DETERMINATION OF FOULING MECHANISMS FOR ULTRAFILTRATION OF OILY WASTEWATER
A thesis submitted to the
Division of Graduate Studies and Research of the University of Cincinnati
in partial fulfillment of the requirements of the degree of
MASTER OF SCIENCE (M.S.)
in the Department of Chemical Engineering of the College of Engineering and Applied Science
2011
By
Leila Safazadeh Haghighi
BS Chemical Engineering, University Of Tehran, Tehran, Iran, 2008
Thesis Advisor and Committee Chair: Professor Rakesh Govind
Abstract
The use of Membrane technology is extensively increasing in water and wastewater treatment, food processing, chemical, biotechnological, and pharmaceutical industries because of their versatility, effectiveness, high removal capacity and ability to meet multiple treatment objectives. A common problem with using membranes is fouling, which results in increasing operating costs due to higher operating pressure losses, membrane downtime needed for cleaning, with associated production loss and manpower costs. In the literature, four different mechanisms for membrane fouling have been studied, which are complete pore blocking, internal pore blinding, partial pore bridging and cake filtration. Mathematical models have been developed for each of these fouling mechanisms.
The objective of this thesis was to investigate the membrane fouling mechanisms for one porous and one dense membrane, during ultrafiltration of an emulsified industrial oily wastewater. An experimental system was designed, assembled and operated at the
Ford Transmission Plant in Sharonville, Ohio, wherein ultrasonic baths were used for cleaning transmission parts before assembly. The oil wastewater, containing emulsified oils and cleaning chemicals was collected in a batch vessel and then pumped through a porous polyethersulfone, monolithic membrane, and through a dense cuproammonium cellulose membrane unit. For the porous membrane, use of a Dupont’s flurosurfactant
(FS 63) and backwashing with permeate and for the dense membrane the use of both the flurorosurfactant and sparged air were investigated to reduce membrane fouling.
ii
For the porous membrane study, it was observed that the permeate flux was strongly dependent on the transmembrane pressure difference, and addition of the flurosurfactant significantly improved the performance of the membrane. The backwashing cleaning efficiency was found to depend on the duration of backwashing and its frequency. An integrated fouling model was developed by combining the individual models for each fouling mechanism, originally published by Hermia [18], and analysis of the experimental data for ultrafiltration of oily emulsion revealed that the primary mechanism for fouling of the porous membrane was cake filtration. With increasing transmembrane pressure, the role of other mechanisms, such as pore blocking and partial pore bridging, increases, although the effect of cake filtration dominates.
Hence, for oily emulsions, methods to disrupt the formation of a cake layer at the membrane surface would have the most impact in increasing the water permeation rates through the membrane.
For the dense membrane study, permeate flux also increased with increasing transmembrane pressure difference, as in the porous membrane, and the major mechanisms for fouling were found to be concentration polarization gel layer formation on the membrane surface. In this case, the use of both sparged air and fluorosurfactant, increased the water permeation rates, but the permeation rate improvement with sparged air alone was significantly higher than with fluorosurfactant only. A mathematical model was developed to derive the mass transfer coefficients under the various operation conditions.
iii
Future studies will concentrate on improving membrane performance by reducing the impact of the dominant fouling mechanisms, found in this study, for both porous and dense membranes.
iv
v
DEDICATION
THIS THESIS IS DEDICATED TO MY FAMILY
vi
Acknowledgements
I would like to express my sincere gratitude to my adviser, Professor Rakesh
Govind. I am deeply indebted to him for giving me the opportunity to work on this project. I’m particularly grateful for his enthusiastic guidance, discussion, understanding, encouragement and numerous hours spent helping me complete this thesis.
I’d also like to express my special thanks to the thesis committee, Professor
Junhang Dong and Professor Joo-Youp Lee for the efforts to provide valuable comments during my proposal presentation, for the their valuable time to review this thesis and for offering me an opportunity to defend in front of them.
My appreciation is also extended to Mr. Lyle Carman and Mr. David Ferguson for helping me with the experimental set-up.
I am really grateful to my dear husband for his company, understanding, continuous encouragement and support all the way. Without his help, these accomplishments would not have been possible.
vii
Table of Contents
Abstract ...... …..…...ii Acknowledgement……………………………………………………………….………vii Table of Contents ...... …..viii List of Figures ...... xi List of Tables ...... …...xiv List of Symbols and Abbreviations………………………………………...…………...xvi Chapter 1: Introduction ...... 1 1.1 Motivation for Research……...... 1 1.2 Membrane Filtration ……………...... 1 1.3 Membrane Fouling and its Mechanisms...... 6 1.3.1 Concentration Polarization………………………………………...... 7 1.3.2 Cake Formation………………………………………………...... ….8 1.3.3 Natural Organic Matter Adsorption……………………………...…..……...8 1.3.4 Calcium, Iron and Manganese Precipitation……………………………..… 9 1.3.5 Fouling Mechanisms………………………..…………………………….... 9 1.4 Factors Affecting Membrane Fouling……………………………….………….……10 1.5 Conventional Membrane Cleaning Methods……………………………………...…11 1.5.1 Backwashing………………………………………………...………….……11 1.5.2 Enhanced Backwashing……………………………………..………….……12 1.5.3 Chemical Cleaning…………………………………………..………………13 1.6 Disadvantages of the Conventional Cleaning Methods………….………...... 13 1.7 Prevention and Reduction of Membrane Fouling……………………………………14 1.8 Pretreatment……………………………………………………………………….....14 1.8.1 Physical Disruption of Concentration Polarization………………………….15 1.9 Use of Surfactants……………………………………………………………………16
viii
1.10 Thesis Outline………………………………………………………………………17 Chapter 2: Literature Review ...... 18 2.1 Ultrafiltration of Oily Emulsions...... 19 2.2 Conventional Membrane Cleaning Methods….…………….……………………….22 2.2.1 Backwashing ...... 23 2.2.2 Gas Sparging ...... 24 2.3 Ultrasonic Cleaning of Membranes...... 27 2.3.1. Effect of Sonication on Polymeric Membranes...... 29 2.4 Fouling Mechanism in Ultrafiltration...... 31 Chapter 3: Thesis Objectives ...... 34 Chapter 4: Materials and Methods ...... 38 4.1 Selection of Membranes...... 38 4.2 Experimental Systems ...... 42 4.3 Cleaning Procedure...... 45 4.4 Theory of Ultrafiltration of Oily Wastewater………….………………………….....48 4.5 Models for Membrane Fouling Mechanism……….………………………………...52 4.5.1. Fouling Mechanisms Involved In UF Using Porous Monolith Polyether Sulfone Membrane…………………………………………………………52 4.5.1.1 Complete Pore Blocking Model (n=2) …………………………….55 4.5.1.2. Internal Pore Blocking Model (n=3/2)…………..…….…………..56 4.5.1.3. Partial pore bridging model (n=1)…………………………..…….56 4.5.1.4. Cake Layer Formation Model (n=0)………………………………57 Chapter 5: Results and Discussions……………………………………………………...59 5.1 Filtration of Oily Wastewater using Porous Monolith Polyether Sulfone Membrane ……………………………………………………………………………………………59 5.1.1Effect of Transmembrane Pressure……………………………….…..……..59 5.1.2Effect of Feed Concentration……………………………………….…….....64 5.1.3Effect of Backwashing on Permeate Flux Recovery…………………...……66 ix
5.1.4Prediction of Permeate Flux by Hermia’s models……………...……………70 5.1.5. Flux Decay Analysis by using a combination of Hermia’s models…...... 73 5.1.6. Mass Balance Analysis……………………………………………………..82 5.2 Filtration of Oily Wastewater using Dense Hollow-Fiber Regenerated Cellulose Ultrafiltration Membrane……………………………………………...…………...84 5.2.1. Effect of Transmembrane Pressure……………………………………….....84 5.2.2. Effect of Air Injection and Surfactant on Membrane Performance………....90 5.2.3. Analysis of Permeate Flux for Ultrafiltration of oily emulsion in Dense Hollow-Fiber Regenerated Cellulose Membrane……………………….….92 5.2.4. Mass Balance Analysis………………………………………………...……99 Chapter 6: Conclusions and Recommendations………………………………………..101 Bibliography ...... 104 Appendix………………………………………………………………………………..111 Appendix1………………………………………………………………………112 Appendix2………………………………………………………………………122
x
List of Figures
Figure 1.1. Membrane filtration application for solute molecules removal (Page 3)
Figure 1.2. Different fouling mechanisms happening in porous membrane (Page9)
Figure 4.1. Tangential cross-flow filtration (Page 38) Figure 4.2. Structure of the porous membrane fibers (Page 40)
Figure 4.3. Image of the cuprammonium regenerated cellulose hollow fibers (Page 41) Figure 4.4. Schematic figure of the experimental set-up (Page 43) Figure 4.5. Photographs of the experimental system, operated at Ford Motor Company Plant, Sharonville, Ohio.(Page 44) Figure 4.6. Schematic figure of backwash system (Page 46) Figure 4.7. Schematic figure of air injection system (Page 47) Figure 4.8. Demonstration of the contact angle of a liquid sample (Page 49) Figure 4.9. Different fouling mechanisms happening in porous membranes (Page 52)
Figure 5.1 Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures (TMP ) with surfactant (Page 60) Figure 5.2 Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures (TMP) without surfactant (Page 60)
Figure 5.3 Effect of transmembrane pressure (kPa) permeate flux ( l/m2.h) for ultrafiltration with surfactant (Pag e 61)
Figure 5.4 Effect of transmembrane pressure (kPa) permeate flux ( l/m2.h) for ultrafiltration without surfactant (Page 62)
Figure 5.5 Variation of permeate flux ( l/m2.h) with time at transmembrane pressure of 289.69 kPa & 186.16 kPa , for ultrafiltration with and without surfactant (Page 64) Figure 5.6 Comparison of flux variation with time at different feed concentration for ultrafiltration with and without surfactant (Page 65) Figure 5.7 Effect of backwashing interval on permeate flux recovery (Page 67) Figure 5.8. Effect of backwashing duration on permeate flux recovery (Page 69) Figure 5.9. Error values for different Hermia’s models for ultrafiltration of oily wastewate with surfactant at different transmembrane pressure (Page 78)
xi
Figure 5.10. Error values for different Hermia’s models for ultrafiltration of oily wastewate without surfactant at different transmembrane pressure (Page 78) Figure 5.11. reduction precentages of error values combined model with respect to the cake filtration model, for ultrafiltration of oily wastewate with surfactant at different transmembrane pressures (Page 79) Figure 5.12. reduction precentages of error values combined model with respect to the cake filtration model, for ultrafiltration of oily wastewate without surfactant at different transmembrane pressures (Page 80) Figure 5.13. contribution precentages of cake filtration and pore blocking mechanisms, for ultrafiltration of oily wastewate with surfactant at different transmembrane pressure (Page 81) Figure 5.14. contribution precentages of cake filtration and pore blocking mechanisms, for ultrafiltration of oily wastewate with surfactant at different transmembrane pressure (Page 81) Figure 5.15. Comparision of the contribution precentages of cake filtration and pore blocking mechanisms, for ultrafiltration of oily wastewate at different transmembrane pressure with and without surfactant (Page 82)
Figure 5.16. Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures for ultrafiltration of oily wastewate without surfactant and without air injection (Page 85)
Figure 5.17. Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures for ultrafiltration of oily wastewate with surfactant and without air injection (Page 86)
Figure 5.18. Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures for ultrafiltration of oily wastewate with surfactant and with air injection (Page 86)
Figure 5.19 Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures for ultrafiltration of oily wastewate without surfactant and with air injection (Page 87) Figure 5.20. Effect of transmembrane pressure on permeate flux for ultrafiltration of oily wastewate without surfactant and without air injection (Page 88) Figure 5.21. Effect of transmembrane pressure on permeate flux for ultrafiltration of oily wastewate with surfactant and without air injection (Page 88) Figure 5.22. Effect of transmembrane pressure on permeate flux for ultrafiltration of oily wastewate with surfactant and with air injection (Page 89)
xii
Figure 5.23. Effect of transmembrane pressure on permeate flux for ultrafiltration of oily wastewate without surfactant and with air injection (Page 89) Figure 5.24 Effect of air injection and surfactant on permeate flux for transmembrane pressure (Page 90) Figure 5.25 Effect of air injection and surfactant on permeate flux variation with time (Page 91) Figure 5.26 The comparison of the values of the mass transfer coefficient for ultrafiltration of oily emulsion using dense membrane under different experimental conditions (Page 99)
xiii
List of Tables
Table.1.1.Thesis Outline (Page 17)
Table 4.1. Design characteristics of the membrane cartridge (Page 39) Table 4.2. Characteristics of the porous membrane (Page 40) Table 4.3. Hollow fiber module and membrane characteristics (Page 41) Table 5.1. Empirical constants for the linear relation between permeate flux and transmembrane pressure (Page 62) Table 5.2. Percentages of flux enhancement after backwashing with intervals of 60 minutes and 90 minutes (Page 67) Table 5.3. Percentage of flux enhancement after backwashing with durations 100s and 200s (Page 69) Table 5.4. Hermia’s model relation for different fouling mechanisms and the simplified equations (Page 71) Table 5.5. K values of Hermia’s models obtained from experimental data for ultrafiltration with surfactant (Page 72) Table 5.6. K values of Hermia’s models obtained from experimental data for ultrafiltration without surfactant (Page 72) Table 5.7. K values of Hermia’s models obtained for ultrafiltration of oily emulsion with surfactant (Page 76) Table 5.8. K values of Hermia’s models obtained for ultrafiltration of oily emulsion without surfactant (Page 76) Table 5.9.Error values measured for Hermia’s models and the combined model, with Surfactant (Page 77) Table 5.10.Error values measured for Hermia’s models and the combined model, without Surfactant (Page 77) Table 5.11. Calculation of Oil Concentration in the Porous Membrane Reject Flow, with Surfactant (Page 83) Table 5.12. Calculation of Oil Concentration in the Porous Membrane Reject Flow, without Surfactant (Page 83) Table 5.13. Operating conditions for the experiments conducted using the dense membrane (Page 84)
xiv
Table 5.14. Experimental data for ultrafiltration oily emulsion using dense membrane (Page 97 & 98) Table5.15.The fitting parameter of experimental data (Page 98) Table 5.16. Calculation of Oil Concentration in the Dense Membrane Reject Flow (Page 100)
xv
List of Symbols & Abbreviations
A : Membrane surface ( m2 )
A0 : Membrane porous surface ( ) Ac : Cross- sectional area of the fiber ( )
Am : Total effective area of membrane ( )
At : Total membrane active area ( )
Al : Fiber lateral area ( )
C1 : Empirical constant
C2 : Empirical constant
Coil : Oil Concentration (wt %)
2 Dwm : Water diffusion coefficient ( m/s )
(m2 /s) Dw : Water diffusivity (m) d fiber : The fiber internal diameter
2 ErrorCombination : The error value obtained for combined model ( l/m .h)
2 ErrorCake : The error value obtained for cake filtration model ( l/m .h) FD : Flux declination percentage (%)
2 J 0 : Initial permeate flux ( l/m .h)
2 J p : Permeate flux ( l/m .h) 2 J f : Final permeate flux ( l/m .h) 2 J 2 : The permeate flux passed through the pores that are internally blocked ( l/m .h) J 3 : The permeate flux passed through the opening of the pores that are partially blocked 2 ( l/m .h)
xvi
J 3 : The permeate flux passed through the pores that are blanketed with a formed gel layer ( l/m2.h)
2 J Experimental : The flux value ( l/m .h), achieved from experiment at time t(s) and transmembrane pressure (kPa). l/m2.h J Estimated : The flux value ( ), predicted by a Hermia model at time t(s) and transmembrane pressure (kPa).
m3 /m2.s J stst : The steady-state flux ( )
K: Phenomenological coefficient (K unit depend on the parameter n in equation (4.8) K A : Parameter in equation 4.10 that represents the membrane surface blocked per unit of m1 the total volume permeated through the membrane ( )
K B : Parameter in equation 4.12 that represents the decrease in the cross-sectional area of the membrane pores per unit of the total volume permeated through the membrane (1/s)
Kc : Constant in equation 4.9 and 4.10 that corresponds to the complete pore blocking 1 model ( m )
KD : Parameter in equation 4.16 that represents the cake layer area per unit of the total 1/m3 volume permeated through the membrane ( ) K : Constant in equation 4.11 and 4.12 that corresponds to the standard pore blocking S 3 model (1/s )
Ki : Constant in equation 4.14 that corresponds to the intermediate pore blocking model 1 ( m )
K gl : Constant in equation 4.16 that corresponds to the cake layer formation model 6 ( s/m )
kc : The mass transfer coefficient ( m/s)
L : Membrane length with end cup (m) L1 : Membrane length without end cup (m) (m) l fiber : The fiber length
n : General index depending on type of fouling xvii
N : Number of fibers in dense membrane module (KPa) Pc : Pressure of the oil droplet or capillary pressure (KPa) Pinlet : Pressure at the membrane inlet port (KPa) Poutlet : Pressure at the membrane outlet port P : Pressure difference across the membrane ( Pa)
Q : Permeate volume flow rate (ml/min-m3/s) q : volume flow rate (gallons/min)
r : Pore effective radius ( m) t : Time (unit is s or hr, based on the related equation) t fiber: Membrane thickness ( m) rfiber : Radius of the membrane fiber ( m)
1 Rt : Total membrane resistance ( m )
1 Rm : Hydraulic membrane resistance ( m ) R 1 f : Fouling resistance ( m ) Rg : Cake layer resistance ( ) R2 : Fouling resistance due to the internal pore blocking fouling mechanism ( )
R3 : Fouling resistance due to the partial pore bridging mechanism ( )
R4 : Fouling resistance due to the cake filtration mechanism ( )
R : The gas constant ( Pa.m3 /mol.K )
Re : The Reynolds number SCFH: Standard cubic feet per hour T : Feed temperature ( C) (m) t fiber : Fiber thickness
TMP: Transmembrane pressure (KPa) xviii
2 u : Water flow velocity inside the fiber (m /s) (liter) V0 : Initial feed volume
3 V : Accumulated permeate volume ( m )
3 : Molar volume of water ( m /gmole) (Pa.s) water : Water viscosity (Pa.s) p is permeate viscosity 3 water 970(kg/ m ) o / w : Interfacial tension between water and oil droplets
o / w : Contact angle of the oil droplet on the membrane surface
: The proportional constant ( m2.s/m3)
xix
Chapter 1: Introduction
1.1. Motivation for Research
Emulsions of oil in water are encountered in many applications including crude oil recovery, oil refining, automotive, metal plating, food, and wastewater treatment. Due to the presence of surfactants and co-surfactants, the oil-in-water emulsions are stable. In recent years, the use of porous and dense membranes for separating the water from the emulsified oil has been studied extensively, and the major challenge is minimizing the deleterious effect of membrane fouling. Initially, it was thought that the major mechanism of membrane fouling was concentration polarization, which resulted in building up a layer of high concentration of solutes and particles that were rejected by the membrane. Various methods of disrupting concentration polarization were developed, such as high shear stress near the membrane surface, ultrasonic vibration of the fluid, etc.
However, a detailed literature search, presented here in the following section, revealed that concentration polarization was only one mechanism that could explain permeate flux decline through a membrane. The objectives of this research work were therefore changed from studying a specific mechanism of disrupting concentration polarization, to determining the various mechanisms for permeate flux decline when separating water from emulsified oily waste streams.
1.2. Membrane Filtration
The membrane filtration process is a process based on the application of semi- permeable membranes. A membrane is a thin layer of material that is capable of separating particles as a function of their physical and chemical properties when a driving
1 force such as pressure, is applied across the membrane. This pressure difference is either applied pressure, or vacuum. Membrane filtration process can be an effective alternative for the traditional separation techniques such as flocculation, sedimentation, extraction, and distillation.
Based on the pore size, which basically shows the membrane rejection ability, membranes are classified into reverse osmosis (RO), nanofiltration (NF), ultrafiltration
(UF), and microfiltration (MF) membranes [1], Figure (1). All these four types of membranes are in the same category based on their applied driving force, which is the pressure difference across the membrane.
Reverse Osmosis and Nanofiltration, also known as hyper-filtration, are typically applied for removing the dissolved contaminants from feed stream. A Reverse Osmosis membrane has a pore diameter less than 1 nanometer and it’s been designed to retain the salts and low-molecular-weight solutes. A typical Nanofiltration membrane, has pore diameters smaller than 5 nanometer, which lies between RO and UF membrane in terms of selectivity of the membrane, and is designed for removal of the multivalent ions (e.g. calcium and magnesium), in softening operation [1]. Generally, NF membranes exhibit much lower rejection of monovalent ions than RO membranes [3].
An Ultrafiltration membrane is a porous membrane, with a pore diameter ranges from 0.005 to 0.4 µm (nominally 0.01 m). Ultrafiltration membranes are typically categorized based on their molecular weight cut off (MWCO) rather than by a particular pore size. In membrane industry, MWCO or the molecular weight of globular protein that is 90% retained by the membrane, and it’s generally given in Daltons or gram-molecular weight. Ultrafiltration membranes can be applied in different industries such as water
2 treatment, food, and pharmaceutical industry, and basically wherever the recovery of the product compensates for the cost of ultrafiltration. These membranes can be used for treatment of a variety of wastewaters, and also for clarifying juice, beer, and broth [3, 4].
Figure 1.1. Membrane filtration application for solute molecules removal [2]
MF membrane is a step bigger than UF membrane; with a general pore size ranges from 0.1 to 0.2 m (nominally 0.1 m). The main application of MF is for clarification of wastewater. MF membranes are often placed prior to other membranes
(UF, NF, RO) to retain small particles, because all other membranes are at risk of getting fouled by microorganisms and colloids which might be in the feed stream.
Membranes are generally made in four main configurations, Plate and frame, spiral wound, tubular, and hollow fiber. The plate and frame configuration includes a number of flat sheet membranes, which are placed in a series or in parallel form, separated by support plates or filtration spacers. The feed circulates between the
3 membranes of two adjacent plates. The packing density of this configuration is about 100
to 400 m2 /m3, which is not high enough to make the unit efficient. The benefit of this
configuration is that, the units are simply disassembled to have access for manual cleaning or membrane replacement. Depending on the design, permeate can is collected
separately from different support plates, and this makes the diagnosis of the faulty
membranes easy [1]. Spiral wound configuration is made up of two flat sheet membranes,
called a leaf wound, enclosing a flexible porous sheet, which collects permeate and is
sealed on three of its ends. The diameter of this module can be up to 30 cm, and its length
can be up to 1.5m. The benefit of this configuration is its high packing density (700 to
1000 m2 /m3), and also its low head loss. However this configuration is more susceptible
to fouling than plate and frame, and can’t be used without a pretreatment of feed [1]. Tubular configuration is the simplest membrane module and consists of a
membrane placed inside the wall of a porous support tube. These tubes with internal
diameter of 6 to 40mm may be placed individually inside a stainless steel or PVC sleeves
or can be potted in bundles of 3 to 151 tubes in a cylindrical housing. The advantage of
this configuration is that it doesn’t require any pre-filtration of feed stream and they are
simple to clean, but the main disadvantage of this module is that they have a low packing
density and therefore a high capital cost. The last configuration that a membrane can be
formed in is hollow fiber. Hollow fiber module is made up of a bundle of thousands, even
millions of fibers. Depending on the position of the feed flow, these modules can be
categorized into (inside-out configuration) for the flow, taking place inside the fibers, or
(outside-in configuration) for flow taking place outside the fibers. One of the advantages
of this configuration is that the packing density inversely depends on the diameter of the
4 module, and therefore these units are very compact, with the packing density of 1000
m2 /m3 in UF modules to 10,000 m2 /m3 in RO modules. The other benefit of this
configuration, which has made the UF and MF hollow fiber modules, very popular in water treatment industry, is that the fibers are self supporting, and therefore, if the
membranes get fouled, they can be backwashed without getting damaged.
Other than these four configurations, membranes can also be made in “Rotating
Disc” configuration. This module has the advantage that it promotes secondary flows to
help depolarize the solute and particle formation at the membrane interface with solution,
which causes the improvement of flux in pressure driven processes. The disadvantage of
this module is its high energy consumption, difficulties in maintenance and scaling up the
capacity of the module [5]. In general, RO and NF membranes are typically made in
spiral wound configuration, while MF and UF membranes are mostly made in hollow
fiber configuration [6].
The use of Membrane technology is extensively increasing in water and
wastewater treatment, food processing, chemical, biotechnological, and pharmaceutical
industries because of their versatility, effectiveness, high removal capacity and ability to
meet multiple treatment objectives [3, 7]. In water and wastewater treatment, membrane
processes provide an alternative approach to conventional systems for desalination, ultra-
pure water production, pathogen removal from water, and solid- liquid separation. With
membranes having high removal thresholds, conventional coagulation / flocculation, and
sedimentation operations can be replaced by a single process [3].
5 However, a common problem encountered during the membrane filtration processes, is membrane fouling. Membrane fouling leads to increase of operating costs due to the higher pressures needed to maintain permeate flux, downtime needed for membrane cleaning and ultimate membrane replacement [7,8,9].
1.3. Membrane Fouling and its Mechanisms
Fouling is the blockage of membrane pores during filtration, which is caused by the combination of sieving and adsorption of particles and compounds onto the surface of membrane or within its pores. This blockage is the limiting phenomenon that is responsible for a flux decline over time while all operation parameters like temperature, pressure, feed concentration and flow rate are kept constant [10]. Fouling is responsible for most of the difficulties encountered in the generalization of membrane technology for filtration processes, as it worsens membrane performance and shortens membrane life [1,
11].
Fouling can be broadly classified into reversible and irreversible. Reversible fouling is referred to those types, which can be removed with applied cleaning methods such as; backwashing, flushing, and chemical cleaning, and if they are not removable by any means, they are considered as irreversible fouling and in these cases the membrane cannot be restored to its original flow rate [1].
Depending on the foulant, membrane fouling can also be classified into four major categories of;
1. Inorganic fouling/scaling
2. Particle/colloidal fouling
6 3. Microbial/biological fouling
4. Organic fouling
Among different types of membrane fouling, colloids are of particular concern
[12]. Colloidal particles are small enough to easily pass through most pretreatment systems, and consequently block the pores of membrane and/or form a compact cake layer or gel layer on the membrane surface [1]. A wide range of colloids in natural waters may lead to fouling during water treatment, including natural organic matter (NOM), silicate, iron oxides, calcite, and clays [10]. When colloids are present in the feed solution, in low-pressure membrane systems, such as ultra- and micro-filtration, concentration polarization, cake formation, pore blocking, and adsorptive fouling appear to be the predominant causes of decreased permeate flux over time [13,14].
1.3.1. Concentration Polarization
Concentration polarization is essentially a consequence of the solute being stopped by a physical barrier of the membrane rather than a fouling phenomenon. When solutes accumulate on the membrane surface, the consequent concentration gradient reduces the permeate flux. Concentration polarization is considered as one of the main limiting phenomena leading to the permeate flux reduction during the membrane filtration [3].
Since stopping the filtration process, results in the disappearance of the solute concentration gradient, this phenomenon can be considered reversible, except in the case, when accumulated solutes on the membrane wall form a gel layer, where a hydraulic or chemical washing process will be required [15, 16].
7 1.3.2. Cake Formation
Accumulation of the colloidal particles, which are retained due to sieving mechanism, results in the formation of a cake layer on the membrane surface. The cake layer may function as a filter and retain smaller particles or it may compact with time.
Cake layer formation along with pore blocking and adsorption of solute particles into the pore walls, could lead to resistances to permeate flux that can potentially exceed the membrane resistance. This type of fouling can be considered partially reversible, with using hydraulic washing techniques such as backwashing or flushing [1].
1.3.3. Natural Organic Matter Adsorption
Natural organic matter (NOM), in water can cause the membrane blockage, either by adsorption on the formed cake or by adsorption on the membrane pores. All these phenomena can be considered as fouling due to adsorption, and they mainly depend on the affinity that these natural organic matters have for the polymeric membrane material.
This type of fouling is irreversible or at least slowly reversible, since it needs desorption of the organic molecules, to be removed [1]. The reversibility is only possible if the
NOM concentration in feed water drops suddenly. The decrease of concentration shifts the adsorption equilibrium toward desorption, which leads to cleaning of the membrane.
The use of oxidizing shocks, such as chlorine shock during membrane backwashing or chemical cleaning, generally improves the possibility of the membrane permeate flux recovery [8, 14, 17].
8 1.3.4. Calcium, Iron and Manganese Precipitation
Ultrafiltration processes don’t have a high retention for dissolved salts and so these dissolved salts lead to mineral precipitation on the membrane. The mineral precipitation is one of the major causes of membrane fouling. This type of fouling is limited to the calcium carbonate precipitation in unbalanced waters (scaling waters), or dissolved metal precipitation metals such as iron and manganese that happens due to oxidation and hydrolysis during the filtration process. This fouling is reversible by using backwashing, but an increase of the cake thickness leads to an increase in longitudinal head loss of the module which can increase the energy consumption for concentrate recycling [1,3].
1.3.5. Fouling Mechanisms
Membrane structure has a significant impact on fouling mechanisms. For instance, if the membrane is porous, and its pores are larger than the size of the solute macromolecules, these oil/particles droplets could enter the pores resulting in irreversible fouling. However, if the pores are smaller than the size of particulates droplets present in the feed, particles could accumulate over the membrane surface causing pore blockage or formation of a cake layer [18]. Four different type of fouling involved in UF processes are shown in Figure 1.2.
Figure 1.2.Different fouling mechanisms happening in porous membrane: (a) Complete Pore Blocking; (b) Internal pore blinding; (c) Partial Pore bridging; and (d) cake filtration[18]
9 For dense membranes, the only fouling mechanism is deposition of material on the dense surface of the membrane, which is similar to cake filtration. In the case of oily emulsions, as the concentration of oil increases, free oil tends to blanket the membrane surface, and since free oil density is less than water, it tends to float in the top section of the membrane and blanket part of the surface.
1.4. Factors Affecting Membrane Fouling
Membrane fouling significantly depends on the operating conditions such as, feed concentration, permeability, cross-flow velocity, back wash interval (permeation period), backwash pressure, and back wash mode [19,20]. One of the most effective ways to control fouling is by adjusting the cross flow velocity. Increasing the cross flow velocity decreases the polarization degree, by increasing the mass transfer or improving other back-transport mechanisms [9]. Generally in a cross flow filtration, the deposition of particles on a surface, is controlled by two competitive processes: fouling caused by deposition of particles, transported by permeate flux from the bulk solution on the membrane surface, and removing and back transport of the particles from the membrane surface to bulk solution by the cross flow [1,12,17]. At steady state, these two competitive processes reach a balance in particle transport [3]. According to this theory, the permeation flux is higher at higher cross flow velocities and lower feed colloid concentrations [3, 8, 20].
In addition, the feed concentration, also affects the extent of membrane fouling.
With this effect, there are changes in transmembrane pressure (TMP), and permeate quality, and with higher feed concentration of solute, transmembrane pressure increases
10 rapidly in a short time. Also, due to the role of concentration polarization, the concentration of the membrane surface solution increases, and the fouling layer is slowly formed. Therefore, in practical operation, the feed should be pretreated to decrease the concentration of feed organics and retard membrane fouling resulting in improved membrane efficiency [3].
Both the backwash interval, and backwash pressure, has effects on transmembrane pressure, but backwash frequency, is not an important factor to influence the transmembrane pressure. Generally as the filtration continues, the transmembrane pressure effect on the permeation flux is such that, its increase improves the permeation flux [10].
1.5. Conventional Membrane Cleaning Methods
Membrane chemical cleaning process, is an integral part of the operation for microfiltration and ultrafiltration systems in water treatment industry, and has important impact on membrane performance. Certain fouling materials are reversible by using hydraulic means, such as backwashing, and more can be removed by enhanced backwashing (EBW), cleaning in place (CIP), or off-line chemical cleaning or soaking
[10,21].
1.5.1. Backwashing
Backwashing, is carried out by placing membrane permeates under a pressure, higher than the feed pressure. Usually a backwash pump, is employed to reverse the permeate flow from the permeate side to the feed side of the membrane, at an effective backwash pressure ranging from 5 to 50 lb /in 2 , depending on the membrane employed.
11 The permeate fluid, is used to clean the membrane surface (i.e., membrane cake), and deconcentrate the system piping. The backwash water is then wasted [1, 10]. A periodic backwash cleans the membrane surface in cross-flow and dead-end UF/MF systems by disturbing the onset of the mass transfer boundary layer near the surface. However, due to backwashing, the overall membrane recovery is decreased as permeate is used [7, 22].
Therefore, backwashing is usually combined with a cross-flushing or forward flushing. A backwash lifts the accumulated material from the membrane surface, while a cross flush transports material out of membrane module. The efficiency of backwashing is dependant on the frequency, process duration, flux, applied pressure, and type of fouling [10,23].
1.5.2. Enhanced Backwashing
In this method, a low dose of oxidant disinfectant, is automatically added to the permeate flow during the backwash, to improve its cleaning efficiency. While cleaning duration for a normal backwash is short (about 15 -30 s), with frequency of (15-45 min), an enhance backwash usually takes long (i.e. approx 10-15 min), and the frequency is usually every 4-6 h [10, 24, 25].
Enhanced backwash includes three steps: first backwashes with permeate to remove the precipitated particles from the membrane surface, second, a short soak with a low dose of disinfectant to remove adsorbed particles from the membrane, and, finally another backwash without oxidant/disinfectant to remove the cleaning chemicals from the system [10].
12 1.5.3. Chemical Cleaning
When normal backwash and enhanced backwash, are not sufficient to remove the fouling layer from the membrane surface, chemical cleaning is used. In chemical cleaning, dose of used chemicals are higher than what is used during the enhanced backwash, and its frequency is usually lower. Chemical cleaning cannot be done automatically, and it involves labor [1, 10].
The chemicals that are used depend on the fouling type and resistance of the membrane to different chemical agents [3]. Once the cause of membrane fouling is recognized, different chemical agents can be applied to remove the fouling materials and recover the membrane flux. The common chemicals used in cleaning of MF/UF systems in the water industry are categorized into five groups of Caustic agents,
Oxidants/disinfectants, Acids, Chelating agents, and surfactants [10, 14, 26]. Since membrane cleaning efficiency is directly conducted to the chemical reaction between cleaning chemicals and fouling materials, so all factors, that have an impact on the mass transfer and chemical reaction, such as temperature, concentration, duration of cleaning process, and hydrodynamic conditions, also affect the efficiency of the applied cleaning method [27].
1.6. Disadvantages of the Conventional Cleaning Methods
Conventional cleaning methods, mentioned in the previous section, are ineffective for reducing the different types of fouling [27]. Membranes exposed to back flushing/backwashing cycles, typically experiences degradation in membrane flux, and require a break in operation to be applied, which increases the labor and complexity of
13 the membrane filtration process [7, 15]. Also, that these methods do not completely remove adherent films or material trapped within the porous substructure of the membrane. For chemical membrane cleaning, using strong chemicals such as acids, detergents, etc. for flux recovery sometimes damages the membrane materials and results in a secondary pollution. Thus, chemical cleaning should be minimized or avoided [22,
28, 29]. In addition, using chemicals for cleaning also increases the operating cost, due to high price of some chemicals, and also the cost of chemical waste disposal [1, 10].
1.7. Prevention and Reduction of Membrane Fouling
Membrane fouling cannot be completely avoided, but its effect can be limited by various methods. The prevention of fouling can improve filtration efficiency and make membrane cleaning simpler. This also reduces the need for a severe cleaning regime, and has the potential to extend the life of polymeric membranes [3].
1.8. Pretreatment
Membrane filtration requires some measures of upstream feed pretreatment.
However, it is important to realize that the applied pretreatment method depends on the quality of the feed and also on membrane application. Pretreatment is the first step to control the fouling and it can be really effective [1, 10]. The simplest form of pretreatment involves microstraining with no chemical addition. However, when surface water is treated, various pretreatment procedures are required such as; pH adjustment, chlorination, addition of coagulants (e.g. alum, polyelectrolytes), sedimentation, clarification, decholorination (e.g., addition of sodium bisulphate), adsorption onto
14 activated carbon, addition of complexing agents (e.g., EDTA, SHMP), pH adjustment, and finally polishing [1].
Some of the factors that are important and must be considered are:
Membrane material structure
Module arrangement
Feed-water quality
Recovery ratio
Final water quality
1.8.1. Physical Disruption of Concentration Polarization
Enhancement of the shear stress near the membrane surface is another effective way to reduce the fouling. This method increases the mass transfer of accumulated materials back into the bulk feed [30,31], and limits the concentration polarization and cake formation. Some of the methods that can increase the local shear rate near the membrane surface are:
• Rotating membranes
• Vibrating membrane modules
In both these methods, the whole module is vibrating and this vibration generates a shear rate at the membrane surface, which decreases the formation of fouling layer or cake layer and reduces the concentration polarization. It’s been proved that rotating disks, and specially dose which are equipped with the vanes, show higher permeate flux, than vibratory shear enhanced process (VSEP) modules, due to the higher maximum shear rate which they can achieve [32,33]. The disadvantage of these methods is that since the
15 whole module is vibrating, the energy consumption is high, and because the range of the applied vibration frequency is low, these methods are not effective enough in removing the small particles that are trapped inside the pores.
Gas sparging or injection of gas into feed stream, has been found effective, in improving the cross flow UF performance, by disrupting the concentration polarization layer. Additional force such as magnetic, electric, ultrasonic and centrifugal, can also be used to enhance the permeate flux [7, 30]. The disadvantages of the continuously use of applied electrical fields are the corrosion of electrodes and high power consumption, but the use of pulsed electrical fields has shown great results.
1.9. Use of Surfactants
Surfactants reduce the surface tension between the oil-water phases, enabling oil to break up into smaller droplets, which are then readily removed from the surface of the membrane. Surfactants also reduce the ability of the oil to adsorb on the membrane surface, which results in blanketing the membrane pores that would have otherwise permeated water. In this study, use of DuPont’s fluorosurfactants, FS63, when added to the influent feed at a 0.01% concentration reduces the surface tension of water to 32 dynes/cm from its normal value of 72 dynes/cm.
16 1.10 Thesis Outline
The main goal of this thesis was to evaluate and determine the fouling mechanisms involved in ultrafiltration processes of oily emulsions using porous and dense membranes. This thesis is organized into six chapters, listed below:
Chapter Title Content
Discuss the problem of fouling and the 1 Introduction different mechanisms involved. Present the findings from a detailed 2 Literature Review review of the literature on membrane fouling Present the objectives of the research 3 Research Objectives work that was conducted Experimental studies conducted in this 4 Materials and Methods research work and methods used for these experiments Analysis of the experimental data and development of an integrated model to 5 Results and Discussion understand the major fouling mechanisms in porous and dense membranes Conclusions and Conclusions of this research work and 6 Recommendations recommendations for future studies Table.1.1.Thesis Outline
17 Chapter 2: Literature Review
A significant number of studies have been conducted on membrane technology that is focused on ultrafiltration membrane performance, and applications. Since every single membrane is susceptible to fouling, several researchers have spent considerable time and effort to verify the fouling mechanisms and develop methods to prevent or at least minimize them.
In the literature, many researchers have investigated the effect of the membrane operating conditions on the performance yield. A study by Hyeok Choi et al [34], showed that during microfiltration and ultrafiltration of biological suspension, the permeate flux had a linear proportionality to cross flow velocity, and that increasing the cross flow velocity had a greater effect on decreasing the MF membrane fouling than that of UF membrane. They also found the optimum cross flow velocity for removing the reversible fouling layer formed on the MF and UF membrane surfaces. In another study by
Seungkwan Hong et al. [35], it’s been revealed that at the transient stages of filtration, the cross flow velocity had no influence on permeate flux, and also that increasing feed particle concentration and transmembrane pressure dramatically decreases the flux. In addition, in a study by Lei Wang et al.[36], the influence of operational conditions on membrane fouling in UF of a synthetic water was evaluated .The results showed that row water quality, membrane types, the set of permeability, membrane surface cross-flow velocity and backwash conditions were significant for keeping the transmembrane pressure stable and maintaining long-term stable permeability. Johannes de Bruijn et al.[37] , studied the performance of zirconium oxide membranes for filtration of juice,
18 and it was indicated that the high feed velocity and low TMP across the membrane resulted high permeate flux.
2.1. Ultrafiltration of Oily Emulsions
Every day, large amount of wastewaters are produced in different industries, which can’t be discharged into the environment or drained into the sewage systems unless they pass through different treatment processes. These effluents typically content a complex combination of different emulsifiers, fatty acids, oil, corrosion inhibitors, bactericides, and other chemicals [38]. An important part of these wastewaters is oil- water emulsions, which is a major pollutant of environment. In wastewater treatment plants many traditional filtration techniques such as dissolved air flotation, centrifuge, coalescence, adsorption, gravity settling (API oil-water separator), skimming, etcetera, are used for separation of oily wastewaters. In addition to these techniques, for treatment of unstable emulsions, which contain oil, droplets with diameters higher than 100 m, chemical filtration techniques such as coagulation and flocculation are applied. However there are some limitations in using the above techniques in treatment of emulsions which contain oil droplets with diameter lower than 20 m, and also, these techniques no longer satisfy the international standards be cause of the high concentrations of oil in their treated waters, and also their low efficiency and high operating costs [39,40].
The membrane technology developed in the last 30 years, have shown to be a great effective alternative treatments for separation of Oily waters. Many researchers have studied the performance of UF, and MF membranes for oily water treatment, and they have shown that both UF and MF are highly effective in oil rejection, and against
19 other traditional techniques they don’t need chemical additives, and they are more economical [40]. A comparative study between the performances of UF membrane and some a biological treatment method, showed that UF was more effective in removing the oil content, TSS and turbidity, and the oil concentration in its permeate was low enough that it could be discharged into the environment [39]. A. Salahi et al. [39], studied the effects of different operating conditions such as transmembrane pressure, cross flow velocity, temperature and pH on the performance, fouling resistance, permeate flux, and rejection of a polysulfone (PS), a polyacrylonitrile (PAN), and an API oil-water separator in treatment of the oily wastewater produced in a refinery. The results showed that the
PAN membrane showed a higher permeate flux, rejection, and less membrane resistance that PS membrane. This study also showed that the best cleaning method to recover PAN membrane efficiency was a combination of SDS (as a surfactant) and EDTA (as a chelating agent). Hong-Jian Li et al. [41], developed a hydrophilic hollow fiber UF membrane, made of a new dope containing cellulose/monohydrate N-methylmorpholine-
N-oxide, for oil-water treatment. The membrane performance and its tolerance to acid/alkali (pH1-14) were studied and fouling resistances were measured by osmotic pressure-adsorption model. The results showed that the cellulose hollow fiber UF membrane developed in this experiment not only was resistant to fouling, but also it tolerated a wide range of pH, and showed to be a feasible and desirable choice for treatment of oily wastewaters. B. Chakrabarty et al. [42], tested the performances of different polysulfone UF membrane for treatment of oily wastewaters, and they indicated that the TMP and feed properties are important factors in permeate flux, and also oil separation. H. Ohya and J.J. Kim [43] studied the influence of the pore size on separation
20 mechanisms of MF of oil-water emulsion by applying a glass tubular membrane. They specified the transition of separation mechanism from blocking to cake formation, and also suggested that the pore sizes of oil concentrations and TMP to be adjusted before the start of the filtration process. Hesampour et al. [44] tested the influences of different operating conditions on UF of oil-water emulsion using Taguchi method. The results showed that among different parameters, Temperature had the most impact on permeate flux. S. Elmaleh et al. [45] used an M9 Carbosep UF membrane for filtration of a mixed suspension containing hydrocarbons and biological solids from an active sludge plant.
The results showed that the inorganic membrane has a high retention efficiency for hydrocarbons and suspended solids, and also that the temperature had no effect on the particle size distribution of the suspension.
K.Karakulski and A.Kozlowski [46], tested the performance efficiency of three tubular UF membrane made of three different materials. They showed that all three membrane with their own certain pore size was able to retain the oil particles to a very good extent that in fact their filtrate had less than 10 mg/l oil content which is an acceptable amount to discharge into the natural environment. J. M. Benito et al. [47], used two UF membranes with different pore sizes, to study the effects of different operating conditions on permeate flux. Their experiments showed that there’s an optimum operating condition, such as TMP, Temperature, cross-flow velocity for each individual membrane, which gave the highest, permeate flux. M. Gryta et al. [48] investigated the oil rejection efficiency of a combination of a UF and membrane distillation. The measured oil concentration was reduced to less than 5 ppm after UF, and passing through the second stage of membrane distillation completely removed the oil
21 contents, as well as remained soluble particles in the feed. N.Ghaffour et al. [38] compared the oil rejection efficiency of four different UF membranes, two organic, made up of Alumine(A) and Alumine zircon(AZ), and two inorganic made up of
Carbonzircon(CZ) and Carbon carbon(CC), in treatment of refinery wastewater containing emulsions of Arabic Aramco crude oil diluted in tap water. The results showed that the tubular UF membrane made up of CZ was a complete barrier for the oil at any feed concentration, and had the highest permeate flux. In this research, the effects of operating conditions were also evaluated, and the results showed that high feed concentration had a severe impact on permeate flux reduction, and transmembrane pressure and shear stress had relatively little effect on permeate flux loss. The severity of the formation of fouling layer on the membrane surface depends mainly on the pressure and the feed concentration, and is independent of the temperature and cross flow velocity due to shearing forces. The results showed that the main parameter in controlling the permeate flux is temperature, since it has a direct influence on the droplet size distribution which specifies the permeation flux.
2.2. Conventional Membrane Cleaning Methods
As mentioned above, fouling is the limiting phenomenon for the wide application of membrane, and it’s the result of concentration polarization and particle accumulation on membrane surface. Many approaches have been studied so far, to minimize membrane fouling, such as pretreatment of water, backwashing the membrane with or without applying chemical agents, hydrodynamic cleaning with high cross-flow velocity and many other mechanical or chemical methods that have been applied individually or in a combinations to enhance membranes performances.
22 2.2.1. Backwashing
W.J.C. Van, et al [24] invented the concept of partial backwashing as an effective method for enhancing the performance of the membrane systems where fouling occurs as the result of either local accumulation of particles at the end of the fiber, or partially formation of concentration polarization layer. They indicated that this cleaning method wouldn’t be effective if whole module is affected by fouling. Heng Liang et al. [50] evaluated the effects of hydraulic and chemical cleaning on membrane performance in filtration process of algae-rich reservoir water. They showed that backwashing followed by multiple forward flushing was more effective in improving membrane performance, than backwashing alone, and that using a combination of NaOH and NaOCl increased the cleaning efficiency. Keith N. Bourgeous et al. [51] indicated that using rapid backwashing alleviated rapid fouling of UF membrane. Paul et al. [52] found that a periodic backwashing could be effective in removing most of the membrane reversible fouling, and he also developed a new control system to optimize the backwashing which led to 40 % reduction in the required backwash permeate. J. Paul Chen et al.[53]developed a statistical factorial design to determine the key elements as well as their interactions in both physical and chemical cleaning of US and reverse osmosis membranes in municipal wastewater treatment. The results indicated that physical cleaning of both modules was highly affected by the filtration duration between cleanings, as well as applied pressure during forward washing, and also cleanings durations. They also realized that chemical cleaning was influenced by temperature, concentration of cleaning solution and backwash after chemical cleaning.
23 Mansoor Kazemimoghadam et al. [54] used chemical cleaning for removing the precipitated milk components on the polysulfone UF membrane surface. They reached to an optimum membrane recovery by cleaning the membrane with a combination of
NaOH, EDTA, and sodium dodecyl sulfate. H. Peng et al. [55] used MF- UF hybrid system for filtration of an oily wastewater. They found that fouling of MF membranes is the result of aggregation of oil droplets inside the pores. They also found that cleaning with methods such as hot water heating, stream cleaning and air backflushing is effective in enhancing membrane performance.
2.2.2. Gas Sparging
Gas sparging is defined as injection of air bubbles into feed stream, and has been found to be an effective, simple and economical technique in improve ultrafiltration in both downward and upward cross-flow filtrations [56]. In an experiment by Cui and
Wright [57, 58], gas sparging effect on permeate flux and membrane rejection of a tubular ultrafiltration membrane was investigated. In this study, the tested feeds were of dextrans and BSA solution, and it was found that gas injecting into feed stream, even at a very low flow rate could significantly increase the permeate flux and membrane rejection ratio. Their results also indicated that gas sparging would have a more significant effect on permeate flux enhancement, when concentration polarization is more severe, and it’d most likely happen at higher transmembrane pressures, lower feed cross flow velocities, and higher feed concentrations. In another study by S. R. Bellara, et al. [59], effect of gas sparging on enhancing the permeate flux and membrane sieving coefficient was evaluated at different TMPs, feed concentrations and gas to liquid flow ratios. In their experimental work with hollow fiber membranes using dextran and albumin as test
24 media, flux enhancement of up to 63% was observed in protein filtration. It was found that in protein ultrafiltration, flux enhancement was insensitive feed concentration, and it’s not highly dependent to TMP either, but it was highly sensitive to gas flow rate, however for dextran ultrafiltration it was seen that gas flow rate didn’t have much impact on the flux enhancement.
R. Ghosh and Z. F. Cui [60] discussed the mechanism of flux enhancement in the special case of upward slug flow in tubular ultrafiltration membrane in terms of enhancement of mass transfer coefficient. In their study, they modeled the gas sparged ultrafiltration based on dividing the membrane surface area into different zones in respect to the hydrodynamic regime in those areas. By evaluating the fluid flow in these regions they found the mass transfer coefficients for each zone and consequently the averaged permeate flux was predicted. The simulation results showed that gas sparging would be more effective at higher TMP and higher feed concentration, while it looses its effectiveness at higher feed flow rate.
It’s known that the mass transfer is related to the wall shear stress, as high shear stresses inducing higher fluxes. In a study by Cui[57], it was claimed that gas sparging could increase the turbulence inside the membrane tubes or fibers, and therefore could increase the shear stress[61]. Experimental results from Cui work showed that the permeate enhancement is a result of a combination of the concentration polarization layer disruption and also increased feed fluid cross-flow velocities at the membrane surface.
Inspired by this study, C. Cabassud et. Al [61], verified the mechanism by which a tangential gas-flow two-phase flow in a hollow fiber ultrafiltration membrane, intended to filtrate drinking water, could influence the formation of a particle layer, by affecting
25 the wall shear stresses. The experimental results indicated that in filtration with hollow fiber UF membranes, air slugs effects on flux enhancement is related to their beneficial effect on enhancing the shear stresses at the wall which could effectively sweep the particle deposit from the surface. They also found that intermittent use of gas injection is far less effective than a continuous injection throughout the filtration, and it was the result of the deposit built up during the air flow interruption.
In another study by Mercier et al [62], using a tubular membrane, they evaluated the effect of gas sparging on flux enhancement, and its sensitivity to the membrane geometry. They realized that the difference between the inner diameters of tubular membranes and hollow fibers results to different hydrodynamic characteristics inside each membrane. This would cause a discrepancy in the gas sparging performance in these two different modules, which suggests that the mechanism of flux enhancement is different between these two membranes. Under the most filtration conditions, the flow pattern is laminar for the hollow fiber membranes and turbulent for tubular ones, and it’s been shown that the less turbulent the fluid is, the more effective the gas sparging would be in enhancing the permeate flux.
The gas sparging method effectiveness mainly corresponds to its effect in disrupting the concentration polarization layer. It does not seem to be an effective in membrane systems where the main mechanism of fouling is pore blocking and cake formation [58].
26 2.3. Ultrasonic Cleaning of Membranes
Ultrasound waves have a frequency range from 16 kHz to 10 MHz. It is above the human hearing range. Ultrasound waves, as with any sound waves, consist of series of compression and expansion waves made in the molecules of the medium though, which it propagates. Compression cycles exert a positive pressure on the liquid – pushing molecules together. Expansion cycles may exceed the attractive forces of the liquid molecules, and cavitation bubbles will form. Cavitation can be defined as: the formation, growth, and implosive collapse of bubbles [63, 64]. And it occurs at frequencies of roughly 20-1000KHZ. Collapse of these bubbles has a significant effect on the chemical and physical properties of the medium.each of these bubbles can behave like a hot spot, that generates temperatures of up to 4000 K, extreme heating or cooling rates of 1010 K/s and pressures exceeding 100 MPa are observed in the transient bubbles while the bulk fluid remains at ambient temperature and pressure. The implosion happens with lifetimes shorter than 0.1 μs [65]. In a heterogeneous liquid-solid system, a collapse near a surface produces a nonsymmetrical inrush of fluid to fill the void, which leads to the formation of the liquid jets. These liquid jets may lead to particle release from a fouled surface as a result of ultrasound irradiation. Along with these microjets, cavitational mechanisms such as microstreaming and mictostreamer are also important in detaching the particles from the surface, while turbulence associated with ultrasound, i.e. acoustic streaming plays a role in the transport of particles away from the surface.
In a study by T.Kobayashi [27] ultrasound effect on UF of Dextran solution with polyacrylonitrile membrane at different frequencies and power intensities was investigated. The result showed that at lower frequencies, the cavitation had more power
27 and would lead to decrease of fouling layer and increase of mass transfer through the membrane. Also they found that increasing the power and decreasing the feed concentration would increase the permeate flux. Beside these factors, direction of ultrasound propagation was also affecting the permeate flux, but its affect was negligible in compare to power intensity and frequency of the ultrasound waves. Xijun Chai et al
[28] studied the effect of ultrasound cleaning on four polymeric membranes made of PS,
PAN8, PAN15, and PVDF in cross-flow filtration of peptone solution. The applied cleaning procedure included three steps: sonication, water cleaning, and water cleaning under sonication. They found that water cleaning under sonication had a better affect on increasing the permeate flux, than either of these two methods separately. Furthermore, they evaluated the effects of different operating temperature and peptone concentration on PS membrane performance .The results indicated that increasing the temperature increases the cleaning efficiency for water filtration under sonication.
The intensity of cavitation can be affected by: feed water properties such as viscosity, amount of dissolved gas, and vapor pressure, Operating conditions like
Temperature and static pressure, ultrasound frequency and intensity. The most important factor in controlling the improving the ultrasound cleaning efficiency is its frequency.
The lower the frequency, the larger the bubbles will be and subsequently its implosion energy release would also be higher.
28 2.3.1. Effect of Sonication on Polymeric Membranes
Although ultrasound is widely used for cleaning of membranes in different areas of industrial process engineering, and it’s led to significant results in permeate flux improvement, but it’s been also observed in some studies, especially those with the polymeric membranes, that ultrasound caused some damage on membrane. Of course the influence of ultrasound irradiation on the membrane is dependant on the factors such as, membrane material, power intensity, distance between membrane surface and ultrasound transducer, but it’s mainly come from the main mechanism behind ultrasound cleaning, which is collapse of the cavitation bubbles near the surface. Although this collapse leads to particle detach from the membrane surface, but it gradually damages the structure of the membrane, and eventually leads to its failure. Verification of the exact cause of damage has been an area of investigation.
Plesset [67] conjectured the irradiated shock waves from bubbles explosion could be the reason behind the damage occurrence, but he also noted that this collapse needs to happen in the vicinity of the substrate and that the bubbles have to be low-gas contained.
Bubble collapse and explosion causes extreme temperature and liquid jet velocities, which could significantly damage the surface structure. It was indicated in Plesset study that the liquid jets that are spread out of collapsed bubbles could also cause surface damage. In a study by Markov and Rosenberg [68], high speed cinematography was used to detect the causes of surface damage in ultrasonic cleaning. Their results indicated that the shock waves formed by cavity explosion resulted in disintegration of the surface.
They also noticed that the gradual peeling of pieces of the membrane occurred due to vapor bubbles penetration between the film and solid surface placed under that.
29 Reviewing literature on membrane cleaning by means of a traditional laboratory ultrasonic bath, we noticed that the cleaned membrane, presented an unexpected behavior after sonication. For instance, a study by Isabelle Masselin, et al [69], and effect of ultrasound on the polymeric membranes was investigated. This study was specifically important for us, since they have shown the effect of ultrasonic cleaning on the PES ultrafiltration membrane, which is identical to what we’re using in our system. Their results showed that over the three polymeric membrane tested in their study; polyether sulfone(PES), polyvinylidene fluoride(PVDF) and polyacrylonitrile (PAN), PES membrane got significantly damaged over its surface , while the others didn’t much. A comparison between the microscopic images of a non-irradiated and irradiated PES100 membrane indicated membrane surface degradation under ultrasonic stress which would consequently led to an increase in pores size for large pores, an overall increase in pore density and membrane porosity and to the formation of large cracks formed between the adjacent pores and at the edges of the membrane.
This result is consistent with what Xiao-li Wang et al. [70] found from their experiment. They used ultrasound on four different polymeric membranes including PES, to evaluate the effect of ultrasound on membrane polymeric structure. They noticed that membrane filtration behavior dramatically changed after irradiation in respect of its permeate flux and the rejection rate. In addition, it was found out that the membranes are affected by sonication over their entire surface, and also that the ultrasonic irradiation leads to formation of large cracks and in some cases large holes on membrane surface.
From all these findings it’s obvious that in spite of their great benefits in enhancing membrane performance and filtration process, ultrasonic waves have to be
30 used with care. The nature of the polymeric membranes and specifically PES membranes makes them so sensitive to the ultrasonic waves at the chosen frequency, as well as irradiation duration and intensity. Since maintaining the membrane integrity is the first priority in a cleaning process, and the main goal of this project, is applying a cleaning procedure with the minimum damaging effect on membrane surface structure, we’ve come to this conclusion that under the operating conditions that we would conduct our tests, and in spite the fact that our tested membranes are polymeric, the backwashing with permeate fluid and also warm water would be a suitable choice for enhancing the membrane performance.
Except the sensitive nature of the PES membrane to the ultrasonic irradiation, our tested membrane is a porous monolith and like any other porous membrane is likely to suffer from the pore blockage during the operation. It’s well known that the ultrasonic cleaning benefits in enhancing the permeate flux is basically the result of the waves effect on disruption of the cake layer, and it’s not highly beneficial to removing the adsorbed foulants on the pores walls or on pores openings. As the result, we didn’t find the ultrasonic cleaning, an effective method for enhancing our tested membrane efficiency.
2.4. Fouling Mechanism in Ultrafiltration
In recent decades many studies have been done on empirical models for the explanation of permeate flux decline with time during an ultrafiltration process. One of the most common models to determine the fouling mechanisms involved in membrane filtration is the Hermia’s modeling. This method of modeling was originally developed for the dead-end ultrafiltration membranes, but further it was modified by R W. Field et
31 al [71] for cross-flow ultrafiltration membranes. Abdolhamid Salahi et al. [18] used the
Hermia’s models to evaluate the fouling mechanisms occurred in UF of a refinery oily wastewater. They found that cake layer formation on the membrane surface followed by partial pore bridging is the best fit to the experimental data and under different tested experimental conditions. They also described the fouling mechanism in more details by dividing the filtration curve into different regions to show the partial contribution of each mechanism. S. T. D. de Barros et al. [72] evaluated the flux behavior of ceramic and polysulfone membranes during cross-flow UF of depectinized pineapple juice. They modeled the observed flux decays in their experiments by means of the modified forms of
Hermia’s mechanisms, and by estimation of the models parameters according to a nonlinear regression optimization method, they found the dominated mechanisms for hollow fiber PS membrane and ceramic membrane to be cake formation and pore blocking respectively.
In another study by Abdolhamid Salahi et al [73], Hermias models were applied for evaluation of fouling mechanisms in cross-flow UF polymeric membranes. The results from UF and MF membranes showed that the best fit to the experimental curve was cake filtration model and the worst one was complete pore blocking. T. Mohammadi et al [74] investigated the fouling mechanisms in reverse osmosis during filtration of an oily emulsion. Using Hermia’s models they found the partial pore bridging to be the best fit to their experimental data. They also believed that complete pore blocking model is oversimplified assuming that molecules block the pores without having superimposition effect on one another. A.L. Lim and Renbi Bai [75] studied the type of fouling mechanisms in MF of activated sludge wastewater. They found that the main types of
32 fouling correspond to initial pore blocking followed by cake formation. They indicated that pore-blocking mechanism is responsible mechanism for initial flux decline and cake formation is responsible for the slower flux decline over filtration duration. M. Cinta
Vincent Vela et al [76], used Hermia’s models to investigate the fouling mechanisms involved in the cross flow UF of polyethylene glycol. They found that the best fit to experimental data corresponds to the cake formation mechanism followed by the partial pore bridging under all tested experimental conditions. They also conducted a more detail investigation on determining the fouling mechanisms by dividing the experimental curve into different regions that attribute to different fouling mechanisms.
33 Chapter 3: Thesis Objectives
The original objectives of the proposed research work were as follows:
1. Conduct a complete survey of current methods to reduce ultrafiltration fouling and
understand their advantages and disadvantages;
2. Study the papers in detail that employ ultrasonic methods to reduce ultrafiltration
fouling;
3. Study the modes of vibration of a rectangular membrane, which can be employed to
reduce ultrafiltration fouling;
4. Conduct experiments in which an ultrasonic flat element will be used to vibrate a flat
membrane at different resonant frequencies;
5. Conduct experiments with various ultrafiltration membrane/solute combinations,
which includes the following:
(a) Regenerated cellulose, 10,000 MWCO and Polyethylene Glycol with average
molecular weight of 10,000
(b) Polysulfone, 100,000 MWCO and Dextran with average 162,000 molecular
weight
(c) Actual laundry water obtained from CINTAS and GE/Osmonics membrane
6. Measure the water flux with and without ultrasonic vibration of the flat membranes,
employing different resonant frequencies of the flat membrane sheet
7. Develop a simple mathematical analysis of the impact of the membrane vibrating
modes on ultrafiltration fouling
8. Write papers and thesis
34 However, when a detailed literature search was conducted, it was found that ultrasound causes significant physical damage to the structure and surface of polymeric membranes. Although the factors such as, membrane material, power intensity, distance between membrane surface and ultrasound transducer, are influential on the efficiency of ultrasound irradiation, but the vital parameter is the quality of the bubble collapses near the membrane surface. Despite the fact that this collapse leads to particle detachment from the membrane surface, but it also gradually damages the structure of the membrane, and eventually leads to its failure.
Based on what’s been reported in literature about the effects of using ultrasound for cleaning of polymeric membranes, it’s obvious that in spite of their great benefits in enhancing membrane performance and filtration process, ultrasonic waves have to be used with care. The nature of the polymeric membranes and specifically PES membranes makes them so sensitive to the ultrasonic waves at the chosen frequency, as well as irradiation duration and intensity. Since maintaining the membrane integrity is the first priority in a cleaning process, and the main goal of this project, is applying a cleaning procedure with the minimum damaging effect on membrane surface structure, we’ve come to this conclusion that under the operating conditions that we would conduct our tests, and in spite the fact that our tested membranes are polymeric, the backwashing with permeate fluid and also warm water would be a suitable choice for enhancing the membrane performance.
Except the sensitive nature of the PES membrane to the ultrasonic irradiation, our tested membrane is a porous monolith and like any other porous membrane is likely to suffer from the pore blockage during the operation. It’s well known that the ultrasonic
35 cleaning benefits in enhancing the permeate flux is basically the result of the waves effect on disruption of the cake layer, and it’s not highly beneficial to removing the adsorbed foulants on the pores walls or on pores openings. As the result, we didn’t find the ultrasonic cleaning, an effective method for enhancing our tested membrane efficiency.
Based on these detailed literature searches, the specific objectives of the proposed research work were changed, to the following:
1. Assemble an experimental apparatus that can be used to test two types of
membranes: (1) Porous membrane; and (2) Dense Membrane;
2. Conduct a detailed literature search to select two commercially available
membranes that can be used for separation of water from an oily-water emulsion,
one which is porous and the second has a dense structure;
3. Collaborate with a local company to conduct experimental testing of the two
membrane types, using a real oily emulsion;
4. Experimentally measure the water permeation rates as a function of inlet feed
pressure, to determine the “optimum” feed pressure that maximizes the water
permeation rates;
5. Operate the membrane unit at the “optimum” feed pressure to determine the
permeate flow rate as a function of time; Take influent and reject flow water
samples to determine the oil concentration and characterize the type of oil in the
oily emulsion; Obtain experimental data with and without use of fluorosurfactant
for the porous membrane and sparged air for the dense membrane;
6. Determine the backwashing procedure and protocol for the porous membrane;
36 7. Develop a model that can quantitate the role of each fouling mechanism (refer to
Figure 1.2) for each type of membrane;
8. Write papers and thesis, presenting the findings of the proposed research work.
37 Chapter 4: Materials and Methods
4.1. Selection of Membranes
As mentioned earlier, two types of membranes were selected for testing in the proposed research work:
(1) Porous Membranes: Porous membranes are manufactured in various geometries, including flat sheets, pleated flat sheets, hollow fibers, tubular, monoliths, and spiral wound. Spiral wound membranes suffer from fouling issues mainly due to lack of sufficient turbulence near the membrane surface, although in recent years, use of wider spacers and higher operating liquid velocities have attempted to overcome this problem.
In most membranes, tangential flow is used as opposed to dead-end filtration, to minimize deposition of particle and rejected solute on the membrane surface, as shown below in Figure 4.1.
Figure 4.1. Tangential cross-flow filtration [77]
In this research work, a monolith structure, porous, membrane, made of hydrophilic polyether sulfone material, was selected, since it exhibited the following advantages: (1) high surface area; (2) high permeation capability for water at low trans-
38 membrane pressure difference; (3) capability to withstand fouling; (4) permanently hydrophilic membrane surface; (5) nominal pore size of 0.02 mm, which does not allow any oil to enter the pores, unless the feed pressure exceeds a high limit; (6) less probability of fiber breakage due to pressure fluctuations, especially during start-up and shutdown of the membrane module; (7) able to be backwashed; and (8) low cost, compared to ceramic membranes.
The cartridge used in our experiments was a cross-flow UF cartridge, a cross flow
UF cartridge, dizzer® XL 0.9 MB 38W manufactured by inge water technologies AG.
The module and membrane specifications are given in Table 4.1 and Table 4.2:
Polyether Sulfone Membrane Module Data
Membrane area 38 m2
Length with end cup (L) 1180 ± 3 mm
Length without end cup (L1) 986 ± 1.5 mm
Outer diameter module (D) 250 mm
Housing material PVC-U, grey
End cap material PVC-U, grey
Pressure max. 70 psi
Temperature range 32-104 °F
Table 4.1. Design characteristics of the membrane cartridge
39 Polyether Sulfone Membrane Data
Membrane Material Polyether sulfone
Capillaries per fiber 7
Membrane Inside Diameter 0.9 mm
Membrane Outside Diameter 4.0 mm
Pore Size Approx. 0.02 μm
Molecular weight cut-off 100 KD
Table 4.2. Characteristics of the porous membrane
Figure 4.2. Structure of the porous membrane fiber [78]
(2) Dense Membranes: The main advantages of dense membranes is their ability to produce a higher quality permeate, since oil cannot penetrate the dense layer and elimination of pore fouling mechanisms, such as pore blocking, internal pore blinding, and pore bridging. Deposition on the membrane surface is still possible, although with high liquid operating velocities, this can be minimized considerably. The main
40 disadvantage of dense membranes is their lower water permeation rates, compared to porous membranes, which increases the needed surface area.
The selected dense membrane is hollow-fiber, manufactured from cupraammonium regenerated cellulose. The module and membrane specifications are shown in Tables 4.3 and 4.4.
Regenerated Cellulose Membrane and Module Data
Total Membrane Active Area 12.9 m2
Module Housing Material 4-inch diameter CPVC
Temperature Range 32 -180 °F
Pressure max. 55 psig
Membrane Material cupraammonium regenerated cellulose
Membrane Internal Diameter: 420 μm
Membrane Outer Diameter: 500 μm
Table 4.3. Hollow-fiber module and membrane characteristics
a) b)
Figure 4.3. Image of the cuprammonium regenerated cellulose hollow fibers [79]
41 4.2. Experimental System
Figure 4.4 shows experimental setup used in all the experiments. The oily wastewater treatment was operated in cross flow batch concentration mode. During filtration process, the feed was pumped from the tank (Tk-101) by means of a horizontal centrifugal pump P-101, to the bottom of the membrane module where it was fed to the membrane channels. The experiments were ran in a complete recycle mode of filtration, where retentate and permeate flows were continuously recycled back into the feed tank by using V-09 and V-10. Total recycling of the retentate and permeate helped to keep the feed concentration almost constant. There was a by-pass prior to the feed inlet V-02 and a by-pass before the membranes inlets V-04, to recycle the extra feed to the tank. The main flow rate and the desired transmembrane pressure were adjusted using the valve V-04 in the bypass flow and valve V-10 in the retentate flow.
The pressures were measured prior to the venturi, and at the inlet and outlet ports of the membrane, with pressure gauges (PGM, Omega) ranges from 0-100Psi (0-690
KPa). To examine permeate during the runs, it was collected in the permeate tank TK-
102, with using V-08.
The cleaning method that was applied for porous polyether sulfone UF membrane, was backwashing using the permeate flow. The backwashing loop included a backwashing tank (TK-102), a horizontal centrifugal pump (P-102) and four valves (V-
05, V-06, V-08, V-09). Permeate collected in tank TK-102, was pressurized by the pump
P-102, and derived to the membrane channels from the permeate side. The volume of the permeate sample collected in tank TK-102, was measured and the permeate flow rate was calculated by dividing the volume of permeate sample by the sampling time. The tank
42 volume was 130 Gallon or 492 Litter. Experiments were conducted at TMPs lower than
70 Psig for polyether sulfone membrane, to prevent membrane compaction, and below 55
Psig for cuprammonium regenerated cellulose membrane. The experiments were done at room temperature of 25 °C, and feed temperature of 40°C.
Figure 4.4. Schematic figure of the experimental set-up
43
Figure 4.5. Photographs of the experimental system, operated at Ford Motor Company Plant, Sharonville, Ohio
44 4.3. Cleaning Procedure
In this project, backwashing with clean warm water of 50 ( C), and also with the permeate itself, were used as a clean-in-place (CIP) method to increase the efficiency of porous monolithic polyether sulfone membrane in filtration of oily wastewater, and for the hollow fiber dense membrane, air injection was applied as a continuous cleaning method throughout the filtration duration.
Backwashing process is defined as a periodic mode in which the membrane fibers are cleaned by pushing pressurized water in the reverse direction of membrane filtration,
Figure 4.6 [1]. The backwashing loop is shown in Figure 4.6 in bold line. As it’s been indicated backwashing was operated in the opposite direction of ultrafiltration by sending the permeate flow through the membrane fibers from permeate side. In order to start backwashing the membrane, the system was shut down completely. The backwash pump
P-102 pushed permeate (or water) flow from the tank Tk-102 to the membrane inlet ports where permeate forced through the fibers. Pushing the permeate from the opposite side through the fibers, made precipitated solids to be lifted off and flushed out of the membrane pores and then out of lumens to the retentate line, where they recycled back to the feed tank.
45 V-12 P I-04
Retentate flow V-10 By-pass flow
Air in Cellulose PESM UF
Feed tank UF Membrane
F Module I I-05 Membrane Tk-101 V-02 Module V-01 V-06 I-01 I-02 I-03 V-05 P P Feed V-04 flow
P-103 V-10 V-03 P-101 V-09 V-07
Tk-102 Permeate flow V-08 P-102 Figure 4.6. Schematic figure of backwash system
For the dense membrane however, the cleaning method that was introduced, was air injection into feed stream, as it’s shown in Figure 4.7 in bold line. When the valve V-
03 is completely open, the entire feed stream preferably flows through the valve instead of the venturi, and by partially closing this valve; a portion of feed flows through the venturi I-02 and as it passes through the constricted section of venturi, it sucks the air.
The volumetric flow rate of air addition can be monitored by the rotameter I-05. No effort was done to observe and control the pattern of gas-liquid two-phase flow inside the membrane fibers. Pressure gauges were mounted before and after the membrane module to measure the inlet and outlet pressure, where the mean of these two pressure values gives the transmembrane pressure (TMP). Different TMPs were applied to the membrane module by adjusting the V-10 valve at the outlet of the module. Permeate flow rate was measured by the timed collection of permeate. The experiments were conducted at room temperature around 25°C.
46
Figure 4.7. Schematic figure of air injection system
In order to evaluate the effect of adding surfactant to the feed stream, all the conducted experiments on two polyether sulfone and cuprammonium regenerated cellulose membranes, were repeated under comparable related operating conditions, with adding surfactant to the feed stream. For this purpose, before starting the filtration, the fluorosurfactant, DuPont, FS-63 Capstone surfactant, was added to the feed tank with concentration of 0.05 ( l/m3) and it was mixed well with the feed by recycling the fluid to the tank for 15 minutes.
47 4.4. Theory of Ultrafiltration of Oily Wastewater
Permeation flux is a critical parameter to assess performance of the membrane,
and it indicates the amount of permeate and the product rate [40]. The permeate flux is
calculated as follow:
Q Jp 3600 Eq.4.1 A0
where: 2 J p is the permeate flux ( l/m .h)
Q is the permeate volumetric flow rate (l/s)
2 A0is the total membrane active area ( m )
The permeate flux can also be expressed by Darcy’s law as:
TMP 9 Jp 3.6 10 Eq.4.2 T Rt
where: TMP is transmembrane pressure ( KPa)
T is feed viscosity ( Pa.s)
1 Rt is total membrane resistance( m )
As it’s been explained in [6], the feed viscosity can be expressed as a function of
temperature as follows:
3 5 6 2 8 3 water (1.784 10 ) (5.75 10 T) (1.110 T ) (10 T ) Eq.4.3
48 where T is water temperature ( C).
In our experiments, the permeate tank was open and exposed to ambient condition. Therefore the transmembrane pressure (TMP) was calculated as the average of
the inlet pressure ( Pinlet) and outlet pressure ( Poutlet) or:
(P P ) TMP inlet outlet Eq.4.4 2
The total membrane resistance can be calculated as:
Rt Rm Rf Eq.4.5
where:
1 Rm is hydraulic membrane resistance( m )
1 R f is fouling resistance( m )
Flux decline percentages were calculated as follows: J FD(%) (1 f ) 100 Eq.4.6 J0
2 2 where Ji is the initial permeate flux ( l/m .h) and J f is the final permeate flux ( l/m .h).
Figure 4.8. Demonstration of the contact angle of a liquid sample [80]
49 Figure 4.8 shows the principle of contact angle. Basically the contact angle is a
quantitative measure of the wettability of a solid surface by a liquid. It is defined
geometrically as the angle formed by a liquid, and the tangent line to the upper surface
where three phase boundaries where a liquid, gas and solid intersect. The contact angle
between any droplet and a solid surface is indicative of a very important characteristic of
that solid surface called “hydrophobicity”. Generally, Surfaces are categorized in to two
major categories of hydrophilic and hydrophobic surfaces. Usually in defining the
hydrophobicity and hydrophilicity, the contact angle range is defined based on the water
droplet on a solid surface, and so when 90 surface is hydrophobic, and when
0 90 surface is hydrophilic. However, in our experiment we are dealing with the case of oil droplet on the solid surface, or membrane surface in our study, this definition is reversed.
In our study, if 90 the membrane is hydrophilic and when 0 90 , the
membrane is hydrophobic, which means water can’t be dispersed over the membrane surface, because the membrane is wet by oil droplets, that have spread out over the
surface like a layer.
Surface chemistry, interactions between solute-solute and solute-membrane
surface and wettability are major parameters in oil-water separation by ultrafiltration, and
can be expressed in terms of some elements, like the interfacial tension between water
and oil droplets ( o/ w ), the contact angle of the oil droplet on the membrane surface
(o/ w ), the pore effective radius (r), and the pressure of the oil droplet ( Pc ) or also called
capillary pressure.
50 Capillary pressure is expressed by following equation:
2 cos P o/ w o/ w Eq.4.7 c r
When the system is ran under high operating pressure, after some time the membrane get fouled and oil droplets spread out over the membrane surface, causing a
change in surface chemistry. And as mentioned above, change of surface chemistry
means a change of critical surface tension, pore size and contact angle of the membrane.
Typically, 90 and the membranes are hydrophilic (leading to high permeate fluxes),
and the capillary pressure of is negative. This keeps the oil droplets from entering the membrane pores against the operating pressure. A major and critical point in preventing
the membrane fouling is keeping the operating pressure bellow the approximate value of
the capillary pressure of oil droplet, because if the operating pressure exceeds the
capillary pressure value, the oil droplets could be deformed and penetrate through the
membrane pores, pollute the permeate, and further adsorb in pores causing membrane
fouling [40,81].
The UF membranes used in our experiments were hydrophobic. During filtration
with porous membrane, small foulant particles mainly oil droplets existed in feed stream,
can enter the membrane pores, coalesce and potentially block the pores. This pore
clogging leads to membrane fouling and therefore permeation flux reduction. In the case
of filtration with dense membrane, although there’s no pores to be blocked, but deposits
can coalesce on the surface and make a resistant layer in way of water transfer through
the membrane wall.
51 4.5. Models for Membrane Fouling Mechanism
Membrane fouling in cross-flow ultrafiltration is the key factor that challenges the technological viability of ultrafiltration process. Typically in ultrafiltration with polymeric membranes permeate flux variation over time is an initial quick decline followed by a long gradual decrease. Therefore, modeling of flux decline to determine the fouling mechanism in ultrafiltration of macromolecules is essential from the technological and economical point of view.
4.5.1. Fouling Mechanisms Involved In UF Using Porous Monolith Polyether Sulfone Membrane
Membrane structure has a significant impact on fouling mechanisms. For instance, if the membrane is porous, and its pores are larger than the size of the solute macromolecules, these oil/particles droplets could enter the pores leading irreversible fouling. However, if the pores are smaller than the size of particles droplets existed in the feed, particles could accumulate over the membrane surface causing pore blockage or formation of a cake layer [18]. Four different type of fouling involved in UF processes are shown in Figure 4.9:
Figure 4.9. Different fouling mechanisms happening in porous membranes [18]
52 In general fouling mechanisms could be the result of a) complete pore blocking: when the pores are sealed; b) Internal pore blinding: when particles not rejected by the pore entrance is adsorbed or trapped on the pore wall or in the membrane support; c) Pore bridging: where the pore entrance is partially obstructed d) particles droplets deposition on the membrane surface forming a cake/gel layer [72]. During UF process these mechanisms may occur simultaneously. As mentioned earlier, the intensity of the membrane fouling is dependent to three major factors: operating conditions, feed characteristics and membrane type. Typically operating conditions are important factors in determining the degree of fouling, in particular transmembrane pressure plays a vital role as its increasing could increase the formed cake layer density and lead to complete pore blocking [18,72].
In the past two decades many studies have been done on empirical models for the explanation of permeate flux decline with time during an ultrafiltration process. In spite of the precision of these empirical models, they are not able to describe adequately the fouling mechanisms occurred in membrane filtration. Theoretical models on the other hand, can explain the fouling mechanism to some extent. However the complete theoretical models developed in the literature suffer from inaccuracy in prediction of flux decline with time, if they are applied without experimental data getting fit to them to estimate at least one of the model constants. As the result, semi-empirical models, those with model constants having actual physical meanings have shown to be suitable solutions to achieve an accurate prediction of the flux decline in UF and also to determine the fouling mechanisms [18].
53 So far, quantification of the impacts of operating parameters such as transmembrane pressure on membrane fouling in UF is not completely known. For this purpose, there is not a completely theoretical model that describes UF dynamics with acceptable accuracy for engineering applications [18]. In this study, the effect of transmembrane pressure on cross flow UF of oily wastewater and involved fouling mechanisms are investigated. The applied empirical models to explain permeate flux behavior and to determine the involved fouling mechanisms are the Hermia’s models
[82]. Hermia developed four empirical models that correlate with four main types of fouling: complete pore blocking, pore bridging, and internal pore blinding and cake formation. The models parameters have a physical meaning and correspond to the comprehension of the fouling mechanisms [76]. Hermia’s models were originally developed for dead-end filtration and based on the constant pressure filtration laws.
However, it’s worthy to be mentioned that in spite of the different sets of applied mass and momentum equations for dead-end and cross-flow filtrations, many researchers have applied the Hermia’s models to cross-flow filtrations. By comparing the model’s predicted values with the results from conducted experiments, they found that the models perfectly predict the models under different operating conditions [74, 76, 83, 84, 85]. In
Theory, it’s assumed that in cross-flow filtration system reaches to a steady-state, but it should be considered that this steady-state is a quasi-steady-state condition not a rigorous one, and this is due to the fact that although permeate flux is almost constant for a long time scale, it actually decreases very slowly over time until a permeate flux of zero is achieved for a very long time scale [18].
54 Hermia’s model is expressed by the following general differential equation:
d2t dt ( ) K( )n Eq.4.8 dV 2 dV
where:
V is accumulated permeate volume ( m3), t is filtration time (s), and K and n are
phenomenological coefficient and general index, respectively, both depending on type of fouling depicted in Figure 4.9. (K unit depend on the parameter n in equation 4.8).
4.5.1.1 Complete Pore Blocking Model (n=2)
When particles sizes are larger than membrane pore size, the portion of the
membrane area, which is reached out by particles, is blocked as a consequence of pore
obstruction with pore sealing. Hermia concluded that n was equal to 2 in this case. For
n=2, Eq.4.8 is expressed in terms of permeate flux versus time as [85]:
ln(Jp ) ln(J0) Kct Eq.4.9
where: 2 J p is the permeate flux ( l/m .h)
2 J0 is the initial permeate flux ( l/m .h) Kc is the equation constant
The parameter Kc can be described as a function of the membrane surface
obstructed per unit of the total permeate volume that permeates through the membrane, KA , and as a function of the initial permeate flux, J0, as it’s shown by Eq.4.10 [86]. As
the result, the active membrane area reduces due to the pores being completely clogged
[18].
55 Kc KA J0 Eq.4.10
4.5.1.2. Internal Pore Blocking Model (n=3/2) When the solute molecule size is smaller than the membrane pore size, pore
blocking occurs inside the pores [74]. This model considers that particles droplets either
adsorb or deposit over the pore walls. Therefore the volume of membrane pores declines
proportionally to the permeate volume permeates through the membrane. As a result, the
cross sectional area of the membrane pore decreases with time, and consequently
membrane resistance increases [73]. It’s been considered that pores lengths and diameters
are constant along the entire membrane. Considering these hypotheses, Hermia [82]
concluded that n was equal to 3/2 in this model.
For the internal pore blocking mechanism, permeate flux is expressed, as a
function of time as follows [80]:
1 1 ( 1/ 2 ) ( 1/ 2 ) Kst Eq.4.11 Jp J0
The parameter Ks is expressed as:
K K 2 B A J1/ 2 Eq.4.12 s 0 A0
4.5.1.3. Partial pore bridging model (n=1)
When the particles sizes are similar to the membrane pore size, Partial pore
bridging occurs. As in the complete pore blocking model, this model considers that, solid
particles or macromolecules that at any time reach an open pore might block it.
56 Nevertheless, dynamic situation of blocking/unblocking may occur. Also, particles may
bridge a pore by blocking the opening but not completely seal it [73]. Considering these
hypotheses, Hermia [82] concluded that n was equal to 1 in this model.
Mohammadi et al. [74] linearized Eq.4.8 for n=1, and expressed permeate flux as
a function of time, resulting in:
1 1 Kit Eq.4.13 Jp J0
The parameter Ki can be expressed as a function of blocked membrane surface
per unit of the total permeate volume that passes through the membrane, KA , Eq.4.14.
The portion of the membrane surface that is not blocked diminishes with time [87]. As
the result, the probability of a molecule obstructing a membrane pore continuously
decreases with time [80].
Ki KA Eq.4.14
4.5.1.4. Cake Layer Formation Model (n=0) As in the case of pore blocking model, solute molecules are greater than the
membrane pore size, and they can’t penetrate inside them [87]. In this model, a cake layer
forms on the surface. Nevertheless, the concentration of the solute molecules is
considerable and they can deposit on the surface and also on the already deposited layer
of solute molecules.
57 For cake layer formation model, permeate flux is given as a function of time by linearized
Eq.4.15:
1 1 ( 2 ) ( 2 ) Kglt Eq.4.15 Jp J0
The parameter Kgl is defined as:
K R K 2 D g Eq.4.16 gl J0 Rm
58 Chapter 5: Results and Discussions
5.1. Filtration of Oily Wastewater using Porous Monolith Polyether Sulfone Membrane In this section, the results are obtained for filtration of oily wastewater with polyether sulfone membrane and the effect of transmembrane pressure on permeate flux has been studied. In addition, backwashing as a cleaning method, was applied using permeate flow, with different cleaning duration and different intervals, to obtain the most efficient cleaning condition.
5.1.1. Effect of Transmembrane Pressure
Figure 5.1 and 5.2 show variation of the permeate flux with time for ultrafiltration of oily emulsion with and without surfactant respectively, at different transmembrane pressures (TMP). The graphs show the typical flux decline during ultrafiltration. As oil droplets and other macromolecules existing in the feed stream accumulate near, on and within the membrane, they reduce the permeate flux by blocking or constricting pores and by forming a layer of additional resistance to water permeation through the membrane.
59
Figure 5.1. Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures (TMP ) with surfactant
Figure 5.2. Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures (TMP) without surfactant
Figure 5.3 and 5.4 indicate the positive effect of the pressure on the permeate flux, with and without surfactant respectively. As shown, the permeate flux increases
60 linearly with increasing transmembrane pressure for both cases, until it attains a condition, where permeate flux is almost independent of the transmembrane pressure.
This condition provides the optimum transmembrane pressure for the membrane. Note that the maximum transmembrane pressure used was significant less than 70 psi, which was the maximum pressure recommended by the manufacturer beyond which membrane compaction begins to occur. The optimum transmembrane pressure was found to be 41
Psig or 282.69 kPa without surfactant addition in the feed and 39 Psig or 268.89 kPa , when surfactant was added to the feed. Formation of a cake layer with accumulation of oil droplets on the membrane surface is the main reason for permeation flux decline. As the filtration continues, more oil droplets precipitate on the cake layer and therefore its thickness increases, until a gel layer is formed. The permeate flux cannot be further increased after gel polarization occurs, and it may eventually decline with further transmembrane pressure increase due to compaction of the gel layer [38, 88].
Flux Vs Transmembrane Pressure (Porous Membrane, with Surfactant) 21
20
19
18
17 PermeateFlux(l/m2.h) 16
15 230 240 250 260 270 280 290 Transmembrane Pressure (kPa)
Figure 5.3. Effect of transmembrane pressure (kPa) permeate flux ( l/m2.h) for ultrafiltration with surfactant
61 Flux Vs Transmembrane Pressure 19 (Porous Membrane,without Surfactant)
18
17
16
15
14 Permeate Permeate Flux(l/m2.h) 13
12 230 240 250 260 270 280 290 300 310 Transmembrane Pressure (kPa)
Figure 5.4. Effect of transmembrane pressure (kPa) permeate flux ( l/m2.h) for ultrafiltration without surfactant
As the results show, the permeate flux can be related to transmembrane pressure
by a linear relation such as Jp C1 TMP C2, while C1 and C2 are two empirical constants. The constant values obtained from experimental data are presented in Table 5.1:
Empirical Constants of the C1 C2 Equation
With surfactant 0.082 -2.012
Without surfactant 0.074 -3.364
Table 5.1. Empirical constants for the linear relation between permeate flux and transmembrane pressure.
62 The total membrane resistance can be measured by applying the Darcy’s law:
TMP 9 Jp 3.6 10 Eq.5.1 water Rt
where: 2 J p is permeate flux ( l/m .h)
TMP is transmembrane pressure ( KPa) water is water viscosity ( Pa.s)
1 Rt is total membrane resistance( m )
Based on Darcy’s law when water is almost 6.04E-4 at feed temperature of 40
oC (Eq.4.3), total membrane resistance or is 6.55 E+13 with surfactant and 7.26E+13 without surfactant addition the feed. Surfactants molecule solubilize the solute macromolecules present in the feed by forming micelles around them, and assists
in dislodging the precipitated particles from surface the membrane which decreases the
total membrane resistance to water permeation through the membrane [39].
Figure 5.5 shows the flux decline during ultrafiltration of oily wastewater with
and without surfactant, and at transmembrane pressures of 41 Psig (282.69 kPa) and 27
(186.16kPa). As the results show at higher transmembrane pressure the permeate flux
decreases faster than at lower transmembrane pressure. The results from experimental
data for ultrafiltration at 282.69 kPa show that permeate flux declined by 36% and 39%
with and without surfactant respectively. The flux decline percentage for ultrafiltration at
186.16 kPa is 34% and 31% with and without surfactant respectively. It indicates that at
higher the transmembrane pressure is, the flux decline rate is higher. At high pressures
63 traces of oil can be seen in permeates, which indicates oil penetration through the pores, although the pore sizes are smaller than oil droplets. High pressure increases the effect of concentration polarization which causes the oil droplets to coalesce on the membrane surface, and when driving pressure exceeds the capillary pressure in some pores, the oil droplets can be deformed and therefore be pushed though the pores and into the permeate
[38].
Flux Vs Time (Porous Membrane) 25 TMP=186.16kPa, without surfactant
20 TMP=186.16kPa, with surfactant
TMP=282.7kPa, with surfactant 15 TMP=282.7kPa, without surfactant 10
Permeate Permeate Flux(l/m2.h) 5
0 0 1 2 3 4 5 Time(hr)
Figure 5.5. Variation of permeate flux ( l/m2.h) with time at transmembrane pressure of 289.69 kPa & 186.16 kPa , for ultrafiltration with and without surfactant
5.1.2. Effect of Feed Concentration
Figure 5.6 compares the permeation flux decline with time for ultrafiltration of oily wastewater with constant and variable feed concentrations of oil. For ultrafiltration of emulsion with constant oil concentration, the experimental set-up works in a complete recycle mode, where both retentate and permeate flows were pumped back to the feed tank and as a result of that the oil concentration was kept constant through the process.
64 However, in ultrafiltration of emulsion with variable oil concentration, the permeate flow was driven out of the system instead of getting recycled back to the feed tank. Therefore, the oil concentration increased gradually with time, and because of that, as it’s shown in
Figure 5.6 the flux decreased with a much faster rate than ultrafiltration with constant oil concentration. This phenomenon can be ascribed to the fact that when concentration of oil increases, the rate of accumulation of particles on the formed cake layer on membrane surface also increases, which gradually thickens the cake layer. The thicker the formed cake layer on the membrane surface, the lower is the permeation flux. At first, most of the oil particles precipitate on membrane surface and then with greater accumulation of oil particles at higher concentrations, they gradually adsorbed onto the membrane pores and plugged them. The higher the oil concentration, the higher is the pore-clogging rate.
After sometime, pore clogging slows down, and the cake layer begins to govern. [42,81].
The results show that with surfactant the flux decline rate is lower as surfactant helps to dislodge the precipitated particles from surface of the membrane.
Flux Vs Time ( Porous Membrane ) 25 Constant Feed Concentration, without Surfactant 20 Variable Feed Concentration, without Surfactant Constant Feed Concentration, 15 with Surfactant Variable Feed Concentration, 10 with Surfactant
Permeate Flux (l/m2.h) PermeateFlux 5
0 0 1 2 3 4 5 Time (hr)
Figure 5.6. Comparison of flux variation with time at different feed concentration for ultrafiltration with and without surfactant
65 5.1.3 Effect of Backwashing on Permeate Flux Recovery
As mentioned above, the use of ultrafiltration membranes in industries is limited to some extent. The major reason for this limitation is the flux decline, which occurs because of fouling. Membrane fouling lowers the treated water production rate, increases the energy consumption for membrane operation and necessitates a frequent membrane washing [89]. In this study, backwashing was applied to improve the membrane performance and also to recover the permeate flux to its initial value. To further investigate the effects of cleaning conditions on ultrafiltration membrane fouling, several experiments were carried out on polyether sulfone membrane to examine the influence of backwash interval and duration on the permeate flux.
In order to evaluate the effect of backwash interval on permeate flux recovery, two sets of experiments were done on the polyether sulfone membrane. In both experiments, ultrafiltration was carried out until the permeate flux declined to its constant value, and then for purpose of comparison, backwashing was applied, once in interval of
60 minutes and once, in interval of 90 minutes. For backwashing the membrane, permeate which had been collected in tank TK-102 for this purpose, was used as the backwashing fluid. Backwashing was done with permeate flux of 8.5 ( l/m2.h), and for duration of 200s. Figure 5.7, shows the results of backwashing with two different intervals, for ultra filtration of oily wastewater with and without surfactant. Ultrafiltration was operated for about 3.5 hours until the permeate flux declined to a constant value. At this point,
66 backwashing was applied with duration of 200s, and in intervals of 60 and 90 minutes.
The flux enhancement percentage for each backwash is presented in Table 5.2:
Backwashing Flux Enhancement Flux Enhancement Flux Enhancement Interval (min) Percentage after the Percentage after the Percentage after the First Backwash (%) Second Backwash Third Backwash (%) (%)
With Surfactant 60 55% 18%, 17%
90 57% 26% 24%
Without Surfactant 60 55% 17% 14%
90 54% 25% 23 %
Table.5.2. Percentages of flux enhancement after backwashing with intervals of 60 minutes and 90 minutes.
25 Flux Vs Time (Porous Membrane)
60min Backwash 20 Interval,with Surfactant
60min Backwash Interval,without 15 Surfactant 90min Backwash Interval,with Surfactant 10 90min Backwash
Interval,without Permeate Flux Permeate Flux (l/m2.h) Surfactant 5
0 0 2 4 6 8 10 Time (hr)
Figure 5.7. Effect of backwashing interval on permeate flux recovery. Smooth lines have been drawn through the data points to demonstrate the trend of permeate flux versus time
As Figure 5.7 shows, a very significant recovery of membrane flux is clearly observed immediately after each backwashing step, while a reduced efficiency of backwashing in recovery of flux over time is also evident. The results show that the backwashing efficiency is higher when surfactant was added to the feed and this shows
67 that the combination of surfactant and backwashing is more effective in recovery of the
flux as well as maintaining the flux at a higher value than just applying the backwashing.
It’s clear from the graphs that despite the fact that the absolute values of fluxes for
both intervals are quite similar , when backwash interval was 60 min, it could greatly
improve membrane filtration efficiency, and this could be due to the fact that the more
frequent backwash could more effectively peel off the formed cake layer on the
membrane surface, which is also consistent with the results that Lei Wang et al got in
their study [25]. Irrespective of the use of surfacant , when backwash was carried out in
90 min interval, the flux decline percentage within an interval was more significant than
that within 60 min interval.These results also suggest that no matter how often the
backwashing is carried out, a frequent backwashing, could increase the permeate flux to
its initial value.
In next two experiments, the effect of backwash duration on permeate flux
recovery was examined.For this purpose, ultrafiltration of oily wastewater was carried for
several hours until the permeate flux declined to a constant value, and then backwashing
was carried out using the permeate flow. Backwashing was done with flux of 8.5
( l/m2.h), and it was repeated three times in 60 min interval. In first test the backwash
duration was 100s and in second test it was 200s. The same tests were ran on oily wastewater with and without surfactant, and results are shown in Figure 5.8. In addition,
the flux recovery percentages are presented in Table 5.3:
68 Backwashing Flux Enhancement Flux Enhancement Flux Enhancement Duration (s) Percentage after the Percentage after the Percentage after the First Backwash (%) Second Backwash Third Backwash (%) (%)
With Surfactant 100 54% 8% 7%
200 53% 7% 6%
Without Surfactant 100 51% 10% 8%
200 55% 7% 6 %
Table 5.3. Percentage of flux enhancement after backwashing with durations 100s and 200s
The results show that the longer the backwashing duration, the higher is the rate of flux recovery. In addition, it should also be noticed that the flux decline within an interval is lower for backwash with higher duration than that for backwash with lower duration, and this could be due to the fact that the longer the backwash duration is, the more effictive is the backwash in dislodging and removing the particles which have been fixed in a form of layer on the membrane surface.
25 Flux Vs Time ( Porous Membrane )
100s Backwash Duration, with Surfactant 20 200s Backwash Duration, with Surfactant
100s Backwash Duration, 15 without Surfactant
200s Backwash Duration, without Surfactant
Permeate Permeate Flux(l/m2.h) 10
5 0 2 4 6 8 10 Time(hr)
Figure 5.8. Effect of backwashing duration on permeate flux recovery. Smooth lines have been drawn through the data points to demonstrate the trend of permeate flux versus time
69 The results from the experimental data for flux enhancement with backwashing
indicates that, backwashing is an effective cleaning method in recovering the permeate
flux. Furthermore, the results from applying the backwashing, along with surfactant,
show that surfactant can significantly improve the cleaning efficiency of backwashing.
5.1.4. Prediction of Permeate Flux by Hermia’s models
Hermia’s models were comprehensively expressed in section 4.5. In this study, in
order to identify the mechanism of fouling during ultrafiltration of oily wastewater, the
model k parameter was estimated by the linear regression method. The adjusted values
of k for n= 0, 1.0, 1.5 and 2.0 were used to solve the respective Hermia’s equations and
also the obtained prediction models in terms of flux decay which are presented in Table
5.4. At the end, in order to find which Hermia model better predicts the flux variation
behavior, the error parameter was calculated from equation 5.2, and the comparison
between the errors values achieved from different models determined the predominant
fouling mechanism, as the model with the lowest error value, fits better to the
experimental data.
The error parameter was defined as:
Error JExperimental JEstimated Eq.5.2
where: 2 JExperimental is the flux value ( l/m .h), achieved from experiment at time t(s) and transmembrane pressure (kPa).
J is the flux value ( l/m2.h), predicted by a Hermia model at time t(s) and Estimated transmembrane pressure (kPa).
70
Model n value Fouling Fouling Hermia’s Model Simplified Relation Mechanism
2 Complete Pore ln(Jp ) ln(J0) Kct Jp J0 exp(Kct) Blocking
1.5 Internal Pore 1 1 J ( ) ( ) K t J 0 Blocking 1/ 2 1/ 2 s p 1/ 2 2 Jp J0 (1 J0 Kst)
1 Partial Pore 1 1 J K t J 0 Bridging J J i p (1 J K t) p 0 0 i 0 Cake Filtration 1 1 J ( ) ( ) K t J 0 J 2 J 2 gl p (1 J 2 K t)1/ 2 p 0 0 gl Table5.4. Hermia’s model relation for different fouling mechanisms and the simplified equations
Table 5.5 presents the numerical values of Hermia’s models constant (k), as well as errors value. This analysis is conducted based on the experimental data from ultrafiltration of oily wastewater with surfactant, at different transmembrane pressures.
As it’s shown in Table 5.5, the lowest values of error for all the transmembrane pressures tested are for the cake layer formation mechanism, also the partial pore bridging mechanism can well fit to the experimental data. Therefore, it can be concluded that the best fitting mechanism to the experimental data is the cake layer formation and the other suitable mechanism is the partial pore bridging for all the transmembrane pressures tested.
The same analysis was done on the experimental data from ultrafiltration of oily wastewater without surfactant to the feed. As the results shown in Table 5.6, the lowest values of error for all the transmembrane pressures tested are related to the cake layer formation mechanism followed by the partial pore bridging mechanism. Therefore, it can be concluded that the best fitting mechanism to the experimental data is the cake layer
71 formation and the other suitable mechanism is the partial pore bridging for all the
transmembrane pressures tested in this study.
Transmembrane Pressure (kPa) 186.16 236.4 241.3 248.2 255.1 262 268.9 275.8 282.7 Complete pore blocking 0.229 0.226 0.226 0.212 0.223 0.24 0.242 0.24 0.228 K*1.00 Internal pore E04 blocking 0.035 0.031 0.029 0.027 0.0285 0.030 0.029 0.029 0.028 Partial pore bridging 0.02 0.018 0.015 0.014 0.015 0.015 0.015 0.014 0.014 Cake filtration 0.004 0.003 0.002 0.002 0.002 0.002 0.002 0.002 0.002 Complete pore blocking 2.959 2.591 3.831 3.733 4.09 5.624 5.664 5.699 6.029 Internal pore Error blocking 2.779 2.361 3.579 3.493 3.812 5.271 5.348 5.394 5.757 Partial pore bridging 2.606 2.150 3.337 3.277 3.546 4.937 5.048 5.103 5.501 Cake filtration 2.275 1.788 2.867 2.861 3.042 4.314 4.482 4.56 5.029 Table5.5. K values of Hermia’s models obtained from experimental data for ultrafiltration with surfactant
Transmembrane Pressure (kPa) 186.16 236.4 241.3 248.2 255.1 262 268.9 275.8 282.7 289.6 Complete pore blocking 0.234 0.258 0.289 0.296 0.297 0.298 0.299 0.271 0.276 0.286 Internal pore K*1.00 blocking 0.039 0.039 0.044 0.044 0.043 0.043 0.042 0.037 0.037 0.039 E04 Partial pore bridging 0.025 0.025 0.027 0.026 0.025 0.025 0.024 0.021 0.020 0.021
Cake filtration 0.006 0.005 0.005 0.005 0.004 0.004 0.004 0.003 0.003 0.003 Complete pore blocking 2.780 3.868 4.180 4.136 4.032 4.429 4.359 5.425 5.503 5.503 Internal pore Error blocking 2.543 3.552 3.761 3.669 3.549 3.929 3.839 4.988 5.005 5.005 Partial pore bridging 2.31 3.247 3.353 3.215 3.076 3.440 3.328 4.558 4.526 4.526
Cake filtration 1.862 2.656 2.547 2.303 2.119 2.457 2.297 3.741 3.592 3.592 Table5.6. K values of Hermia’s models obtained from experimental data for ultrafiltration without surfactant
72 5.1.5. Flux Decay Analysis by using a combination of Hermia’s models
In this section, the permeate flux decay is analyzed by applying the combination
of the constant pressure blocking filtration laws (Hermia’s models) with the measurement
of the membrane fouling resistances in series. In order to define the fouling resistances, it
was considered that the membrane fouling is mainly caused by either pore blocking or
cake/gel layer formation. When pore blocking happens, it can occur inside the pores
(internal pore blocking) or outside (partial pore bridging or complete pore blocking).
Analyzing the blocking filtration laws in terms of the resistances of the membrane, allows
a better understanding of the actual phenomena involved in membrane fouling, and the
significance of both internal and external fouling resistances [90].
As it’s been widely discussed in the literature [72], typically, in first couple of
minutes of ultrafiltration and microfiltration, complete pore blocking predominates and
then there’s a transition to the cake filtration model as the dominated phenomenon
involved in fouling. In other words, by the start of filtration a portion of the pores is
immediately blocked because of complete pore blockage, and then the flux declines
gradually as the result of other three phenomena of cake filtration, partial pore bridging,
and internal pore blocking involved in membrane fouling [18, 91]. Therefore the total
membrane resistance can be expressed as:
1 R R total 1 1 4 R2 R3 Eq.5.3
Where: R2,R3,R4 are resistances due to the internal pore blocking, partial pore bridging
and cake filtration, respectively.
73 As mentioned earlier, Darcy’s law, the permeate flux can be expressed in terms of total
membrane resistance as, Eq.5.1:
TMP 9 Jp 3.6 10 Eq.5.1 p Rt
where: 2 J p is permeate flux ( l/m .h)
TMP is transmembrane pressure ( KPa) p is permeate viscosity ( Pa.s)
1 Rt is total membrane resistance( m )
Substituting the resistance s in Eq. 5.3 by their equivalents from Eq.5.1, we obtain:
1 TMP R Eq.5.4 t 1 1 J p 4 TMP TMP
p J2 p J3
or more simplified as follows: 1 TMP TMP TMP R R Eq.5.5 t J J t p 2 p 3 p J4 p J2 p J3 p J4 TMP Therefore we obtain Eq. 5.6:
TMP 1 1 Rt ( ) Eq.5.6 p J2 J3 J4
or
R 1 1 p t ( ) Eq.5.7 TMP J2 J3 J4
74 If we substitute the fluxes term by the suggested form from Darcy’s law we’ll have:
1 R p t Eq.5.8 JP TMP Therefore, equation 5.4 can be written as: 1 1 1 ( ) Eq.5.9 Jp J2 J3 J4
According to the Hermia’s models, permeate fluxes of J2,J3 and J4 are defined as:
1 1 1 K t J ( )2 2 2 1 J2 J0 K2t J0
1 1 1 K t J 3 3 1 J3 J0 K3t J0
1 1 1 K t J 2 2 4 4 1 J4 J0 2 K4 t J0
By plugging these terms into Eq.5.9 we obtain:
1 1 1 Eq.5.10 J 1 2 1 P ( ) 1 1 1 1 K2t K3t K t J J 2 4 0 0 J0
Based on equation 5.10, an optimization was performed by using program
MATLAB; for each set of flux versus time experimental data, minimum error was
calculated for different values of Hermia’s models constants ( k2,k3,k4 ). Optimization
space for these coefficients was defined from zero up to 10 times the K values, which
were obtained in the previous section, by fitting each Hermia’s model to the experimental
75 data. Table 5.7 and 5.8; show the values of minimum error and the corresponding values
of k for ultrafiltration at different pressures, with and without surfactant, respectively.
Transmembrane Pressure (kPa) 186.16 236.4 241.3 248.2 255.1 262 268.9 275.8 282.7 K values obtained for different models fitted to experimental data individually Complete pore blocking K1 (1.0E-04) 0.2285 0.2263 0.2261 0.2115 0.223 0.24 0.242 0.24 0.2277 Internal pore blocking K2 (1.0E-04) 0.0354 0.0314 0.0295 0.0273 0.0285 0.0302 0.0298 0.0297 0.0282 Partial pore bridging K3 (1.0E-04) 0.022 0.0175 0.0154 0.0141 0.0146 0.0152 0.0147 0.0147 0.014 Cake filtration K4 (1.0E-04) 0.0043 0.0027 0.0021 0.0019 0.0019 0.0019 0.0018 0.0018 0.0017 K values obtained for different models in combination Complete pore blocking K1 (1.0E-04) 0.4571 0.4256 0.4506 0.4596 0.5387 0.6264 0.5786 0.5814 0.6146 Internal pore blocking K2 (1.0E-04) 0.2239 0.1146 0.1577 0.1165 0.2851 0.1508 0.2105 0.1877 0.1995 Partial pore bridging K3 (1.0E-04) 0.0023 0.0018 0.0016 0.0014 0.0021 0.0025 0.0026 0.0028 0.0027 Cake filtration K4 (1.0E-04) 0.0035 0.0022 0.0017 0.0016 0.0015 0.0014 0.0013 0.0012 0.0012 Table5.7. K values of Hermia’s models obtained for ultrafiltration of oily emulsion with surfactant
Transmembrane Pressure (kPa) 186.16 236.4 241.3 248.2 255.1 262 268.9 275.8 282.7 289.6 K values obtained for different models fitted to experimental data individually Complete K1(1.0E- pore blocking 04) 0.2342 0.2575 0.2885 0.2958 0.2965 0.298 0.2994 0.2714 0.2763 0.2856 Internal pore K2(1.0E- blocking 04) 0.0385 0.0399 0.0435 0.0435 0.043 0.0426 0.0423 0.0373 0.0373 0.039 Partial pore K3(1.0E- bridging 04) 0.0253 0.0248 0.0264 0.0257 0.0251 0.0245 0.024 0.0206 0.0202 0.0214 Cake K4(1.0E- filtration 04) 0.0055 0.0048 0.0049 0.0045 0.0043 0.0041 0.0039 0.0032 0.003 0.0032 K values obtained for different models in combination Complete K1 (1.0E- pore blocking 04) 0.48 0.5923 0.6382 0.6065 0.6255 0.6428 0.6316 0.6446 0.6442 0.6722 Internal pore K2 (1.0E- blocking 04) 0.0727 0.1262 0.1646 0.1539 0.2151 0.1348 0.122 0.2156 0.0906 0.0919 Partial pore K3 (1.0E- bridging 04) 0.0025 0.0027 0.0028 0.0027 0.0036 0.004 0.0039 0.0036 0.0035 0.0038 Cake K4 (1.0E- filtration 04) 0.0046 0.004 0.004 0.0037 0.0033 0.003 0.0028 0.0023 0.0022 0.0023 Table5.8. K values of Hermia’s models obtained for ultrafiltration of oily emulsion without surfactant
76 In addition, Table 5.9 and 5.10 and also the Figure 5.9 and 5.10, show the error
values obtained for the proposed model, as well as the error values obtained from fitting
each Hermia’s model individually, to the experimental data for ultrafiltration with and
without surfactant, respectively. The results indicate that the lowest error values
correspond to the combination of Hermia’s models, which shows the great consistency of
the proposed model.
Transmembrane Pressure (kPa) 186.16 236.4 241.3 248.2 255.1 262 268.9 275.8 282.7 Complete pore blocking 2.9595 2.5908 3.8307 3.7334 4.09 5.6236 5.6641 5.6988 6.0294 Internal pore blocking 2.7792 2.3611 3.5797 3.4928 3.8116 5.2707 5.3484 5.3926 5.7567 Error Partial pore bridging 2.6059 2.1503 3.3367 3.2765 3.5456 4.9366 5.0481 5.1027 5.5008
Cake filtration 2.2749 1.7876 2.8671 2.8612 3.0424 4.314 4.4823 4.56 5.0293
Combined model 2.1207 1.6481 2.5448 2.5393 2.517 3.7047 3.7218 3.8714 4.2311 Table 5.9.Error values measured for Hermia’s models and the combined model(with Surfactant)
Transmembrane Pressure (kPa) 186.16 236.4 241.3 248.2 255.1 262 268.9 275.8 282.7 289.6 Complete pore blocking 2.7803 3.8683 4.1801 4.1356 4.0318 4.4297 4.3587 5.4245 5.3805 5.5025 Internal pore blocking 2.5434 3.5519 3.7606 3.6699 3.549 3.9295 3.8385 4.9819 4.9021 5.0051 Error Partial pore bridging 2.313 3.247 3.3531 3.2149 3.0755 3.4402 3.3281 4.558 4.4402 4.5261
Cake filtration 1.8621 2.6555 2.5468 2.303 2.1196 2.457 2.2966 3.7405 3.5381 3.5918
Combined model 1.8429 2.4905 2.5196 2.2797 2.0704 2.3778 2.2313 3.4138 3.2528 3.3575 Table 5.10.Error values measured for Hermia’s models and the combined model(without Surfactant)
77
Figure 5.9. Error values for different Hermia’s models for ultrafiltration of oily wastewate with surfactant at different transmembrane pressure
Comparison of error for different Fouling Mechanisms (Porous Membrane, Without Surfactant) Complete Pore 6 Blocking Internal Pore Blocking 5 Partial Pore Bridging Cake Filteration 4 Combination
3 Error (l/m2.h) Error 2
1
0 186.16 236.4 241.3 248.2 255.1 262 268.9 275.8 282.7 289.6
Transmembrane Pressure (kPa)
Figure 5.10. Error values for different Hermia’s models for ultrafiltration of oily wastewate without surfactant at different transmembrane pressure
78 As shown earlier, after the combined model, the best consistency corresponds to
the cake filtration model. Figure 5.11 and 5.12 show the percentage of error reduction
with respect to the error values obtained by fitting cake filtration model . The error
reduction is obtained by:
(Error Error ) reduction(%) Combination Cake 100 Eq.5.11 ErrorCake
where :
ErrorCombination is the error value obtained for combined model
ErrorCake is the error value obtained for cake filtration model
Figure 5.11. Reduction precentages of error values combined model with respect to the cake filtration model, for ultrafiltration of oily wastewate with surfactant at different transmembrane pressures
79
Figure 5.12. Reduction precentages of error values combined model with respect to the cake filtration model, for ultrafiltration of oily wastewate without surfactant at different transmembrane pressures
During the ultrafiltration process, as the transmembrane pressure increases, the gel layer that’s formed on the membrane surface becomes denser, and if the pressure value exceeds the capillary pressure in some pores, the oil droplets can be deformed and therefore be pushed though the pores and block them [38] . Looking at this fact, from the point of the involved fouling mechanisms, it implies that with the increase of transmembrane pressure, the percentage of the contribution of internal pore blocking and pore bridging model, and in contrast the percentage of the contribution of the cake filtration model decreases. This is clearly seen in Figure 5.13 and 5.14 for ultrafiltration of oily emulsion, with and without surfactant. In order to verify the effect of surfactant on the involved fouling mechanism, the contribution percentages of cake filtration and pore blocking models are compared in Figure 5.15. As the results show, the surfactant didn’t have a sensible impact on contribution of fouling mechanisms.
80
Figure 5.13. contribution precentages of cake filtration and pore blocking mechanisms, for ultrafiltration of oily wastewate with surfactant at different transmembrane pressure
Figure 5.14. contribution precentages of cake filtration and pore blocking mechanisms, for ultrafiltration of oily wastewate with surfactant at different transmembrane pressure
81
Figure 5.15. Comparision of the contribution precentages of cake filtration and pore blocking mechanisms, for ultrafiltration of oily wastewate at different transmembrane pressure with and without surfactant
5.1.6. Mass Balance Analysis
Mass balance for oil gives thef olowing equation:
(Coil q)Feed (Coil q)Permeate (Coil q)reject Eq.5.12
where:
Coil is the oil concentration (wt%)
q is volumetric feed flow rate (Gallon/min)
The oil concentration of feed and reject measured by the EPA method (refer to Appendix
2), we can calculate the oil concentration of reject flow using equation 5.12.The results are presented in Table 5.11 and 5.12 for ultrafiltration process with and without surfactant ,respectively:
82
Mass Balance Analysis for the Porous Membrane ( Ultrafiltration with Surfactant). Transmembrane Pressure:268.9kPa,Feed Flow Rate:41.86 (gallons/min) Experimental Calculated Oil Experimental Permeate Flow Reject Flow Oil Concentration Value of Oil Rate Rate Concentration in Reject Flow Concentration (Gallon/min) (Gallon/min) in Feed Flow (wt%) in Reject Flow (wt%) (wt%) Start of 3.40 37.88 Filtration 2.7 0.3 2.943 0.9 2.9 0.3 End of 2.40 38.84 Filtration 2.9 0.3 3.079 0.9 3.4 0.3 Table 5.11. Calculation of Oil Concentration in the Porous Membrane Reject Flow (with Surfactant)
Mass Balance Analysis for the Porous Membrane ( Ultrafiltration without Surfactant). Transmembrane Pressure:282.7kPa, Feed Flow Rate:41.24 (gallons/min) Experimental Calculated Oil Experimental Permeate Flow Reject Flow Oil Concentration Value of Oil Rate Rate Concentration in Reject Flow Concentration (Gallon/min) (Gallon/min) in Feed Flow (wt%) in Reject Flow (wt%) (wt%) Start of 2.98 38.88 Filtration 3.3 0.3 3.553 1.1 3.7 0.3 End of 1.90 39.96 Filtration 3.9 0.3 4.085 1.2 4.2 0.3 Table 5.12. Calculation of Oil Concentration in the Porous Membrane Reject Flow (without Surfactant)
The calculated values for the oil concentration in the membrane reject flow agrees quite well with the experimental measured values, as obtained by Cardinal Laboratories, Inc.
(Wilder, KY), using EPA Method (Appendix 2).
83 5.2. Filtration of Oily Wastewater using Dense Hollow-Fiber Regenerated Cellulose Ultrafiltration Membrane
In order to evaluate the performance of the cupraammonium regenerated cellulose membrane for filtration of oily wastewater, different experiments categorized into four main cases were conducted on the membrane. As mentioned above, for this membrane, the applied cleaning technique was gas sparging. Running the same experiments under comparable operating conditions such as feed concentrations, temperature and feed cross flow velocity, helped to achieve a better understanding of membrane performance and also effect of surfactant and air sparging on flow rate enhancement. These four categories are displayed in Table 5.13:
Experiment Condition Feed Surfactant Volume Air Sparging Fraction ( m3 /m3) Volumetric Flow Rate (SCFH)
Ultrafiltration without Oily wastewater of 0 0 surfactant/without air Ford motor Co. Plant Ultrafiltration with Oily wastewater of 4.4 105 0 surfactant/without air Ford motor Co. Plant
Ultrafiltration Oily wastewater of 4.4 105 0.5 withsurfactant/with Ford motor Co. Plant air
Ultrafiltration without Oily wastewater of 0 0.5 surfactant/with air Ford motor Co. Plant
Table 5.13. Operating conditions for the experiments conducted using the dense membrane 5.2.1. Effect of Transmembrane Pressure
Figure 5.16-5.19 show that permeate flow rate versus time at different transmembrane pressures and under different experimental conditions presented in
Table5.13. As typically is the case in conventional cross-flow ultrafiltration, the permeate
84 flow rate, decreases with time. During the first few minutes of filtration, the permeate flow rate decline is high, which is mainly due to accumulation of oil droplets at the top of the membrane module. As oil is rejected at the membrane surface, the emulsion breaks releasing free, insoluble oil, which rises to the top of the membrane module, due to lower density, thereby blanketing the surface area at the top of the membrane module, and reducing the filtration rate.
Once a steady-state amount of free oil has accumulated at the top of the membrane module, the permeate flow reaches a steady-state, constant value. The constant permeate flow rate indicates the fact that the fouling resistance due to the precipitated particles is constant with time.
Permeate Flow Rate Vs Time (Dense Membrane,without 0.8 Surfactant,without Air, Membrane Area=12.9 m2)
0.7 TMP(137.9kPa)
0.6 TMP(144.8kPa)
0.5 TMP(151.7kPa) TMP(158.6kPa) 0.4 TMP(165.5kPa) 0.3 TMP(172.4kPa) 0.2
Permeate (gallons/min) Rate Permeate Flow TMP(179.3kPa) 0.1
0 0 0.5 1 1.5 2 2.5 3 3.5 Time(hr)
Figure 5.16. Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures for ultrafiltration of oily wastewate without surfactant and without air injection
85 Permeate Flow Rate Vs Time (Dense Membrane, with Surfactant,without Air, Membrane Area=12.9m2) 0.8 TMP(131 kPa)
0.7 TMP(137.9kPa) TMP(144.8kPa) 0.6 TMP(151.7kPa) 0.5 TMP(158.6kPa) 0.4 TMP(165.5kPa) 0.3 TMP(172.4kPa)
0.2 Permeate Rate(gallons/min) Permeate Flow 0.1
0 0 0.5 1 1.5 2 2.5 3 3.5 Time(hr)
Figure 5.17. Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures for ultrafiltration of oily wastewate with surfactant and without air injection
Permeate Flow Rate Vs Time (Dense Membrane, with Surfactant,without Air, Membrane Area=12.9m2) 0.8 TMP(131 kPa) 0.7 TMP(137.9kPa) 0.6 TMP(144.8kPa) 0.5 TMP(151.7kPa)
0.4 TMP(158.6kPa) TMP(165.5kPa) 0.3 TMP(172.4kPa)
0.2 Permeate Rate(gallons/min) Permeate Flow 0.1
0 0 0.5 1 1.5 2 2.5 3 3.5 Time(hr)
Figure 5.18. Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures for ultrafiltration of oily wastewate with surfactant and with air injection
86 Permeate Flow Rate Vs Time ( Dense Membrane, without Surfactant, with Air, Membrane Area=12.9m2) 0.8 TMP(137.9kPa) 0.7 TMP(144.8kPa)
0.6 TMP(151.7kPa) TMP(158.6kPa) 0.5 TMP(165.5kPa)
0.4 TMP(172.4kPa)
0.3 Permeate Rate(gallons/min) Permeate Flow 0.2 0 0.5 1 1.5 2 2.5 3 3.5 Time(hr)
Figure 5.19. Variation of permeate flux ( l/m2.h) with time at different transmembrane pressures for ultrafiltration of oily wastewate without surfactant and with air injection
Figure 5.20-5.23 represent the permeate flux versus transmembrane pressure, for different experimental conditions. Results show that high transmembrane pressure leads to higher driven force but also flux decline due to higher fouling resistance; however the average flux for all cases showed a significant enhancement with transmembrane pressure. This implies that flux enhancement rate due to higher transmembrane pressure
(TMP) overcomes the decrease in flux due to fouling resistance.
87 Permeate Flux Vs Transmembrane Pressure (Dense Membrane, without Surfactant and without Air)
16 14 12 10 8 6
Permeate Flux Permeate (l/m2.h) Flux 4 2 0 130 140 150 160 170 180 190
Transmembrane Pressure (kPa)
Figure 5.20. Effect of transmembrane pressure on permeate flux for ultrafiltration of oily wastewate without surfactant and without air injection
Permeate FLux Vs Transmembrane Pressure ( Dense Membrane, with Surfactant, without Air) 16 14 12 10 8 6
4 Permeate (l/m2.h) FluxPermeate 2 0 125 135 145 155 165 175 Transmembrane Pressure (kPa)
Figure 5.21. Effect of transmembrane pressure on permeate flux for ultrafiltration of oily wastewate with surfactant and without air injection
88 Permeate Flux Vs Transmembrane Pressure (Dense Membrane, with Surfactant, with Air) 18
16
14
12
10
8 Permeate Permeate Flux (l/m2.h) 6
4 135 140 145 150 155 160 165 170
Transmembrane Pressure (kPa)
Figure 5.22. Effect of transmembrane pressure on permeate flux for ultrafiltration of oily wastewate with surfactant and with air injection
Permeate Flux Vs Transmembrane Pressure ( Dense Membrane, without Surfactant, with Air) 16
14
12
10
8
6
Permeate Permeate FLux (l/m2.h) 4
2
0 135 140 145 150 155 160 165 170 Transmembrane Pressure (kPa)
Figure 5.23. Effect of transmembrane pressure on permeate flux for ultrafiltration of oily wastewate without surfactant and with air injection
89 5.2.2. Effect of Air Injection and Surfactant on Membrane Performance
Figure 5.24 shows the permeate flux versus transmembrane pressure for ultrafiltration of oily wastewater with and without surfactant, and with and without air sparging. The experiments were conducted under the same operating conditions such as feed concentration and temperature.
Permeate Flux Vs Transmembrane Pressure (Dense Membrane) 16
sparged/with Surfactant 14 unsparged/with Surfactant 12 sparged/without Surfactant
10 unsparged/without Surfactant 8
Permeate Flux Permeate Flux (l/m2.h) 6
4 135 140 145 150 155 160 Transmembrane Pressure (kPa)
Figure 5.24. Effect of air injection and surfactant on permeate flux for transmembrane pressure
For conventional unsparged filtration, permeate flux increased gradually with increasing of transmembrane pressure, and higher permeate flux achieved at higher transmembrane pressure. Gas injection caused significant flux enhancements, and in terms of percentage increase in permeate flux, gas sparging can lead to a higher enhancement where concentration polarization is expected to be severe, which is, at higher transmembrane pressure [92]. As the figure implies the flux enhancement, is higher for filtration with surfactant than without surfactant, and it shows the effective role of surfactant in enhancement of membrane filtration efficiency.
90 Figure 5.25 shows the permeate flow rate versus time for ultrafiltration of oily wastewater at optimized condition, with and without surfactant, and with and without air sparging.
Permeate Flow Rate Vs Time ( Dense Membrane)
1
0.9 sparged/without Surfactant sparged/with Surfactant 0.8 unsparged/with Surfactant 0.7 unsparged/without Surfactant 0.6
0.5
0.4
Permeate Flow Rate ( Gallon/min) ( Rate PermeateFlow 0.3 0 1 2 3 4 Time (hr)
Figure 5.25. Effect of air injection and surfactant on permeate flux variation with time
As the graph shows, the initial permeate flow rate as well as steady state permeate flow rate is higher for ultrafiltration with surfactant and air sparging. The comparison between the permeate flow rate values clearly shows that air injection keeps the permeate flow rate higher throughout the process and this is mainly due to the fact that air sparging decreases the membrane fouling resistance by disrupting the formed concentration polarization layer [92]. The results also indicate that combination of surfactant and air injection is more effective in enhancing the flux than either air sparging or surfactant, although the effect of air sparging alone is higher than with surfactant only the surfactant reduces the surface tension between oil and water, and with forming micelles around the oil droplets, it prevents formation of dense oil aggregates, and makes it easier for the air
91 slugs to get through the deposits layer, disrupting it with creating turbulence, which significantly increases water mass transfer through the membrane which is seen as higher permeate flow rate.
5.2.3. Analysis of Permeate Flux for Ultrafiltration of oily emulsion in Dense Hollow-Fiber Regenerated Cellulose Membrane
The hollow fiber module tested here, is consisted of N fibers of same size, in which the membrane is formed on the inside of N tiny dense tubes. In ultrafiltration with a dense membrane, water permeation occurs by diffusion through the membrane in contrast to diffusion through the pores, which happens during ultrafiltration with a porous membrane. Therefore, water molecules need to diffuse through the emulsion and then through the membrane wall, in order to transfer to the permeate side of the membrane.
As mentioned earlier, ultrafiltration is a pressure-driven process, and for a small applied pressure, it’s been observed that the permeate flux through the membrane is proportional to the pressure difference across the membrane. It’s like a Darcy’s law permeability for flow through porous membrane. However, as the transmembrane pressure increases the permeate flux begins to decline bellow the value which would result from a linear-pressure behavior. According to Ho-Ming Yeh and Jin-Hong Dong
[93], the water permeation flux in a cross-flow ultrafiltration is related to the transmembrane pressure by this equation:
92 P J p RT 1 t ( fiber ) P kc Dwm Eq.5.13 1 Jstst
where: 3 2 J p is water (permeate) flux( m /m .s)
kc is the mass transfer coefficient( m/s)
2 Dwm is water diffusion coefficient( m/s ) t fiberis membrane thickness( m) Pis pressure difference across the membrane ( Pa)
is proportional constant ( m2.s/m3) 3 2 Jstst is the steady-state flux( m /m .s)
R=8.314 ( Pa.m3 /mol.K ) &T =40 oC
is molar volume of water at 40 oC and it’s equal to 1.81E -5 ( m3 /gmole)
Equation 5.13 in fact is an expression of the resistance-in-series model, where
1/kc denotes the mass transfer resistance, t /Dwm is the membrane intrinsic resistance and
P is the resistance due to the concentration polarization which will be proportional to
the amount and specific hydraulic resistance of the compressible layer formed on the
membrane and can be assumed to be a linear function of transmembrane pressure with
as a proportional constant.
With the use of the results from the experimental data of permeate flux for pure
water at 379 kPa, a straight line of (1/Jp )exp versus (1/P)exp could be constructed by the
93 least-squares method [93,94,95]. Thus the intrinsic resistance of the hollow fiber
membrane module employed in this study can be determined from the experimental data
of permeate flux for pure water and by using the following equation, which can be
modified from Eq.5.13 by setting zero for pure water.
1 RT 1 t 1 ( fiber ) Eq.5.14 (J ) k D P p exp c wm exp
The measured value of RT(1/kc t fiber /Dwm)/ for the membrane system employed in this study was determined graphically as [96]: RT 1 t ( fiber ) 5 107(Pa.m2.s/m3) v kc Dwm
Since the pure water flow regime inside the fiber is laminar, the mass transfer resistance of pure water was obtained from this equation:
k d d2 u c fiber 1.62( fiber )1/ 3 Eq.5.15 Dw lfiber Dw
where: kc is mass transfer coefficient (m/s)
9 2 Dw is water diffusivity( 3.2 10 m /s)
d fiber is the fiber internal diameter (m)
lfiber is fiber length (m)
u is the water flow velocity inside the fiber (m2 /s)
94 The Reynolds number is defined as:
ud Re fiber & n water Eq.5.16 n water
Where, water is the pure water viscosity, water is the pure water density. The water flow velocity inside the fiber u can be obtained by: 2 Q/N d fiber u & Ac Eq.5.17 Ac 4 Where:
3 Q is the feed volume flow rate ( m /s)
2 Ac is the cross- sectional area of the fiber ( m ) N is the number of fibers and can be expressed as: A N t Eq.5.18 A l where:
At is the total membrane active area ( )
Al is the fiber lateral area ( )
lfiber is fiber length ( m)
The value of N is obtained as follows:
6 6 d fiber 420 10 (m) 2 rfiber rfiber 210 10 (m)
2 7 2 Ac (rfiber) 1.4 10 (m )
lfiber 0.986(m)
A (2r ) l 2 210 106 0.986 1.31103 (m2) l fiber fiber 12.9 N 9847 (1.31103)
95 Having the value of Reynolds number, pure water viscosity and density, we found the
feed velocity inside the fiber as:
ud Re fiber & n water n water Re 2000 6 water 670 10 (Pa.s) 3 water 970(kg/m )
The water flow velocity inside the fiber obtained as:
u=16.8(m/s)
and furthermore the mass transfer coefficient was determined from equation 5.15 as:
kc 9.16 (m/s)
As mentioned above, the value of the total membrane resistance determined graphically for pure water was:
RT 1 t ( fiber ) 5 107(Pa.m2.s/m3) v kc Dwm
Therefore, by having the mass transfer coefficient of water we can determine the
diffusivity of water in membrane and further the membrane intrinsic resistance t fiber /Dwm :
96 while : 1 1 0.109(s / m) kc 9.164 and 6 t fiber 8010 (m) R 8.314(Pa.m3 / gmole.K) T 313(k) v 1.81105 (m3 / gmole)
t 1.81105 fiber (5 107 ) 0.109 0.239(s/m) Dwm 8.314 313 and 4 2 Dwm 3.35 10 (m /s)
Furthermore, the experimental data obtained in ultrafiltration of emulsion in this study is also applied to Eq.5.13, so:
1 RT 1 t 1 ( fiber ) Eq.5.19 Jp v kc Dwm P
Therefore, from a straight line plot of (1/Jp )expversus (1/P)expat constant feed concentration, the experimental values of (the intersection at the ordinate) and RT(1/kc t fiber /Dwm)/v (the slope), as well as1/kc , were determined graphically from Table 5.14 as function of surfactant and air. The results are presented in Table 5.15 as well as Figure 5.26:
97 Transmembrane 5 6 Ultrafiltration Pressure 10 Permeate Flux 10 Operating Conditions (Pa) ( m3 /m2.s) Without Surfactant & 1.379 1.563 Without Air 1.448 2.149 1.517 2.393 1.586 2.638 1.655 3.175
With Surfactant & 1.379 2.247 Without Air 1.448 2.345 1.517 2.687 1.586 3.175 1.655 3.615
With Surfactant & 1.379 2.491 With Air 1.448 3.077 1.517 3.419 1.586 4.005 1.655 3.761
Without Surfactant & 1.379 2.345 With Air 1.448 2.687 1.517 3.126 1.586 3.761 1.655 3.566 Table5.14. Experimental data for ultrafiltration oily emulsion using dense membrane
(RT(1/k t /D )/v) 1011 3 k 104 107 c fiber wm 1/kc 10 c 2 3 (Pa.m .s/m ) (s/m) (m/s) (s/m) 5.83 0.119 Without Surfactant & 2.465 1.71 Without Air With Surfactant & 1.474 1.02 9.76 0.061 Without Air
With Surfactant & With Air 1.170 0.814 12.3 0.047
Without Surfactant & With Air 1.336 0.929 10.8 0.055
Table5.15.The fitting parameter of experimental data, t fiber /Dwm 0.238(s/m)
98 As expected, as it’s shown in Table 5.15 and Figure 5.26, the mass transfer
resistance as well as the coefficient of the resistance due to concentration polarization
decreases when ultrafiltration is ran with surfactant and with air injection, and this
indicates that the present model easily describes the relationships of permeate flux with
the transmembrane pressure.
1.40E-03 Mass Transfer Coefficient Value for Different Ultrafiltration Experiments using Dense Membrane 1.20E-03
1.00E-03
8.00E-04
6.00E-04
4.00E-04
Mass Transfer(m/s) Mass Coefficient 2.00E-04
0.00E+00 Without With With Surfactant,With Without Surfactant,Without Surfactant,Without Air Surfactant,With Air Air Air Experimental Condition for Ultrafiltration with Dense Membrane Figure 5.26. The comparison of the values of the mass transfer coefficient for ultrafiltration of oily emulsion using dense membrane under different experimental conditions
5.2.4. Mass Balance Analysis
Recalling from section 5.1.6, and according to the law of mass balance we have:
(Coil q)Feed (Coil q)Permeate (Coil q)reject Eq.5.12
Having the values of the oil concentration in the feed flow at the start and the end of filtration process, we calculated the oil concentration in the reject flow with the use of Eq.5.19 and the results are presented in Table 5.16:
99
Mass Balance Analysis for the Dense Membrane (Ultrafiltration without Surfactant/without air),Transmembrane Pressure: 172.4 kPa, Feed Flow Rate: 21.33 (gallons/min)
Permeate Flow Experimental Oil Calculated Value of Reject Flow Rate Concentration in Oil Concentration in Rate (gallons/min) (gallons/min) Feed Flow (wt%) Reject Flow (wt%)
Start of Filtration 0.7 20.63 3.1 0.3 3.21 1.0 End of Filtration 0.39 20.94 3.5 0.3 3.57 1.1 Table 5.16. Calculation of Oil Concentration in the Dense Membrane Reject Flow
The oil concentration in the feed flow is slightly lower than the oil concentration in the
reject flow due to the low permeation rates of water obtained through the dense
membrane.
100 Chapter 6: Conclusions and Recommendations
A comprehensive study has been performed on ultrafiltration system using the oily wastewater, in order to determine the involved fouling mechanism and predict the permeate flux behavior. The ultrafiltration membranes tested in this study included a porous monolithic polyether sulfone membrane and a dense hollow fiber cellulose membrane. The results from the experimental data for each membrane are presented here:
1) Porous Ultrafiltration Membrane:
It was observed that the permeate flux is a strong function of transmembrane pressure,
as it’s consistent with the empirical correlation cited in Eq.4.2. High transmembrane
pressure resulted in high permeate flux, but high flux declination rate as the result of
gel layer formation and intensive pore blocking.
The results indicated that addition of surfactant to the feed, significantly improved the
membrane performance by enabling oil to break up into smaller droplets, which are
then readily removed from the surface of the membrane.
High transmembrane pressure aggregated fouling, that required membrane cleaning.
Backwashing was found to be an effective way to enhance the permeate flux, but
combination of backwashing and surfactant achieved a higher flux recovery, so the
combination cleaning is a more effective cleaning method.
It was noted that the backwashing duration and interval could significantly affect the
backwashing cleaning efficiency. The results showed that the longer was the
backwashing duration, the higher was the flux recovery percentage. It was also found
that the more frequent backwash could more effectively peel off the formed cake layer
on the membrane surface and consequently enhance the permeate flux.
101 In order to determine the fouling mechanism, flux decay was analyzed by using the
combination of the Hermia’s models with the measurments of membrane resistances
arranged in series and parralel format. It was found that permeate flux was governed
by two major fouling mechanisms: the complete pore blocking that occurrs at the
initial stages of filtration, followed by an external surfaca fouling conforming to the
cake filtration model.It was shown that the predominant fouling resistance arises from
the formation of a fouling layer over the membrane surface.
The results indicated that with the increase of transmembrane pressure the influence of
the pore blocking mechanisms such as partial pore bridging and internal pore blocking
becomes more significant, while the effect of cake filtration mechanism weekens,
however the predominant fouling mechanism is still the cake filtration model.
2) Dense Ultrafiltration Membrane:
The results showed that the permeate flux was highly enhanced by increase in
transmembrane pressure. However, gel layer formation on the membrane surface, as
well as concentration polarization led to high flux declination rate.
The permeate flux enhanced by addition of surfactant to the feed, as it happened for
porous membrane.
Air was injected into the feed stream, to reduce concentration polarization and
membrane fouling. The results from experimental data showed a significant
enhancement in permeate flux by air injection and the enhancement rate was even
higher when a combination of air injection and surfactant was applied; The effect of
air sparging only was higher than with surfactant by itself.
The resistance in series model, proposed by Ho-Ming Yeh and Jin-Hong Dong [92],
102 was applied to describe the relationship of permeate flux with transmembrane pressure
and operating conditions. Further, as the results presented in Table 5.15 indicate, the
mass transfer coefficient is higher when either air is injected or surfactant is added to
the feed stream. This indicates the important role of air and surfactant in decreasing
the fouling resistance and enhancing permeate flux.
Future work recommended includes testing the porous and dense membranes for other types of wastewaters, such as waters produced in the plating industry, which typically contains oils, greases, metal precipitates and dirt, or domestic and industrial wastewaters containing high chemical oxygen demand (COD) contaminants and particulates.
103 Bibliography:
[1] J. Mallevialle, P.E. Odendaal, L. Des eaux-Dumez, M.R. Wiestner, Water Treatment Membrane Processes, McGraw-Hill,1996. [2]http://www.memos-filtration.de/cms/en/membranfiltration.php [3] D. Chen, Ultrasonic control of ceramic membrane fouling caused by silica particles and dissolved organic matter, Diss, The Ohio State University, 2005.
[4] The desalting and water treatment membrane manual. U.S. Department of the interior Bereau of Reclamation Denver office, research and laboratory services division applied sciences branch. [5] Belfort,G., R. H. Davis, and A. L. Zydney, 1994 “ the behavior of suspensions and macromolecular Solutions in Crossflow Microfiltration,”J.Membrane Sci., 96:1-58. [6] EPA, Membrane filtration guidance manual, 815-R-06-009.
[7] D. Chen, L.K. Weavers, H.W. Walker, Ultrasonic control of ceramic membrane fouling by particles:Effect of ultrasonic factors, Ultrasonics sonochemistry 13 (2006) 379-387. [8] I.C. Escobar, E.M. Hoek, C.J. Gabelich, F.A. DiGiano, et al. Recent advances and research needs in membrane fouling, American water works Association Journal; August 2005; 97,8; ABI/INFORM trade & Industry pg.79. [9] A.G. FANE, C.J.D.FELL, A Review of Fouling and Fouling Control in Ultrafiltration, Desalination 62 (1987) 117-136. [10] N.N. Li, A.G. Fane, W. S. Winston Ho, T. Matsuura, Advanced Membrane Technology and Applications, WILEY, 2008. [11] M. Gryta, Fouling in direct contact membrane distillation process, Journal of Membrane Science 325 (2008) 383-394. [12] J. Kim, Z. Cai, M.M. Benjamin, Effect of adsorbents on membrane fouling by natural organic matter, Journal of Membrane Science 310 (2008) 356-364. [13] D. Jermann, W. Pronk, S. Meylan, M. Boller, Interplay of different NOM fouling mechanisms during ultrafiltration for drinking water production, Water research 41(2007) 1713-1722. [14] A.R. Costa, M.N. de Pinho, M.Elimelech, Mechanisms of colloidal natural organic matter fouling in ultrafiltration, Journal of Membrane Science 281 (2006) 716-725. [15] M.O. Lamminen, H.W. Walker, L.K. Weavers, Mechanisms and factors influencing the ultrasonic cleaning of particle-fouled ceramic membranes, Journal of Membrane science 237 (2004) 213-223.
104 [16] L.M. Vane, F.R. Alvarez, Vibrating pervaporation modules: Effect of module design on performance, Journal of Membrane Science 255 (2005) 213-224. [17] J. Kim, Z. Cai, M.M. Benjamin, A serial filtration investigation of membrane fouling by natural organic matter, Journal of Membrane Science 294 (2007) 115-126. [18] Abdolhamid Salahi, Mohsen Abbasi, Toraj Mohammadi, Permeate flux decline during UF of oily wastewater: Experimental and modeling, Desalination 251 (2010) 153–160. [19] T. Kobayashi, X. Chai, N. Fujii, Ultrasound enhanced cross-flow membrane filtration, Separation and Purification Technology 17 (1999) 31-40. [20] L. Wang, X. Wang, K.I. Fukushi, Effect of proposal conditions on ultrafiltration membrane fouling, Desalination 229 (2008) 181-191. [21] Z.F. Cui, S. Chang, A.G. Fane, The use of gas bubbling to enhance membrane processes, Journal of Membrane Science 221 (2003) 1-35. [22] R.S. Juang, K.H. Lin, Flux recovery in the ultrafiltration of suspended solutions with ultrasound, Journal of Membrane Science 243 (2004) 115-124. [23] P. Vaisanen, M.R. Bird, M. Nystrom, Treatment of UF membrane with simple and formulated Cleaning agents, Transition Inistitute of Chemical Engineers. 80 (Part C ) June 2002. [24] W.J.C. van de Ven, I.G.M Punt, A. Zwijinenburg , et al. Hollow fiber ultrafiltration:The concept of partial backwashing, Journal of membrane science 320 (2008) 319-324. [25] L.M. Vane, F.R. Alvarez, E.L. Giroux, Reduction of concentration polarization in pervaporation using vibrating membrane module, Journal of Membrane Science 153 (1999) 233-241.
[26] H. Lee, G. Amy, J. Cho, Y. Yoon, et al. Cleaning Strategies For Flux Recovery Of An Ultrafiltration Membrane Fouling By Natural Organic Matter, Water Research 35 (2001) 3301-3308. [27] T. Kobayashi, T. Kobayashi, Y. Hosaka, N. Fujii, Ultrasound-enhanced membrane- cleaning processes applied water treatment: influence of sonic frequency on filtration treatments, Ultrasonics 41 (2003) 185-190. [28] X. Chai, T. Kobayashi, N. Fujii, Ultrasound-associated cleaning of polymeric membranes for water treatment, Separation and Purification Technology 15 (1999) 139-146. [29] S. Muthukumaran, K. Yang, A. seuren, S. Kentish, M. Ashokkumar, G.W. Stevens, F. Grieser, The use of ultrasonic cleaning for ultrafiltration membranes in the dairy industry, Separation and Purification Technology 39 (2004) 99-107. [30] C. Williams, R.J. Wakeman, Membrane fouling and alternative techniques for its alleviation, Membrane Technology, 124 (2000) 4-14.
105 [31] M. Dekker, M.R. Boom, Improving membrane filtration processes, Trends in Biotechnology, 13 (1995) 129-131.
[32] M.Y. Jaffrin, L.H. Ding, O. Akoun, A. Brou, A hydrodynamic comparison between rotating disk and vibratory dynamic filtration systems, Journal of Membrane science 242 (2004) 155-167. [33] M.D. Petala, A.I. Zouboulis, Vibratory shear enhanced processing membrane filtration applied for the removal of natural organ matter from surface waters, Journal of Membrane science 269 (2006) 1-14. [34] Hyeok Choi, Kai Zhang, Dionysios D. Dionysiou, Daniel B. Oerther, George A. Sorial, Influence of cross-flow velocity on membrane performance during filtration of biological suspension, Journal of Membrane Science 248 (2005) 189-199. [35] Seungkwan Hong,Ron S. Faibish, Menachem Elimelech, Kinetics of permeate flux decline in cross-flow membrane filtration of colloidal suspensions, Journal of Colloid and Interface Science 196 (1997) 267-277.
[36] X.L. Wang, X.F. Li, X.Q. Fu, R. Chen, B. Gao, Effect of ultrasound irradiation on polymeric microfiltration membranes, Desalination 175 (2005) 187-196. [37] Johannes de Bruijna, Alejandro Venegasb, Rodrigo Borquezc, Influence of cross-flow ultrafiltration on membrane fouling and apple juice quality, Desalination 148 (2002) 131 -136.
[38] N. Ghaffoura, M.W. Naceurb, N. Drouichec, H. Mahmoudid, Use of ultrafiltration membranes in the treatment of refinery wastewaters, Desalination and Water Treatment 5 (2009) 159–166.
[39] Abdolhamid Salahia, Toraj Mohammadia, Ali Rahmat Pourb, Fatemeh Rekabdarb, Oily wastewater treatment using ultrafiltration, Desalination and Water Treatment 6 (2009) 289– 298. [40] P. Srijaroonrat, E. Julien* and Y. Aurelle, Unstable secondary oil/water emulsion treatment using ultrafiltration: fouling control by backflushing, Journal of Membrane Science 159, Issues 1-2 (1999) 11-20.
[41] Hong-Jian Li, Yi-Ming Cao, Jian-Jun Qui, Xing-Ming Jie, Tong-Hua Wang, Jian-Hui Liu, Quan Yuan, Development and characterization of anti-fouling cellulose hollo fiber UF membranes for oil-water separation, Journal of Membrane Science 279(2006) 328-335.
[42] B. Chakrabarty, A.K. Ghoshal, M.K.Purkait, Ultrafiltration of stable oil-in-water emulsion by polysulfone membrane, Journal of Membrane Science 325 (2008) 427-237. [43] H. Ohya, J.J. Kim. A. Chinen, M. Aihara, S.I. Semenova, Y. Negishi, O. Mori, M. Yasuda, Effects of pore size on separation mechanisms of microfiltration of oily water, using porous glass tubular membrane, Journal of Membrane Science 145 (1998) 1-14.
106 [44] Mehrdad Hesampour, Agnieszka Krzyzaniak, Marianne Nystrom, The influence of different factors on the stability and ultrafiltration of emulsified oil in water, Journal of Membrane Science 325 (2008) 199-208.
[45] S. Elmaleh, N. Ghaffor, Corss-flow ultrafiltration of hydrocarbon and biological solid mixed suspensions, Journal of Membrane Science 118 (1996) 111-120.
[46] K. Karakulski, A. Kozlowski, A. W. Morawski, Purification of oily wastewater by ultrafiltration, Separations Technology 5 (1995) 197-205.
[47] J.M. Benito, S. Ebel, B. Gutierrez, C. Pazos and J. Coca, Ultrafiltration of a waste emulsified cutting oil using organic membranes, Water, Air, and Soil Pollution 128 (2001) 181-195.
[48] M. Gryta, K. Karakulski, A. W. Morawski, Purification of oily wasteater by hybrid UF/MD, Wat. Res. Vol. 35, No. 15, (2001) 3665–3669.
[49] A. Nel, Removal of Organic Foulants from Capillary Ultrafiltration Membranes by Use of Ultrasound , Diss, University of Stellenbosch, 2005.
[50] Heng Liang, Weijia Gong, Cleaning of fouled ultrafiltration (UF) membrane by algae during reservoir water treatment, Desalination 220 (2008) 267-272.
[51] Keith N. Bourgeous, Jeannie L. Darby and George Tchobanoglous, Ultrafiltration of wastewater: effects of particles, mode of operation and backwash effectiveness, Wat. Res. Vol. 35, No. 1, (2001) 77-90.
[52]J. Paul Chen, S.L. Kim, Y.P. Ting, Optimization of membrane physical and chemical cleaning by a statistically designed approach, Journal of Membrane Science 219 (2003) 27-45. [53] Paul James Smith, Saravanamuth Vigneswaran, Huu Hao Ngo, Roger Ben-Aim, Hung Nguyen, A new approach to backwash initiation in membrane systems, Journal of Membrane Science 278 (2006) 381-389.
[54] Mansoor Kazemimoghadam, Toraj Mohammadi, Chemical cleaning of ultrafiltration membranes in the milk industry, Desalination 204 (2007) 213-218.
[55] H. Peng, A.Y. Tremblay, Membrane regeneration and filtration modeling in treating oily wastewaters, Journal of Membrane Science 324 (2008) 59-66.
[56] Taha Taha, Z.F. Cui, CFD modelling of gas-sparged ultrafiltration in tubular membranes, Journal of Membrane Science 210 (2002) 13-27. [57] Z.F. Cui, K.I.T. Wright, Gas-liquid two-phase cross-flow ultrafiltration of BSA and dextran solutions, Journal of Membrane Science 90 (1994) 183-189. [58] Z.F. Cui, K.I.T. Wright, Flux enhancements with gas sparging in downwards crossflow
107 ultrafiltration: performance and mechanism, Journal of Membrane Science 117 (1996) 109- 116.
[59] S.R. Bellara , Z.F. Cui, D.S. Pepper, Gas sparging to enhance permeate flux in ultrafiltration using hollow fiber membranes, Journal of Membrane Science 121 (1996) 175- 184.
[60] R. Ghosh, Z.F. Cui, Mass transfer in gas-sparged ultrafiltration: upward slug flow in tubular membranes, Journal of Membrane Science 162 (1999) 91-102.
[61] C. Cabassud, S. Laborie, J.M. Laine, How slug flow can improve ultrafiltration flux in organic hollow fibers, Journal of Membrane Science 128 (1997) 93-101.
[62] M. Mercier, C. Fonade and C. Lafforgue-Delorme, How slug flow can improve ultrafiltration flux in mineral tubular membranes, Journal of Membrane Science 128 (1997)103-113.
[63] Pandit, A.B. and Moholkar , V.V. (1996) Harness Cavitation to Improve Processing. Chemical Engineering Progress, July, 57-69.
[64] Pall corporation (2006). Membrane chemical cleaning: From art to science. Available : http://www.pall.com/pdf/mtcpaper.pdf, accessed Sept. 5,2006.
[65] Mason, T.J. and Cordemans, E.D. (1996) Ultrasonic Intensification of Chemical Processing and Related Operations: A review. Transactions of the Institution of Chemical Engineers, 74, Part A, 511-516.
[66] A. Simon, L. Penpenic, N. Gondrexon, S. Taha, G. Dorange, A comparative study between classical stirred and ultrasonically-assisted dead-end Ultrafiltration, Ultrasonics Sonochemistry 7 (2000) 183-186.
[67] Plesset, M.S 1966, Shockwaves from Cavity Collapse, Phil. Trans. Royal Soc., vol. 260A (1996), p. 241.
[68] Markov, L. O., and Rosenberg, L. D., 1957, On the Mechanism of Ultrasonic Cleaning, Soviet Physics-Acoustics, vol.3, p. 403.
[69] Isabelle Masselin, Xavier Chasseray, Laurence Durand-Bourlier, Jean-Michel Lainé, Pierre-Yves Syzaret and Daniel Lemordant, Effect of sonication on polymeric membranes, Journal of Membrane Science 181 (2001) 213-220.
[70] Xiao-li Wang, Xiu-fen Li, Xue-qi Fu, Rui Chen , Bo Gao, Effect of ultrasound irradiation on polymeric microfiltration membranes, Desalination 175 (2005) 187-196.
[71] R.W. Field, D. Wu, J.A. Howell, B.B. Gupta, Critical Flux Concept for Microfiltration Fouling, Journal of Membrane Science 100 ( 1995) 250-272.
108
[72] S.T.D. de Barros, C.M.G. Andrade, E.S. Mendes, L. Peres, Study of fouling mechanism in pineapple juice clarification ultrafiltration, Journal of Membrane Science 215 (2003) 213- 224.
[73] Abdolhamid Salahi, Ali Gheshlaghi, Toraj Mohammadi, Sayed Siavash Madaeni, Experimental performance evaluation of polymeric membranes for treatment of an industrial oily wastewater, Desalination 262 (2010) 235-242.
[74] Toraj Mohammadi, M. Kazemimoghadam, M. Saadabadi, Modeling of membrane fouling and flux decline in reverse osmosis during separation of oil in water emulsions, Desalination 157 (2003) 369-375.
[75] A. L. Lim, Renbi Bai, Membrane fouling and cleaning in microfiltration of activated sludge wastewater, Journal of Membrane Science 216 (2003) 279-290.
[76] M. Cinta Vincent Vela, Silvia Alvarez Blanco, Jaime Lora Garcia, Enrique Bergantinos Rodriguez, Analysis of membrane pore blocking models applied to the ultrafiltration of PEG, Separation and purification Technology 62 (2008) 489-498.
[77] http://www.spectrumlabs.com/filtration/Edge.html
[78] http://www.inge.ag/index_en.php
[79] http://www.separationdynamics.com/
[80] http://www.attension.com/?id=1092&cid
[81] Nina Zhou, Parametric study of ultrafiltration membrane system & development of fouling control mechanism, Diss. Purdue University, August 2010.
[82] J. Hermia, Transactions of the Institution of Chemical Engineers 60 (1982) 183–187.
[83] H. Susanto, Y. Feng, M. Ulbricht, Fouling behavior during ultrafiltration of aqueous solutions of polyphenolic compounds, J. Food Eng. 91 (2009), 333-340.
[84] T. Mohammadi, A. Esmaeelifar, Wastewater treatment of a vegetable oil factory by a hybrid ultrafiltration-activated carbon process, Journal of Membrane Science 254 (2005) 129– 137. [85] K.J. Hwang, C.Y. Liao, K.L. Tung, Effect of membrane pore size on the particle fouling in membrane filtration Desalination 234 (2008) 16–23.
[86] A.L. Lim, R. Bai, Membrane fouling and cleaning in microfiltration of activated sludge wastewater, J. Membr. Sci. 216 (2003) 279–290.
[87]K.-J.Hwang, T.-T. Lin, Effect of morphology of polymeric membrane on the performance
109 of cross-flow microfiltration, J. Membr. Sci. 199 (2002) 41–52.
[88] Huang, L., Morrisey, M. T. (1998). Fouling of Membranes During Microfiltration of Surimi Wash Water: Roles of Pore Blocking and Surface Cake Formation. Journal of Membrane Science, 144, 113. [89] Shuji Nakatsuka, Ichiro Nakate, Tadaaki Miyano, Drinking water treatment by using ultrafiltration hollow fiber membranes, Desalination 106 (1996) 55-61.
[90] P. Blanpain and M. Lalande, Investigation of fouling mechanisms governing permeate flux in the cross flow microfiltration of beer, Journal of Filtration & Separation Volume 34, Issue 10, December 1997, Pages 1065-1069. [91] Taha Taha, Z.F. Cui, CFD modelling of gas-sparged ultrafiltration in tubular membranes, Journal of Membrane Science 210 (2002) 13–27.
[92] Ho-Ming Yeh and Jin-Hong Dong, Further Analysis of Permeate Flux for Membrane Ultrafiltration Along a Solid-Rod Tubular Module, Tamkang Journal of Science and Engineering, Vol. 6, No. 1, pp. 1-7 (2003).
[93] Yeh, H. M.; Chen, H. Y.; Chen, K. T, Membrane ultrafiltration in a tubular module with a steel rod inserted concentrically for improved performance. Journal of Membrane Science 168 (2000) 121-133.
[94] Yeh, H. M.; Wu, H. H, Membrane ultrafiltration in combined hollow-fiber module systems , Journal of Membrane Science 124 (1997) 93-105.
[95] Yeh, H. M.; Tsai, J. W, Membrane ultrafiltration in multipass hollow-fiber modules , Journal of Membrane Science 142 (1998) 61-73.
[96] Dong, J. F. Permeate Flux Analysis for Membrane Ultrafiltration, M. S. Thesis, Tamkang University, Tamsui, Taiwan, R.O.C. 2001.
110 Appendix
Appendix 1: Raw Data
Appendix 2: EPA Analytical Methods - Method 1664, Revision A: N-Hexane Extractable Material (HEM; Oil and Grease) and Silica Gel Treated N-Hexane Extractable Material (SGT-HEM; Non-polar Material) by Extraction and Gravimetry
111 Appendix1: Raw Data
Transmembrane Pressure (kPa)
186.16 236.40 241.30 248.20 255.10 262.00 268.90 275.80 282.70
Time(hr) Permeate Flux (l/m2.h)
0.05 12.54 15.64 17.55 17.91 18.63 19.40 20.30 20.12 20.00
0.50 11.64 14.39 16.54 16.42 16.82 17.31 18.51 18.21 17.91
1.00 11.10 13.79 15.52 15.64 15.90 16.20 17.01 16.83 16.66
1.50 10.33 12.95 14.69 15.04 15.06 15.40 16.18 16.00 15.70
2.00 9.67 12.54 13.85 14.27 14.51 14.74 15.34 14.92 15.10
2.50 9.37 11.94 13.31 13.43 13.70 13.91 14.57 14.51 14.39
3.00 9.19 11.64 13.01 13.37 13.55 13.79 14.45 14.39 14.27
3.50 9.13 11.34 12.89 13.25 13.37 13.73 14.39 14.21 14.21
4.00 9.13 11.34 12.89 13.25 13.37 13.73 14.33 14.21 14.21
Experimental Results shown in Figure 5.1
Transmembrane Pressure (kPa)
186.16 236.40 241.30 248.20 255.10 262.00 268.90 275.80 282.70 289.60
Time(hr) Permeate Flux (l/m2.h)
0.00 11.46 13.31 14.33 15.16 15.64 16.12 16.54 17.25 17.79 17.61
0.50 10.69 12.12 13.07 13.79 14.15 14.57 15.04 15.34 16.12 15.82
1.00 10.03 11.16 12.06 12.72 13.07 13.55 13.85 14.15 14.92 14.45
1.50 9.49 10.39 11.28 11.76 12.24 12.60 12.83 13.07 13.97 13.55
2.00 8.89 9.79 10.51 11.16 11.52 11.82 12.18 12.42 13.07 12.66
2.50 8.48 9.49 9.85 10.51 10.86 11.16 11.46 11.94 12.36 12.00
3.00 8.30 9.19 9.55 10.03 10.39 10.63 10.98 11.58 11.82 11.58
3.50 8.06 8.95 9.25 9.73 10.03 10.21 10.57 11.10 11.64 11.34
4.00 7.94 8.83 9.13 9.55 9.79 10.09 10.33 11.04 11.46 11.16
4.50 7.94 8.66 9.01 9.43 9.67 9.97 10.21 10.98 11.34 10.98
Experimental Results shown in Figure 5.2
112
Transmembrane Pressure (kPa) premeate flux(l/m2.h)
282.70 19.88
275.80 19.88
268.91 19.90
262.01 19.70
255.12 19.16
248.22 18.51
241.33 17.91
234.43 17.13
Experimental Results shown in Figure 5.3
Transmembrane Pressure (kPa) Permeate Flux (lmh)
234.43 14.03
241.33 14.69
248.22 15.16
255.12 15.64
262.01 16.12
268.91 16.54
275.80 17.25
282.70 17.70
289.59 17.67
296.49 17.67
303.38 17.67
Experimental Results shown in Figure 5.4
113
With Surfactant Without Surfactant
TMP=186.16kPa TMP=282.7kPa TMP=186.16kPa TMP=282.7kPa
Time(hr) Permeate Flux (l/m2.h)
0.05 12.54 20.00 11.46 17.79
0.5 11.64 17.91 10.69 16.12
1 11.10 16.66 10.03 14.92
1.5 10.33 15.52 9.49 13.97
2 9.67 14.98 8.89 13.07
2.5 9.37 14.39 8.48 12.36
3 9.19 14.27 8.30 11.82
3.5 9.13 14.21 8.06 11.64
4 9.13 14.21 7.94 11.46
Experimental Results shown in Figure 5.5
With Surfactant Without Surfactant Constant Feed Variable Feed Constant Feed Variable Feed Concentration Concentration Concentration Concentration
Time(hr) Permeate Flux (l/m2.h)
0.05 20.30 20.00 17.79 17.79
0.50 18.39 17.91 16.12 15.40 1.00 16.83 16.12 14.92 13.55 1.50 15.82 14.63 13.97 11.76
2.00 15.04 13.61 13.07 10.45
2.50 14.57 12.89 12.36 9.37
3.00 14.21 12.36 11.82 8.66
3.50 13.97 11.88 11.46 8.18
4.00 13.73 11.70 11.10 7.70
Experimental Results shown in Figure 5.6
114
With Surfactant Without Surfactant
60min Backwash 90min Backwash 60min Backwash 90min Backwash Interval Interval Interval Interval
Time(hr) Permeate Flux (l/m2.h)
0 19.94 17.73 20.36 17.61
0.5 18.51 16.18 18.80 15.94
1 16.95 14.86 17.43 14.86
1.5 15.58 13.91 16.30 13.79
2 14.69 13.01 15.16 13.13
2.5 13.73 11.94 14.15 12.24
3 12.83 11.28 13.07 11.58
3.5 12.83 11.10 12.77 11.10
3.55 19.88 17.19 20.00 17.07
4.5 16.71 14.33 17.01 14.45
4.55 19.76 16.71 16.77 14.27
5 17.91 15.22 15.52 13.19
5.05 17.79 15.10 19.52 16.54
5.5 16.42 14.15 17.85 15.28
5.55 19.28 16.12 17.25 14.80
6 17.73 15.28 16.24 13.79
6.5 16.30 13.91 15.46 13.01
Experimental Results shown in Figure 5.7
115
With Surfactant Without Surfactant
100s Backwash 200s Backwash 100s Backwash 200s Backwash Duration Duration Duration Duration
Time (hr) Permeate Flux (l/m2.h)
0 19.94 20.00 17.73 17.55
0.5 18.51 18.92 16.12 15.82
1 16.95 17.61 14.74 14.98
1.5 15.94 16.42 13.91 13.97
2 14.80 15.16 12.95 13.13
2.5 13.73 14.15 12.24 12.12
3 12.83 13.07 11.58 11.16
3.5 12.83 13.07 11.10 11.04
3.53 19.76 20.06 16.71 17.07
4.5 17.97 18.63 14.92 15.64
4.53 19.34 19.88 16.48 16.71
5.5 17.67 18.27 14.63 15.22
5.53 18.98 19.40 15.82 16.12
6.5 17.37 17.91 14.15 14.74
7 16.36 16.89 13.31 13.85
7.5 15.28 15.82 12.42 12.89
8 14.09 14.74 11.76 12.24
Experimental Results shown in Figure 5.8
With Transmembrane Pressure (kPa) Surfactant
186.16 236.40 241.30 248.20 255.10 262.00 268.90 275.80 282.70 289.60
Error Reduction (%) 6.78 7.81 11.24 11.25 17.27 14.12 16.97 15.10 15.87
Results Obtained from modeling for Porous Membrane, shown in Figure 5.11
116
Without Transmembrane Pressure (kPa) Surfactant
186.16 236.40 241.30 248.20 255.10 262.00 268.90 275.80 282.70 289.60
Error Reduction (%) 1.03 6.21 1.07 1.01 2.32 3.22 2.84 8.73 8.06 6.52
Results Obtained from modeling for Porous Membrane, shown in Figure 5.12
Transmembrane Pressure (kPa)
186.2 236.4 241.3 248.2 255.1 262.0 268.9 275.8 282.7 289.6
Without Relative 90.35 89.44 89.62 89.82 86.00 84.38 84.42 83.20 83.53 82.97 Surfactant Contribution Percentage (%) 9.65 10.56 10.38 10.18 14.00 15.62 15.58 16.80 16.47 17.03 of Fouling With Mechanism 89.72 90.13 90.01 90.07 85.59 84.04 82.70 81.39 81.23 Surfactant 10.28 9.87 9.99 9.93 14.41 15.96 17.30 18.61 18.77
Results Obtained from modeling for Porous Membrane, shown in Figure 5.15
Transmembrane Pressure (kPa)
137.9 144.8 151.7 158.60 165.5 172.4 179.3
Time (hr) Permeate Flow Rate (Gallon/min)
0.01 0.38 0.44 0.49 0.54 0.65 0.7 0.67
0.5 0.3 0.36 0.38 0.42 0.48 0.54 0.5
1 0.24 0.28 0.31 0.34 0.38 0.44 0.41
1.5 0.21 0.246 0.28 0.30 0.35 0.4 0.38
2 0.19 0.24 0.27 0.29 0.34 0.39 0.36
2.5 0.19 0.24 0.27 0.29 0.324 0.385 0.35
3 0.19 0.24 0.27 0.29 0.33 0.385 0.35
Experimental Results shown in Figure 5.16
117
Transmembrane Pressure (kPa)
131 137.9 144.8 151.7 158.6 165.5 172.4
Time (hr) Permeate Flow Rate (Gallon/min)
0.01 0.31 0.4 0.48 0.55 0.64 0.72 0.66
0.5 0.26 0.33 0.4 0.43 0.52 0.55 0.5
1 0.22 0.3 0.35 0.38 0.44 0.47 0.43
1.5 0.2 0.27 0.335 0.36 0.41 0.44 0.4
2 0.197 0.263 0.33 0.35 0.4 0.435 0.39
2.5 0.195 0.26 0.325 0.347 0.398 0.435 0.39
3 0.193 0.259 0.325 0.347 0.398 0.435 0.39
Experimental Results shown in Figure 5.17
Transmembrane Pressure (kPa)
137.9 144.8 151.7 158.60 165.5 172.4
Time (hr) Permeate Flow Rate (Gallon/min)
0.01 0.51 0.63 0.7 0.82 0.77 0.74
0.5 0.44 0.51 0.56 0.65 0.6 0.58
1 0.4 0.43 0.46 0.54 0.52 0.49
1.5 0.38 0.41 0.438 0.52 0.495 0.47
2 0.37 0.399 0.435 0.51 0.483 0.46
2.5 0.36 0.4 0.43 0.5 0.474 0.45
3 0.364 0.4 0.43 0.5 0.474 0.445
Experimental Results shown in Figure 5.18
118 Transmembrane Pressure (kPa)
137.9 144.8 151.7 158.60 165.5 172.4
Time (hr) Permeate Flow Rate (Gallon/min)
0.01 0.48 0.57 0.66 0.77 0.75 0.7
0.5 0.41 0.46 0.51 0.62 0.6 0.57
1 0.36 0.41 0.45 0.53 0.51 0.48
1.5 0.33 0.375 0.41 0.49 0.47 0.44
2 0.32 0.36 0.4 0.482 0.463 0.435
2.5 0.31 0.35 0.4 0.47 0.457 0.428
3 0.31 0.35 0.4 0.475 0.45 0.42
Experimental Results shown in Figure 5.19
Transmembrane Permeate Flux Pressure (kPa) (l/m2.h) 137.90 6.68
144.80 7.74
151.70 8.97
158.60 10.20
165.50 11.43
172.40 12.31
179.30 11.78
Experimental Results shown in Figure 5.20
119 Transmembrane Permeate Flux Pressure (kPa) (l/m2.h)
131.00 5.63
137.90 7.03
144.80 8.44
151.70 9.67
158.60 11.25
165.50 12.66
172.40 11.61
Experimental Results shown in Figure 5.21
Transmembrane Permeate Flux Pressure (kPa) (l/m2.h)
137.90 8.79
144.80 11.08
151.70 13.01
158.60 14.42
165.50 13.54
Experimental Results shown in Figure 5.22
Transmembrane Permeate Flux Pressure (kPa) (l/m2.h)
137.90 8.44
144.80 10.02
151.70 11.61
158.60 13.54
165.50 13.19
Experimental Results shown in Figure 5.23
120
With Surfactant Without Surfactant
unsparged sparged unsparged sparged
Transmembrane pressure (kPa) Permeate Flux (l/m2.h)
137.90 7.03 8.97 6.51 8.44
144.80 8.44 11.08 7.74 10.02
151.70 9.67 13.01 8.62 11.61
158.60 11.25 14.95 9.50 13.54
Experimental Results shown in Figure 5.24
Time (hr) Permeate Flow Rate ( Gallon/min)
0.01 0.7 0.72 0.82 0.77
0.5 0.54 0.56 0.65 0.62
1 0.43 0.47 0.54 0.52
1.5 0.4 0.44 0.52 0.485
2 0.39 0.435 0.518 0.482
2.5 0.385 0.435 0.518 0.48
3 0.385 0.435 0.518 0.48
Experimental Results shown in Figure 5.25
Ultrafiltration with Dense Membrane Mass Transfer Coefficient (m/s)
Without Surfactant/Without Air 5.83E-04
With Surfactant/Without Air 9.76 E-04
With Surfactant/With Air 12.3 E-04
Without Surfactant/With Air 10.8 E-04
Results shown in Figure 5.26
121 Appendix2: EPA Analytical Methods - Method 1664, Revision A: N-Hexane Extractable Material (HEM; Oil and Grease) and Silica Gel Treated N-Hexane Extractable Material (SGT- HEM; Non-polar Material) by Extraction and Gravimetry
EPA-821-R-98-002; February 1999
A.2.1. Introduction:
Method 1664 is a performance-based method applicable to aqueous matrices that requires the use of n-hexane as the extraction solvent and gravimetry as the determinative technique. Alternative extraction and concentration techniques are allowed, provided that all performance specifications are met. In addition, QC procedures designed to monitor precision and accuracy have been incorporated into Method 1664.
A.2.2. Summary of Method:
A 1-L sample is acidified to pH <2 and serially extracted three times with n-hexane in a separatory funnel. The extract is dried over sodium sulfate. The solvent is distilled from the extract and the HEM is desiccated and weighed. If the HEM is to be used for determination of SGT-HEM, the HEM is redissolved in n-hexane. For SGT-HEM determination, an amount of silica gel proportionate to the amount of HEM is added to the solution containing the redissolved HEM to remove polar materials. The solution is filtered to remove the silica gel, the solvent is distilled, and the SGT-HEM is desiccated and weighed. Quality is assured through calibration and testing of the extraction, distillation, and gravimetric systems.
A.2.3. Definitions:
HEM and SGT-HEM are method-defined analytes; i.e., the definitions of both HEM and SGT- HEM are dependent on the procedure used. The nature of the oils and/or greases, and the presence of extractable non-oily matter in the sample will influence the material measured and interpretation of results. Definitions for terms used in this method are given in the glossary at the end of the method.
A.2.4. Procedure:
This method is entirely empirical. Precise and accurate results can be obtained only by strict adherence to all details.
NOTE: The procedure below is based on the preparation, extraction, and analysis of a 1-L sample. If a smaller volume is collected for analysis, the laboratory should dilute the sample to 1 L with reagent water so that results across the IPR, blank, OPR, MS, and, if performed, the MSD, are consistent. It is also important that all glassware surfaces be rinsed with n-hexane to effect a quantitative transfer of the constituents in the sample and of the hexadecane/stearic acid in the IPR, OPR, MS, and, if performed, the MSD.
122 A.2.4.1. Preparation of the analytical batch:
Bring the analytical batch of samples, including the sample aliquots for the MS (and MSD), to room temperature. Place approximately 1000 mL (950-1050 mL) of reagent water in a clean sample bottle to serve as the laboratory blank. Prepare the OPR using the PAR standard Either mark the sample bottle at the water meniscus or weigh the bottle for later determination of sample volume. Weighing will be more accurate. Mark or weigh the MS (and MSD).
A.2.4.2. pH verification:
A.2.4.2.1 Verify that the pH of the sample is less than 2 using the following procedure:
Dip a glass stirring rod into the well mixed sample. Withdraw the stirring rod and allow a drop of the sample to fall on or touch the pH paper.
NOTE: Do not dip the pH paper into the bottle or touch it to the sample on the lid.
Rinse the stirring rod with a small portion of n-hexane that will be used for extraction (to ensure that no extractable material is lost on the stirring rod). Collect the rinsate in the separatory funnel to be used for sample extraction.
A.2.4.2.2 If the sample is at neutral pH, add 5-6 mL of HCl or H2SO4 solution to the 1-L sample. If the sample is at high pH, use a proportionately larger amount of HCl or H2SO4 solution. If a smaller sample volume was collected, use a proportionately smaller amount of HCl or H2SO4 solution.
A.2.4.2.3 Replace the cap and shake the bottle to mix thoroughly. Check the pH of the sample using the procedure in Section 11.2.1. If necessary, add more acid to the sample and retest.
A.2.4.2.4 Add the appropriate amount of HCl or H2SO4 solution to the blank, OPR, MS (and MSD) to adjust the pH of these solutions to <2.
NOTE: The procedure detailed below is for separatory funnel liquid-liquid extraction. Solid-phase extraction (SPE) may be used at the discretion of the discharger/generator and its laboratory. However, if SPE is used, it is the responsibility of the discharger/generator and laboratory to assure that results produced are equivalent to results produced by the procedure below.
A.2.4.3. Extraction:
A.2.4.3.1 Tare a clean boiling flask containing 3-5 boiling chips as follows:
Place the flask containing the chips in an oven at 105-115 C for a minimum of 2 h to dry the flask and chips. Remove from the oven and immediately transfer to a desiccator to cool to room temperature. When cool, remove from the desiccator with tongs and weigh immediately on a calibrated balance (Section 10).
A.2.4.3.2 Pour the sample into the separatory funnel.
123 A.2.4.3.3 Add 30 mL of n-hexane to the sample bottle and seal the bottle with the original bottle cap. Shake the bottle to rinse all interior surfaces of the bottle, including the lid of the bottle cap. Pour the solvent into the separatory funnel.
A.2.4.3.4 Extract the sample by shaking the separatory funnel vigorously for 2 minutes with periodic venting into a hood to release excess pressure.
A.2.4.3.5 Allow the organic phase to separate from the aqueous phase for a minimum of 10 minutes. If an emulsion forms between the phases and the emulsion is greater than one-third the volume of the solvent layer, the laboratory must employ emulsion-breaking techniques to complete the phase separation. The optimum technique depends upon the sample, but may include stirring, filtration through glass wool, use of solvent phase separation paper, centrifugation, use of an ultrasonic bath with ice, addition of NaCl, or other physical methods. Alternatively, solid-phase extraction (SPE), continuous liquid-liquid extraction, or other extraction techniques may be used to prevent emulsion formation, provided that the requirements in Section 9.1.2 are met.
A.2.4.3.6 Drain the aqueous layer (lower layer) into the original sample container. Drain a small amount of the organic layer into the sample container to minimize the amount of water remaining in the separatory funnel.
NOTE: The amount of water remaining with the n-hexane must be minimized to prevent dissolution or clumping of the sodium sulfate in the solution drying process.
A.2.4.3.7 Place a filter paper in a filter funnel adds approximately 10 g of anhydrous Na2SO4, and rinse with a small portion of n-hexane. Discard the rinsate.
NOTE: The specific properties of a sample may necessitate the use of larger amounts of Na2SO4.
A.2.4.3.8 Drain the n-hexane layer (upper layer) from the separatory funnel through the Na2SO4 into the pre- weighed boiling flask containing the boiling chips.
NOTE: It is important that water be removed in this step. Water allowed to filter through the Na2SO4 will dissolve some of the Na2SO4 and carry it into the boiling flask compromising the determination.
A.2.4.3.9 Repeat the extraction twice more with fresh 30-mL portions of n-hexane, combining the extracts in the boiling flask.
A.2.4.3.10 Rinse the tip of the separatory funnel, the filter paper, and the funnel with 2-3 small (3-5 mL) portions of n-hexane. Collect the rinsings in the flask.
NOTE: For samples that are expected to contain a high concentration of salt (e.g., waters from oil production facilities), it may be prudent to collect the extract in a 250-mL separatory funnel and back-extract with reagent water. After back-extraction, the extract should be drained through Na2SO4 to remove all traces of water.
A.2.4.3.11 A milky extract indicates the presence of water. If the extract is milky, allow the solution to stand for up to one hour to allow the water to settle. Decant the solvent layer (upper layer) through sodium sulfate to remove any excess water as in Sections A.2.4.3.7 and A.2.4.3.8. Rinse the glassware and sodium sulfate with small portions of n-hexane to affect a quantitative transfer.
A.2.4.3.12 If only SGT-HEM is to be determined, proceed to Section 11.5.
124 A.2.4.4. Solvent distillation:
A.2.4.4.1 Connect the boiling flask to the distilling head apparatus and distill the solvent by immersing the lower half of the flask in a water bath or a steam bath. Adjust the water temperature as required to complete the concentration in less than 30 minutes. Collect the solvent for reuse.
A.2.4.4.2 When the temperature in the distilling head reaches approximately 70C or the flask appears almost dry, remove the distilling head. Sweep out the flask for 15 seconds with air to remove solvent vapor by inserting a glass tube connected to a vacuum source. Using tongs, immediately remove the flask from the heat source and wipe the outside surface dry to remove moisture and fingerprints.
NOTE: The laboratory should carefully monitor the flask during the final stages of distillation to assure that all of the solvent is removed and to prevent loss of the more volatile sample constituents.
A.2.4.4.3 Inspect the residue in the boiling flask for crystals. Crystal formation is an indication that sodium sulfate may have dissolved and passed into the boiling flask. This may happen if the drying capacity of the sodium sulfate is exceeded or if the sample is not adjusted to low pH. If crystals are observed, redissolve the extract in n-hexane, quantitatively transfer through a filter into another tared boiling flask, and repeat the distillation procedure.
A.2.4.4.4 Dry the boiling flask for 30 - 45 minutes in an oven maintained at 70 ± 2 C. Cool to room temperature in a desiccator and maintain in the desiccator for 30 minutes minimum. Remove with tongs and weigh immediately. Repeat the cycle of drying, cooling, desiccating, and weighing until the weight loss is less than 4 % of the previous weight or less than 0.5 mg, whichever is less.
If the extract was from the HEM procedure, determine the HEM (Wh) by subtracting the tare weight from the total weight of the flask. If the extract was from the SGT-HEM procedure , determine the weight of SGT-HEM (Ws) by subtracting the tare weight from the total weight of the flask.
A.2.4.4.5 Determine the original sample volume (Vs) in liters by filling the sample bottle to the mark with water and measuring the volume of water in a 1- to 2-L graduated cylinder. If the sample weight was used , weigh the empty bottle and cap and determine Vs by difference, assuming a sample density of 1.00.
A.2.4.5. SGT-HEM determination:
A.2.4.5.1 Silica gel capacity--To ensure that the capacity of the silica gel will not be exceeded, the amount of HEM must be less than 100 mg or, if above 100 mg, must be known.
If it is known that the amount of HEM is less than 100 mg, the laboratory may proceed with the determination of SGT-HEM per Sections A.2.4.5.3-A.2.4.5.5 without determination of HEM. If, however, the amount of HEM is not known, HEM must first be determined using the procedure in Sections A.2.4.3-A.2.4.4.
A.2.4.5.2 Extractable materials in silica gel--Because the capacity of silica gel is not known for all substances, it is presumed that 3 g will normally adsorb 100 mg of all adsorbable materials. Therefore, for samples containing 1000 mg HEM, 30 g of silica gel will be needed. The amount of silica gel that can be used for adsorption in the SGT-HEM procedure below has been limited to 30 g because of concerns about possible extractable impurities in the silica gel. If the amount of HEM in the sample is greater than 1000 mg, split the extract per the following procedure:
Add 85-90 mL of n-hexane to the boiling flask to redissolve the HEM. If necessary, warm the solution to completely redissolve the HEM.
125 Quantitatively transfer the extract to a 100-mL volumetric flask. Dilute to the mark with n-hexane. Calculate the extract volume that contains 1000 mg of extractable material according to the following equation:
1000Vt Va Wh Eq.A.2.1
Vt is the total volume of solvent used in section A.2.5.5 (ml) Wh is the weight of extractable material HEM measurement (mg) Va is the volume of aliquat to be withdrawn (ml) Using a calibrated pipet, remove the volume to be withdrawn (Va) and return to the boiling flask. Dilute to approximately 100 mL with n-hexane. A.2.4.5.3 Adsorption with silica gel
Add 3.0 ± 0.3 g of anhydrous silica gel to the boiling flask for every 100 mg of HEM, or fraction thereof, to a maximum of 30 g of silica gel. For example, if the weight of HEM is 735 mg, add 3 x 8 = 24 g of silica gel. Add a fluoropolymer-coated stirring bar to the flask and stir the solution on a magnetic stirrer for a minimum of 5 minutes.
A.2.4.5.4 Filter the solution through n-hexane moistened filter paper into a pre-dried, tared boiling flask containing several boiling chips. Rinse the silica gel and filter paper with several small amounts of n-hexane to complete the transfer.
A.2.4.5.5 Distill the solution and determine the weight of SGT-HEM per Section A.2.4.4.
A.2.5 Data Analysis and Calculations:
A.2.5.1 n -Hexane extractable material--Calculate the concentration of HEM ("oil and grease") in the sample per the following equation:
W (mg) HEM(mg/L) h Eq.A.2.2 Vs(L)
Wh is the weight of extractable material from section A.2.4.4 Vs is sample volume from section A.2.4.5
A.2.5.2 Silica gel treated n-hexane extractable material--Calculate the concentration of SGT-HEM ("non-polar material") in the sample per the equation above, substituting Ws for Wh. If the extract was split to decrease the total amount of material to 1,000 mg, determine the corrected total weight of SGT-HEM in the un-split extract (Wc) using the following equation:
126 Vt We (mg) Wd (mg) Eq.A.2.3 Va
We is the weight in the portion of the extract split for adsorption from section A.2.5.2 and A.2.4.4 Vt and Va are defined in Eq.A.2.1
Use the corrected total weight of SGT-HEM in the unsplit extract (Wc) to determine the total SGT-HEM in the sample by substituting Wc for Wh in Equation 5. A.2.5.3 Reporting--Report results to three significant figures for HEM and SGT-HEM found at or above 10 mg/L, and report results to two significant figures for HEM and SGT-HEM found below 10 mg/L.
Samples--Report results for HEM and SGT-HEM found below the ML as < 5.0 mg/L, or as required by the permitting aut hority or permit. Blanks--Report results for HEM and SGT-HEM found below the MDL as < 1.4 mg/L, or as required by the permitting authority or permit. Do not report results below the MDL unless required by the permitting authority or permit. Results from tests performed with an analytical system that is not in control must not be reported or otherwise used for permitting or regulatory compliance purposes but do not relieve a discharger or permittee of timely reporting.
A.2.6. Glossary of Definitions and Purposes:
The definitions and purposes are specific to this method but have been conformed to common usage to the extent possible.
A.2.6.1 Units of weight and measure and their abbreviations
Symbols C: degrees Celsius <: less than %: percent ±: plus or minus Alphabetical g: gram h: hour L: liter mg: milligram mg/g: milligram per gram mg/L: milligram per liter mg/mL: milligram per milliliter mL: milliliter No. : Number rpm: revolutions per minute Definitions, acronyms, and abbreviations Analyte: The HEM or SGT-HEM determined by this method.
127 Analytical batch: The set of samples started through the extraction process in a 12-hour shift, to a maximum of 20 field samples. Each analytical batch of 20 or fewer samples must be accompanied by a laboratory blank (Section 9.4), an ongoing precision and recovery sample (OPR, Section 9.6), and a matrix spike, (Section 9.3), resulting in a minimum of four analyses (1 sample, 1 blank, 1 OPR, and 1 MS) and a maximum of 23 analyses (20 field samples, 1 blank, 1 OPR, and 1 MS) in the batch. If greater than 20 samples are to be extracted in a 12- hour shift, the samples must be separated into analytical batches of 20 or fewer samples.
Discharge (matrix type): A sample medium with common characteristics across a given industrial subcategory (40 CFR parts 403-500). For example, C-stage effluents from chlorine bleach mills in the Pulp, Paper, and Paperboard industrial category; effluent from the Continuous Casting subcategory of the Iron and Steel industrial category; publicly owned treatment work (POTW) sludge; and in-process streams in the Atlantic and Gulf Coast Hand-shucked Oyster Processing subcategory are each a matrix type.
Field blank: An aliquot of reagent water that is placed in a sample container in the laboratory or in the field and treated as a sample in all respects, including exposure to sampling site conditions, storage, preservation, and all analytical procedures. The purpose of the field blank is to determine if the field or sample transporting procedures and environments have contaminated the sample.
HEM: See n-Hexane extractable material. n-Hexane extractable material: Material that is extracted from a sample and determined by this method (oil and grease). This material includes relatively non-volatile hydrocarbons, vegetable oils, animal fats, waxes, soaps, greases, and related matter.
IPR: See initial precision and recovery.
Initial precision and recovery (IPR): Four aliquots of the diluted PAR analyzed to establish the ability to generate acceptable precision and accuracy. An IPR is performed the first time this method is used and any time the method is modified.
Laboratory blank (method blank): An aliquot of reagent water that is treated exactly as a sample including exposure to all glassware, equipment, solvents, reagents, internal standards, and surrogates that are used with samples. The laboratory blank is used to determine if analytes or interferences are present in the laboratory environment, the reagents, or the apparatus.
Laboratory control sample (LCS): See Ongoing precision and recovery standard (OPR).
Matrix spike (MS) and matrix spike duplicate (MSD): Aliquots of an environmental sample to which known quantities of the analytes are added in the laboratory. The MS and MSD are prepared and/or analyzed exactly like a field sample. Their purpose is to quantify any additional bias and imprecision caused by the sample matrix. The background concentrations of the analytes in the sample matrix must be determined in a separate aliquot and the measured values in the MS and MSD corrected for background concentrations.
May: This action, activity, or procedural step is neither required nor prohibited.
May not: This action, activity, or procedural step is prohibited.
Method Detection Limit: The lowest level at which an analyte can be detected with 99 percent confidence that the analyte concentration is greater than zero.
Minimum Level (ML): The lowest level at which the entire analytical system gives a recognizable signal and acceptable calibration point for the analyte. It is equivalent to the concentration of the lowest calibration standard, assuming that all method-specified sample weights, volumes, and cleanup procedures have been employed.
Must: This action, activity, or procedural step is required.
128
Ongoing precision and recovery standard (OPR, also called a laboratory control sample): A laboratory blank spiked with known quantities of analytes. The OPR is analyzed exactly like a sample. Its purpose is to assure that the results produced by the laboratory remain within the limits specified in this method for precision and accuracy.
OPR: See ongoing precision and recovery standard.
PAR: See precision and recovery standard.
Precision and recovery standard: Secondary standard that is diluted and spiked to form the IPR and OPR.
Quality control sample (QCS): A sample containing analytes of interest at known concentrations. The QCS is obtained from a source external to the laboratory or is prepared from standards obtained from a different source than the calibration standards. The purpose is to check laboratory performance using test materials that have been prepared independently from the normal preparation process.
Quantitative transfer: The process of transferring a solution from one container to another using a pipet in which as much solution as possible is transferred, followed by rinsing of the walls of the source container with a small volume of rinsing solution (e.g., n-hexane), followed by transfer of the rinsing solution, followed by a second and third rinse and transfer.
Reagent water: Water demonstrated to be free from HEM and SGT-HEM and potentially interfering substances at or above the minimum level of this method.
Regulatory Compliance Limit: A limit on the concentration or amount of a pollutant or contaminant specified in a nationwide standard, in a permit, or otherwise established by a regulatory authority. SGT-HEM: See Silica gel treated n-hexane extractable material.
Should: This action, activity, or procedural step is suggested but not required.
Silica gel treated n-hexane extractable material: Components of n-Hexane extractable material (HEM) that are not adsorbed by silica gel; i.e., non-polar material (NPM).
Stock solution: A solution containing an analyte that is prepared using a reference material traceable to EPA, the National Institute of Science and Technology (NIST), or a source that will attest to the purity and authenticity of the reference material.
A.2.7. References:
1. "Methods for Chemical Analysis of Water and Wastes," 3rd Edition, Environmental Protection Agency, Environmental Monitoring Systems Laboratory-Cincinnati (EMSL-Ci), Cincinnati, Ohio 45268, EPA- 600/4-79-020, Method 413.1, (1983). 2. Ibid., Method 418.1 3. Guidelines Establishing Test Procedures for the Analysis of Oil and Grease and Non-polar Materials; Final Rule; Preamble, Responses to Comments, and Docket, as referenced in the Final Rule. 4. "Carcinogens - Working With Carcinogens," Department of Health, Education, and Welfare, Public Health Service, Center for Disease Control, National Institute for Occupational Safety and Health, Publication No. 77-206, August 1977. 5. "OSHA Safety and Health Standards, General Industry," (29 CFR 1910), Occupational Safety and Health Administration, OSHA 2206 (Revised, January 1976). 6. "Safety in Academic Chemistry Laboratories," American Chemical Society, Committee on Chemical Safety, 3rd Edition, 1979. 7. "Standard Practices for Sampling Water," ASTM Annual Book of Standards, Part 31, D3370-76, American Society for Testing and Materials, 1916 Race Street, Philadelphia, PA 19103-1187, 1980.
129 8. "Handbook of Analytical Quality Control in Water and Wastewater Laboratories," USEPA, EMSL-Ci, Cincinnati, OH 45268, EPA-600/4-79-019, March 1979. 9. Report of the Method 1664 Validation Studies, April 1995. Available from the Sample Control Center (operated by DynCorp I&ET), 6101 Stevenson Avenue, Alexandria, VA 22304, (703) 461-2100.
130