrane systems, the most significant being their lower transmembrane pressure and lower energy consumption (Ueda et al., 1997).  2.1  Fouling in membrane systems Fouling remains a major impediment for submerged HF systems, driving up  energy and maintenance costs and thus the cost per volume of filtered permeate. Fouling can be particularly severe in membrane bioreactors, which is one of the most promising markets for submerged HF technology. For microfiltration and ultrafiltration, membrane fouling appears to be predominantly caused by cake formation and pore blocking (Hilal et al., 2005). Cake formation is a process whereby insoluble material, such as bacterial cells and particulate too large to pass through the membrane pores, collect at the membrane surface, forming a compacted layer which provides additional resistance to water passing through the membrane. Pore blocking, as the name suggests, is the clogging of membrane pores by foulants small enough to enter the pores, thereby reducing the number of pores available for the passage of water. Besides acting as a physical barrier to the passage of water, cake formation appears to promote the growth of biofilms (Busch et al., 2007), microbial communities which develop upon the membrane surface, which are in turn thought to play a significant role in fouling, especially within MBR systems. A number of factors affect the severity of fouling mechanisms such as cake formation and pore blocking, including pH, ion concentration, transmembrane pressure, solids concentration, membrane type, and characteristics of the solids being filtered (Hilal et al., 2005).  3  2.2  Approaches to fouling control Hilal et al. provide an extensive look at the many methods used for fouling  control in their 2005 review paper, and summarize the four key approaches as: modification of the membrane surface, pretreatment of source water, modification of operating conditions, and chemical cleaning. Of these four methods, modification of operating conditions has perhaps the greatest opportunity for optimization and impact on energy consumption within low-pressure membrane systems. Relaxation time, crossflow, gas-sparging, and system configuration are four operating condition parameters that have been observed to impact membrane fouling (Cui et al. 2003). Relaxation time, a technique which simply allows the membrane ‘to rest’ without any applied pressure or vacuum, has been observed to reverse reductions in flux caused by fouling; however, as this approach involves significant periods without filtration, it decreases the total permeate flux and thus increases the total surface area of membrane required (Bérubé et al., 2006b). Cross-flow, the practice of inducing a velocity gradient parallel to the membrane surface, has been widely demonstrated to significantly decrease membrane fouling (Chang et al., 2002). Cross-flow generates shear stress along the membrane surface which increases the flow of material away from the membrane and back into the bulk fluid. Several authors have reported that increased cross-flow leads to decreased filtration resistance, though there appears to be an optimum above which further increases either have no effect or may even increase fouling (Vyas et al., 2000). Within tubular membrane systems, cross-flow is predominantly achieved using recirculation pumps, whereas in submerged HF systems it is achieved through gas-sparging. Gas-sparging refers to the injection of gas (usually air, or re-circulated methane in anaerobic systems) to create a two-phase flow on the concentrate side of the membrane, a practice widely used to reduce fouling in submerged HF systems. While gas-sparging is used partly to generate cross-flow, caused by the upward flow of expanding gas bubbles, many authors have noted that the flux-enhancement brought about by gas-sparging exceeds that attributable to the generated cross-flow. Chang and Fane (2000) note that in 4  their experiments, gas hold-up, the volume of gas within a given volume containing two phase flow, was more important for fouling reduction than the measured cross-flow velocity. Chang and Fane (2001) noted that the use of gas-sparging allowed a 30% increase in permeate flux from a single hollow fiber compared to the same cross-flow under single phase flow conditions. Bérubé and Lei (2006) found similar results (a 2060% increase in flux compared to single phase flow) as did Bellara et al. (1996) & Cui et al. (2003) (a 10-60% increase) using multi-fiber systems. Increases in sparging rate, like increases in cross-flow velocity, provide improved fouling reduction compared to lower flow rates, though the benefit of each individual escalation is less compared to the preceding increase, such that eventually a plateau is reached above which increases in sparging intensity have no further positive effect on fouling reduction (Chang and Fane, 2000; Bérubé and Lei, 2006). Importantly, by reducing reversible fouling (the layer of insoluble debris that accumulates at the membrane surface as a ‘cake’ layer), gas sparging also reduces irreversible fouling (adhered microbial films and material strongly adsorbed to the membrane) (Chang and Fane, 2000), and thus the frequency of chemical cleaning required to maintain acceptable membrane performance. Cui et al., in their substantial 2003 review paper on gas-sparging, note that while gas-sparging is widely accepted as a means of fouling control, the extent to which gassparging is able to reduce fouling depends on numerous factors: type of membrane system, solids concentration within the bulk fluid, applied transmembrane pressure, geometry/orientation of the system, bubble size and quantity, and applied cross-flow velocity (in some systems, such as tubular systems, gas-sparging may not be the dominant method used for generating cross-flow). Altering the system configuration refers to modification of the physical elements of the membrane apparatus, such as the spacing and tension of membrane fibers. Ueda et al. (1997) found that by concentrating HF membrane modules and gas sparging over a smaller area within his pilot apparatus, fouling was reduced without increasing the total rate of gas sparging (m3/hr). However, Chan et al. (2007) suggest that there may be a  5  limit to this concentration of membrane fibers above which rising bubbles may be unable to reach fibers shielded by surrounding fibers. Fiber tension has also been investigated as an important factor controlling fouling in submerged HF systems. Chang and Fane (2000, 2001) note that fiber sway is likely a considerable variable in fouling control, while the observations of Bérubé and Lei (2006) and Chan et al. (2007) more specifically suggest that the importance of fiber sway is its ability to induce bubble penetration deep into the module, and the induction of inter-fiber rubbing.  2.3  Measuring shear stress and hydrodynamics As fluid movement has long been recognized as playing a significant role in  fouling reduction within membrane systems, many methods have been used to examine and quantify this parameter. Simple conductivity probes can be used to provide basic information regarding gas hold-up, while hot wire and hot film probes can be used to provide similar information with additional insight into liquid cross-flow velocity (Nakoryakov et al., 1989). Several more advanced techniques have been used successfully in membrane systems to characterize two-phase flow conditions, the measurement capabilities of each of these methods are summarized in Table 2.1.  Table 2.1  Techniques for measuring hydrodynamics in membrane systems.  Measurement technique Conductivity probe Hot wire/ hot film Electromagnetic flow velocity meter Particle image velocimetry Acoustic doppler velocimetry Optical probe Electrochemical shear probe  Measurement capability Shear stress Fluid Gas at Examples cross-flow hold-up membrane of use: velocity surface No No Nakoryakov 1989 Yes No No No  Yes  Yes Yes Yes Yes  Yes  No  No No No No  Yes  No  Yes  Yes  Ueda 1997 Gaucher 2002 Tacke 2008 Nguyen 2008 Chan 2007  6  From Table 2.1 it can be noted that the hot wire and electrochemical techniques are uniquely well suited for characterizing shear stress at the membrane surface; however, the hot wire technique is not capable of capturing the high shear events of passing bubbles under two-phase flow conditions (Farrar et al., 1995), whereas the electrochemical method is particularly well suited for such measurements. Shear stress is a key variable controlling the mass transfer at the membrane surface, thus its measurement should be constructive in elucidating the basic mechanisms controlling fouling in membrane systems.  2.4  Electrochemical shear probes Electrochemical shear probes are not only uniquely well suited for measuring  shear stress within two-phase flow systems, but have the additional benefits of rapid response and (relatively) simple construction. Electrochemical shear probes are small metal electrodes submersed in a reversible ionic system. An extremely fast voltagedriven reaction occurs at the electrode surface, and as the reaction involves electrons from the probe surface, the reaction can be measured as an electrical current being passed through the probe. For example, when using the ferri/ferrocyanide ionic couple with an electrochemical shear probe the following reaction occurs at the probe (cathode) surface, with the reverse reaction occurring at the anode (which is not used for measurements). Fe(CN)63- + e- → Fe(CN)64(e- , an electron, derives from the electrical current passing through the probe) As the reaction step is far faster than the diffusion rate of reactive ions to the probe surface, the reaction rate becomes analogous to the diffusion rate near the probe surface. The diffusion rate can in turn be related to the mass transfer coefficient and shear stress at the electrode surface using the equations A4.1 to A4.4 (given in Appendix 4).  7  The metal used for probe construction depends somewhat on the reversible ionic solution used. Selman and Tobias (1978) note that few ionic couples are used for mass transfer measurements, as for a system to be useful it must meet a number of criteria: chemical stability, high solubility, an electrode potential different from hydrogen, and reasonable cost. Ferri/ferrocyanide is one of the most commonly used reactions, with Reiss and Hanratty (1962) pioneering the use of this reaction with electrochemical probes for the measurement of shear stresses at surfaces.  2.4.1 The concept of limiting current When a voltage is applied between two electrodes submerged in a solution containing an electrochemical couple, such as the commonly used ferri/ferrocyanide couple, the ions are able to accept electrons from the cathode and donate electrons to the anode such that an electric current is established (as long as the voltage applied is greater than the standard reduction potential (Eo) of the redox reaction involved in electron transfer). When a sufficiently high voltage is applied, the concentration of the reacting ion (ferricyanide) at the cathode surface approaches zero, causing the reaction to become limited by the rate of ion diffusion to the cathode surface rather than by the applied voltage (somewhat confusingly this is known as ‘limiting current conditions’, as the current is limited by the rate of diffusion) (Gordon et al., 1965; Mitchell and Hanratty, 1966). Within the range of voltages capable of generating limiting conditions, the current passing through the probe (cathode) remains constant, as the electrode surface is depleted of ions capable of transferring electrons (Rajeshwar and Ibanez, 1997). Under limiting current conditions the magnitude of the current passing through the electrode is a consequence of just three variables: (1) the diffusion rate of ions to/from the probe surface, (2) the probe surface area, and (3) the concentration of the active ionic species within the solution (Kissinger and Heineman, 1996). To prevent the reaction at the probe surface from being limited by the reaction rate at its counter-electrode (usually the probe is operated as the cathode), the counter-electrode (anode) must have a much larger active surface area than the probe (Nakoryakov et al., 1981; Reiss and Hanratty, 1963).  8  There is a critical fluid flow rate above which the electrochemical reaction occurring at the probe surface can no longer be assumed to be infinite (compared to the rate of diffusion) (Mizushina, 1971), but this becomes a concern only at very high flow rates, above those expected to occur in submerged membrane systems. The limiting current plateau stretches over only a small range of voltages (specific to the standard reduction potential of the electrochemical couple used), beyond which slight increases in voltage result in dramatic increases in current due to the hydrogen overpotential (hydrolysis) and other side reactions (Mizushina, 1971). Theoretically (based on Pourbaix diagrams and the Nernst equation) the hydrogen overpotential becomes a concern at applied voltages above 0.41 volts, but Rajeshwar and Ibanez (1997) note that water is somewhat more stable than these theoretical equations suggest. Most investigators have found that optimal limiting current conditions are achieved at applied voltages between 0.2 and 0.5 volts (some as high as 0.9 volts) (Sobolik et al., 1998; Reiss and Hanratty, 1962) for ferri/ferrocyanide systems, a relatively wide voltage range, which is one of the main reasons for the popularity of this ionic couple for mass transfer experiments (Reiss and Hanratty, 1962).  2.4.2 The electrochemical method As the diffusion of reacting ions to the electrode surface is the controlling element under limiting current conditions, “the mechanisms involved in ion transport can be directly linked to the limiting current” (Berger and Ziai, 1983, p.377). Ion movement near the probe surface is by diffusion, not convection, as ion movement directly adjacent to the electrode surface occurs within the Nernst diffusion layer (Reiss and Hanratty, 1962; Mitchell and Hanratty, 1966). The Nernst diffusion layer is itself contained within a viscous sublayer at the solid/liquid interface (Nakoryakov et al., 1981; Mitchell and Hanratty, 1966), where viscous forces ensure a linear velocity gradient near the wall despite whatever turbulence may be occurring in the bulk fluid (Nakoryakov et al., 1981). However, the thickness of the viscous sublayer is dependent on convective movement within the bulk fluid, such that increased fluid movement in the vicinity of the electrode  9  decreases the thickness of the diffusion layer and allows more rapid transmission of ions between the ion-rich bulk fluid and the ion-deficient region near the electrode surface. Each reacting ion accepts an electron at the probe/cathode surface (or donates one at the anode surface), allowing the rate of diffusion to be measured as an electrical current passing through the probe. The electrochemical method thus “measures fluctuations in the flow indirectly through fluctuations in the gradient of concentration at the wall” (Reiss and Hanratty, 1962, p.245). Increases in fluid movement in the vicinity of the probe (decreasing the thickness of the diffusion layer) generate increases in electrical current passing through the probe, measured as the fluctuating voltage drop across a resistor (Reiss and Hanratty, 1962). As electrochemical probes take their measurements from within the extremely thin diffusion layer, they require only an extremely thin layer of fluid to operate (Nakoryakov et al., 1981). A thin layer of solution remains on the probe surface even during the passage of bubbles in close proximity to the probe, thus electrochemical probes function equally well under both single-phase and two-phase flow conditions. In its simplest form, the measured limiting current can be related to the shear stress experienced at the surface of the probe using equation 2.1 (Cognet et al., 1984). IL = α . (τ / µ)(1/3)  (2.1)  where IL is the measured limiting current (A), τ is the shear stress (Pa), µ is the dynamic viscosity of the fluid (Pa.s), and α is a constant specific to the probe used, but whose value depends on probe surface area (m2), diffusivity of the active ionic species (m2/s), and concentration of the active ionic species within the bulk fluid (mol/m3). For a given probe, and assuming constant solution viscosity and ion diffusivity, the measured limiting current is proportional to the cube root of the shear stress experienced at the probe surface (see Figure A1.1 in Appendix 1). As long as temperature is controlled (as diffusivity and viscosity depend on temperature), the only measurements required for the electrochemical method are: the concentration of the reacting ion species, the surface areas of all shear probes, and the limiting current measured by those probes (Berger and Ziai, 1983).  10  An excess of unreactive but highly conductive electrolyte, such as potassium chloride or sodium hydroxide, is typically used both to reduce ohmic resistance between cathode and anode (this resistance can contribute to signal noise) and to reduce the gradient in electrical potential which occurs near the electrode surface. The reduction of this potential gradient through the use of a non-reactive electrolyte reduces undesirable ion migration at the electrode surface to a negligible level (~1% of total current signal) (Mizushina, 1971), ensuring that diffusion is the dominant factor in ion movement at the electrode surface.  2.4.3 Previous uses of electrochemical probes in membrane systems Several researchers have used electrochemical probes to investigate shear stress within membrane systems. Ducom et al. (2002) used electrochemical probes to demonstrate an association between shear stress intensity and flux enhancement at the surface of a gas-sparged flat sheet nanofiltration membrane. Cabassud et al. (2003) performed similar research, investigating the effect of two-phase flow on shear stress in both a flat sheet and an inside-out HF membrane system (both bench scale), determining that shear stress significantly increased flux performance for the flat sheet system but not for the inside-out HF system. Bérubé et al. (2006a) pioneered the use of electrochemical probes within submerged (outside-in) HF systems, demonstrating that two phase flow was characterized by both sporadic peaks in shear stress measurements and a higher flux, compared to single phase flow conditions with a comparable cross-flow velocity. Chan et al. (2007) investigated the importance of fiber tension, fiber packing density, and sparging rate on shear stress within a bench-scale submerged HF system using the electrochemical method. Chan observed that increasing sparging rate increased both cross-flow velocity (‘baseline shear’) as well as the frequency of high shear events, while fiber tension and packing density affected both the ability of bubbles to penetrate and to stay within the fiber bundle. Importantly, Chan et al. (2007), using a high speed camera together with electrochemical probes measuring at 1000 Hz, showed that the high shear events caused by the passage of bubbles under bubbly flow in a multi-fiber system were  11  approximately 0.02-0.1 seconds in duration. The probes used by Chan et al. (2007), similar to those used in the present study, could readily identify the passage of these bubbles near the probe surface.  2.5  The knowledge gap: Scalable shear stress data for submerged hollow-fiber systems The present research builds upon recent bench-scale research by Bérubé et al.  (2006) and Chan et al. (2007) by using electrochemical shear probes within test fibers to map shear stresses on full-scale hollow fiber membrane modules in a pilot-scale air sparged submerged membrane system (GE-Zenon Zeeweed-500c). Numerous studies have demonstrated the importance of shear stress on fouling control and flux enhancement, but the majority of these studies has been performed at the bench-scale, and few of these have looked at shear stress within submerged HF systems. Though shear stress is a fundamental mechanism controlling fouling within submerged HF systems, data reflecting the shear stress experienced within a full-scale treatment system are not available. Without such data designers are left to rely predominantly on flux and transmembrane pressure (TMP) data, only guessing at the fundamental mechanisms that control these variables. Furthermore, while previous work by Bérubé et al. (2006) and Chan et al. (2007) has investigated the effects of numerous variables on shear stress, such as fiber tension and sparging rate, it is uncertain whether these observations reflect the reality within full-scale systems. The objective of the present study was to determine average shear stresses within full-scale submerged membrane modules, and to display these average shear stress values within visually accessible ‘shear maps’, ‘sheet averages’ (ie. average shear stresses experienced within cross sections of the cassette), and ‘cassette averages’ (ie. average shear stress experienced within the cassette). These shear maps and shear averages were used to remark on differences among investigated experimental configurations. The data produced in this study will subsequently be used to examine the affect of experimental configurations on specific shear event patterns, and to begin to develop computational  12  fluid dynamic (CFD) models to simulate the complex environment in submerged air sparged systems.  13  CHAPTER 3 MATERIALS AND METHODS  3.1  The structure of an electrochemical shear probe Electrochemical probes are not only uniquely well suited for measuring shear  stress within two-phase flow systems, but have the additional benefits of rapid response and (relatively) simple construction. It is desirable to use very small probes, as measurements reflect the average of the shear stresses experienced across the entire probe surface, which for large probes may prevent an accurate reading of the real shear stresses experienced at any specific point (Reiss and Hanratty, 1963). Nickel and platinum have both commonly been used as electrode materials with the ferri/ferrocyanide couple for mass transfer measurements, but results of Taama et al. (1996) indicate that platinum is far superior for this purpose, as it requires less frequent cleaning to maintain its ability to provide accurate readings. The electrochemical shear probes used in this study were essentially the same as those used by Bérubé et al. (2006), a 0.5 mm diameter round platinum wire sanded flush with the outer surface of a 1.8 mm hollow Teflon tube (see Figures 3.1 and 3.2). Once completed, test fibers were similar in shape and flexibility to Zeeweed-500 PVDF membrane fibers (see Figure 3.2). Details of the procedures used to manufacture the sixty probes required for this experiment are provided in Appendix 2.  14  (a)  Figure 3.1  (b)  Top (a) and side (b) views of a shear probe on a test fiber.  The probe diameter is approximately 0.50 mm. The bubbles seen in image (b) are within hardened epoxy resin, which holds the probe in place.  Figure 3.2  A test fiber shown with a Zeeweed-500 PVDF membrane fiber.  15  3.2  Reagents The solution used for all experiments consisted of water, 0.003M potassium  ferricyanide, 0.006M potassium ferrocyanide, and 0.3M potassium chloride (a nonreacting electrolyte). De-ionized water was used for probe standardization tests, while tap water was used for pilot-scale tests. For probe standardization experiments, oxygen was stripped from solution using compressed nitrogen gas immediately subsequent to the addition of chemical reagents. For the pilot tests, oxygen and residual chlorine were stripped from the tap water using compressed nitrogen gas prior to the addition of chemical reagents. Based on the work of Chan et al. (2007), Sobolik et al. (1998), Reiss and Hanratty (1962), and preliminary tests, the present study used an applied voltage of 0.25 volts ( ± 0.002 volts) to achieve limiting current for all probe standardization experiments, and a voltage of 0.31 volts ( ± 0.002 volts) for all pilot-scale work. One of the few negative qualities of the ferri/ferrocyanide couple is that steps must be taken to exclude oxygen from the system, mainly to prevent the gradual oxidation of ferrocyanide and the development of oxide films on the surface of electrodes, but also to prevent oxygen from reacting at the probe (cathode) surface (Reiss and Hanratty, 1962; Mitchell and Hanratty, 1966; Berger and Ziai, 1983), though the latter concern is refuted by Berger and Ziai (1983). While the presence of oxygen is not catastrophic, results of Berger and Ziai (1983) illustrate that it has the undesirable effect of making the limiting current mildly responsive to changes in voltage, casting doubt on whether the equations relating limiting current measurements to shear stress (ie. equations A4.1 to A4.4) can still be used appropriately. The present study went to great lengths to exclude oxygen (detailed in sections 3.3 and 3.4). Ferri/ferrocyanide reagents are gradually degraded through exposure to UV light, and likely through various uncharacterized slow reactions with plastics and other constituents commonly used in experiments, thus Selman and Tobias (1978) recommended replacing the electrolyte solution every few days as a basic quality control procedure. Preliminary tests by the author indicated that the ferri/ferrocyanide solution functioned well for three to four days, but to be conservative, fresh solution was prepared  16  every second day during testing. For the pilot-scale tests (section 3.4) a single large batch of solution was used continuously for 43 hours.  3.3  Side loop membrane apparatus The signal obtained from a probe is related to its surface area. Although care was  taken to manufacture all probes such that they had a similar surface area, it was not possible to make the areas of all probes identical. In order that the signals from all probes could be compared, signals from each probe were normalized to that of an ideal probe with an area of 1.96 x 10-7 m2 (i.e. a diameter of 0.00050 m). A side loop membrane system was used to determine the voltage correction factors for the probes on each of the test fibers manufactured for the present study (Figures 3.3 and 3.4). The side loop system allowed the author to subject each probe to a series of controlled hydrodynamic conditions, thereby allowing observation of the electrical current response of each probe. Sobolik et al. (1998) noted that it is difficult to generate high shear rates under highly controlled conditions; however, for the present study it was deemed unnecessary to test probes under very high shear conditions, as shear probes of 0.5 mm diameter have already been demonstrated to perform well even with shear stresses as high as 800 Pa (Sobolik et al., 1998), approximately two orders of magnitude greater than those likely to be experienced within a gas-sparged system. The work of Bérubé et al. (2006) also indicates that electrochemical probes similar to those used in the present work give results in agreement with theoretical calculations for cross-flow velocities of 0.2-0.4 m/s (see Appendix 3 for calculations), which is similar to the range of cross-flow velocities previously recorded within a submerged HF system by Ueda et al. (1997).  17  3.3.1 Side loop mechanical systems The side loop apparatus used for probe standardization tests consisted of a Plexiglas cylinder 1.2 m in length with an inner diameter of 0.025 m, within which ferri/ferrocyanide solution could be circulated using a Masterflex peristaltic pump (see Figures 3.3 and 3.4). The flow could be varied from 0-3.5 liters/min, though only flow rates of 0.5-2.0 liters/min were used (equal to superficial cross-flow velocities of 0.0170.065 m/sec) to ensure laminar flow conditions. The Plexiglas cylinder was constructed such that a single test fiber (with shear probe) could be accurately placed at the centerline of the cylinder, resulting in an annular type geometry. A reference probe (of dimensions comparable to the probes on the test fibers) was also placed on the inside wall of the Plexiglas cylinder.  Figure 3.3  Side loop apparatus used for all probe standardization tests.  The cathodes are the probes used for shear stress measurements.  18  Figure 3.4  Photo of the Plexiglas column used in the side loop apparatus.  The column (height 1.2m, inner diameter 0.025m) was designed to allow a test fiber to be placed at its centerline, allowing it to operate as an annulus. A reference probe was built into the wall of a test cell, visible as the slightly-wider section near the center of the column.  19  3.3.2 Side loop electrical system and data acquisition During probe standardization testing, an applied potential of 0.252 V ( ± 0.002 V) was used to generate limiting current conditions at the surface of the probes. The anode used was a 3 cm section of stainless steel pipe (3/8” inner diameter), located at the top outlet of the annulus (see Figure 3.3). Figure 3.5 illustrates the electrical circuit used for collecting measurements using the side loop apparatus. Measurements from each probe (on test fibers) and from the reference probe were collected at 100Hz using a computer with Signal Express software (National Instruments, New York City). Each probe was exposed to five different flow rates, with data collection carried out over one minute for each flow rate.  20  Figure 3.5  Electrical circuit used for probe standardization tests.  The cathodes are the probes used for shear stress measurements.  21  3.3.3 Side loop experimental procedures As the geometry of the side-loop reactor was relatively simple (i.e. an annulus), for a specific hydrodynamic condition it was possible to calculate the theoretical shear stress on a test fiber (i.e. probe) (Appendix 3), as well as the theoretical voltage that would be generated by an ideal probe (i.e. with a diameter of exactly 0.0005 m) on this test fiber (Appendix 4). The ratio of the theoretical voltage of an ideal probe to that measured using a probe on an actual test fiber, for different hydrodynamic conditions, was used to develop a voltage correction factor for each of the probes (on test fibers) manufactured for the present study. Details of the procedure used to develop the voltage correction factors are presented in Appendix 4. The voltage correction factors for the probes on the 60 test fibers manufactured for the present study ranged from 0.85 to 1.18 (Table 4.1). A similar standardization procedure has been used previously by Rode et al. (1994) and recommended by Sobolik et al. (1998). A diffusion coefficient (D) of 8.36 x 10-10 m2/sec was assumed for all probe standardization tests, based on the rotating disk electrode results of Gaucher et al. (2002b). It should be noted that this diffusion coefficient is only an approximation for the present study, as the solution used in the present study differed somewhat from Gaucher’s in both composition and temperature (Gaucher et al. used 30ºC and potassium sulfate as the inactive electrolyte). However, for the purposes of the present study it was sufficient to assume a reasonable value of the diffusion coefficient which was then treated as a constant in all calculations, as all probes in the present work were exposed to solution of the same concentration and temperature during probe standardization procedures. Nonetheless, it is worth noting that Legrand et al. (1999) found a diffusion coefficient of 7.5 x 10-10 m2/sec at 25ºC using the rotating disk electrode and unsteady diffusion methods, a value 10% lower than that used in the present study. A dynamic viscosity of 8.1 x 10-4 Pa/sec was assumed for all probe standardization calculations (Gaucher et al., 2002b).  22  3.3.4 QA/QC The goal of testing using the side loop membrane apparatus was to provide an accurate estimation of the voltage signal provided by each probe when exposed to a known set of experimental parameters. For measurements using electrochemical probes, the relevant parameters which required control were: dissolved oxygen, temperature, applied voltage, solution flow rate, and cleanliness of the probe surface (keeping it free of both grime and gas bubbles). Compressed nitrogen gas was used to strip dissolved oxygen from the system prior to each set of measurements, as well as to dislodge any small bubbles which could be seen adhered near the probe (a visual inspection was done prior to each test). A water bath was used to control temperature, as the limiting current is an approximately linear function of temperature, with Berger and Ziai (1983) reporting that a 3ºC change in temperature results in a 20% rise in limiting current. The present study used a set-point of 25ºC, with a thermometer (immersed within the electrolyte solution) used to ensure that temperature was controlled to within 0.5ºC (equivalent to ± 3.3% change in limiting current based on results from Berger and Ziai (1983)). The applied voltage was measured using a voltmeter prior to each set of probe measurements, ensuring that the applied voltage remained accurate to within ± 0.002 volts. Solution flow rate was determined using the LCD screen of the Masterflex peristaltic pump. The pump was calibrated periodically (after every twenty sets of probe measurements) using a measuring column to ensure that its readings were consistently accurate to within four percent. Probes were rinsed with de-ionized water and polished using damp Kimwipes® (laboratory tissue paper) both before and after each set of measurements, both to keep probes clean and to prevent poisoning of the platinum surface by breakdown products of the reagents, such as cyanide (Berger and Ziai, 1983). Cathodic activation, a procedure suggested by Reiss and Hanratty (1963) and Berger and Ziai (1983) to increase probe 23  accuracy, was not considered necessary in the present study, as the applied voltage was controlled to within ± 0.002V during all measurements. However, it is possible that the use of this procedure may have increased the accuracy of probe measurements by several percent (2-5% assuming Re < 200,000, which is probable given the dimensions of the pilot membrane system used, given in Figures 3.13 and 3.14) (Berger and Ziai, 1983). A five-sided Faraday cage made of steel screen (essentially a box without a front face, to allow quick access to the equipment inside) was constructed around the apparatus to reduce electrical interference from the surrounding environment; however, preliminary tests indicated that this was likely excessively precautious, as even the use of electromagnetic equipment (solenoid valves, pumps) appeared to have no measurable effect except when used in extreme proximity (<0.3m) to the signal conditioning equipment (data not shown). A ‘reference probe’ was built into the side of the Plexiglas cylinder and was used to take measurements simultaneously with the probe on the test fiber. The purpose of the reference probe was to provide a reference signal for monitoring the electrochemical solution and alerting the author to any aberrations or trends in measurements collected using the side loop membrane apparatus. The side loop membrane apparatus was used to collect measurements in duplicate for each probe prior to the use of these probes within the pilot-scale system. Duplicate testing involved collecting two successive series of measurements for each probe, with cleaning of the probes and anode (using wet Kimwipes® and wet cotton swabs) performed between the two series of tests. A second duplicate set of measurements (performed at random) was collected from ten percent of probes to allow greater estimation of variance among duplicate measurements. Subsequent to use with the pilotscale system, all probes were again subjected to duplicate measurements using the side loop membrane apparatus, to allow observation of whether the probe voltage correction factors (Table 4.1, and Appendix 4) had changed during pilot testing.  24  3.4  Pilot-scale membrane system with full-scale membrane modules  3.4.1 Mechanical systems Pilot-scale tests were performed using a GE-Zenon Zeeweed-500c system, provided by GE-Zenon. Previous research has indicated that the performance of this pilot system, which contains full-scale membrane modules, can be scaled to reflect the performance of full-scale Zeeweed-500 systems of similar module type (source: correspondence with GE-Zenon). The pilot system (see Figures 3.6 and 3.7) contained one cassette consisting of three full-scale membrane modules submerged within a stainless steel membrane submersion tank, with each module containing several hundred individual membrane fibers.  Figure 3.6  Pilot apparatus mechanical systems.  25  Figure 3.7  Photo of pilot apparatus mechanical systems: Zeeweed-500c cassette.  The supporting frame with and without the Zeeweed-500c membrane cassette. The stainless steel frame was constructed with built-in pneumatics (pressurized gas lines can be seen running down the side of the frame) to allow module spacing and fiber tension to be easily controlled remotely.  A pneumatic control system was used to control both the spacing of the modules and the tension of the membrane fibers by varying the distance between the top and bottom bulkheads of the cassette. A pneumatic control box was used to control the supply of pressurized nitrogen gas to composite plastic actuator cylinders (Figure 3.8) located both above and below the membrane cassette on a supporting stainless steel frame (Figure 3.7). Such an externally controlled system was required to ensure quick changes in configuration without having to remove the membrane cassette from the system. The three module spacings and fiber tensions that could be generated by the pneumatic control system are shown in Table 3.1, while the position of the modules with respect to the gas diffusers is shown in Figure 3.9. The fiber tensions and module spacings investigated were selected based on those recommended by GE-Zenon (Cui et al., 2003; personal correspondence with GE-Zenon representatives).  26  Figure 3.8  Close up of a single pneumatic actuator cylinder.  The cylinders used during pilot testing were Premair ® composite cylinders made from plastic and stainless steel, operated using compressed nitrogen gas. Note that the reddened metal shown in the picture is paint, not rust.  27  Table 3.1  Fiber tensions, module spacings, and sparging settings for the pilot-scale membrane apparatus.  Condition FiberExamined Tension  Descriptor Loose5  Medium5 between modules) Tight (Vertical distance  Module Spacing2 Wide (horizontal distance Medium between modules) Narrow5 Sparging Settings High5  Set-point 160 cm (95.4%)1 164 cm (97.8%)1 167 cm (99.6%)1 12.3 cm 8.26 cm 6.35 cm 9 cfm (15.3 m3/hr)  Medium  6 cfm (10.2 m3/hr)  Low  3 cfm (5.1 m3/hr) 3 sec  Fast alternating3  Slow alternating3 6 sec 3 sec on, 3 sec off Fast pulse4 Slow pulse4  6 sec on, 6 sec off  Notes: 1. Percent tension values were calculated as the distance between the top and bottom bulkheads of the cassette divided by the maximum fiber length (167.6 cm). 2. The horizontal module spacing was measured between the center of the inner module and the center of one of the outer modules. The width of each module was approximately 5 cm. 3. Alternating flow between the two diffusers 4. Alternating flow on and off (both diffusers on, then both diffusers off) 5. Typical (manufacturer suggested) operating configurations for Zeeweed-500c systems: fiber slack of 1-5%, narrow spacing (6.0 cm, center to center), high sparging rate (~15 cfm (25.5 m3/hr)).  Gas sparging of the system was accomplished using two diffusers, provided with the Zeeweed-500c pilot system. Diffusers were made from PVC pipes, each perforated with eleven 0.5 cm diameter holes, lain 0.20 meters below the membrane cassette. Nitrogen gas was provided to the diffusers by a rotary vane compressor which recirculated nitrogen gas from the headspace of the membrane submersion tank. The gas  28  recirculation rate was controlled using a variable frequency drive connected to the compressor motor, which allowed control of the total sparging rate from 3 to 9 cfm (5.1 m3/hr to 15.3 m3/hr). Sparging rates were selected largely because they were similar to those suggested by GE-Zenon for the pilot system used (~15 cfm), but also because they were significantly different from one another, and thus more likely to generate discernible differences in shear stress within the membrane cassette. Solenoid valves located within the pipes conducting gas to the diffusers allowed for each of the two diffusers to be operated independently (Figure 3.6), allowing the effects of several different sparging patterns to be explored (see Table 3.3). Two different types of sparging patterns were used: pulse sparging (a period of sparging followed by a period without any sparging) and alternating sparging (sparging alternated between the two diffusers, with sparging provided to only one diffuser at a time). These two patterns were selected because each has a distinctive benefit: pulse flow uses only half the total volume of gas of continuous sparging, and alternating flow provides each diffuser with twice the gas flow rate of continuous sparging while using approximately the same total volume of gas. Each of these two patterns was tested with two different time set points, three seconds and six seconds, with equal periods of on/off (for simplicity). Three seconds was the shortest set point used when investigating the effects of sparging pattern, as it was observed that three seconds was required to allow gas flow to pass from the solenoid valves down to the diffusers and then up to the water surface as bubbles. Six seconds was selected as the longest set point because it was observed to be long enough in duration to allow flow circulation to develop within the pilot apparatus, but short enough that several cycles could be performed within each 45 second measurement. The stainless steel membrane submersion tank was sealed at the top with a plastic lid bolted to the tank, and though the system was nearly airtight, compressed nitrogen cylinders were used to maintain a slight positive pressure within the system throughout testing to ensure oxygen was continuously excluded from the system.  29  Figure 3.9  Top view (looking down) of the diffusers and membrane modules.  All three module spacings are illustrated: narrow (A), medium (B), and wide (C). Only half of the system is shown, the other half is a mirror image on the other side of the cassette centerline. The diffusers were located 0.20 m below the bottom of the modules.  30  3.4.2 Pilot electrical system and data acquisition The electrochemical circuit within the pilot apparatus consisted of sixty shear probes (operated as cathodes) and two platinum anodes. The anodes, made from two 8 cm platinum wires, had a combined surface area 210 times that of the total active cathode area to ensure that the current density was much higher at the cathodic surfaces than at the anode. The sixty test fibers were placed within the membrane cassette, while the two anodes were attached to the non-conducting plastic surface of one of the pipes used to supply gas to the diffusers (Figure 3.10). Each cathode was thus within 0.45 meters of an anode. A potential of 0.310 V ( ± 0.002 V) was applied between the cathodes and anodes to generate limiting current conditions at the surface of the probes. The stainless steel membrane submersion tank was placed on non-conductive material to ensure that it was electrically floating and not an active component of the electrical circuit. Figure 3.10 illustrates the electrical system used to collect measurements using the pilot apparatus, while Figure 3.11 provides a schematic of the electrical system.  31  Figure 3.10  Diagram of the electrical components of the pilot apparatus.  For simplicity, only four channels are shown per switch (rather than ten), and only three electrical connections (rather than sixty).  Great care was taken to ensure that only stainless steel and plastic components were in contact with the electrolyte solution, as suggested by Son and Hanratty (1969) and Reiss and Hanratty (1962). It should be noted that the use of other metals (such as aluminum) in the system can not only cause rapid corrosion, but can greatly interfere with electrical current measurements collected using the electrochemical method. To minimize potential impacts of corrosion, all stainless steel components in contact with the electrolyte solution (including the inside of the membrane submersion tank), were sprayed with two coats of clear acrylic paint (Krylon crystal clear®). A computer with Signal Express software (National Instruments, New York City) was used to collect and store data at 500 Hz, with one minute of data being collected from each probe for each configuration tested. Due to circuit limitations within the signal conditioning equipment, a switch box with six switches, each with ten channels, was used 32  (see Figures 3.11 and 3.12). The switchbox was used to follow a set sequence which determined which six probes were concurrently measuring shear stress during any given minute. As six probes were active at any given moment, and each probe collected data over a period of one minute, it took ten minutes to collect data from all sixty probes for each of the configurations tested.  33  Figure 3.11  Schematic of the electrical circuit within the pilot apparatus.  Measurements were collected from six probes concurrently, with a switchbox used to allow measurement collection from all sixty probes.  34  Figure 3.12  Photo of the 60-channel switch box connected to the 60 test fibers.  The switch box allowed all 60 probes to be tested sequentially, with 6 probe signals simultaneously directed to the conditioning apparatus.  3.4.3 Probe locations Shear probes (on test fibers) were inserted into the Zeeweed-500c membrane cassette in a five-layered grid (see Figures 3.14, 3.15, and A9.2). Test fibers were held in place at the upper and lower bulkheads using small cable ties and epoxy to adhere each test fiber to the base of the adjacent membrane fiber. The sixty probes were all placed within one quadrant of the cassette, as system symmetry allowed interpolation of results within the other three quadrants of the cassette (see Figure 3.13). The free length of each test fiber was carefully controlled to be equal to that of the adjacent membrane fibers, such that the tension of each test fiber was comparable to that of its surrounding fibers. It was not practical to position the probes such that they faced in any specific direction, due to the length and flexibility of the test fibers. Thus the directional orientations of the probes were random.  35  Figure 3.13  Probes were placed within one corner of the cassette.  Due to system symmetry, measurements from one corner of the membrane cassette could be used to interpolate the shear stress throughout the entire system. As sixty probes were used, this gave a total of 240 points. The dimensions of the tank, within which the cassette was submerged, were 2.16 m (Y) by 0.85 m (X) by 0.47 m (Z).  36  Figure 3.14  Side views of probe locations within the cassette.  Probes were placed in five vertical planes (sheets) within the cassette. Sheets 3-5 within one of the outer modules, and sheets 1-2 within the inner module. Heights (above bottom bulkhead): R4= 0.10 m, R3= 0.59 m, R2= 1.08 m, R1= 1.57 m. Total distance between top and bottom bulkhead when fibers fully extended= 1.67 m. ‘A’ was located at the cassette edge, ‘B’ 0.18 m from ‘A’, and ‘C’ (at the cassette centerline) 0.36 m from ‘A’.  Figure 3.15  Top view of probe locations within the cassette.  Probes within vertical sheets 2, 3, and 5 were located within the layer of membrane fibers nearest the edge of their respective modules. Probes in sheets 1 and 4 were located directly in the center of their respective modules, and thus surrounded on all sides by membrane fibers (each module was approximately ten membrane fibers in width, with fibers epoxied in place within a 2.5 cm region in the center of each module bulkhead).  37  3.4.4 Pilot apparatus experimental procedures All experiments were performed over a single continuous forty-three hour period in order to minimize the degradation of the ferri/ferrocyanide electrolyte solution. All experimental configurations tested are summarized in Tables 3.2 and 3.3. A total of 35 (27+8) experimental configurations were examined. Each configuration was tested in triplicate as part of a randomized testing procedure. Before the start of the experiments, the probe surfaces were thoroughly washed using de-ionized water and Kimwipes® (non-abrasive laboratory-grade paper towel) to ensure their cleanliness. The membrane cassette was lowered into the stainless steel membrane submersion tank, which was then filled with tap water and sealed with a plastic lid bolted to the top of the tank. The system was sparged with nitrogen gas for 24 hours to strip oxygen and chlorine prior to the addition of concentrated ferricyanide, ferrocyanide, and potassium chloride solutions. Sparging with nitrogen was continued for an additional three hours until the dissolved oxygen within the side-loop system was lowered below 7% of saturation. Dissolved oxygen measurements for the solution in the membrane submersion tank were made using the side-loop apparatus (see Section 3.4.5) As the temperature within the pilot apparatus was 16.5 ºC, a diffusion coefficient of 6.6 x 10-10 m2/sec and a dynamic viscosity of 0.001 Pa•sec were used. The viscosity at 16.5 ºC was determined from the viscosity at 25ºC (0.00081 Pa•sec) using tables from Munson et al. (1990), and the diffusion coefficient at 16.5 ºC was determined from the diffusion coefficient at 25 ºC (8.36 x 10-10 m2/sec) using equation 3.1 (Legrand et al., 2000).  D 298K µ 298 K D 289.5 K µ 289.5 K = T 298 K T 289.5 K  (3.1)  While a decision was made to present most shear stress data in volts to avoid reducing the accuracy of collected data (discussed further in chapter 4), certain voltage data were converted to Pascals using equations A4.1 to A4.4 given in Appendix 4.  38  Table 3.2  Configurations examining the effect of module spacing, fiber tension, and sparging rate on shear stress. Tension: Spacing: Gas flow: Loose Narrow Low Medium High Medium Low Medium High Wide Low Medium High Medium Narrow Low Medium High Medium Low Medium High Wide Low Medium High Tight Narrow Low Medium High Medium Low Medium High Wide Low Medium High Total:  Name: LNL LNM LNH LML LMM LMH LWL LWM LWH MNL MNM MNH MML MMM MMH MWL MWM MWH TNL TNM TNH TML TMM TMH TWL TWM TWH  27 configurations  *(gas flow continuous for all configurations listed here)  39  Table 3.3  Configurations examining the effect of sparging pattern on shear stress. Sparging Pattern: Fast Alternating  Module spacing: Narrow Wide Slow Alternating Narrow Wide Fast Pulse Narrow Wide Slow Pulse Narrow Wide Total:  Fiber Gas flow: tension: Medium Medium Medium  Medium  Medium  Medium  Medium  Medium  Name: FA.MNM FA.MWM SA.MNM SA.MWM FP.MNM FP.MWM SP.MNM SP.MWM  8 Configurations  Definitions: •alternating = gas provided to one diffuser, then the other •pulse = gas provided to both diffusers, then neither •fast = 3 second intervals, slow = 6 second intervals  40  3.4.5 QA/QC For all experimental configurations examined, tests were performed in triplicate. All tests were essentially performed in a randomized testing procedure. However, tests with discontinuous gas sparging (Table 3.3) were performed after those with continuous sparging (Table 3.2) were completed, as the tests with discontinuous gas sparging required the use of 100% compressed nitrogen gas for sparging since the rotary vane gas compressor could not operate in an on-off mode. The Plexiglas column used in the side loop membrane system (described in Section 3.3) was hooked up to the pilot-scale membrane system to monitor the condition of the ferri/ferrocyanide electrolyte solution. The column was utilized in a manner similar to that used during the probe standardization procedure, with the column subjected to cross-flows of 0.017 to 0.065 m/sec with voltage measurements collected from the reference probe for each flow rate considered. Any change in the signal from the reference probe in the side loop system for a given cross-flow rate would indicate that the characteristics of the ferri/ferrocyanide electrolyte solution had changed. The characteristics of the ferri/ferrocyanide electrolyte solution were monitored after every 10th experimental configuration had been tested (approximately once every 4 hours). The electrical system (i.e. cathode and anode) in the side-loop column was disconnected when not being used to prevent possible interference with the electrochemical circuit within the pilot-scale membrane system. A thermometer and a dissolved oxygen probe (DO probe) were also installed in this side loop so that the temperature and dissolved oxygen concentration of the ferri/ferrocyanide electrolyte solution could be monitored. Ten ‘blank’ tests (gas sparging turned off in the pilot apparatus) were performed within the randomized testing procedure. The purpose of these tests, which otherwise followed the same procedures as all other pilot tests, was to identify gas leaks within the pneumatics as well as trends in probe voltage measurements collected over the forty-three hour testing period.  41  CHAPTER 4 EXPERIMENTAL RESULTS  4.1  Measurements taken within the pilot apparatus As previously discussed, measurements were collected for 6 probes at a time, for  a period of 1 minute, taking 10 minutes to collect measurements from all 60 probes, for each experimental configuration. Considering that the measurements were collected at a rate of 500 Hz, for 35 different experimental configurations, each of which was tested in triplicate, a total of 189,000,000 data points was collected during the pilot-scale tests. Note that for each of the one minute measurement periods, the first ten seconds and last five seconds were excluded from the data analysis presented in Chapter 5. Typical measurements collected over a ten minute period are presented in Figure 4.1. Note that the voltage spikes in Figure 4.1 which occur every minute correspond to switching measurement acquisition between different probes using the switch box (Figure 3.12). The measurements within the dashed red line correspond to an example of those that were used for data analysis. A close-up of this area is provided in Figure 4.2.  42  Figure 4.1  Typical raw unsorted voltage data.  Data shown were from one of the six circuits used for taking measurements within the pilot apparatus. Note the ten distinct signals, each one minute in length and each corresponding to a single shear probe. The data indicated by the dotted red line is expanded in Figure 4.2.  43  Figure 4.2  Typical 45 seconds of voltage data collected from a single shear probe.  The data shown correspond to the highlighted section of Figure 4.1.  The voltage readings collected by the shear probes can be converted to shear stress (Pa) using equations A4.1 to A4.4 given in Appendix 4. However, as the calculations used to convert voltage signals (V)v to shear stress (Pa) values involve the use of a cubic term, which greatly amplifies any error within the data, it was decided to express the majority of shear stress measurements in volts. Thus, when reviewing the shear stress results presented in Chapters 4 and 5, it should be realized that relatively small differences in voltage measurements can indicate relatively large differences in shear stresses experienced at the membrane surface (see Appendix 1 for the relationship between voltage and shear stress for the system used during pilot testing).  44  The scope of the present project was such that the vast amount of collected data was largely summarized as mean (average) values. While information is lost in the averaging process (see Figure 4.2 for details from a single probe measurement), this allowed a single average voltage value to be derived for each probe for each configuration. It should be noted that because voltage is proportional to the cube-root of the shear stress, the shear stress calculated using the mean voltage measurement does not equate to the mean shear stress, but instead to a value somewhat lower than the mean shear stress. Nonetheless, as we are predominantly concerned with the observation of trends, the use of mean voltage data is appropriate. For observation of trends, data were visualized as shear stress contour maps (Appendix 9a and 9b) using Matlab software, and as summarizing bar graphs using Microsoft Excel (see Appendix 9c for bar graphs, Appendix 8 for Matlab codes). All presented data are the average of triplicate measurements. The average shear stresses observed corresponded well with the findings of Bérubé et al. (2006a) and Chan et al. (2007), who measured surface shear stresses in lab and bench scale gas-sparged submerged hollow fibers membrane systems, with average measurements generally between 0.3 V and 0.5 V (0.3-1.2 Pa). Peak shear measurements were generally observed to be near 1 V (9.8 Pa), similar to the 6 Pa peak measurements reported by Bérubé et al. (2006a) and the 8 Pa peak measurements reported by Chan et al. (2007). Thus, the majority of the peak shear events observed within the submerged HF system were approximately one order of magnitude higher than the average shear stress. These peak measurements likely coincided with the passing of bubble vortices in the vicinity of the probes, as observed by Chan et al. (2007) and Nakoryakov et al. (1989), though some may have been be due to physical contact between fibers, a mechanism observed by Bérubé and Lei (2006).  45  4.2  QA/QC results  4.2.1 Probe standardization using the side loop membrane apparatus The probe voltage correction factors, derived using the methods described in Appendix 4 and 3.3.3, are given below in Table 4.1. Table 4.1  Table of probe voltage correction factors.  Probe number  Probe name  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30  R1A R1C R1F R1G R1H R1i R1J R1L R1M R1N R1P R1Q R1R R1T R1U R2A R2B R2C R2D R2E R2F R2H R2i R2J R2K R2L R2M R2N R2O R2P  Voltage correction factor 1.03 0.99 1.00 damaged damaged 1.08 0.95 damaged 1.08 1.04 1.08 1.01 1.06 0.91 1.09 damaged 0.85 1.00 0.99 1.06 0.96 0.94 1.05 0.99 damaged 1.03 1.10 0.99 1.01 1.11  Probe number  Probe name  31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60  R3 beta R3i R3J R3M R3N R3O R3P R3R R3S R3T R3U R3V R3Y R3Z R3 alpha R4C R4D R4E R4F R4G R2Q R4i R4J R4K R4L R4W R4N R4P R4Q R1V  Voltage correction factor 0.98 0.92 0.92 0.98 1.05 0.96 1.04 1.11 1.07 1.07 1.08 1.13 1.12 1.14 1.20 1.00 0.90 1.04 0.93 0.94 1.02 0.98 1.16 1.13 1.02 1.10 1.09 1.12 1.18 1.07  46  Five of the sixty shear probes were suspected to be damaged. None of the measurements from these five probes were used in any of the presented results. Within the shear maps given in Appendix 9, data for the locations of these five probes were replaced with interpolated points, which were calculated as the average of the nearest five probes (the nearest two within the same horizontal sheet, and the nearest three within the vertical sheet). While these ‘damaged’ probes performed appropriately prior to being used in pilot testing, they behaved erratically during the set of probe standardization tests which were undertaken subsequent to pilot testing. These probes tended to display either extremely high voltages (many times higher than normal) or none at all. While the exact cause of malfunction can only be speculated upon, the most likely causes were leaks around the edge of the probe face and breaks in the solder where the platinum wire was soldered to the electrical wire which conducted electrical current measurements to the signal conditioning apparatus (see Figures 3.1 and 3.3). The former malfunction would greatly increase the probe surface area by allowing the electrolyte to access the sides of the probe, while the latter would simply disconnect the probe from the signal conditioning apparatus. Leaks near the probe also could have allowed the electrolyte solution to contact where the platinum was soldered to the signal-conducting electrical wire (though this would have been rare due to the abundance of epoxy). The reader may also note from the discontinuity in probe names within Table 4.1 that a substantial number of probes were discarded prior to pilot testing, due to their observed malfunction during the set of probe standardization experiments which were undertaken prior to pilot testing (in all 85 probes were manufactured, while only 60 were deemed adequate for use in pilot testing). Duplicate measurements using the side loop apparatus indicated that the voltage correction factors determined prior to pilot testing had a standard deviation of approximately 3% of the mean (the mean of duplicate measurements were used to calculate the voltage correction factors). Similar results were observed for the voltage correction factors determined using measurements collected subsequent to pilot testing. These results indicate that the probes on each of the test fibers gave consistent readings both before and after pilot testing. However, there were significant differences between the sets of voltage correction factors determined before and after pilot testing, with 47  differences for individual probes ranging from approximately -10% to +50% of the original value for the sixty test fiber probes. The source of these differences is uncertain. It is unlikely that poisoning (i.e. fouling) of the probes occurred, since this would have predominantly reduced the voltage correction factors. It is more likely that the probes were inadvertently modified during the numerous trials that were performed using the pilot-scale system, before actual measurements began. These trials were aimed at ‘debugging’ the mechanical and electrical components of the pilot-scale system. Approximately five trials were performed over a period of about three months during this ‘debugging’ process. In hindsight, it would have been more prudent to perform these trials before the voltage correction factors were initially determined. Since the voltage correction factors determined immediately after pilot testing gave consistent readings, they were used to adjust the readings obtained during pilot testing (Table 4.1). As mentioned, the values given in Table 4.1 are mean values with standard deviations of approximately 3% of the mean. Methods which can be used to rapidly estimate voltage correction factors in situ, such as voltage-step techniques (Gaucher et al., 2002a; Sobolik et al., 1998), should be considered in future research.  4.2.2 QA/QC during pilot testing Dissolved oxygen measurements collected using the side-loop apparatus indicated that dissolved oxygen remained below 6% of saturation (saturation = 8.6 mg/l during experimental conditions) throughout pilot testing, except for two brief peaks (of 14% and 17% of saturation) which corresponded to times when the system was briefly exposed to atmospheric pressure to allow fittings to be changed within the tubing which directed nitrogen to the diffusers (see Figure 3.6). The peaks in dissolved oxygen did not discernibly affect the side loop voltage measurements, so were not believed to be a concern. Temperature (measured with the side loop) fluctuated between 16.5 ºC and 17 ºC throughout the stage of testing which used continuous sparging, so no temperature  48  correction factor was applied to these voltage measurements. However, the temperature was consistently lower (16.0 ºC) during the sparging tests (tests using pulse and alternating sparging). Voltage measurements collected from the reference probe in the side loop were on average 3% lower during sparging tests in comparison to average measurements collected using continuous sparging. This observation is consistent with those of Berger and Ziai (1983), who reported a drop in limiting current of 6.7 % for each decrease of 1 ºC. To allow the results of sparging tests (tests using pulse and alternating sparging) to be compared with those from tests using continuous sparging, a correction factor was applied to increase all voltage measurements by 3.3 % for tests which used discontinuous sparging. The results of the ten randomized blank tests (described in section 3.4.5) indicated that there was a minor nitrogen gas leak (a few bubbles per minute) from one of the pneumatic actuators below the cassette. However, this leak was negligible compared to the gas flow rates used for sparging, and did not impact the operation of the actuators (verified visually subsequent to pilot testing), thus it was not a concern.  4.3  Data limitations Review of raw data measurements, visualized using Matlab software, (see Figure  4.2 for an example) revealed that many of the peak probe measurements (typically in the 1-2 V range) could be attributed to high frequency noise events of uncertain origin (see Figure 4.3). As these events occurred infrequently (generally 0-5 times per one-minute probe measurement) and were extremely short in duration (0.002-0.01 seconds) their effect on the mean signal measurements was negligible (less than 0.1% of the mean signal measurement), thus the raw data were not passed through a low-pass filter to eliminate these infrequent events prior to analysis. While the source of this noise is not known, it occurred more frequently during the pulse and alternating sparging patterns and at a recurrence of approximately 3 seconds, suggesting that the probable source was the on/off charging cycle of the solenoid valves used to control gas flow to the diffusers (see Figure 3.6, and Figure A7.4 in Appendix 7).  49  Figure 4.3  High frequency noise in the raw voltage data, an example.  The figure illustrates 7 seconds of data from one probe, indicating the clear difference between ‘true’ and ‘false’ (noise) peak voltage measurements.  When comparing the triplicate measurements for any given configuration, the standard deviation for the average voltage measurement of each probe was relatively high (generally 10-15% of the probe-average voltage for each probe), indicating that longer periods of measurement (more than one minute) would have been valuable to even-out the fluctuating voltage signals and provide a more accurate estimation of mean voltage. The largest uncertainty within the data likely lies in the voltage correction factors used to normalize all probe measurements (Table 4.1). Duplicate measurements taken during probe standardization procedures indicated that voltage correction factors could only be determined to within ± 6% (two standard deviations, data not shown), despite the fact that considerable effort was made to minimize this source of error.  50  CHAPTER 5 DATA ANALYSIS AND DISCUSSION Trends observed in the data presented in Appendix 9 are summarized here in chapter 5. However, the reader is encouraged to look through the data in Appendix 9, as great effort has been made to make these data visibly accessible through the use of contour ‘shear maps’ which contrast various configurations side-by-side to allow straightforward identification of trends and differences. Note that while nearly all presented data are in volts, due to the established relationship between voltage and shear stress (see appendices 1 and 4) the terms “voltage” and “shear stress” are used interchangeably within this chapter. Three levels of averages were used during trend analysis: (1) the individual probeaverage voltages, which are the averages of the 45 seconds of measurements collected from each probe; (2) the sheet-average voltages, which are the average of all probe measurements within a sheet (cross-sectional plane); and (3) the cassette-average voltages, which are the average of all probe measurements within the entire cassette. These three levels of averages, from the most local (individual probes) to the most general (the entire cassette), were then used to compare the thirty-five experimental configurations tested, allowing differences and trends to be identified. The most general data (cassette wide) are presented in Table 5.1 and Figure 5.1. The full inventory of sheet-averages (averages within cross-sectional planes through the membrane cassette) are given in Appendix 9c as bar graphs. The full inventory of individual probe averages, which provide the most localized information regarding shear stress within the cassette, are shown in appendices 9a and 9b within colour shear stress contour maps (units in volts). Typical examples of data found in Appendix 9 have been included in the relevant sections of chapter 5; however, the reader is directed to the first few pages of appendix 9 for a brief guide which provides instructions on how to read shear stress maps. The variability of shear stress within the cassette (ie. variation among all sixty individual probe measurements) was estimated for each configuration using the standard deviation of the cassette-average voltages. Use of standard deviations was deemed to be reasonable for the estimation of variability of shear stress within the cassette, as for 80%  51  of the experimental configurations investigated the sixty probe means could be approximated as normally distributed (the average skew for the thirty five configurations investigated was 0.32, with an average kurtosis of 0.16); however, for 20% of the experimental configurations investigated the sixty probe means were slightly right skewed (for these seven configurations skew varied from 0.66 to 0.93) (data not shown). The standard deviations of the cassette-average voltages, presented in Table 5.1, were analyzed using the Wilcoxon paired-sample test (Zar, 1999) to determine the effect of each experimental configuration on the variation of the shear stress within the cassette. The 27 tested configurations that used continuous sparging are ranked in Table 5.1 and Figure 5.1, both by cassette-average shear stress (ie. average shear stress magnitude) and by standard deviation of the cassette-average shear stress (ie. evenness of distribution of shear stress within the cassette). Ranking in this manner provides rapid identification of 2 key results: (a) which experimental configurations generated the highest shear stress, and (b) which generated the most evenly distributed shear stress within the membrane cassette.  52  Figure 5.1  Cassette-averaged shear stress for each tested configuration.  Cassette-averaged shear stress (Pa) in Figure 5.1 was calculated from voltage averages (given in Table 5.1) and equations A4.1 to A4.4 given in Appendix 4 (for a temperature of 16.5ºC, diffusion coefficient and viscosity given in section 3.4.4). Note that error bars (two standard deviations) reflect the variation in average shear stress values (Pa) among the triplicate measurements collected for each configuration, not the range of shear stresses experienced across the cassette (ie. among the 60 probes).  53  Table 5.1  Membrane cassette configuration  Ranking configurations by cassette-average voltage and standard deviation of cassette-average voltage Cassetteaverage voltage (Volts) -0.48 -0.47 -0.44 -0.43 -0.43 -0.43 -0.42 -0.41 -0.40 -0.39 -0.38 -0.38 -0.38 -0.37 -0.37 -0.36 -0.35 -0.35 -0.35 -0.34 -0.34 -0.34 -0.33 -0.31 -0.31 -0.30 -0.28  Rank (from highest to lowest)  Parameter  Range  Abbreviation  Fiber tension:  loose medium tight narrow medium wide low medium high  mwh lwh mwm lwm lmh twh twm mmh tmh mwl lmm lwl mmm mnh lnh tmm twl tnh mnm lml lnm mml lnl tnm tml mnl tnl  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27  Membrane cassette configuration mml lnl lnh mnh lmm lnm tnh tnl mmm mmh tnm tmm mnm mnl tmh tml lmh lwm lwl lml mwl twl lwh twm mwm mwh twh  Standard deviation among all probes (Volts)  Standard deviation among all probes (%)  0.08 0.07 0.08 0.07 0.08 0.09 0.08 0.07 0.10 0.09 0.09 0.08 0.09  22% 22% 22% 22% 23% 23% 23% 24% 24% 24% 25% 25% 26%  0.11 0.08 0.09 0.12 0.12 0.10 0.11 0.13 0.12 0.17 0.15 0.16 0.18 0.17  27% 27% 28% 28% 28% 29% 29% 34% 34% 35% 35% 37% 38% 39%  Rank (from lowest to highest) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27  Legend:  Module spacing: Gas flow rate:  L M T N M W L M H  Value 95.4% 97.8% 99.6% 6.35 cm 8.26 cm 12.38 cm 3 cfm 6 cfm 9 cfm  The configuration naming sytem is: "fiber tension/ module spacing/ gas flow rate", example: TNM = tight fibers/ narrow module spacing/ medium gas flow rate  The cassette-average voltage is the average of measurements from all sixty probes, and is an approximation of the overall shear stress experienced within the cassette. The standard deviation of the cassette-average voltage (given in percent of the cassette average voltage) reflects the variation among the mean voltage measurements from all sixty probes, and thus reflects the distribution of shear stress within the cassette.  54  From Table 5.1 and Figure 5.1 it is immediately apparent that there were significant differences in the cassette-average shear stress measured under different experimental conditions. These significant differences, which cannot be attributed solely to differences in sparging rate, indicate that an opportunity exists to optimize system configuration to provide higher and more evenly distributed shear stress. For example, configuration MWM exhibited approximately three times the cassette-average shear stress of configuration TNM (see Figure 5.1), though both configurations used the same gas flow rate for gas sparging (see the bottom of Table 5.1 for a legend to the configuration naming system). Sections 5.1 and 5.2 and their subsections discuss the observed effects of each experimental configuration on shear stress: sparging rate, module spacing, fiber tension, sparging pattern, and location within the cassette. Each section begins with the examination of the most general data (cassette-average voltages) and then discusses more localized observations. Trends were identified visually and their significance investigated using the Wilcoxon paired-sample test. The reader should note that within this document the term ‘significant’ is used exclusively to refer to statistical significance, and should not be mistaken as commentary regarding the magnitude of the observed trends.  5.1  Experimental configurations with continuous sparging  5.1.1 Relationship between average shear stress and variation of the shear stress for individual probe measurements For the experimental configurations investigated that utilized continuous sparging, a linear relationship between the mean shear stress and the variability (i.e. standard deviation) in the shear stress was observed. For individual probes, typical means and standard deviations are presented together in Figure 5.2 for ten different experimental configurations which used continuous sparging. As previously discussed in  55  section 2.2, there is presently uncertainty regarding whether high average shear stress or highly variable shear stress is preferable for the removal of foulants in membrane systems. The results of Figure 5.2 indicate that in submerged HF systems with continuous sparging, the use of mean shear stress is a reasonable surrogate for estimating the variability of the shear stress.  Standard deviation of probe measurements (Pa)  7  6  5  4  3  y = 0.68x  2  2  R = 0.91 1  0 0  1  2  3  4  5  6  7  Means of probe measurements (Pa)  Figure 5.2  Typical mean shear stresses vs. their standard deviations for configurations using continuous sparging.  Each point represents the mean and standard deviation of 45 seconds of measurement from a single shear probe (600 points shown = 60 probes x 10 configurations). Note the clear trend of increasing standard deviation as the mean shear signal increases (slope of approximately 0.7:1).  To generate Figure 5.2, all measurements (volts) for ten tests (which used continuous sparging) were converted to Pascals using equations A4.1 to A4.4 (for a temperature of 16.5ºC, diffusion coefficient and viscosity given in section 3.4.4) (see Appendix 4 for equations, Appendix 8 for Matlab code). Means and standard deviations of these means were subsequently derived for each probe from these measurements.  56  5.1.2 The effect of sparging rate The cassette-average results in Figure 5.3 indicate that configurations with a higher sparging rate exhibited higher cassette-average shear stress than those that used a lower sparging rate. Use of the Wilcoxon paired-sample test with the standard deviations of cassette-averages (ie. the variation among the average measurements of all 60 probes, Table 5.1) indicated that increases in shear stress caused by increases in sparging rate were evenly distributed across the cassette (ie. there was no significant difference (α~0.5) among the standard deviations of the cassette-averages for the different experimental sparging rates). Thus within the range of sparging rates investigated, the rate of sparging did not significantly affect the variability of the shear stress profile within the cassette. The trend of increasing shear stress as sparging rate was increased was also observed at the sheet-average level, as presented in Figure 5.4 (see also Figures A9.1c to A9.9c). As the sparging rate was increased, the shear stress in all vertical and horizontal sheets increased. The Wilcoxon paired-sample test indicated that this trend was significant both when sparging rate was increased from 3 cfm to 6 cfm (α<0.001), and from 6 cfm to 9 cfm (α<0.001), indicating that within the range of sparging rates used, increased sparging rate consistently generated greater shear stress. When considering probe-average shear stresses, increases in sparging rate appeared to cause especially large increases in shear stress within several localized regions of the cassette, as can be noted from Figure 5.5 (see also A9.1a-A9.9a and A9.1bA9.9b). Nonetheless, in general, increases in sparging rate increased probe-average shear stresses across the entire cassette, with the exception of the lower region of the inner module, for which increases in sparging rate did not result in increased probe-average shear stress (see bottoms of vertical sheets 1 and 2 in Figure 5.5, as well as section 5.1.6). The results of Nguyen et al. (2008) illustrate that in submerged systems increased sparging rate results in increased bubble velocity and fluid velocity, increased bubble number, and a slight increase in gas holdup. All three of these factors likely contributed to the observation in the present study that increased sparging rate resulted in increased shear stress.  57  Engineering Implications: Increasing sparging rate increases shear stress at the membrane surface, with increases in sparging rate having an insignificant effect on the variability of the shear stress profile within the cassette.  Cassette average shear stress (Pa)  1.2  LM LN  1  LW  0.8 MM MN  0.6  MW  0.4 TM  0.2  TN TW  0 low  medium  high  Gas flow rate  Figure 5.3  Effect of sparging rate on cassette average shear stress  In the legend in Figure 5.3 the first letter indicates the fiber tension (loose, medium, tight), while the second letter indicates the module spacing (medium, narrow, wide). Error bars (omitted from Figure 5.3 as they distract from the presented data) are shown in Figure 5.1 (a bar chart which displays the same data illustrated in Figure 5.3)  58  (a)  Medium tension, medium spacing, three gas flow rates -0.70  MML MMM MMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  (b)  Medium tension, medium spacing, three gas flow rates -0.70  MML MMM MMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure 5.4  Typical sheet-averages, illustrating effect of increasing sparging rate  Illustrated data are averages of measurements from probes within vertical (a) and horizontal (b) cross sections (‘sheets’) of the membrane cassette. See Figures A9.2 and A9.3 for diagrams describing sheet locations within the cassette. Error bars (two standard deviations) reflect the variation among the randomized triplicate measurements. 59  Figure 5.5  Typical shear maps illustrating the effect of increasing sparging rate.  Top row = low sparging rate; middle row = medium sparging rate; bottom row = high sparging rate. Shear map axis units are in meters. See Figure A9.2 and the first few pages of Appendix 9 for a guide to reading shear maps such as those shown in Figure 5.5.  60  5.1.3 The effect of module spacing The cassette-average results in Figure 5.6 indicate that configurations with wider spacing between modules exhibited higher cassette-average shear stress than those that used narrower spacing. Use of the Wilcoxon paired-sample test with the standard deviations of cassette-averages (ie. the standard deviation among all 60 probes, Table 5.1) indicated that wide module spacing produced a significantly more heterogeneous shear profile than either medium (α~0.01) or narrow (α~0.005) spacing. This indicates that increases in shear stress caused by increasing module spacing were not distributed evenly within the cassette, with some regions of the cassette experiencing proportionally larger increases in shear stress than others, as the distance between modules was increased. The trend of increasing shear stress as module spacing was increased was also observed at the sheet-average level, as presented in Figure 5.7 (see also Figures A9.10c to A9.18c). As module spacing was increased, the shear stress in all vertical and horizontal sheets increased, with the exception of vertical sheet 5 (the sheet closest to the wall of the tank within which the cassette was submerged). Use of the Wilcoxon paired-sample test indicated that this trend was significant both when module spacing was increased from narrow spacing to medium spacing (α<0.05) and from medium spacing to wide spacing (α<0.05), indicating that within the range of module spacings used, increased spacing between modules consistently provided greater shear stress. Figure 5.7 (and Figures A9.10c to A9.18c) indicates that the most substantial benefits of increased module spacing were experienced predominantly by sheets 1 to 4. As the two bottom diffusers remained stationary, the repositioning of modules had the effect of repositioning vertical sheets 3-5 (sheets in the outer module) with respect to the diffuser (see Figure 3.9). Thus it is not surprising that sheets within the outer module should have experienced greater shear stress with medium module spacing as compared to narrow spacing, as with medium spacing, the diffusers were located directly below the outer membrane modules. However, both the outer and inner module exhibited a further increase in shear stress when wide spacing was used (as compared with medium spacing), which placed the diffusers between the inner and outer modules (Figure 3.9). The shear stress experienced  61  by vertical sheet 5 (the outermost sheet) did not change to a great extent as module spacing increased, despite the fact that with a narrow spacing the diffuser was located directly beneath this sheet. This suggests that the changes in flow pattern induced by increases in module spacing were, in general, beneficial, and not simply a matter of shifting the scouring effect of bubbles from one region of the cassette to another. When considering probe-average shear stresses, increases in sparging rate appeared to cause especially large increases in shear stress within several localized regions of the cassette, as can be noted from Figure 5.8 (see also A9.10a-A9.18a and A9.10b-A9.18b). Nonetheless, in general, increases in module spacing increased the probe-average shear stresses, with the exception of the lower region of the inner module, for which increases in module spacing did not result in increased probe-average shear stress (see bottoms of vertical sheets 1 and 2 in Figure 5.8). Engineering Implications: Increasing module spacing increases shear stress at the membrane surface, though the increases in shear stress caused by increasing the distance between membrane modules are not distributed evenly within the cassette.  62  1.2 Cassette average shear stress (Pa)  L_L 1  L_M L_H  0.8 M_L 0.6  M_M M_H  0.4 T_L 0.2  T_M T_H  0 narrow  medium  wide  Module spacing  Figure 5.6  Effect of module spacing on cassette average shear stress  In the legend in Figure 5.6 the first letter indicates the fiber tension (loose, medium, tight), while the second letter indicates the sparging rate (low, medium, high). Error bars (omitted from Figure 5.6 as they distract from the presented data) are shown in Figure 5.2 (a bar chart which displays the same data illustrated in Figure 5.6)  63  (a) Medium tension, three module spacings, medium gas flow -0.70  MNM MMM MWM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  (b)  Medium tension, three module spacings, medium gas flow -0.70  MNM MMM MWM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure 5.7  Typical sheet averages illustrating effect of increasing module spacing.  Illustrated data are averages of measurements from probes within vertical (a) and horizontal (b) cross sections (‘sheets’) of the membrane cassette. See Figures A9.2 and A9.3 for diagrams describing sheet locations within the cassette. Error bars (two standard deviations) reflect the variation among the randomized triplicate measurements.  64  Figure 5.8  Typical shear maps illustrating the effect of increasing module spacing.  Top row = narrow spacing; middle row = medium spacing; bottom row = wide spacing. Shear map axis units are in meters. See Figure A9.2 and the first few pages of Appendix 9 for a guide to reading shear maps such as those shown in Figure 5.8.  65  5.1.4 The effect of fiber tension  The cassette-average results in Figure 5.9 indicate that configurations with tight fibers (ie. high fiber tension) experienced lower cassette-average shear stress than those that used medium or loose fiber tension. Use of the Wilcoxon paired-sample test with the standard deviations of cassette-averages (ie. the standard deviation among all 60 probes, Table 5.1) indicated that, though the trend was not statistically significant (α~0.07), loose fibers (4.6% fiber slack) appeared to provide a more even distribution of shear stress within the cassette than did tight fibers (0.4% fiber slack). The trend of decreased shear stress for tight fiber configurations was also observed at the sheet-average level, as presented in Figure 5.10 (see also Figures A9.19c to A9.27c). .Use of the Wilcoxon paired-sample test with the vertical-sheet average data confirmed that loose fiber configurations provided higher average shear stress than tight fibers (α<0.001), and that medium fiber configurations provided higher average shear stress than tight fibers (α<0.001). However, there was no statistically significant benefit of loose fibers over fibers with medium tension (α ~0.5), suggesting that 2-3% fiber slack may be sufficient to achieve the benefit of increased shear stress due to fiber looseness. GE-Zenon’s suggestion of 1-5% fiber slack (Cui et al., 2003) for operation of Zeeweed500 systems is thus likely sufficient to achieve the benefit of loose fibers. Note that percent slack is calculated as: 100% - 100*(distance between fiber ends) ÷ (maximum distance between fiber ends). When considering probe-average shear stresses, increased fiber slack appeared to have a homogenizing effect on shear stress within the cassette, decreasing the variability of the shear stress profile within the cassette, as observed in Figure 5.11 (see also A9.19aA9.27a and A9.19b-A9.27b). These results are consistent with those by Chan et al. (2007), who noted that looser fibers tended to allow rising bubbles to penetrate deeper into the module. Bérubé and Lei (2006) theorized that the benefit of looser fibers may be due both to increased fiber sway and to infrequent high-shear stress events caused by inter-fiber rubbing, which  66  they observed within their bench-scale submerged HF system. They conjectured that such inter-fiber rubbing could be a primary reason multi-fiber systems have been observed to provide up to 40% higher flux than single fiber systems (Bérubé and Lei, 2006). Analysis of infrequent high shear stress events (such as those potentially caused by inter-fiber contact) was beyond the scope of the present study. Engineering Implications: The use of tight fibers (less than 1% fiber slack) provides less shear stress than the use of fibers with more than 2% fiber slack. The use of fibers with more than 2% slack appears to provide more evenly distributed shear stress within the cassette (as compared to fibers with less than 1% slack), though this observation was not statistically significant.  67  1.2 Cassette average shear stress (Pa)  _NL 1  _NM _NH  0.8 _ML 0.6  _MM _MH  0.4  _WL 0.2  _WM _WH  0 loose  medium  tight  Fiber tension  Figure 5.9  Effect of fiber tension on cassette average shear stress  In the legend in Figure 5.9 the first letter indicates the module spacing (medium, narrow, wide), while the second letter indicates the sparging rate (low, medium, high). Error bars (omitted from Figure 5.6 as they distract from the presented data) are shown in Figure 5.1 (a bar chart which displays the same data illustrated in Figure 5.9).  68  (a) Three fiber tensions, medium spacing, medium gas flow -0.70  LMM MMM TMM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  (b)  Three fiber tensions, medium spacing, medium gas flow -0.70  LMM MMM TMM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure 5.10  Typical sheet-averages, illustrating effect of increasing fiber tension  Illustrated data are averages of measurements from probes within vertical (a) and horizontal (b) cross sections (‘sheets’) of the membrane cassette. See Figures A9.2 and A9.3 for diagrams describing sheet locations within the cassette. Error bars (two standard deviations) reflect the variation among the randomized triplicate measurements.  69  Figure 5.11  Typical shear maps illustrating the effect of increasing fiber tension.  Top row = tight fibers; middle row = medium tension; bottom row = loose fibers. Shear map axis units are in meters. See Figure A9.2 and the first few pages of Appendix 9 for a guide to reading shear maps such as those shown in Figure 5.11.  70  5.1.5 The effect of vertical location within the membrane cassette The horizontal sheet data presented in Figures A9.1c-A9.27c clearly demonstrate that the measured shear stress increased with vertical distance from the diffusers, and was greatest nearest the water surface (Figure 5.7 is a typical example). However, as the observed differences were small, the Wilcoxon paired-sample test was used to analyze the horizontal-sheet-average data. Results confirmed that each horizontal sheet of probes (corresponding to elevations of 0.1 m, 0.59 m, 1.08 m, and 1.57 m above the bottom of the cassette) experienced greater shear stress than the horizontal sheets below it (α<0.005). That shear stress increased from the bottom to the top of the cassette is consistent with the results of Bérubé et al. (2006a) who suggested that this effect may be due to the acceleration of bubbles as they rise within the system. The present study suggests that while bubble acceleration is likely a key variable, lateral dispersion of bubbles (away from the bottom diffusers) as bubbles move upward within the system is likely also a factor contributing to increased shear stress nearer the water surface. Engineering Implications: Shear stress is higher near the water surface than at lower levels of the cassette.  71  5.1.6 General observations Probes near the bottom of the inner membrane module (vertical sheets 1 and 2) generally measured very low shear stress, and unlike the rest of the cassette, these probes were relatively unaffected by increases in sparging rate or module spacing (Figures 5.8, 5.11, and Figures A9.1a-A9.27a). This observation suggests that the 8.5 cm horizontal distance between the inner module and the diffusers (located 20 cm below the cassette) was too great to allow the bottom of the inner module to experience direct bubble scouring; however, as the submersion tank used to contain the membrane system was not transparent, this hypothesis could not be visually confirmed. Lower shear stresses were typically measured by probes located on the edges of the cassette (i.e. row A, see Figure 3.14) than by probes in most other regions of the cassette (see Figures 5.5 and 5.8). These results suggest that to some extent, sparged bubbles converged towards the center of the tank within which the cassette was submerged. Similar results were reported by Nguyen et al. (2008), who observed that sparged bubbles converged towards the center of the membrane submersion tank and a downward flow pattern developed at the edges of the membrane submersion tank. It should be noted that the geometry of the membrane submersion tank used by Nguyen et al. (2008) was different from that used in the present study (i.e. included baffles to promote a recirculating flow).  Engineering Implications: Each membrane module should be located in close horizontal proximity to a gas diffuser to ensure that the lower regions of the cassette do not fail to benefit from direct gas scouring. The development of flow circulation patterns caused by rising gas bubbles may cause regions near the central vertical axis of the cassette to experience higher shear stresses than other regions of the cassette.  72  5.2  Experimental configurations with discontinuous sparging The term ‘discontinuous sparging’ is used in this document to refer to both pulse  and alternating sparging patterns, to differentiate these patterns from ‘continuous sparging’ where gas flow was provided to both diffusers continuously. As described previously in section 3.4.1, the term ‘pulse sparging’ is used to describe sparging patterns which involved a period of sparging followed by a period without any sparging, with gas flow for sparging provided to both diffusers at once. The term ‘alternating sparging’ is used to refer to sparging patterns where sparging (ie. gas flow) was alternated between the two diffusers, with sparging provided to only one diffuser at a time.  5.2.1  Relationship between average shear stress and variation of the shear stress, individual probe measurements For the experimental configurations investigated using discontinuous sparging, a  linear relationship between the mean shear stress and the variability (i.e. standard deviation) in the shear stress was observed. For individual probes, typical means and standard deviations are presented together in Figure 5.12 for ten different experimental configurations which used discontinuous sparging. By comparing Figures 5.2 and 5.12 it can be noted that for configurations using discontinuous sparging the standard deviation of the mean measurement of each probe was comparatively higher than for configurations using continuous sparging. Examination of the raw voltage (i.e. shear) measurements suggested that this is because discontinuous sparging patterns cause many probes (notably those closest to the gas diffusers) to experience greater variation in shear stress due to the existence of periods with relatively low shear (i.e. when sparging is off in that region) followed by periods of higher shear (i.e. when sparging is on in that region). As previously discussed in section 2.2, there is uncertainty regarding whether high average shear stress or highly variable shear stress are preferable for the removal of foulants in membrane systems. The present results indicate that in submerged HF systems with discontinuous sparging (or continuous sparging, as shown in section 5.1.1), the use of mean shear stress is a reasonable surrogate for estimating the variability of the shear stress. The shear stress data (Pascals) presented in Figure 5.12 were derived from raw voltage measurements using the methods outlined previously in section 5.1.1. 73  Standard deviation of probe measurements (Pa)  4  3.5  3 2.5  2  y = 1.1x 2  R = 0.77  1.5  1 0.5  0 0  0.5  1  1.5  2  2.5  3  3.5  4  Means of probe measurements (Pa)  Figure 5.12  Typical mean shear stresses vs. their standard deviations for configurations using discontinuous sparging (ie. pulse and alternating flow).  Each point represents the mean and standard deviation of 45 seconds of measurement from a single shear probe (600 points shown = 60 probes x 10 configurations). Note the clear trend of increasing standard deviation as the mean shear signal increases (slope of approximately 1.1:1). 5.2.2  The effect of sparging pattern The cassette-average results in Figure 5.13 and Table 5.2 indicate that there was  remarkably little difference in cassette-average shear stress among the five different sparging patterns investigated (i.e. continuous sparging and the four modes of discontinuous sparging presented in Table 3.3). However, for a given module spacing, configurations using continuous sparging and alternating sparging consistently had higher cassette-average shear stress than configurations using pulse air flow (Table 5.2). The standard deviations of cassette-averages (ie. the standard deviation of the mean of the 60 probes, Table 5.2) indicate that, for a given module spacing, discontinuous sparging (ie. an alternating or a pulse pattern) consistently distributed shear stress more evenly within the cassette than the use of continuous sparging (see Table 5.2 and Figures 5.14 and 74  5.15); however, this trend was not statistically significant based on the data collected (α>0.05). Use of the Wilcoxon paired-sample test with vertical sheet-average data, such as those presented in the top half of Figure 5.14 (see also Figures A9.28c to A9.35c) indicate that continuous sparging generated higher shear stress than alternating sparging (α<0.05) or pulse sparging (α<0.05), and that in turn alternating sparging generated higher shear stress than pulse sparging (α<0.05). It is uncertain why continuous sparging generated higher shear stress than sparging patterns using discontinuous sparging, though Figure 5.15 indicates that continuous sparging may have generated a more marked flow circulation pattern which may have allowed for higher cross-flow velocities in some regions of the cassette. That pulse sparging patterns provided somewhat lower shear stress than the other investigated sparging patters was as expected, as continuous sparging and alternating sparging patterns (which alternated sparging between the two bottom diffuser) provided twice as much gas to the cassette as pulse sparging patterns (which used a both diffusers on / both diffuser off pattern). It is thus interesting that pulse sparging generated shear stress of similar magnitude to that generated by continuous and alternating sparging patterns, which provided approximately double the volume of gas to the diffusers. Note that pulse sparging may have provided somewhat more than half the gas flow of the other sparging patterns used, as it took approximately 2 seconds for gas flow to stabilize during pulse sparging, nonetheless considerably less gas was used for pulse sparging. When looking at evenness of shear stress distribution (approximated using the standard deviation of the sheet-average shear stresses), use of the Wilcoxon paired-sample test demonstrated that both pulse and alternating flow patterns provided more distributed shear stress than continuous sparging (α<0.01), consistent with the observations for the cassette-average data. The evenness of the shear stress distribution provided by alternating flow was not significantly different from that provided by pulse flow (α~0.08) for the set points (on/off timing of 3 and 6 seconds) used in this study. When considering probe-average shear stresses (Figure 5.16, and Figures A9.28a to A9.35a), continuous sparging generated higher localized shear stresses than either  75  alternating or pulse flow. On the other hand, alternating and pulse flow patterns generated shear stress which was visibly much more evenly distributed across the cassette (Figures 5.15 and 5.16). Figure 5.15 (and Figures A9.28 to A9.31) demonstrates that shear stress patterns generated by fast alternating and slow alternating patterns were very similar, as were slow pulse and fast pulse patterns. Cui et al. (2003) note that alternating sparging between the two sides of the system decreases the density of the gas-liquid emulsion within one half of the system with respect to the other, causing flow to migrate through the cassette, which corroborates the present study’s observation that alternating flow provided shear stress which was significantly more evenly distributed than continuous sparging. Engineering Implications: Continuous sparging provides higher shear stress than discontinuous sparging (based on the discontinuous frequencies used in this study, 3 sec and 6 sec). Discontinuous sparging patterns provide shear stress which is more evenly distributed within the cassette than continuous sparging. Alternating sparging between diffusers provides higher (average) shear stress than pulse sparging (however, the pulse sparging pattern in this study used approximately 50% as much gas (total volume) as either continuous or alternating sparging patterns).  76  Medium tension, wide spacing, medium gas flow, five sparging patterns  MWM FA.MWM FP.MWM SA.MWM SP.MWM  -0.5 -0.45 -0.4  Volts  -0.35 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0  Horizontal sheets (*error bars = 2 standard deviations)  Figure 5.13  Typical effect of sparging pattern on cassette average shear stress (volts)  The cassette average shear stress is the average of measurements from all 60 test fibers. The error bars in the figure illustrate the variation (approximated as two standard deviations) among the triplicate measurements which were collected for each experimental configuration. See Tables 3.2 and 3.3 for a legend to the cassette configuration naming system. Figure 5.13 illustrates only configurations using wide module spacing. Legend Definitions: MWM describes the cassette configuration (medium fiber tension, wide module spacing, medium gas flow). FA= fast alternating, FP= fast pulse, SA= slow alternating, SP= slow pulse. Definitions: •alternating = gas provided to one diffuser, then the other •pulse = gas provided to both diffusers, then neither •fast = 3 second intervals, slow = 6 second intervals  77  Table 5.2 Ranking sparging patterns based on cassette-average voltage and standard deviation of cassette-average voltage Rank Membrane Average (from cassette voltage highest to lowest) configuration (volts) MWM -0.44 1 SA.MWM -0.41 2 SP.MWM -0.39 3 FA.MWM -0.39 4 FP.MWM -0.39 5 SA.MNM -0.34 6 MNM -0.34 7 SP.MNM -0.34 8 FA.MNM -0.33 9 FP.MNM -0.32 10  Legend: Parameter Fiber tension: Module spacing: Gas flow rate: Sparging pattern:  Abbreviation Range loose L medium M tight T narrow N medium M wide W low L medium M high H slow alternating SA fast alternating FA slow pulse SP fast pulse FP  Standard Membrane deviation cassette among all configuration probes (volts) FA.MNM 0.07 SA.MNM 0.08 FA.MWM 0.09 FP.MNM 0.08 SP.MNM 0.09 SA.MWM 0.11 MNM 0.10 SP.MWM 0.12 FP.MWM 0.12 MWM 0.16  Standard deviation among all probes (%) 21% 23% 23% 24% 26% 27% 29% 30% 32% 37%  Rank (from lowest to highest) 1 2 3 4 5 6 7 8 9 10  Value 95.4% 97.8% 99.6% 6.35 cm 8.26 cm 12.38 cm 3 cfm 6 cfm 9 cfm 6 sec 3 sec 6 sec 3 sec  *the configuration naming systems is: Sparging pattern/ fiber tension/ module spacing/ gas flow rate eg. FA.MNM = "fast alternating/ medium tension/ narrow spacing/ medium gas flow"  The ‘cassette-average voltage’ is the average of measurements from all sixty probes, indicating the overall average shear experienced by the cassette. The standard deviation (given as percent of the cassette average voltage) reflects the deviation among the mean voltage measurements from all sixty probes, and thus reflects the distribution of shear stress within the cassette. As only two configurations (MNM and MWM) were tested using the five different sparging patterns, data involving sparging patterns other than continuous sparging were not included in Table 5.1 but were instead examined separately.  78  (a)  Medium tension, wide spacing, medium gas flow, five sparging patterns  MWM FA.MWM FP.MWM SA.MWM SP.MWM  -0.70 -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette center  middle cassette, interior face  outer cassette, outer cassette outer cassette interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  (b)  Medium tension, wide spacing, medium gas flow, five sparging patterns  MWM FA.MWM FP.MWM SA.MWM SP.MWM  -0.70 -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure 5.14  Typical sheet-averages illustrating effect of sparging pattern  Illustrated data are averages of measurements from probes within vertical (a) and horizontal (b) cross sections (‘sheets’) of the membrane cassette. See Figures A9.2 and A9.3 for diagrams describing sheet locations within the cassette.  79  Figure 5.15  Typical shear maps illustrating the effect of sparging pattern.  Top row = continuous sparging; middle row = slow alternating sparging; bottom row = slow pulse sparging. Shear map axis units are in meters. See Figure A9.2 and the first few pages of Appendix 9 for a guide to reading shear maps such as those shown in Figure 5.15.  80  Figure 5.16  Alternating sparging between diffusers vs. continuous sparging.  Configuration shown is medium tension, narrow spacing, medium sparging rate. Top row = continuous sparging; Middle row= slow alternating (6 sec); Bottom row= fast alternating (3 sec). Shear map axis units are in meters.  81  5.3  Assumptions regarding optimal shear stress As it was not the focus of the present study to determine the effect of shear stress  on membrane flux, the results are discussed with the assumption that higher average shear stress and more uniformly distributed shear stress are preferable. These two assumptions are reasonable, given the results of Cabassud et al. (2003) and Ducom et al. (2002) who investigated the effect of shear stress on fouling within inside-out HF and flatsheet systems. Nonetheless, other authors have noted that other aspects of shear stress may impact the performance of membrane systems. Busch et al. (2007) note that shear stress magnitude can affect biofloc and particle size within membrane bioreactors, with smaller biofloc (generated by high shear stress) potentially generating increased cakelayer resistance. Bérubé et al. (2006b) also note that higher shear stress may increase internal membrane fouling by effectively removing the ‘protective’ cake layer and allowing smaller particles to reach the membrane surface. Busch et al. (2007) remark that cake layer resistance is largely attributable to the dense bottom layer of the cake closest to the membrane surface, an observation which may indicate that high peak shear stresses may be more effective at reducing the cake layer resistance than high average shear stress. The results of Gaucher (2002a) indicate that permeate flux may increase and homogenize the tangential liquid velocity gradient (and thus shear stress) adjacent to the membrane surface. However, most submerged systems do not withdraw permeate during backwash, so the relevance of this observation to submerged HF systems is debatable (Cui et al. 2003).  82  CHAPTER 6 CONCLUSIONS 1. The relationship between mean (average) shear stress and the variability (standard deviation) in shear stress was approximately linear within the pilot-scale submerged HF system for all configurations investigated. 2. Increases in sparging gas flow rate increased the average shear stress measured throughout the cassette, with the exception of lower regions of the cassette which were laterally distant from the diffusers and which may not have benefited from direct scouring by rising gas bubbles. 3. Increasing the spacing between modules greatly increased the average shear stress experienced within the cassette. However, the benefits of increasing module spacing accrued heterogeneously to certain regions of the cassette, possibly due to the repositioning of the cassette with respect to the diffusers. The increases in shear stress observed when module spacing was increased from 0.06 m to 0.012 m exceeded those observed when sparging rate was raised from 3 cfm to 9 cfm (see Figures A9.4c and A9.13c). This indicates that module spacing and diffuser location are of tremendous importance for optimal shear stress generation in submerged HF systems. 4. Loose and medium fiber tensions provided higher average shear stress than tight fibers, and appeared to provide more evenly distributed shear stress (though this was not statistically significant). As there was little observable difference between results from configurations using loose and medium fiber tensions, it is likely that 2-3% slack is sufficient to garner the benefits of fiber looseness. 5. All five investigated sparging patterns provided very similar cassette-averaged shear stress. However, discontinuous sparging (pulse and alternating) provided shear stress which was more uniformly distributed across the cassette. The performance of pulse sparging (both diffusers on followed by both diffusers off) was particularly interesting, as this sparging pattern required roughly half as much gas flow as either continuous sparging or alternating sparging between diffusers. 83  6. Shear stress within the cassette was greatest near the water surface, and decreased towards the bottom of the cassette where the diffusers were located. Although statistically significant, the observed differences in shear stress among the different vertical elevations of the cassette were relatively small.  6.1  Suggestions for membrane design and configuration Based on the results of this study, the suggested optimal configuration for a three-  module Zeeweed-500 system would have the following characteristics: •  Module spacing of at least 0.12 m (center of inner module to center of outer module)  •  Fiber tension of 95-97% (3-5% slack)  •  Diffusers located less than 0.08 m laterally from each module (given a diffuser depth of 0.2 m below the cassette)  •  Pulse gas sparging pattern or alternating sparging pattern (for decreased gas requirement and increased evenness of shear stress among locations within the cassette)  •  The sparging rate required for optimal sparging would likely depend on the nature of the source water, though increases in sparging gas flow rate were observed to raise the average shear stress within the cassette without greatly altering its evenness of distribution (within the range of sparging rates tested)  •  As the upper regions of the cassette were observed to experience higher shear stress, it is likely that a higher permeate flux can be sustained at the top of the cassettes. Therefore, permeate should be withdrawn from the top of the cassette.  84  It should be noted that the liquid viscosity is substantially different between the solution used in the present study (which was similar to that of pure water) and the more viscous mixed liquor found in membrane bioreactors. As “bubble rise velocity is inversely proportional to the liquid viscosity” (Cui et al., 2003, p.3), and shear stress is also linearly proportional to viscosity, designers and operators of submerged HF systems should be cautious in applying the findings of the present study directly to membrane bioreactors.  6.2  Future research The data produced in this study will subsequently be used to examine the affect of  experimental configurations on specific shear event patterns, and to begin to develop computational fluid dynamic (CFD) models to simulate the complex environment in submerged air sparged systems.  85  REFERENCES  Bellara, S. R., Cui, Z. F., & Pepper, D. S. (1996). Gas sparging to enhance permeate flux in ultrafiltration using hollow fibre membranes. Journal of Membrane Science, 121(2), 175-184. Berger, F. R., & Ziai, A. (1983). Optimization of experimental conditions for electrochemical mass transfer measurements. Chemical Engineering Research & Design, 61(6), 377-382. Bérubé, P. R., Afonso, G., Taghipour, F., & Chan, C. C. V. (2006). Quantifying the shear at the surface of submerged hollow fiber membranes. Journal of Membrane Science, 279(1-2), 495-505. Bérubé, P. R., Hall, E. R., & Sutton, P. M. (August 2006). Parameters governing permeate flux in an anaerobic membrane bioreactor treating low-strength municipal wastewaters: A literature review. Water Environment Research, 78, 887-896(10). Bérubé, P. R., & Lei, E. (2006). The effect of hydrodynamic conditions and system configurations on the permeate flux in a submerged hollow fiber membrane system. Journal of Membrane Science, 271(1-2), 29-37. Beszedits, S., and Walker, L. (1994) Treatment of cyanide wastes by chemical oxidation. Toronto, ON: B&L Information Services. Busch, J., Cruse, A., & Marquardt, W. (2007). Modeling submerged hollow-fiber membrane filtration for wastewater treatment. Journal of Membrane Science, 288(12), 94-111. Cabassud, C., Ducom, G., & Laborie, S. (2003). Measurement and comparison of wall shear stresses in a gas/liquid two-phase flow for two module configurations. Proc. IMSTEC, Sydney. (paper 163) 6.  86  Chan, C. C. V., Bérubé, P. R., & Hall, E. R. (2007). Shear profiles inside gas sparged submerged hollow fiber membrane modules. Journal of Membrane Science, 297(12), 104-120. Chang, I., Le Clech, P., Jefferson, B., & Judd, S. (2002). Membrane fouling in membrane bioreactors for wastewater treatment. Journal of Environmental Engineering, 128(11), 1018-1030. Chang, S., & Fane, A. G. (2001). The effect of fibre diameter on filtration and flux distribution — relevance to submerged hollow fibre modules. Journal of Membrane Science, 184(2), 221-231. Chang, S., & Fane, A. G. (2000). Characteristics of microfiltration of suspensions with inter-fiber two-phase flow. Journal of Chemical Technology & Biotechnology, 75(7), 533-540. Cognet, G., Lebouche, M., & Souhar, M. (1984). Wall shear measurements by electrochemical probe for gas-liquid two-phase flow in vertical duct. AICHE Journal, 30(2), 338-341. Cui, Z. F., Chang, S., & Fane, A. G. (2003). The use of gas bubbling to enhance membrane processes. Journal of Membrane Science, 221(1-2), 1-35. Ducom, G., Puech, F. P., & Cabassud, C. (2002). Air sparging with flat sheet nanofiltration: A link between wall shear stresses and flux enhancement. Desalination, 145(1-3), 97-102. Elimelech, M., Chen, W. H., & Waypa, J. J. (1994). Measuring the zeta (electrokinetic) potential of reverse osmosis membranes by a streaming potential analyzer. Desalination, 95, 269-286. Farrar, B., Samways, A. L., Ali, J., & Bruun, H. H. (1995). A computer-based hot-film technique for two-phase flow measurements. Measurement Science and Technology, 6(10), 1528-1537. 87  Gaucher, C., Jaouen, P., Legentilhomme, P., & Comiti, J. (2002). Suction effect on the shear stress at a plane ultrafiltration ceramic membrane surface. Separation Science and Technology, 37(10), 2251-2270. Gaucher, C., Legentilhomme, P., Jaouen, P., Comiti, J., & Pruvost, J. (2002). Hydrodynamics study in a plane ultrafiltration module using an electrochemical method and particle image velocimetry visualization. Experiments in Fluids, 32(3), 283-293. Gordon, S. L., Newman, J. L., & Tobias, C. W. (1966). The role of ionic migration in electrolytic mass transport: Diffusivities of [fe(CN6)]3− and [fe(CN6)]4− in KOH and NaOH solutions. Berichte Der Bunsengesellschaft Für Physikalische Chemie, 70, 414-420. Hilal, N., Ogunbiyi, O. O., Miles, N. J., & Nigmatullin, R. (2005). Methods employed for control of fouling in MF and UF membranes: A comprehensive review. Separation Science and Technology, 40(10), 1957. Igarashi, C., & Nishizawa, K. (2007). Effect of turbulent flow condition on streaming potential measurement using cylindrical cells. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 303(3), 235-240. Kissinger, P. T., & Heineman, W. R. (1996). Laboratory techniques in electroanalytical chemistry (2nd ed.). New York: Mercel Dekker Inc. Legrand, J., Dumont, E., Comiti, J., & Fayolle, F. (2000). Diffusion coefficients of ferricyanide ions in polymeric solutions — comparison of different experimental methods. Electrochimica Acta, 45(11), 1791-1803. LMNO Engineering, Research, and Software ltd. (2001). Moody friction factor calculator Retrieved March 1, 2009 from: http://www.lmnoeng.com/moody.htm  88  Mitchell, J. E., & Hanratty, T. J. (1966). A study of turbulence at a wall using an electrochemical wall shear-stress meter. Journal of Fluid Mechanics Digital Archive, 26(01), 199-221. Mizushina, T. (1971). The electrochemical method in transport phenomena. In T. P. Irvine, & J. P. Hartnett (Eds.), Advances in heat transfer (pp. 87-181). New York: Academic Press. Munson, B., Young, D., & Okiishi, T. (1990). Fundamentals of fluid mechanics (2nd ed.). New York: Wiley. Nakoryakov, V. (1986). Experimental study of gas-liquid slug flow in a small diameter pipe. International Journal of Multiphase Flow, 12(3), 337-355. Nakoryakov, V., Kashinsky, O., Petukhov, A., & Gorelik, R. (1993). Study of local hydrodynamic characteristics of upward slug flow. Mineralium Deposita, 28(4), 560-566. Nakoryakov, V. E., Kashinsky, O. N., Burdukov, A. P., & Odnoral, V. P. (1981). Local characteristics of upward gas-liquid flows. International Journal of Multiphase Flow, 7(1), 63-81. Nguyen Cong Duc, E., Fournier, L., Levecq, C., Lesjean, B., Grelier, P., & Tazi-Pain, A. (2008). Local hydrodynamic investigation of the aeration in a submerged hollow fibre membranes cassette. Journal of Membrane Science, 321(2), 264-271. Rajeshwar, K., & Ibanez, J. G. (1997). Environmental electrochemistry: Fundamentals and applications in pollution abatement Academic Press. Reiss, L. P., & Hanratty, T. J. (1962). Measurement of instantaneous rates of mass transfer to a small sink on a wall. AICHE Journal, 8(2), 245-247. Reiss, L. P., & Hanratty, T. J. (1963). An experimental study of the unsteady nature of the viscous sublayer. AICHE Journal, 9(2), 154-160.  89  Rode, S., Midoux, N., Latifi, M. A., Stork, A., & Saatdjian, E. (1994). Hydrodynamics of liquid flows in packed beds: An experimental study using elec-trochemical shear rate sensors. Chemical Engineering Science, 49, 889-900. Rode, S., Latifi, M., Storck, A., & Midoux, N. (1994). An experimental study of the frequency response of electrochemical sensors in nonhomogeneous flow in packed beds. Journal of Applied Electrochemistry, 24(7), 639-646. Selman, J. R., & Tobias, C. W. (1978). Mass transfer measurements by the limitingcurrent technique. In T. B. Drew, G. R. Cokelet, J. W. Hoopes & T. Vermeulen (Eds.), Advances in chemical engineering (pp. 211-275). New York: Academic Press. Sobolik, V., Tihon, J., Wein, O., & Wichterle, K. (1998). Calibration of electrodiffusion friction probes using a voltage-step transient. Journal of Applied Electrochemistry, 28(3), 329-335. Son, J. S., & Hanratty, T. J. (1969). Velocity gradients at the wall for flow around a cylinder at reynolds numbers from 5 &times; 103 to 105. Journal of Fluid Mechanics Digital Archive, 35(02), 353-368. Taama, W. M., Plimley, R. E., & Scott, K. (1996). Influence of supporting electrolyte on ferricyanide reduction at a rotating disc electrode. Electrochimica Acta, 41(4), 549551. Tacke, D., Pinnekamp, J., Prieske, H., & Kraume, M. (2008). Membrane bioreactor aeration: Investigation of the velocity flow pattern. Water Science & Technology, 57(4), 559-565. Ueda, T., Hata, K., Kikuoka, Y., & Seino, O. (1997). Effects of aeration on suction pressure in a submerged membrane bioreactor. Water Research, 31(3), 489-494.  90  United States Environmental Protection Agency, Office of Ground Water and Drinking Water (2003). Membrane filtration guidance manual proposal draft. EPA document number 815-D-03-008. Vyas, H. K., Bennett, R. J., & Marshall, A. D. (2000). Influence of operating conditions on membrane fouling in cross-flow microfiltration of particulate suspensions. International Dairy Journal, 10(7), 477-487. Zar, J. H. (1999). Biostatistical analysis (4th ed.) Prentice Hall Inc.  91  APPENDIX 1: VOLTAGE vs. SHEAR STRESS  Voltage vs. Shear Stress Corresponding Shear Stress (Pa) (16.5ºC)  10.00 9.00 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 0  -0.1  -0.2  -0.3  -0.4  -0.5  -0.6  -0.7  -0.8  -0.9  -1  Measured Voltage (Volts)  Figure A1.1  The relationship between measured voltage and shear stress.  The relationship illustrated in Figure A1.1 is based on equations from Reiss & Hanratty (1962) and Gaucher et al. (2002). Note that the voltage reading is simply the current passing through the probe multiplied by 100,000: (current (A) * 100Ω resistor * 1000 gain amplifier = voltage reading). Figure A1.1 is for a probe with diameter 0.50mm (the diameter to which all probes in this study were normalized), using a diffusion coefficient of 6.6x10-10 m^2/sec and a dynamic viscosity of 0.001 Pa•sec (ie. at 16.5ºC).  92  APPENDIX 2: MANUFACTURING SHEAR PROBES  Probes were manufactured using: •  Teflon plastic tubing (outer diameter 1.8 mm), clear, flexible, hollow  •  Platinum wire (99.95% pure, 0.508 mm diameter)  •  30 gauge (0.25 mm diameter) Kynar electrical wire  A highspeed drillpress was used to drill a 0.50 mm diameter hole into a piece of Teflon tubing, and a razorblade was used to cut a 5 mm slit into the tubing directly behind the drilled hole. A sewing needle, tied to a long thread, was placed into one end of the hollow tubing. The needle was pulled through the tubing using a magnet (external to the tube), and pulled out the 5 mm slit. The thread, running from one end of the tubing to the slit near the 0.50mm hole, was then tied to the 30 gauge electrical wire, which was then guided through the tubing by pulling the thread through the 5 mm slit. Once one end of the wire had been pulled through the slit, a soldering iron was used to solder the electrical wire to a 5 mm section of platinum wire, which had been bent into an “L” shape. The un-soldered end of the platinum wire was then pushed through the 0.50 mm diameter hole from the inside out, with the remaining platinum wire and the electrical wire snugly fitted into the hollow center of the tubing. Epoxy was injected into the 5 mm slit using a syringe and allowed to harden, ensuring that the platinum wire was held in place and that no fluid should be able to enter the tubing through the slit. The ends of the tubing were also sealed with injected epoxy to prevent fluid entry. The following day, once the epoxy had hardened, the end of the platinum wire sticking through the hole in the tubing was sequentially sanded with sandpaper and emery paper with progressively finer grit size, before being polished for several minutes with a wet Kimwipe® to remove any dirt. Each finished probe was then examined and photographed beneath a stereo microscope, and if no aberrations were noted then the probe was passed to the probe standardization procedures outlined in section 3.3.  93  APPENDIX 3: SHEAR STRESS IN ANNULAR FLOW By taking the derivative of the annular flow velocity equation given in Munson et al. (1990), it can be noted that the shear at the inner wall (the test fiber surface) is 3.05 times (approximated as 3) that at the outer wall (the Plexiglas pipe surface). Note that as the system is operated under laminar flow, knowledge of the roughness coefficients of the surfaces is not required for the derivation of shear stresses.  Shear Profile in Annulus 0.018 0.016  Plexiglas pipe  0.012  Teflon Tube  Shear (Pa)  0.014  0.01 0.008 0.006 0.004 0.002 0 0  0.002  0.004  0.006  0.008  0.01  0.012  0.014  Distance from center of annulus (m) Figure A3.1  Profile of shear stress within an annulus with laminar flow.  Note that the shear stress experienced on the inner cylinder (the test fiber) is three times the shear stress on the outer cylinder (the Plexiglas cylinder) for the annulus geometry used for probe standardization tests. The data illustrated was derived using equations from Munson et al. (1990).  94  Definition of terms for equations A3.1 to A3.8: µ = dynamic viscosity [Pa•Sec] xI = diameter inner pipe [m] xo = diameter outer pipe [m] L = pipe length to point x [m] where shear stress is to be determined DH = hydraulic diameter [m] = (xo - xI ) ρ = density [kg/m3] v = superficial velocity [m/sec] = (volumetric flowrate)/(pipe cross sectional area) vx = tangential velocity [m/sec] at a distance x [m] from the center of the annulus Re = Reynolds number [dimensionless] f f = friction factor [dimensionless]  hL = headloss [m] τx = shear stress [Pa] at a distance x [m] from the center of the annulus VFR = volumetric flow rate [m3/min]  The shear stress at the outer wall of the annulus can be derived using the following equations from Munson et al. (1990): The superficial velocity in an annulus can be determined from the hydraulic diameter of the annulus using equation A3.1:  (  (  ))  v = (VFR ) / πDH / 4 ÷ (60 sec/ min ) 2  (A3.1)  where v is the superficial liquid velocity [m/sec], VFR is the volumetric flow rate [m3/min], and DH is the hydraulic diameter of the annulus [m].  95  The Reynolds number is related to the superficial velocity by equation A3.2:  Re = ρvDH / µ  (A3.2)  where Re is the Reynolds number [dimensionless], ρ is the solution density [kg/m3], v is the superficial liquid velocity [m/sec] derived from equation A3.1, DH is the hydraulic diameter of the annulus [m], and µ is the dynamic viscosity of the solution [Pa•Sec]. The friction factor (for laminar flow in an annulus with an xI/xo ratio of 0.072) is related to the Reynolds number by equation A3.3: f f = 87 / Re  (A3.3)  where f f is the friction factor [dimensionless] and Re is the Reynolds number [dimensionless] derived from equation A3.2. The frictional headloss is related to the friction factor and annulus geometry by equation A3.4:  h L = f f ⋅ L ⋅ v 2 / (D H ⋅ 2 g )  (A3.4)  where hL is the frictional headloss [m], f f is the friction factor [dimensionless] derived from equation A3.3, L is the annulus length to point x [m] where shear stress is to be determined, v is the superficial velocity [m/sec] from equation A3.1, DH is the hydraulic diameter [m], and g is the gravitational constant [9.8 m/s2]. The shear stress at the outer wall of the annulus is related to the frictional headloss by equation A3.5:  τ = hL ⋅ g ⋅ DH ⋅ ρ ÷ ( 4 L )  (A3.5)  where τ is the shear stress [Pa] at the outer wall of the annulus, hL is the headloss [m] derived from equation A3.4, g is the gravitational constant [9.8 m/s2], DH is the hydraulic  96  diameter [m] of the annulus, ρ is the solution density [kg/m3], and L is the pipe length to point x [m] where shear stress is to be determined. Velocity in an annulus at distance x from the centre of the inner pipe of the annulus is related to the headloss, viscosity, and system geometry by equation A3.6:  (  )  v x = Abs (h L / 4 µ ) ⋅ [ x 2 − x o + {( x I − x o ) / (ln( x o / x I ) )} ⋅ ln( x / x o )]  (A3.6)  where vx is the tangential velocity [m/sec] at a distance x [m] from the center of the annulus, hL is the headloss [m] derived from equation A3.4, µ is the dynamic viscosity of the solution [Pa•Sec], xI is the diameter of the inner pipe [m], xo is the diameter of the outer pipe [m]. As shear stress can be calculated as the product of viscosity and the derivative of the tangential velocity with respect to the radial distance (ie. τ = dv/dx * µ, Munson et al.  (1990)) the shear stress at a distance x from the center of the annulus can be determined using equation A3.7, the derivative of equation A3.6 multiplied by viscosity:  (  {(  )  } )  τ x = Abs (hL / 4 µ ) ⋅ [2 x + x I 2 − x o 2 / (ln( x o / x I ) ) / x ⋅ µ  (A3.7)  Using equation A3.7 it can be determined that the shear stress at the inner wall of the annulus used for probe standardization experiments (ie. the test fiber surface within the side loop apparatus) is 3 times the shear stress at the wall of the outer pipe of the annulus (ie. the wall of the Plexiglas cylinder shown in Figure 3.4. (see Figure A3.1 for shear stress profile within an annulus). The side loop apparatus used in the present study was operated using laminar conditions, thus a friction factor of 87/Re could be used (based on tables from Munson et  al. (1990). Bérubé et al. (2006) used similar shear probes at higher cross-flow velocities (0.2-0.4 m/sec) than those used in the present study (0.017-0.065 m/sec). Theoretical shear stress calculations for the apparatus of Bérubé et al. (2006) are similar to those for the apparatus used in the present study (ie. equations A3.1 to A3.7), except that a turbulent flow friction factor (Ff, unitless) is required, as the apparatus of Bérubé et al. (2006) was operated under turbulent flow conditions. The turbulent friction factor used  97  for calculating the theoretical shear stress within the apparatus of Bérubé et al. (2006) was derived using equation A3.8 (LMNO Engineering, 2001). As mentioned in section 3.3, the measured voltage results of Bérubé et al. (2006) compare favourably with the theoretical voltage determined using equations A3.1 to A3.8 with equations A4.1 to A4.4 (replacing the use of equation A3.3 for laminar flow with equation A3.8 for turbulent flow). ff =  1.325   e 5.74  ln + 0. 9   3.7 D p Re         2  (A3.8)  In equation A3.8 f f is the turbulent friction factor [dimensionless], e is surface roughness ( ≅ 0 for plastic pipe), Dp is the pipe diameter (0.07m), and Re is the Reynolds number (unitless).  98  APPENDIX 4: DETERMINING SHEAR PROBE VOLTAGE CORRECTION FACTORS Definition of terms for equations A4.1 to A4.4: τ = shear stress [Pa]  S = mean wall velocity gradient [sec-1] µ = dynamic viscosity [Pa•Sec] k = the mass transfer coefficient [m/sec] D = the diffusion coefficient [m2/sec] de = shear probe diameter [m]. IL = electrical current [A] Co = bulk concentration of ferri-cyanide ions [mol/m3] F = the Faraday constant [96500 C/mol]  V = voltage [volts] R = electrical resistance [Ω] A = amplification [dimensionless] ve = number of reacting electrons per active ion [1 electron]  99  Table A4.1  Values of controlled variables within the side loop apparatus  Controlled variables  Symbol Value  Units  Diameter of outer pipe (column)  xo  0.025  [m]  Diameter of inner pipe (test fiber) xI  0.0018  [m]  Volumetric flow rates used  VFR  0.0005 to 0.002 [m3/min]  Dynamic viscosity of solution 1  µ  0.00081  [Pa•Sec]  Diffusion coefficient 1  D  8.36 x 10-10  [m2/sec]  Magnitude of resistor used  R  100  [Ω]  Magnitude of amplifier used  A  1000  [unitless]  Notes: 1. Values for the dynamic viscosity and the diffusion coefficient were assumed (Gaucher et al., 2002) based on a controlled temperature of 25ºC during all probe standardization tests completed using the side loop apparatus.  Given the theoretical shear stress profile within an annulus of inner pipe diameter 1.8mm and outer pipe diameter of 25mm (see Appendix 3 for these calculations), the theoretical voltage signal expected from a circular shear probe of diameter de [m] was calculated using equations A4.1 to A4.4.  100  The shear stress can be related to the mean wall velocity gradient using equation A4.1 (Munson et al., 1990):  τ = Sµ  (A4.1)  where τ is the shear stress [Pa], S is the mean wall velocity gradient [sec-1], and µ is the dynamic viscosity [Pa/sec]. The mass transfer coefficient can then be determined using equation A4.2 (Reiss & Hanratty, 1962): 1/ 3   D2 S   k = 0.862  d  e   (A4.2)  where k is the mass transfer coefficient [m/sec], D is the diffusion coefficient [m2/sec], and de is the diameter of the shear probe [m]. The diffusion limited electrical current can be related to the mass transfer coefficient by equation A4.3 (Gaucher et al., 2002b):   πd e 2  C o k I L = ve F    4   (A4.3)  where IL is the diffusion limited current passing through the probe [A], F is the Faraday constant [96500 C/mol], Co is bulk concentration of ferri-cyanide ions in solution [mol/m3], and ve is the number of reacting electrons per ferri-cyanide ion [1 electron]. The diffusion limited electrical current is related to the measured voltage by equation A4.4:  V = IL ⋅ R ⋅ A  (A4.4)  where R is the magnitude of the resistor used (Ω), and A is the magnitude of the amplifier used to magnify the signal (unitless).  101  For given hydrodynamic conditions (see Table A4.1 for system geometry), for which the theoretical surface shear stress can be calculated (τTi) at the test fiber surface, the corresponding theoretical voltage (VTi) can be determined using equations A4.1 to A4.4, assuming an ideal circular probe with a diameter (de) of 0.0005m. For all probes, which were made using platinum wires with a diameter of 0.0005m, the slope of the line formed by plotting the theoretical voltages (VTi) vs. the measured voltages (VMi), for different hydrodynamic conditions, should have been 1. However, as presented in Figure A4.1, the slope of the theoretical voltages vs. the measured voltages, was not typically exactly 1. This discrepancy was assumed to be due to the fact that the majority of probes did not have a surface area that corresponded exactly to that of an ideal probe with a diameter of 0.0005m. Rode et al., (1994) also made use of controlled hydrodynamic conditions in a column to calibrate their electrochemical probes, while Sobolik et al .(1998) commented that calibration using controlled hydrodynamic conditions is prudent. So that the signals from all probes could be compared with each other, the signal from each probe was normalized to that of a probe with a diameter of 0.0005m. This was achieved by identifying a voltage correction factor (for each probe) that generated a slope of 1 when plotting the theoretical voltages vs. the corrected measured voltages (see Figure A4.1). Note that the use of a voltage correction factor is similar to applying a multiplier to the diameter of the probe (ie. a surface area correction factor). However, it would be insufficient to simply use surface area correction factors in the place of voltage correction factors, as the relationship between measured current and probe surface area is non-linear (see Reiss 1962 for details regarding why the relationship is non-linear). Both voltage correction factors and surface area correction factors, each calculated using the method outlined above using equations A4.1 to A4.4, are illustrated in Table A5.1, so that the similarity/disparity between these two types of correction factors. Only voltage correction factors were used to normalize measurement data, as per equation A4.5: VTi = (V Mi ) ⋅ (VoltageCorrectionFactor )  (A4.5)  where VTi is the theoretical voltage calculated using equations A3.1 to A3.7 and the data in Table A4.1, VMi is the measured voltages collected using the side loop apparatus  102  (section 3.3), and the ‘Voltage Correction Factor’ is a unitless multiplier specific to each probe (Table A5.1)  theoretical (x) vs. theoretical (y) (slope of 1) Measured (x) vs. theoretical (y) Measured*(Voltage correction factor) (x) vs. theoretical (y) 310  Theoretical voltage (mV)  290 270 250 230 210 190 170 150 150  170  190  210  230  250  270  290  310  Measured data (mV), corrected for channel bias  Figure A4.1  Typical determination of a voltage correction factor.  Excel goalseek function was used to determine a voltage correction factor (a constant unitless multiplier specific to each probe). When used with the measured voltage data, the voltage correction factor aligns the measured voltage signal with the theoretical voltage from an ideal circular probe of effective diameter (de) 0.0005m, causing the measured voltage vs. theoretical voltage to have a slope of 1. For the probe shown (R1m), the voltage correction factor was determined to be 1.08. Note that the surface area correction factor for this probe was 1.11 (Table A5.1), slightly higher than the voltage correction factor, as expected according to the rationale of Reiss & Hanratty (1962).  103  APPENDIX 5: TABLE OF PROBE SURFACE AREA CORRECTION FACTORS Table A5.1  Probe Probe number name 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30  R1A R1C R1F R1G R1H R1i R1J R1L R1M R1N R1P R1Q R1R R1T R1U R2A R2B R2C R2D R2E R2F R2H R2i R2J R2K R2L R2M R2N R2O R2P  Table of probe surface area and voltage correction factors. Surface Voltage area correction correction factor factor 1.04 1.00 1.00 damaged damaged 1.11 0.93 damaged 1.11 1.06 1.11 1.02 1.08 0.87 1.13 damaged 0.80 1.00 0.98 1.09 0.94 0.92 1.07 0.98 damaged 1.04 1.14 0.99 1.02 1.15  1.03 0.99 1.00  1.08 0.95 1.08 1.04 1.08 1.01 1.06 0.91 1.09 0.85 1.00 0.99 1.06 0.96 0.94 1.05 0.99 1.03 1.10 0.99 1.01 1.11  Probe Probe number name 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60  R3 beta R3i R3J R3M R3N R3O R3P R3R R3S R3T R3U R3V R3Y R3Z R3 alpha R4C R4D R4E R4F R4G R2Q R4i R4J R4K R4L R4W R4N R4P R4Q R1V  Surface Voltage area correction correction factor factor 0.97 0.89 0.89 0.97 1.07 0.95 1.06 1.16 1.10 1.10 1.11 1.18 1.17 1.20 1.30 0.99 0.87 1.06 0.90 0.92 1.02 0.98 1.24 1.19 1.03 1.14 1.12 1.17 1.26 1.11  0.98 0.92 0.92 0.98 1.05 0.96 1.04 1.11 1.07 1.07 1.08 1.13 1.12 1.14 1.20 1.00 0.90 1.04 0.93 0.94 1.02 0.98 1.16 1.13 1.02 1.10 1.09 1.12 1.18 1.07  Surface area and voltage correction factors were determined using the procedures described in Appendix 4. Voltage correction factors were used to normalize measurements from all probes to those expected from an ‘ideal’ circular probe of diameter 0.50 mm. Surface area correction factors are given simply to illustrate to the reader that the magnitude of the voltage correction factors is related to probe surface areas (smaller probes require larger voltage correction factors to normalize their signals).  104  APPENDIX 6: COMPLICATIONS OF THE ELECTROCHEMICAL METHOD As in any experiment, the present study had several minor setbacks during the stages leading up to pilot testing. The author believes it worthwhile to note these minor setbacks here in the off-chance that this information might benefit other researchers looking to adapt the use of electrochemical shear probes to their own work. In brief, the setbacks encountered in this study can be summarized as: •  Corrosion  •  Streaming/leak current  •  Using pneumatics in submerged systems  •  Treatment of ferri/ferrocyanide waste (costs and methods) The importance of corrosion cannot be understated. While authors such as Reiss  & Hanratty (1962) and Son & Hanratty (1969) have noted the great importance of using only stainless steel and plastic components within apparatuses designed for use with the electrochemical method, these authors used highly corrosive sodium hydroxide as their inactive electrolyte. However, regardless of the electrolyte used (in the present study it was 0.3M potassium chloride solution), corrosion is a serious concern. In the present study this issue arose when it was discovered that the end caps of the stainless steel actuator cylinders (used for adjusting the configuration of the membrane cassette) were actually aluminum. Within only a few hours of submersion within the highly saline potassium chloride solution (initial tests were done without ferri/ferrocyanide) the endcaps had severely corroded, requiring the costly and time-consuming process of replacing all actuator cylinders with new cylinders made entirely from stainless steel and plastic composite. The corrosion was likely particularly acute because of the galvanic cell set up between the submerged aluminum and stainless steel. This galvanic potential could actually be measured using the shear probes (unintentionally), and was increased  105  by the movement of gas bubbles, presumably because the high shear stresses generated on the aluminum surfaces removed aluminum oxide films and allowed the galvanic process to proceed more quickly. While streaming/leak current did not impact measurements in this study, its discovery was somewhat perplexing and it took several weeks to correctly identify based on observed aberrations in data collected during preliminary tests. Streaming current (also known as streaming potential) is caused whenever fluid moves across a charged surface, and in aqueous systems surfaces are nearly always charged except at a very specific pH known as the point of zero charge (Elimelech et al., 1994). The surface charge creates a layer of counterions (ions of opposite charge to the charged surface), and it is the movement of these charged ions in the liquid phase across the stationary surface which generates streaming potential (Igarashi & Nishizawa, 2007). Within the present study streaming potential was observed as short bursts of electrical current (‘noise’) passing through the shear probe despite the lack of reagents within the solution. It was noted that this noise was of greater magnitude when the distance between the anode and cathode was increased, as the length of plastic tubing (whose surface was generating the streaming potential) was also increased. The addition of potassium chloride decreased the observed streaming potential to negligible levels, and when the distance between the anode and cathode was kept below 1m the electrical noise could no longer be observed. The use of pneumatics within the submerged system was highly successful, as it allowed very precise rapid changes in configuration. However, leaky seals on several of the pneumatic cylinders required considerable effort to fix, and the threat of a cylinder beginning to leak during pilot testing was a serious concern. To ensure that no oxygen could enter into the system from the pneumatics, nitrogen gas was used to drive the actuators. Ferricyanide and ferrocyanide, though they are not considered very hazardous to human health, must be disposed of properly. As a considerable amount of chemical waste (1500 liters) was generated during the course of experimental testing, the author  106  investigated means of treating the waste onsite. While two methods were identified with the potential to break down the waste solution, UV-ozone and UV-hydrogen peroxide (Beszedits & Walker, 1994), fears of liberating cyanide (highly toxic) convinced the author to pay for treatment offsite. The cost of waste treatment should thus be added to the cost of reagents when considering the use of the ferri/ferrocyanide couple for mass transfer measurements. While potassium chloride is frequently used as the excess inactive electrolyte in ferri/ferrocyanide systems, the author suggests experimenting with potassium carbonate (which was not used in the present study), as the presence of the chloride ion has the potential to poison platinum probes should chlorine gas begin to form within the system due to an unwanted side-reaction at the anode. As few difficulties have been observed with potassium chloride it remains a reagent of choice, though the author notes that this sodium hydroxide was also considered a reagent of choice, and is still used with the ferri/ferrocyanide couple despite being highly caustic and easily replaceable by potassium chloride.  107  APPENDIX 7: PHOTOS OF THE EXPERIMENTAL APPARATUS  Figure A7.1  Close-up of the Zeeweed-500 cassette with all three hollow-fiber modules visible. Fiber length was 1.67 meters, and fiber diameter 1.8mm.  Figure A7.2  The signal conditioning apparatus, with controlled DC power supply.  108  Figure A7.3  Close up of the pneumatic control box used to control the fiber tension and module spacing of the membrane cassette.  109  Figure A7.4  Close up of the top of the pilot apparatus  The top of the pilot apparatus was sealed to the top of the stainless steel membrane submersion tank using bolts and a fitted silicone ring.  110  APPENDIX 8: MATLAB CODES USED FOR DATA ANALYSIS Code for rendering text files of raw data into small tables of basic statistics:  % (Note: % at the beginning of a line indicates author comments, not code) %raw data files stored as P1 to P115, the following code sequentially %accesses each file and reduces it to a series of statistics that are %then output to an excel file  clear; firstFile = 1; lastFile = 115; nSamples = 115; columns = 7; for n = firstFile : 1 : lastFile number = num2str( n ); fileIn = strcat('P',number,'.txt'); load ProbeVoltageCorrectionFactors_PostPilot.txt; load MeanChannelBiases.txt; load Sheet_for_eliminating_dead_probes.txt; %PVCF are used to correct for the different surface areas of each probe %MCB is used to correct for minor biases which are constant for each channel %Sfedp is required to exclude data for the five probes suspected to be damaged  PVCF=ProbeVoltageCorrectionFactors_PostPilot; MCB=MeanChannelBiases; Sfedp=Sheet_for_eliminating_dead_probes; %give the loaded matrix name "A"  A(:,:) = load( fileIn ); %rearrange the matrix into 60 columns each with 30,000 rows (1 minute of data) %from a matrix of 7 columns (column 1 is time) each with 300,000 rows (10 min)  111  for i=1:10 B(1:30000,1+6*(i-1):6+6*(i-1))=A(1+30000*(i-1):30000+30000*(i-1),2:7); end %erasing first five and last ten seconds from each run to allow for the %development of limiting current conditions and for channel switching using swithbox  C=B(2500:25000,:); Standard=std(C)'; Mean=mean(C)'; Minimum=min(C)'; Maximum=max(C)'; M=(Mean-MCB).*PVCF; Min=(Minimum-MCB).*PVCF; Max=(Maximum-MCB).*PVCF; S=Standard.*Sfedp; %concatenates a matrix with four columns (mean,standard dev,min,max) and 60 rows, %and exports the matrix to an excel worksheet within the file “MatlabFile.xls”  Stats=[M S Min Max]; xlswrite('C:\Documents and Settings\Matlab Processed data\MatlabFile.xls', Stats, number); clear; end;  112  Code for generating the shear stress maps (only code for one vertical and one horizontal sheet are shown, as the others are simply minor variations): % (Note: % at the beginning of a line indicates author comments, not code)  clear; %TableOfMeans contains mean probe data for each configuration (the average of %triplicate measurements)  load TableOfMeans.txt; %note that the file 'TableOfMeans' has already accounted %for the voltage correction factors and channel biases (as shown in previous code)  %give the matrix name "A"  A=TableOfMeans_Mar3_09; %use the line below to select the column you wish to examine, %see the excel file 'table of means' for a legend to this data %the negative sign in front of the A is needed to allow the graphs to %function properly, note that the magnitude of the signals is not changed  M(1:60,1)=-A(1:60,2); %non-functional probes= 4,5,8,16,25 -for these points interpolate their %values using average of the five nearest probes (3 within that vertical %plane, and its two closest neighbours in the horizontal plane).  I4= [M(19) M(20) M(6) M(1) M(7)]; Interp4=mean(I4); I5= [M(20) M(21) M(6) M(19) M(2) M(8)]; Interp5=mean(I5); I8= [M(7) M(9) M(23) M(11) M(6)]; Interp8=mean(I8); I16= [M(1) M(17) M(31) M(19)]; Interp16=mean(I16); I25= [M(10) M(26) M(40) M(22) M(28)]; Interp25=mean(I25); %creating 5 vertical "sheets" of probes that can be viewed individually (only one shown) %Vert1 is the innermost set of probes, center of the inner module  for i=1 Vert1=[M(i,1) M(i+1,1) M(i+2,1) M(i+1,1) M(i,1);  113  Interp16 M(i+16,1) M(i+17,1) M(i+16,1) Interp16; M(i+30,1) M(i+31,1) M(i+32,1) M(i+31,1) M(i+30,1); M(i+45,1) M(i+46,1) M(i+47,1) M(i+46,1) M(i+45,1)]; end for i=0 Horiz1=[M(i+13,1) M(i+14,1) M(i+15,1) M(i+14,1) M(i+13,1); M(i+10,1) M(i+11,1) M(i+12,1) M(i+11,1) M(i+10,1); M(i+7,1) Interp8 M(i+9,1) Interp8 M(i+7,1); NaN NaN NaN NaN NaN; Interp4 Interp5 M(i+6,1) Interp5 Interp4; M(i+1,1) M(i+2,1) M(i+3,1) M(i+2,1) M(i+1,1); Interp4 Interp5 M(i+6,1) Interp5 Interp4; NaN NaN NaN NaN NaN; M(i+7,1) Interp8 M(i+9,1) Interp8 M(i+7,1); M(i+10,1) M(i+11,1) M(i+12,1) M(i+11,1) M(i+10,1); M(i+13,1) M(i+14,1) M(i+15,1) M(i+14,1) M(i+13,1)]; end %the vectors below are needed for proper spatial plotting of probe %locations, X1 & Y1 for the vertical sheets, X2 & Y2 for horizontal cross sections  X1 = [0 0.18 0.36 0.54 0.72]; Y1= [1.57 1.08 0.59 0.10]; X2 = [0 18 36 54 72]; %the y values below are approximate, based on bulkhead widths and a %wide module spacing (~8.24cm, thin~6.35cm, wide~12.38cm)  Y2= [30 27.5 25 21.25 17.5 15 12.5 8.75 5 2.5 0]; %only the output code for the vertical sheets is shown below, as the horizontal is similar %use FileNumber to avoid having to write in the number for each jpeg file  FileNumber = strcat('TMM') %10 refers to the number of contour lines  114  Contourmap = contourf(X1,Y1,Vert1,10); colormap jet caxis([0 1]); xlabel('Horizontal axis (meters)','FontWeight','bold') ylabel('Vertical axis (meters)','FontWeight','bold') title('High tension, medium spacing, 6cfm spargingrate - Vertical sheet 1','FontWeight','bold') h=colorbar; xlabel(h,'Volts'); fileOut = strcat('C:\Documents and Settings\Administrator\Desktop\Matlab output\',FileNumber,'Vsheet1'); saveas(gcf,fileOut,'jpg');  Code for converting all individual data points (in volts) to Pascals using iteration, and subsequent output of basic statistics (mean and standard deviation). The data presented in Figure 5.2 was generated in this way.  clear; firstFile = 91; lastFile = 100; nSamples = 10; columns = 7; for n = firstFile : 1 : lastFile number = num2str( n ); fileIn = strcat('P',number,'.txt'); load VoltageCorrectionFactors_April2_09.txt; %give the matrix name "A" A(:,:) = load( fileIn ); PVCF=VoltageCorrectionFactors_April2_09; %rearrange the matrix into 60 columns each with 30000 rows (1 minute) %from a matrix of 7 columns (column 1 is time) each with 300000 rows for i=1:10; B(1:30000,1+6*(i-1):6+6*(i-1))=A(1+30000*(i-1):30000+30000*(i-1),2:7); end; %erasing first five secs and last ten seconds from each run to allow for %development of limiting current conditions and channel switching C=B(2500:25000,:); 115  %applying voltage correction factors, takes several minutes for j=1:22500; for k=1:60; D(j,k)=C(j,k)*PVCF(k,1); end; end; %converting from volts to pascals, see Matlab master3 april26 sheet title %'mean vs. stdev' for how the calculation was derived using eqns from %Reiss and Gaucher for l=1:22500; for m=1:60; E=D(l,m); F=E^3*9.77; G(l,m)=F; end; end; %outputting stats in pascals Stdev=std(G)'; Mean=mean(G)'; %concatenates a matrix with six columns (mean,stdev,min,max) and 60 rows Stats=[Mean Stdev]; xlswrite('C:\Documents and Settings\Administrator\Desktop\Matlab output\ShearStress_Vs_Stdev_May17,09.xls', Stats, number); end;  116  APPENDIX 9: SHEAR STRESS MAPS  Probe placement: Vertical sheet 1  1 16 31 46  2 17 32 47  3 18 33 48  Vertical sheet 2  4 19 34 49  5 20 35 50  6 21 36 51  Vertical sheet 3  7 22 37 52  8 23 38 53  9 24 39 54  10 25 40 55  11 26 41 56  12 27 42 57  13 28 43 58  14 29 44 59  15 30 45 60  Vertical sheet 4  Vertical sheet 5  Horizontal Sheet 1  1 4 7 10 14  2 5 8 11 14  3 6 9 12 15  Horizontal Sheet 2  16 19 22 25 29  17 20 23 26 29  18 21 24 27 30  Horizontal Sheet 3  31 34 37 40 44  32 35 38 41 44  33 36 39 42 45  Horizontal Sheet 4  46 49 52 55 59  47 50 53 56 59  48 51 54 57 60  Figure A9.1 Guide to probe placement within the cassette. Shear probes were placed within the hollow fiber membrane cassette in a multilayered grid. The numbers above refer to the probe used at each location (total 60 probes used). All probes were placed within one corner of the cassette (see section 3.4), but due to system symmetry the probe data can be used to elucidate shear stress across the entire cassette.  117  Figure A9.2 Description of the vertical shear maps. Voltage data collected from the sixty shear probes can most easily be visualized as a series of cassette vertical and horizontal cross sections (‘sheets’). Five vertical sheets illustrate the shear stress from the center of inner module (vertical sheet 1) to the outside of the outer module (vertical sheet 5). Vertical sheets are illustrated in figures A9.1a-A9.35a. See also figs. 3.8 to 3.10 for description of probe placement.  118  119  Figure A9.3 Description of the horizontal shear maps. Voltage data collected from the sixty shear probes can most easily be visualized as a series of cassette vertical and horizontal cross sections (‘sheets’). Four horizontal sheets illustrate the shear stress from near the top of the cassette (horizontal sheet 1) to near the bottom of the cassette (horizontal sheet 4). Horizontal sheets are illustrated in figures A9.1bA9.35b. See also figs. 3.8 to 3.10 for description of probe placement.  120  121  Figure A9.4 Interpolation of shear data outside the quadrant of measurement. Due to system symmetry, measurements from one corner of the membrane cassette could be used to interpolate the shear stress within the entire system (see figure 3.8). For vertical sheets half of the data presented is from measurements, while the other half is a reflection of the measured data. For horizontal sheets one-quarter of the data presented is from measurements, while the remainder was generated through reflections across the system centerlines. 122 Axis units are given in meters.  Table A9.1  Index to figures within appendix 9 (sections a, b, and c).  Figures are designed to maximize the ability to identify trends, as the effect of any single variable can be observed while maintaining the other variables constant. Sections A9a, A9b, and A9c all present the same data, but in three different formats. The effect of gas flow rate  Figures: A9.1a, A9.1b, A9.2a, A9.2b, A9.3a, A9.3b, A9.4a, A9.4b, A9.5a, A9.5b, A9.6a, A9.6b, A9.7a, A9.7b, A9.8a, A9.8b, A9.9a, A9.9b,  A9.1c A9.2c A9.3c A9.4c A9.5c A9.6c A9.7c A9.8c A9.9c  Fiber tension  Module spacing  Gas flow rate  tight tight tight medium medium medium loose loose loose  narrow medium wide narrow medium wide narrow medium wide  v aried v aried v aried v aried v aried v aried v aried v aried v aried  Fiber tension  Module spacing  Gas flow rate  tight tight tight medium medium medium loose loose loose  v aried v aried v aried v aried v aried v aried v aried v aried v aried  low medium high low medium high low medium high  Fiber tension  Module spacing  Gas flow rate  v aried v aried v aried v aried v aried v aried v aried v aried v aried  narrow medium wide narrow medium wide narrow medium wide  low medium high low medium high low medium high  Fiber tension  Module spacing  Gas flow rate  medium medium medium medium medium medium medium medium medium medium  narrow wide narrow wide narrow wide narrow wide narrow wide  medium medium high low medium high low medium high high  The effect of module spacing  Figures: A9.10a, A9.10b, A9.11a, A9.11b, A9.12a, A9.12b, A9.13a, A9.13b, A9.14a, A9.14b, A9.15a, A9.15b, A9.16a, A9.16b, A9.17a, A9.17b, A9.18a, A9.18b,  A9.10c A9.11c A9.12c A9.13c A9.14c A9.15c A9.16c A9.17c A9.18c  The effect of fiber tension  Figures: A9.19a, A9.19b, A9.20a, A9.20b, A9.21a, A9.21b, A9.22a, A9.22b, A9.23a, A9.23b, A9.24a, A9.24b, A9.25a, A9.25b, A9.26a, A9.26b, A9.27a, A9.27b,  A9.19c A9.20c A9.21c A9.22c A9.23c A9.24c A9.25c A9.26c A9.27c  The effect of sparging pattern  Figures: A9.28a, A9.28b A9.29a, A9.29b A9.30a, A9.30b A9.31a, A9.31b A9.32a, A9.32b A9.33a, A9.33b A9.34a, A9.34b A9.35a, A9.35b A9.28c A9.29c  Sparging patterns Slow alternating vs. fast alternating vs. continuous Slow alternating vs. fast alternating vs. continuous Slow pulse vs. fast pulse vs. continuous Slow pulse vs. fast pulse vs. continuous Slow alternating vs. slow pulse v s. continuous Slow alternating vs. slow pulse v s. continuous Fast alternating vs. fast pulse vs. continuous Fast alternating vs. fast pulse vs. continuous all five sparging patterns all five sparging patterns  123  Figure A9.1a The effect of increasing gas flow rate on average surface shear (measured in volts) distribution in a vertical sheet Experimental configuration: High tension, narrow spacing, three gas flow rates. 124 Top row= low gas; Middle row= medium gas; Bottom row= high gas. Axis units in [m].  Figure A9.2a The effect of increasing gas flow rate on average surface shear (measured in volts) distribution in a vertical sheet Experimental configuration: High tension, medium spacing, three gas flow rates. 125 Top row= low gas; Middle row= medium gas; Bottom row= high gas. Axis units in [m].  Figure A9.3a The effect of increasing gas flow rate on average surface shear (measured in volts) distribution in a vertical sheet Experimental configuration: High tension, wide spacing, three gas flow rates. 126 Top row= low gas; Middle row= medium gas; Bottom row= high gas. Axis units in [m].  Figure A9.4a The effect of increasing gas flow rate on average surface shear (measured in volts) distribution in a vertical sheet 127 Experimental configuration: Medium tension, narrow spacing, three gas flow rates. Top row= low gas; Middle row= medium gas; Bottom row= high gas. Axis units in [m].  Figure A9.5a The effect of increasing gas flow rate on average surface shear (measured in volts) distribution in a vertical sheet Experimental configuration: Medium tension, medium spacing, three gas flow rates. 128 Top row= low gas; Middle row= medium gas; Bottom row= high gas. Axis units in [m].  Figure A9.6a The effect of increasing gas flow rate on average surface shear (measured in volts) distribution in a vertical sheet Experimental configuration: Medium tension, wide spacing, three gas flow rates. 129 Top row= low gas; Middle row= medium gas; Bottom row= high gas. Axis units in [m].  Figure A9.7a The effect of increasing gas flow rate on average surface shear (measured in volts) distribution in a vertical sheet Experimental configuration: Low tension, narrow spacing, three gas flow rates. 130 Top row= low gas; Middle row= medium gas; Bottom row= high gas. Axis units in [m].  Figure A9.8a The effect of increasing gas flow rate on average surface shear (measured in volts) distribution in a vertical sheet 131 Experimental configuration: Low tension, medium spacing, three gas flow rates. Top row= low gas; Middle row= medium gas; Bottom row= high gas. Axis units in [m].  Figure A9.9a The effect of increasing gas flow rate on average surface shear (measured in volts) distribution in a vertical sheet 132 Experimental configuration: Low tension, wide spacing, three gas flow rates. Top row= low gas; Middle row= medium gas; Bottom row= high gas. Axis units in [m].  Figure A9.10a The effect of increasing spacing distance between modules on average surface shear (measured in volts) distribution in a vertical sheet. High tension, three module spacings, low gas flow rate. 133 Top row= narrow spacing; Middle row= medium; Bottom row= wide. Axis units in [m].  Figure A9.11a The effect of increasing spacing distance between modules on average surface shear (measured in volts) distribution in a vertical sheet. 134 High tension, three module spacings, medium gas flow rate. Top row= narrow spacing; Middle row= medium; Bottom row= wide. Axis units in [m].  Figure A9.12a The effect of increasing spacing distance between modules on average surface shear (measured in volts) distribution in a vertical sheet. 135 High tension, three module spacings, high gas flow rate. Top row= narrow spacing; Middle row= medium; Bottom row= wide. Axis units in [m].  Figure A9.13a The effect of increasing spacing distance between modules on average surface shear (measured in volts) distribution in a vertical sheet. 136 High tension, three module spacings, low gas flow rate. Top row= narrow spacing; Middle row= medium; Bottom row= wide. Axis units in [m].  Figure A9.14a The effect of increasing spacing distance between modules on average surface shear (measured in volts) distribution in a vertical sheet. 137 Medium tension, three module spacings, medium gas flow rate. Top row= narrow spacing; Middle row= medium; Bottom row= wide. Axis units in [m].  Figure A9.15a The effect of increasing spacing distance between modules on average surface shear (measured in volts) distribution in a vertical sheet. Medium tension, three module spacings, high gas flow rate. 138 Top row= narrow spacing; Middle row= medium; Bottom row= wide. Axis units in [m].  Figure A9.16a The effect of increasing spacing distance between modules on average surface shear (measured in volts) distribution in a vertical sheet. 139 Low tension, three module spacings, low gas flow rate. Top row= narrow spacing; Middle row= medium; Bottom row= wide. Axis units in [m].  Figure A9.17a The effect of increasing spacing distance between modules on average surface shear (measured in volts) distribution in a vertical sheet. 140 Low tension, three module spacings, medium gas flow rate. Top row= narrow spacing; Middle row= medium; Bottom row= wide. Axis units in [m].  Figure A9.18a The effect of increasing spacing distance between modules on average surface shear (measured in volts) distribution in a vertical sheet. 141 Low tension, three module spacings, high gas flow rate. Top row= narrow spacing; Middle row= medium; Bottom row= wide. Axis units in [m].  Figure A9.19a The effect of fiber tension on average surface shear (measured in volts) distribution in a vertical sheet. 142 Three fiber tensions, narrow module spacing, low gas flow rate. Top row= high tension; Middle row= medium; Bottom row= loosest. Axis units in [m].  Figure A9.20a The effect of fiber tension on average surface shear (measured in volts) distribution in a vertical sheet. 143 Three fiber tensions, medium module spacing, low gas flow rate. Top row= high tension; Middle row= medium; Bottom row= loosest. Axis units in [m].  Figure A9.21a The effect of fiber tension on average surface shear (measured in volts) distribution in a vertical sheet. 144 Three fiber tensions, wide module spacing, low gas flow rate. Top row= high tension; Middle row= medium; Bottom row= loosest. Axis units in [m].  Figure A9.22a The effect of fiber tension on average surface shear (measured in volts) distribution in a vertical sheet. 145 Three fiber tensions, narrow module spacing, medium gas flow rate. Top row= high tension; Middle row= medium; Bottom row= loosest. Axis units in [m].  Figure A9.23a The effect of fiber tension on average surface shear (measured in volts) distribution in a vertical sheet. 146 Three fiber tensions, medium module spacing, medium gas flow rate. Top row= high tension; Middle row= medium; Bottom row= loosest. Axis units in [m].  Figure A9.24a The effect of fiber tension on average surface shear (measured in volts) distribution in a vertical sheet. 147 Three fiber tensions, wide module spacing, medium gas flow rate. Top row= high tension; Middle row= medium; Bottom row= loosest. Axis units in [m].  Figure A9.25a The effect of fiber tension on average surface shear (measured in volts) distribution in a vertical sheet. 148 Three fiber tensions, narrow module spacing, high gas flow rate. Top row= high tension; Middle row= medium; Bottom row= loosest. Axis units in [m].  Figure A9.26a The effect of fiber tension on average surface shear (measured in volts) distribution in a vertical sheet. 149 Three fiber tensions, medium module spacing, high gas flow rate. Top row= high tension; Middle row= medium; Bottom row= loosest. Axis units in [m].  Figure A9.27a The effect of fiber tension on average surface shear (measured in volts) distribution in a vertical sheet. 150 Three fiber tensions, wide module spacing, high gas flow rate. Top row= high tension; Middle row= medium; Bottom row= loosest. Axis units in [m].  Figure A9.28a Alternating sparging between diffusers compared to continuous gas flow, effect on average surface shear (volts) distribution in a vertical sheet. Medium tension, narrow spacing, medium gas flow rate. 151 Top row= continuous gas flow; Middle row= slow alternating (6 sec); Bottom row= fast alternating (3 sec). Axis units are in [m].  Figure A9.29a Alternating sparging between diffusers compared to continuous gas flow, effect on average surface shear (volts) distribution in a vertical sheet. 152 Medium tension, wide spacing, medium gas flow rate. Top row= continuous gas flow; Middle row= slow alternating (6 sec); Bottom row= fast alternating (3 sec). Axis units are in [m].  Figure A9.30a Pulse sparging (both on/ both off) compared to continuous gas flow, effect on average surface shear (volts) distribution in a vertical sheet. 153 Medium tension, narrow spacing, medium gas flow rate. Top row= continuous gas flow; Middle row= slow pulse (6 sec); Bottom row= fast pulse (3 sec). Axis units are in [m].  Figure A9.31a Pulse sparging (both on/ both off) compared to continuous gas flow, effect on average surface shear (volts) distribution in a vertical sheet. 154 Medium tension, wide spacing, medium gas flow rate. Top row= continuous gas flow; Middle row= slow pulse (6 sec); Bottom row= fast pulse (3 sec). Axis units are in [m].  Figure A9.32a Three sparging patterns compared, effect on average surface shear (volts) distribution in a vertical sheet. 155 Medium tension, narrow spacing, medium gas flow rate. Top row= continuous gas flow; Middle row= slow alternating (6 sec); Bottom row= slow pulse (6 sec). Axis units are in [m].  Figure A9.33a Three sparging patterns compared, effect on average surface shear (volts) distribution in a vertical sheet. 156 Medium tension, wide spacing, medium gas flow rate. Top row= continuous gas flow; Middle row= slow alternating (6 sec); Bottom row= slow pulse (6 sec). Axis units are in [m].  Figure A9.34a Three sparging patterns compared, effect on average surface shear (volts) distribution in a vertical sheet. Medium tension, narrow spacing, medium gas flow rate. Top row= continuous gas flow; Middle row= fast alternating (3 sec); Bottom row= fast pulse (3 sec). Axis units are in [m].  157  Figure A9.35a Three sparging patterns compared, effect on average surface shear (volts) distribution in a vertical sheet. Medium tension, wide spacing, medium gas flow rate. Top row= continuous gas flow; Middle row= fast alternating (3 sec); Bottom row= fast pulse (3 sec). Axis units are in [m].  158  Figure A9.1b The effect of increasing gas flow rate on average surface shear (volts) distribution in a horizontal sheet. High fiber tension, narrow module spacing, three gas flow rates. First column= low gas flow; Second= medium gas; Third= high gas. Axis units are in [m] 159  Figure A9.2b  The effect of increasing gas flow rate on average surface shear (volts) distribution in a horizontal sheet. High fiber tension, medium module spacing, three gas flow rates. First column= low gas flow; Second= medium gas; Third= high gas. Axis units are in [m] 160  Figure A9.3b The effect of increasing gas flow rate on average surface shear (volts) distribution in a horizontal sheet. High fiber tension, wide module spacing, three gas flow rates. First column= low gas; Second= medium gas; Third= high gas. Axis units are in [m]. 161  Figure A9.4b The effect of increasing gas flow rate on average surface shear (volts) distribution in a horizontal sheet. Medium fiber tension, narrow module spacing, three gas flow rates. 162 First column= low gas; Second= medium gas; Third= high gas. Axis units are in [m].  Figure A9.5b The effect of increasing gas flow rate on average surface shear (volts) distribution in a horizontal sheet. Medium fiber tension, medium module spacing, three gas flow rates. 163 First column= low gas; Second= medium gas; Third= high gas. Axis units are in [m].  Figure A9.6b The effect of increasing gas flow rate on average surface shear (volts) distribution in a horizontal sheet. Medium fiber tension, wide module spacing, three gas flow rates. 164 First column= low gas; Second= medium gas; Third= high gas. Axis units are in [m].  Figure A9.7b The effect of increasing gas flow rate on average surface shear (volts) distribution in a horizontal sheet. Low fiber tension, narrow module spacing, three gas flow rates. 165 First column= low gas; Second= medium gas; Third= high gas. Axis units are in [m].  Figure A9.8b The effect of increasing gas flow rate on average surface shear (volts) distribution in a horizontal sheet. Low fiber tension, medium module spacing, three gas flow rates. 166 First column= low gas; Second= medium gas; Third= high gas. Axis units are in [m].  Figure A9.9b The effect of increasing gas flow rate on average surface shear (volts) distribution in a horizontal sheet. Low fiber tension, wide module spacing, three gas flow rates. 167 First column= low gas; Second= medium gas; Third= high gas. Axis units are in [m].  Figure A9.10b The effect of increasing spacing distance between modules on average surface shear (volts) distribution in a horizontal sheet. High fiber tension, three module spacings, low gas flow rate. First column= narrow spacing; Second= medium; Third= wide. Axis units are in [m].168  Figure A9.11b The effect of increasing spacing distance between modules on average surface shear (volts) distribution in a horizontal sheet. High fiber tension, three module spacings, medium gas flow rate. First column= narrow spacing; Second= medium; Third= wide. Axis units are in [m].169  Figure A9.12b The effect of increasing spacing distance between modules on average surface shear (volts) distribution in a horizontal sheet. High fiber tension, three module spacings, high gas flow rate. First column= narrow spacing; Second= medium; Third= wide. Axis units are in [m].170  Figure A9.13b The effect of increasing spacing distance between modules on average surface shear (volts) distribution in a horizontal sheet. Medium fiber tension, three module spacings, low gas flow rate. First column= narrow spacing; Second= medium; Third= wide. Axis units are in [m].171  Figure A9.14b The effect of increasing spacing distance between modules on average surface shear (volts) distribution in a horizontal sheet. Medium fiber tension, three module spacings, medium gas flow rate. First column= narrow spacing; Second= medium; Third= wide. Axis units are in [m].172  Figure A9.15b The effect of increasing spacing distance between modules on average surface shear (volts) distribution in a horizontal sheet. Medium fiber tension, three module spacings, high gas flow rate. First column= narrow spacing; Second= medium; Third= wide. Axis units are in [m].173  Figure A9.16b The effect of increasing spacing distance between modules on average surface shear (volts) distribution in a horizontal sheet. Low fiber tension, three module spacings, low gas flow rate. First column= narrow spacing; Second= medium; Third= wide. Axis units are in [m].174  Figure A9.17b The effect of increasing spacing distance between modules on average surface shear (volts) distribution in a horizontal sheet. Low fiber tension, three module spacings, medium gas flow rate. First column= narrow spacing; Second= medium; Third= wide. Axis units are in [m].175  Figure A9.18b The effect of increasing spacing distance between modules on average surface shear (volts) distribution in a horizontal sheet. Low fiber tension, three module spacings, high gas flow rate. First column= narrow spacing; Second= medium; Third= wide. Axis units are in [m].176  Figure A9.19b The effect of fiber tension on average surface shear (volts) distribution in a horizontal sheet. Three fiber tensions, narrow module spacing, low gas flow rate. First column= high tension; Second= medium; Third= loosest. Axis units are in [m].177  Figure A9.20b The effect of fiber tension on average surface shear (volts) distribution in a horizontal sheet. Three fiber tensions, medium module spacing, low gas flow rate. First column= high tension; Second= medium; Third= loosest. Axis units are in [m].178  Figure A9.21b The effect of fiber tension on average surface shear (volts) distribution in a horizontal sheet. Three fiber tensions, wide module spacing, low gas flow rate. First column= high tension; Second= medium; Third= loosest. Axis units are in [m].179  Figure A9.22b The effect of fiber tension on average surface shear (volts) distribution in a horizontal sheet. Three fiber tensions, narrow module spacing, medium gas flow rate. First column= high tension; Second= medium; Third= loosest. Axis units are in [m].180  Figure A9.23b The effect of fiber tension on average surface shear (volts) distribution in a horizontal sheet. Three fiber tensions, medium module spacing, medium gas flow rate. First column= high tension; Second= medium; Third= loosest. Axis units are in [m].181  Figure A9.24b The effect of fiber tension on average surface shear (volts) distribution in a horizontal sheet. Three fiber tensions, wide module spacing, medium gas flow rate. First column= high tension; Second= medium; Third= loosest. Axis units are in [m].182  Figure A9.25b The effect of fiber tension on average surface shear (volts) distribution in a horizontal sheet. Three fiber tensions, narrow module spacing, high gas flow rate. First column= high tension; Second= medium; Third= loosest. Axis units are in [m].183  Figure A9.26b The effect of fiber tension on average surface shear (volts) distribution in a horizontal sheet. Three fiber tensions, medium module spacing, high gas flow rate. First column= high tension; Second= medium; Third= loosest. Axis units are in [m].184  Figure A9.27b The effect of fiber tension on average surface shear (volts) distribution in a horizontal sheet. Three fiber tensions, wide module spacing, high gas flow rate. First column= high tension; Second= medium; Third= loosest. Axis units are in [m].185  Figure A9.28b Alternating sparging between diffusers compared to continuous gas flow, effect on average surface shear (volts) distribution in a horizontal sheet. Medium tension, narrow spacing, medium gas flow rate. First column= continuous gas flow; second= slow alternating (6 sec); 186 third= fast alternating (3 sec). Axis units are in [m].  Figure A9.29b Alternating sparging between diffusers compared to continuous gas flow, effect on average surface shear (volts) distribution in a horizontal sheet. Medium tension, wide spacing, medium gas flow rate. First column= continuous gas flow; second= slow alternating (6 sec); 187 third= fast alternating (3 sec). Axis units are in [m].  Figure A9.30b Pulse sparging (both on/ both off) compared to continuous gas flow, effect on average surface shear (volts) distribution in a horizontal sheet. Medium tension, narrow spacing, medium gas flow rate. First column= continuous gas flow; second= slow pulse (6 sec); 188 third= fast pulse (3 sec). Axis units are in [m].  Figure A9.31b Pulse sparging (both on/ both off) compared to continuous gas flow, effect on average surface shear (volts) distribution in a horizontal sheet. Medium tension, wide spacing, medium gas flow rate. First column= continuous gas flow; second= slow pulse (6 sec); 189 third= fast pulse (3 sec). Axis units are in [m].  Figure A9.32b Three sparging patterns compared, effect on average surface shear (volts) distribution in a horizontal sheet. Medium tension, narrow spacing, medium gas flow rate. First column= continuous gas flow; second= slow alternating (6 sec); 190 third= slow pulse (3 sec). Axis units are in [m].  Figure A9.33b Three sparging patterns compared, effect on average surface shear (volts) distribution in a horizontal sheet. Medium tension, wide spacing, medium gas flow rate. First column= continuous gas flow; second= slow alternating (6 sec); 191 third= slow pulse (3 sec). Axis units are in [m].  Figure A9.34b Three sparging patterns compared, effect on average surface shear (volts) distribution in a horizontal sheet. Medium tension, narrow spacing, medium gas flow rate. First column= continuous gas flow; second= fast alternating (3 sec); 192 third= fast pulse (3 sec). Axis units are in [m].  Figure A9.35b Three sparging patterns compared, effect on average surface shear (volts) distribution in a horizontal sheet. Medium tension, wide spacing, medium gas flow rate. First column= continuous gas flow; second= fast alternating (3 sec); 193 third= fast pulse (3 sec). Axis units are in [m].  High tension, narrow spacing, three gas flow rates -0.70  TNL TNM TNH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  High tension, narrow spacing, three gas flow rates -0.70  TNL TNM TNH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.1c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 194 Tables 3.2 and 3.3 explain the experimental configuration naming system.  High tension, medium spacing, three gas flow rates -0.70  TML TMM TMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  High tension, medium spacing, three gas flow rates -0.70  TML TMM TMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.2c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 195 Tables 3.2 and 3.3 explain the experimental configuration naming system.  High tension, wide spacing, three gas flow rates -0.70  TWL TWM TWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  High tension, wide spacing, three gas flow rates -0.70  TWL TWM TWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.3c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 196 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Medium tension, narrow spacing, three gas flow rates -0.70  MNL MNM MNH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Medium tension, narrow spacing, three gas flow rates -0.70  MNL MNM MNH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.4c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 197 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Medium tension, medium spacing, three gas flow rates -0.70  MML MMM MMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Medium tension, medium spacing, three gas flow rates -0.70  MML MMM MMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.5c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 198 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Medium tension, wide spacing, three gas flow rates -0.70  MWL MWM MWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Medium tension, wide spacing, three gas flow rates -0.70  MWL MWM MWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.6c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 199 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Low tension, narrow spacing, three gas flow rates -0.70  LNL LNM LNH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Low tension, narrow spacing, three gas flow rates -0.70  LNL LNM LNH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.7c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 200 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Low tension, medium spacing, three gas flow rates -0.70  LML LMM LMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Low tension, medium spacing, three gas flow rates -0.70  LML LMM LMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.8c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 201 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Low tension, wide spacing, three gas flow rates -0.70  LWL LWM LWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Low tension, wide spacing, three gas flow rates -0.70  LWL LWM LWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.9c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 202 Tables 3.2 and 3.3 explain the experimental configuration naming system.  High tension, three module spacings, low gas flow -0.70  TNL TML TWL  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  High tension, three module spacings, low gas flow -0.70  TNL TML TWL  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.10c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 203 Tables 3.2 and 3.3 explain the experimental configuration naming system.  High tension, three module spacings, medium gas flow -0.70  TNM TMM TWM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  High tension, three module spacings, medium gas flow -0.70  TNM TMM TWM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.11c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 204 Tables 3.2 and 3.3 explain the experimental configuration naming system.  High tension, three module spacings, high gas flow -0.70  TNH TMH TWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  High tension, three module spacings, high gas flow -0.70  TNH TMH TWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.12c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 205 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Medium tension, three module spacings, low gas flow -0.70  MNL MML MWL  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Medium tension, three module spacings, low gas flow -0.70  MNL MML MWL  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.13c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 206 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Medium tension, three module spacings, medium gas flow -0.70  MNM MMM MWM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Medium tension, three module spacings, medium gas flow -0.70  MNM MMM MWM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.14c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 207 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Medium tension, three module spacings, high gas flow -0.60  MNH MMH MWH  -0.50  Volts  -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Medium tension, three module spacings, high gas flow -0.70  MNH MMH MWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.15c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 208 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Low tension, three module spacings, low gas flow -0.70  LNL LML LWL  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Low tension, three module spacings, low gas flow -0.70  LNL LML LWL  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.16c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 209 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Low tension, three module spacings, medium gas flow -0.60  LNM LMM LWM  -0.50  Volts  -0.40  -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Low tension, three module spacings, medium gas flow -0.70  LNM LMM LWM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.17c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 210 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Low tension, three module spacings, high gas flow -0.60  LNH LMH LWH  -0.50  Volts  -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Low tension, three module spacings, high gas flow -0.70  LNH LMH LWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.18c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 211 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Three fiber tensions, narrow spacing, low gas flow -0.70  LNL MNL TNL  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Three fiber tensions, narrow spacing, low gas flow -0.70  LNL MNL TNL  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.19c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 212 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Three fiber tensions, medium spacing, low gas flow -0.70  LML MML TML  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Three fiber tensions, medium spacing, low gas flow -0.70  LML MML TML  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.20c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 213 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Three fiber tensions, wide spacing, low gas flow -0.60  LWL MWL TWL  -0.50  Volts  -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Three fiber tensions, wide spacing, low gas flow -0.70  LWL MWL TWL  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.21c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 214 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Three fiber tensions, narrow spacing, medium gas flow -0.70  LNM MNM TNM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Three fiber tensions, narrow spacing, medium gas flow -0.70  LNM MNM TNM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.22c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 215 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Three fiber tensions, medium spacing, medium gas flow -0.70  LMM MMM TMM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Three fiber tensions, medium spacing, medium gas flow -0.70  LMM MMM TMM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.23c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 216 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Three fiber tensions, wide spacing, medium gas flow -0.70  LWM MWM TWM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, interior face  outer cassette, center  outer cassette, outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Three fiber tensions, wide spacing, medium gas flow -0.70  LWM MWM TWM  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.24c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 217 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Three fiber tensions, narrow spacing, high gas flow -0.70  LNH MNH TNH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Three fiber tensions, narrow spacing, high gas flow -0.70  LNH MNH TNH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.25c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 218 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Three fiber tensions, medium spacing, high gas flow -0.70  LMH MMH TMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Three fiber tensions, medium spacing, high gas flow -0.70  LMH MMH TMH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.26c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 219 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Three fiber tensions, wide spacing, high gas flow -0.70  LWH MWH TWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette, center  middle cassette, interior face  outer cassette, outer cassette, outer cassette, interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Three fiber tensions, wide spacing, high gas flow -0.70  LWH MWH TWH  -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.27c Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 220 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Medium tension, narrow spacing, medium gas flow, five sparging patterns  MNM FA.MNM FP.MNM SA.MNM SP.MNM  -0.70 -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette center  middle cassette, interior face  outer cassette, outer cassette outer cassette interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Medium tension, narrow spacing, medium gas flow, five sparging patterns  MNM FA.MNM FP.MNM SA.MNM SP.MNM  -0.70 -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.28c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 221 Tables 3.2 and 3.3 explain the experimental configuration naming system.  Medium tension, wide spacing, medium gas flow, five sparging patterns  MWM FA.MWM FP.MWM SA.MWM SP.MWM  -0.70 -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10  middle cassette center  middle cassette, interior face  outer cassette, outer cassette outer cassette interior face center outer face  0.00  Vertical sheets (*error bars = 2 standard deviations)  Medium tension, wide spacing, medium gas flow, five sparging patterns  MWM FA.MWM FP.MWM SA.MWM SP.MWM  -0.70 -0.60  Volts  -0.50 -0.40 -0.30 -0.20 -0.10 top  mid high  mid low  bottom  0.00  Horizontal sheets (*error bars = 2 standard deviations)  Figure A9.29c  Bar charts presenting average voltage measurements from within vertical and horizontal cross sections (‘sheets’) within the membrane cassette. See figures A9.2 and A9.3 for diagrams describing sheet locations. 222 Tables 3.2 and 3.3 explain the experimental configuration naming system.  

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}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            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:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0063129/manifest

Comment

Related Items