CHARACTERIZATION OF THE MIXING OF WHEAT STRAW SLURRIES THROUGH ELECTRICAL RESISTANCE TOMOGRAPHY (ERT) by: HIVA MOVAFAGH

...................................................................................................................... iii ACKNOWLEDGMENTS.................................................................................................. iv TABLE OF CONTENTS ................................................................................................... vi LIST OF TABLES ............................................................................................................... x LIST OF FIGURES............................................................................................................ xi INTORDUCTION................................................................................................................ 1 1. LITERATURE REVIEW ............................................................................................. 3 1.1 Wheat Straw ......................................................................................................... 3 1.1.1 Ethanol Production from Wheat Straw ............................................................ 4 1.2 Wheat Straw Suspension Rheology ................................................................... 10 1.3 Fiber Networks ................................................................................................... 14 1.4 Power Consumption ........................................................................................... 15 1.4.1 Power Requirements for Mixing of non-Newtonian Fluids ........................... 16 vii 1.5 Impeller Type ..................................................................................................... 17 1.5.1 Impeller Clearance ......................................................................................... 19 1.5.2 Impeller Diameter .......................................................................................... 20 1.6 Fiber Suspension Mixing ................................................................................... 20 1.6.1 Effect of Fiber Size ........................................................................................ 21 1.6.2 Effect of Fiber Concentration ......................................................................... 22 1.7 Cavern ................................................................................................................ 23 1.7.1 Laser Doppler Anemometry (LDA) ............................................................... 23 1.7.2 Ultrasonic Doppler Velocimetry (UDV) ........................................................ 24 1.7.3 Positron Emission Particle Tracking (PEPT) ................................................. 25 1.7.4 X-Ray Technique ........................................................................................... 25 1.7.5 Electrical Resistance Tomography (ERT) ..................................................... 26 1.7.6 Numerical Model ........................................................................................... 26 1.8 Cavern Mathematical Models ............................................................................ 27 1.8.1 The Spherical Model ...................................................................................... 27 1.8.2 The Cylindrical Model (Elson’s Model) ........................................................ 28 1.8.3 The Torus-shaped Model ............................................................................... 29 1.9 Research Objectives ........................................................................................... 30 2. Electrical Resistance Tomography (ERT) .................................................................. 31 2.1 Principal elements of the ERT system ............................................................... 32


INTORDUCTION
Mixing of non-Newtonian fluids in stirred vessels is a common operation encountered in the chemical, biochemical, pharmaceutical, polymer, mineral, food, and wastewater treatment industries. A main group of non-Newtonian fluids exhibits the shear-thinning behaviour with yield stress (Skelland, 1967), such as pulp suspensions, certain polymers, biopolymer solutions, and wastewater sludge . The shear-thinning fluids have low apparent viscosities at high shear rates and high apparent viscosities in low shear rates.
The design of the mixing systems for non-Newtonian fluids is more challenging than for Newtonian fluids due to the complex rheology of non-Newtonian fluids. Mixing of fibrous suspensions, which are non-Newtonian fluids with a yield stress, causes a wellmixed zone around the impeller called cavern with stagnant regions elsewhere (Wichterle and Wein, 1975). The stagnant regions within the mixing vessel result in ineffective heat and mass transfer (Amanullah et al., 1997). To improve the mixing efficiency of a reactor, it is always beneficial to eradicate such undesired stagnant zones in the mixing of non-Newtonian fluids with yield stress.
The size of the cavern has significant effect on the mixing quality. Numerous experimental techniques, such as laser Doppler anemometry (LDA), X-ray photography, hotwired anemometry (HWA), planner laser induction (Arratia et al., 2006), ultrasonic Doppler velocimetry (UDV), positron emission particle tracking (PEPT) and electrical resistance tomography (ERT), and the numerical technique called computational fluid dynamics (CFD) have been applied to study the formation of cavern generated in the mixing of yield stress fluids.
In the literature, there seems to be no study on the cavern evaluation of wheat straw suspensions. The recent and rapid development of non-intrusive ERT seems to provide an efficient tool for the analysis and control of mixing processes especially in the case of non-Newtonian fluids (either opaque or transparent). Electrical resistance tomography (ERT), which is an emerging technology, can be the best alternative to measure the cavern size and to analyze fluid flow in 2D and 3D.
This study will try to employ ERT to study the mixing of wheat straw slurries, which are yield-pseudoplastic fluids, with the A200, A310, and A100 impellers in terms of cavern shape and size. In this study, for the first time, the yield stress of the wheat straw slurries was estimated by applying the cavern size obtained from ERT tomograms.

Wheat Straw
Wheat straw is an agricultural residue made of stalks, stem, and dried leaves of the wheat plant. It contains about 58-78 wt % total sugars depending on the species (Himmel et al., 2007).
International wheat production statistics from years 2000 to 2010 show that wheat straw is the largest biomass feed stock in Europe (FAOSTAT Database, 2011). More than 21% of the world's food depends on the wheat crop, so, in order to satisfy the great demand of human's need, the production of wheat crop need to be increased.
Wheat straw is a complex mixture of three main components: cellulose, hemicellulose and lignin. Generally, wheat straw contains of 33-40 (wt%) of cellulose, 20-25 (wt%) of hemicellulose, and 15-20 (wt%) of lignin (Prasad et al., 2007). Cellulose is a linear monopolymer chain comprised of D-glucose units attached to each other via -1 4-glycosidic bonds. Hemicellulose consists of long chain of xylose molecules, etc. Lignin is an amorphous polymer consisting of phenylpropane units, and it has three aromatic alcohols (monolignols) precursors of p-coumaryl, coniferyl and sinapyl alcohols, which are shown in Figure 1.1 (Buranov and Mezza , 2008). Lignocellulosic materials are a good source for ethanol production (Taherzadeh and Karimi, 2007). The main objective of this chapter is to conduct a literature review on the mixing performance in wheat straw, which is applied in ethanol production. This literature review includes the ethanol production process, wheat straw suspension rheology, power consumption, impeller type, and cavern formation; finally, some ERT applications in mixing are presented.

Ethanol Production from Wheat Straw
Wheat straw is the most attractive low cost feedstock for production of fuel alcohol due to its abundance, renewability and low lignin content (Buranov and Mezza, 2008).

Pre-treatment
Pre-treatment is a significant process in the utilization of lignocellulosic material to achieve large amounts of fermentable sugars. Pre-treatment is performed to break the lignin seal, decrease crystallinity of the cellulose and solubilize hemicellulose (Patel et al., 2009). Pre-treatment techniques are categorized in four groups of physical, chemical, physicochemical, and biological.
Physical pre-treatment: chipping, milling and grinding are main types of physical pretreatments. These treatments reduce both the crystallinity and the degree of polymerization (DP) of cellulose. In addition, they increase the specific surface area (Sun and Cheng, 2002).
Chipping reduces the biomass size to 10-30 mm while grinding and milling can decrease the particle size to 0.2-2 mm. Particle size and cellulose crystallinity are reduced in grinding and milling more than chipping probably because of the shear forces created during milling and grinding. Increasing specific surface area and reducing both cellulose crystallinity and degree of polymerization (DP) are all dependent on the type and duration of milling as well as the type of biomass.
Chemical pre-treatment: chemicals such as acids, alkali, cellulose / organic solvents, and ionic liquids can affect lignocellulosic biomass structure (Swatloski et al., 2002). In acid pretreatment (typically H 2 SO 4 ), most of the hemicellulose is removed and converted to soluble sugars (Gil et al., 2010;Guo et al., 2008). Alkali pre-treatment such as NaOH, KOH, Ca(OH) 2 , hydrazine and anhydrous ammonia can break the lignin structure, increase the internal surface of biomass and reduce both the degree of polymerization and crystallinity (Chandra et al., 2007;Chang and Holtzapple, 2000;Galbe and Zacchi, 2007). The most effective chemical pre-treatment for biomass with low lignin content such as wheat straw is alkali pre-treatment (Agbor et al., 2011). Combined cellulose/organic solvent pre-treatment separates lignocellulose components by using a cellulose solvent such as concentrated phosphoric acid, organic solvent like acetone, and water (Zhang et al., 2007;Zhu et al., 2009). Ionic liquids (ILs) have been used as novel non-derivatizing media for the dissolution of carbohydrates including cellulose (Swatloski et al., 2002;Zhao et al., 2008). Many ILs, especially those containing halide, acetate, formate, and phosphate anions, dissolve cellulose quickly (Ohno and Fukaya, 2009).
Biological pre-treatment: fungi are used in biological pre-treatment. Fungi are able to produce enzymes that degrade lignin, hemicellulose, and polyphenols. White and soft-rot fungi are capable of degrading lignocellulose material. However, brown rot fungi can degrade cellulose structure. White-rot fungi are known as the most effective biological pre-treatment (Lee, 1997;Sun and Cheng, 2002). This method is not used in industry due to several factors such as the need for careful growth environment, a residence time of 10-14 days and the need of large biological reactors (Agbor et al., 2011).

Enzymatic Hydrolysis
Cellulose slowly degrades into glucose by a group of enzymes (cellulases) with various functions during enzymatic hydrolysis of pre-treated lignocellulosic biomass. Cellulase refers to a group of enzymes produced by fungi, bacteria, and protozoans that catalyze cellulolysis (i.e. the hydrolysis of cellulose) (Watanabe et al., 2010). All these enzymes can hydrolyze cellulose by creating new attainable sites for each other. The incomplete conversion of cellobiose to glucose and inhibition of cellobiohydrolase (CBH) are happening because Trichoderma reesei secretes low levels of β-glucosidase enzyme.
Therefore, in order to obtain complete conversion of cellobiose to glucose, it is always required to use supplemental β-glucosidase, such as Novozyme 188 as shown in Table 1.1 (Zhang et al., 2010;Kumar and Wyman, 2009).

Acid Hydrolysis
Dilute acid hydrolysis is also another way to cleave ether bonds between hemicellulose and lignin complexes. Dilute acid (0.75% H 2 SO 4 ) at 120-180 C can both solubilize hemicellulose and convert solubilized hemicellulose to fermentable sugars (Grohmann et al., 1985;Saha et al., 2005). Studies show that during the organosolv and aqueous ethanol delignification of wheat straw, a major factor in lignin breakdown is the cleavage of -aryl and -aryl ether linkages in lignin precursors. The cleavage of -aryl ether bonds is faster than that of -aryl ether bonds when acid catalyst is added (Papatheofanous et al., 1998;Xu et al., 2006;Hongzhang and Liying, 2007). Experiments show that during organosolv pulping of wheat straw in presence of formic-acetic acid-water (30/60/10) system, more than 94.1% of original lignin is removed (Xu et al., 2006). Hydrolyze the xylan and make the cellulose more accessible to cellulase Zhang et al., 2010 ;Kumar and Wyman, 2009;Berlin et al., 2007 Cellulase+β-glucosidase+pectinases Remove the pectin that coat cellulose fibers Zhang et al., 2010 andBerlin et al., 2007 Cellulase+xylanase+βxylosidase Hydrolyze the xylan and eliminate the inhibition of xylobiose and higher xylooligomers Kumar and Wyman, 2009 Cellulase+endoxylanase+α-L-arabinofuranosidase Remove the arabinofuranosyl group that limits the access of xylanases to xylan backbone Sorensen et al., 2007 andAlvira et al., 2011

Fermentation
Ethanol fermentation is a biological process in which sugars such as glucose, xylose, and sucrose are converted into cellular energy and thereby produces ethanol and carbon dioxide as metabolic waste products. Several studies have been done to develop efficient pentose (especially xylose) fermenting microorganisms, such as recombinant Saccharomyces cerevisiae, Zymomonas mobilis and Escherichia coli strains Yonge et al., 2010).  This transport is significant because it appears through nonspecific hexose transporters.
Recently, some heterologous xylose transporters have been found in recombinant S.
cerevisiae. These transporters such as Gxf1 (Hagerdal, 2009), Sut1  are used to increase the absorption of xylose. Ethanologenic microorganisms are useful to convert cellulosic biomass into ethanol (Figure 1.2). There are two ways to convert cellulosic biomass into ethanol. The first way is to use both cellobiose transporter and intracellular β-glucosidase into microorganisms (Galazka et al., 2010) while the second way is to produce extracellular β-glucosidase from ethanologenic microorganisms (Nakamura et al., 2008).

Wheat Straw Suspension Rheology
One important factor in the mixing of non-Newtonian fluids is the rheological behaviour of the fluid (Pakzad et al., 2008). Rheology is the study of deformation and flow of matter, especially in soft solids and liquids with a complex flow behavior (Viamajala et al., 2009). The rheological behavior is significant in the pulp and paper and in the biomass processing industries.
Some studies have been done so far on the rheological properties of cellulose fiber suspensions (Chaussy et al., 2011;Derakhshandeh et al., 2010;and Hui et al., 2009) and also pre-treated lignocellulosic materials, especially corn stover (Primenova et al., 2004).
Yield stress is one of the most important rheological properties of pulp suspensions in designing process equipment for the pulp and paper industry (Derakhshandeh et al., 2011). The yield stress of a fiber suspension is a key parameter in pulp manufacturing since the suspension flows once the yield stress is overcome.
Rheological studies on the slurries of pre-treated corn stover and various pulp fibers have also shown that the yield stress is a function of the fiber mass concentration through a power-law relationship as shown in Equation (1.1) (Lavenson et al., 2011) where a and b are empirical constants. Lavenson et al. (2011) compared fiber types to observe the direct effect of fiber properties on rheology. Bennington et al. (1990) measured the yield stress of pulp and synthetic fiber suspensions using an equation based on network theory, which included fiber aspect ratio and Young's modulus: where A is the fiber aspect ratio, E is the fiber's Young's modulus, c is a constant and C v is the volume concentration of the pulp suspension.


Chen et al. (2002) studied the flow behavior of pulp suspensions and identified three flow regimes. In the first regime, Newtonian flow was observed at low shear rates. Unstable flow was found in the second region, with jumps in the shear stress dependent on shear rate. In the third region, Newtonian behavior at high shear rates was observed, which was called dynamic equilibrium zone. This kind of behavior requires a flexible mixing system to cover the wide range of rheological behavior. With increasing hydrolysis temperature and decreasing particle size, the yield stress decreased. They believed that with solids concentration up to a point, yield stress increased and then became independent of concentration at the high concentrations. The reason for the plateau at high concentrations is not clear.
The study of pre-treated lignocellulose (sawdust slurries) by Dasari and Berson (2007) showed that the rheology of pre-treated lignocellulose is affected by the fiber size. They reported that the yield stress of the slurry increased with an increase in fiber length. Rosgaard et al. (2007) investigated the effect of solids content and enzymatic hydrolysis on the apparent viscosity of barley straw biomass slurries, with solids mass fraction varying from 5 wt% to 15 wt%. It was proved in this study that the apparent viscosity increased with solids mass fraction, and decreased with time during enzymatic hydrolysis. Bashir (2008) measured the rheological properties of wheat straw suspensions at concentrations between 5.0-20.0 wt%. Wheat straw fibers were made by grinding wet or dry wheat straw and were divided into four sizes: 8, 12, 20, and 40 mesh. The yield stress was found to increase with concentration as well as with the size of the wheat straw fibers as shown in Table 1.2. The yield stress data were fitted to the power law equation proposed by Kerekes et al. (1985) for pulp fibers. It was found that only the yield stress of the slurry of 40 mesh size wheat straw fibers agreed with this power law correlation.

Fiber Networks
The large aspect ratio of pulp fiber brings significant contact among fibers (Derakhshadeh et al., 2011).This has a strong effect on suspension rheology. For fibers to form flocs or coherent networks, every fiber must be in contact with at least three other fibers (Dodson, 1996;Meyer and Wahren, 1964;Soszynski and Kerekes, 1988a). This contact regime has been described by a crowding number (N), which is the average number of fibers in a spherical volume swept out by the length of the single fiber (Kerekes et al., 1985).
The concentration of fiber suspensions can be characterized by the crowding number (Kerekes and Schell, 1992;Dodson, 1996). Samaniuk et al. (2011) study showed that cellulose fibers in water form networks that give rise to an apparent yield stress, especially at high solids contents.

Power Consumption
Power consumption is a significant parameter for mixing system. It can be defined as the energy transformed from an impeller to a fluid per unit time (Tatterson, 1991). Having knowledge about power consumption is very important in order to operate the impeller ( Chhabra and Richardson, 1999). The impeller power in a homogeneous liquid depends on the geometry of impeller and tank, the density and viscosity of the fluid, the impeller speed and the gravitational force as shown in the following equation (Tatterson, 1991): where P,µ, ρ, N, D, T, and g are power, fluid viscosity, fluid density, impeller speed, impeller diameter, tank diameter, and gravitational acceleration, respectively. Applying the dimensional analysis, Equation (1.4) is obtained (Skelland, 1967): This equation shows that the dimensionless power coefficient P/ρN 3 D 5 depends on both Reynolds number and Froude number for a fluid. The dimensionless groups are as follows:

Power Requirements for Mixing of non-Newtonian Fluids
The key objective of any mixing process is to maximize the degree of homogeneity of a property such as concentration, viscosity, color, and temperature (Chhabra and Richardson, 2008). For a stirrer system and for a non-Newtonian fluid, the Reynolds number is defined as Equation (1.8): (1.8) In non-Newtonian fluids, apparent viscosity depends on the shear rate. Shear rate decreases with the distance from the impeller. It may fall to zero in stagnant zones (Gabelle et al., 2011). Metzner and Otto (1957) believed that the fluid motion near the impeller could be explained by averaged shear rate, , which is linearly related to the rotational speed of the impeller as in equation (1.9): The constant k s, depends on the impeller and tank configuration. The value of k S reported by Metzner and Otto (1957) was 13 for a disk flat-blade turbine. Skelland (1967) reported the experimental values for k S for different kinds of impellers such as turbines, propellers, paddles, and anchors. In case of pseudoplastic fluids, for most of the impeller types, k S was in the range of 10-13, while for anchors and helical ribbons larger values of 25-30 was reported (Bakker and Gates, 1995).

Impeller Type
The selection of a suitable impeller is a critical design parameter for mixing processes utilized in chemical industries (Tahvildarian, 2010).
Mixing is used in the process industries to achieve different objectives such as blending miscible liquids in reactors, dispersing of gases or immiscible liquids into a liquid phase, and suspension of solids. Impellers related to the generated flow are classified into three groups of axial-flow, radial-flow, and close-clearance impellers.
Axial-flow impellers such as marine propellers and hydrofoil impellers discharge flow along the axis of the impeller as shown in Figure 1.3a (Oldshue, 1983). These types of impellers are used for blending, solid-liquid suspension, and heat transfer (Paul et al., 2004).
Radial-flow impellers such as disk turbines (Rushton) and hollow blade turbines (Scaba) release fluid to the wall of the tank in radial direction as shown in Figure 1.3b (Oldshue, 1983). Radial flow impellers exert shear stresses to the fluid to remove the boundary layer between various phases such as the mixing of immiscible fluids (Thring and Edwards, 1990).
Close-clearance impellers such as anchors, helical ribbons, and helical screws are used in high viscosity applications (Chhabra and Richardson, 2008). Patel et al. (2012) suggested that the rapidly emerging Maxblend impeller is more efficient than conventional closeclearance impellers for mixing of Newtonian and non-Newtonian fluids. Fradette et al.

Impeller Clearance
The distance between the bottom of the tank and the impeller is called clearance.
A large clearance creates a huge vortex, which results in air entrainment into the system and also causes the splash of fluid around (Paul et al., 2004). When an impeller is placed very close to the bottom of the vessel, the axial-flow pattern created with the downward pumping impeller is similar to the radial-flow pattern. In this situation, the bottom of the vessel becomes clear from the suspension of the particles and it leads to the reduction of the level of homogeneity in whole vessel.

Impeller Diameter
Impeller to tank diameter ratio (D/T) is a significant ratio in mixing tank design.
The diameter of the impellers which produce bulk motion such as helical ribbons, screws, and anchors should be close to the tank diameter (Tatterson, 1991).

Fiber Suspension Mixing
Mixing is a process which increases the degree of homogeneity and the rate of mass transfer (Paul et al., 2004;Bhole and Bennington, 2010). To optimize the fiber suspension mixing process, many design parameters such as the type of impeller, impeller clearance, impeller diameter, impeller speed, fiber concentration and fiber size must be , and was ground in a juice blender. The results showed that the feeding time of pre-treated corn straw using the helical impeller was at least 2 hours shorter than that using the Rushton impeller. Thus, the shorter feeding time indicated that the mixing using the helical impeller was better than that using the Rushton impeller.

Effect of Fiber Size
The size of fiber has a prominent effect on the mixing performance. Samaniuk et al.
(2011) measured the rheological properties of concentrated lignocellulose biomass, corn stover. They observed that the yield stress increased with the fibre length. Chaussy et al.
(2011) examined the rheological behavior of cellulose fiber suspensions; they also observed that the suspension viscosity increased with the fiber length. Viamajala et al.
(2009) assessed the rheology of acid hydrolyzed corn stover and reported that the yield stress decreased with a decrease in fiber size. Bashir (2008) measured the rheological properties of wheat straw slurries at different concentrations. Based on his results, the yield stress increased with the size of the wheat straw fibers. Dasari and Berson (2007) examined the rheology of the corn stover and observed that with increasing fiber size, the yield stress increased.

Effect of Fiber Concentration
The literature review showed that the fiber concentration has a significant effect on the

Cavern
The formation of cavern around the impeller is a characteristic of the mixing of viscoplastic fluids with yield stress such as pulp suspensions, polymers, and ceramic pastes. This class of fluids has a relatively high apparent viscosity at low shear rates.
Therefore, the region around the impeller has intensive motion while elsewhere in the mixing tank is almost stagnant (Wichterle and Wein, 1975). The presence of the stagnant zones in non-Newtonian fluids are harmful to the mixing process since it leads to poor heat and mass transfer (Amanullah, 1997). Therefore, it is important to determine the cavern size and shape in these mixing operations (Adams and Barigou, 2007).

Laser Doppler Anemometry (LDA)
This technique was first developed by Yeh and Cummins in 1964. LDA is used for measuring velocity in a non-intrusive manner by using seeded particles conveyed by a fluid flow (Drain, 1980). This is based on the measurement of laser light scattered by particles that pass through a pattern of light and dark surface. LDA measures the velocity at one point or in a very small volume, which is formed by the cross section from twolaser beams (Chaouki et al., 1997). Laser Doppler anemometry has been used to characterize the flow structure in stirred vessels since late 1970s (Durst et al., 1976;Drain, 1980). The earlier works on caverns, using LDA, had considerable practical value to industry Jaworski et al. 1993). However, some of the fundamental concepts and the definition of the cavern boundary based on LDA measurements (Hirata et al., 1991) require reconsideration. The spatial and temporal resolution for this technique is very high and it can cover a huge range of flow velocities (magnitude and direction). Hirata et al. (1994) measured the cavern size using LDA in a shear-thinning plastic fluid agitated by the Rushton turbine. They observed the cavern shape as a circular cylinder.

Ultrasonic Doppler Velocimetry (UDV)
Ultrasonic Doppler Velocimetry (UDV) is a non-invasive fluid flow measurement technique, which has been used for the measurement of fluid velocity profiles (Williams, 1986;and McClements et al., 1990). Pulsed ultrasound echography and detection of the instantaneous Doppler shift frequency are applied in UDV to measure fluid velocity (Takeda, 1991). This technique gives an exact determination of the flow curves of complex fluids such as pulp suspensions (Derakhshandeh et al., 2010).

Positron Emission Particle Tracking (PEPT)
Positron emission particle tracking (PEPT) is a relatively new technique allowing the quantitative study of flow phenomena in three dimensions in opaque systems that could not be studied by optical methods such as laser Doppler anemometry (LDA). Chiti et al. (2011) showed that the PEPT technique could be used to obtain accurate velocity data throughout the entire complex three-dimensional turbulent flow field in an agitated baffled vessel. The PEPT technique was successfully employed to visualize the cavern formed during the mixing of a slurry using a 250 m neutral density tracer (Adams et al., 2008).

X-Ray Technique
The X-ray technique is utilized to visualize the flow patterns in the mixing of opaque fluids. Elson and Cheesman (1986) used the X-ray technique to study the flow patterns and cavern size during the mixing of opaque fluids exhibiting yield stress. Absorption of what happens when only a fraction of the radiation passes through an absorber (a heavy metal tracer). Therefore, a number of photons are lost in the absorption process. In order to use X-ray, an absorber should be added to the mixing tank. Elson (1986) measured dimensions of the cavern created around a disc turbine impeller in the mixing of xanthan gum solutions and characterized the shape of the cavern using a right circular cylinder.

Electrical Resistance Tomography (ERT)
ERT is a non-invasive imaging technique that measures the distribution of conductivity within a region of interest (Mann et al., 1997). This technique has been used in several studies, such as the formation of cavern in the mixing of pseudoplastic fluids (Pakzad et al., 2008), investigation of mixing processes (Kim et al., 2006), solid-liquid filtration processes (Vlaev et al., 2000), multiphase processes like solid-liquid (Lucas et al., 1999;Recard et al., 2005), gas-liquid (Wang et al., 2000), and liquid-liquid (Kaminoyama et al., 2007).

Cavern Mathematical Models
Several mathematical models were developed over the years to predict the cavern size as a function of mixing conditions and fluid properties, such as spherical (Solomon et al., 1981), cylindrical (Elson et al., 1986 and1988;Aoshima, 1994 and and toroidal (Amanullah et al., 1998;. In all cases, the models were developed for the isolated cavern that does not make contact with the vessel walls (Hui et al., 2009).
For the models mentioned above, researchers considered different flow regimes inside the cavern as well as different forces (tangential and axial), and used different rheological models such as Herschel-Bulkley and Bingham plastic. Solomon et al. (1981)

The Spherical Model
where D C is the cavern diameter, D is the impeller diameter, P o is the power number, N is the impeller speed, ρ is fluid density and τ y is fluid yield stress. The term N 2 D 2 ρ/τ y on the right hand side of the equation (1.10) is called the yield stress Reynolds number, and is usually shown as Re y .

The Cylindrical Model (Elson's Model)
The cylindrical model was proposed by Elson et al. (1986 and1988  where in this equation, D c /D is dimensionless cavern diameter, N 2 D 2 ρ/τ y is the yield stress Reynolds number, and H c is the cavern height, and P 0 is the power number. Amanullah et al. (1998) reported that both spherical and cylindrical models predicted the cavern diameter equally well. However, they believed that the cylindrical model was a better representation of the cavern shape.

The Torus-shaped Model
A mathematical (axial force) model was created by Amanullah et al. (1998). This model assumes entire momentum passed on by the impeller as the sum of both tangential and axial shear components, transmitted to the cavern boundary by the pumping action of the impeller. The equation given by Amanullah et al. (1998) is as follows: where N f = F a /ΡN 2 D 4 is the dimensionless axial force number, F a is the axial force imparted by the impeller and P is the power. This model can also be applied to the caverns generated by radial flow impellers if N f = 0, as follows (Bhole and Bennington, 2010): In the literature, there seems to be no study on the cavern evaluation of wheat straw suspensions. The recent and rapid development of non-intrusive ERT seems to provide an efficient tool for the analysis and control of mixing processes, especially in the case of non-Newtonian fluids (either opaque or transparent). Electrical resistance tomography (ERT), which is an emerging technology, can be the best alternative to measure the cavern size and analyze fluid flow in 2D and 3D.

Research Objectives
From the literature review, one can say that the information regarding mixing of wheat straw slurry is still inadequate. This study will try to employ ERT to study the mixing of this pseudoplastic fluid with axial impellers (A100, A200, and A310). The main contributions of this work are:  To employ Electrical Resistance Tomography (ERT) to predict the cavern dimensions of the cavern formed around the impellers  To study the effects of wheat straw slurry concentration on the cavern size  To study the effects of wheat straw fiber size on the cavern size  To understand the flow field generated by the A100, A200, and A310 impellers.

Electrical Resistance Tomography (ERT)
As described in chapter 1, many techniques are available to measure the cavern size.
However, those techniques have their own restrictions such as limitation in working with opaque fluids and changing the local flow field. To solve these issues, a non-intrusive technique called electrical resistance tomography (ERT) can be utilized to measure the cavern size.
ERT is a measurement technique which is used in several studies such as the investigation of mixing processes (Kim et al., 2006), examination of solid-liquid filtration processes (Vlaev et al., 2000), and also observation of multiphase processes such as solid-liquid (Lucas et al., 1999;Recard et al., 2005), gas-liquid (Wang et al., 2000) and liquid-liquid (Kaminoyama et al., 2007).

Principal elements of the ERT system
The objective in electrical tomography technique is to measure electrical signals sent from sensing electrodes and to reconstruct the conducting properties of the fluid inside of the mixing vessel (William et al., 1993;Holden et al., 1998;and Tahvildarian et al., 2011).
The goal of ERT system is to obtain the resistance distribution in the domain of interest.
The ERT system consists of three main components: the sensors, the data acquisition system (DAS) and the image reconstruction system.

Sensing System
Electrodes are the heart of the ERT system. They must be in continuous contact with the fluid inside the vessel. The position of the electrodes is important since the reconstruction algorithm is based on the electrodes being located at exactly defined intervals in order to outline the maximum amount of information from inside of the vessel (Kaminoyama et al., 2007). The size of the electrodes is another important factor in measuring the electric field distribution (Mann et al., 1996).
In this method, multiple electrodes are located around the wall of the mixing vessel (Zhao et al., 2008;Pakzad et al., 2008). An electrical current is applied to two adjacent electrodes and the voltages are measured between the other adjacent electrode pairs (Williams and Beck, 1995). To receive accurate measurement, the electrodes must be more conductive than the fluid (Tahvildarian et al., 2011). The electrodes, which are located around the boundary of the vessel, make the electrical contact with the fluid inside the vessel and are connected to the data acquisition system (DAS) by co-axial cables to reduce electromagnetic noise and interference (Dickin and Wang, 1996).
Electrodes can be fabricated from gold, platinum, stainless steel, brass, or silver (Paulson et al., 1992).

Data Acquisition System (DAS)
A data acquisition system (DAS) is connected to the electrodes and communicates with the host image reconstruction computer.
DAS is responsible for signal measurement, filter and control, demodulation, multiplexer control, waveform generation, synchronization, and power supply (Holden et al., 1998). It is necessary to select the scheme that has a good distinguishability and high sensitivity to conductivity changes in the fluid.   The digital "stir-case" generators are used to generate staircase wave. A high-speed digital to analogue converter (DAC) converts the digital pattern to analogue and subsequently filters to reduce unwanted harmonics. The major part of DAS is voltage generator. The output of this part is a sine wave voltage that is sent to a voltage-to-current convertor. Multiplexers (MUX) are necessary to share the current source and voltage measurement stages between any numbers of electrodes (Beck et al., 1993). The data acquisition system is responsible to perform the desired measurement protocol.
As can be seen in Figure 2.2, an adjacent measurement protocol is used to apply an AC current between two adjacent electrodes and to measure the voltage between all other pairs of adjacent electrodes. The AC current is then applied to the next pair of electrodes and the voltage is measured for all other electrode pairs (Barber et al., 1983).  (Pakzad, 2007)

Data Collection Strategies
There are four main strategies to examine conductivity distribution in a tank and to gain maximum information, as follow:  Adjacent strategy  Opposite strategy  Diagonal strategy  Conducting boundary strategy

The Adjacent Strategy
This strategy is used for sensors with insulating boundaries. In this strategy, current is applied through two neighbouring electrodes and the voltage differences are measured using all other pairs of neighbouring electrodes. To repeat this process, the current is applied to all other possible pairs of neighbouring electrodes. In this strategy, the total number of independent measurements (M) achieved is given by equation (2.1): where N is number of electrodes.
This strategy entails minimal hardware to implement and image reconstruction. It is very delicate to measure error and noise, based on the non-uniformity of the current distribution and the low current density at the center of the vessel (Mann et al., 1997;Kaminoyama, 2005).

The Opposite Strategy
In this strategy, current is applied through diametrically opposed electrodes. The voltage reference is that to the electrode adjacent to the current-injecting electrode. The voltages are measured with respect to the reference, at all electrodes except the current-injecting ones. The whole process is repeated in the clockwise direction until all independent measurements have been made. In this case, the total number of independent measurements M is calculated as follows (Viergever and Todd-Pokropek, 1988): Compared to the adjacent strategy, the opposite strategy is less sensitive to conductivity changes at the boundary because most of the current runs through the central part of the region (Hua et al., 1993). The opposite strategy has less image resolution than the adjacent strategy, due to the reduced number of independent current projections (Abdullah, 1993).

The Diagonal Strategy
In this strategy, a current is applied between electrodes separated by large dimensions.
Electrodes 1 and 2 are fixed as the current reference and the voltage reference, respectively. Then, the current is applied to electrodes 3, 5, 7, etc., and the voltage is measured from electrodes to the left of electrode number 2. Then the current reference and voltage reference are changed to electrode 4 and electrode 3, respectively. Sending current through electrodes 2, 6, 8, etc., the voltage is measured on all other electrodes except the current-injecting ones. This strategy, compared with the adjacent strategy, has a good sensitivity over the entire vessel and gives better quality (Hua et al., 1993).

The Conducting Boundary Strategy
In this strategy, both current and voltage are used between two electrodes. The main benefit of this strategy is its low common-mode voltage component. The large surface area of the conducting boundary is used as the current sink to reduce the common-mode voltage across the electrodes doing the measurement (Gisser et al., 1987).
The common-mode voltage for conducting boundary strategy is greatly reduced compared with that of the adjacent strategy (Dickin and Wang, 1996).

Image Reconstruction System
The adjacent strategy provides 104 individual voltage measurements for each plane with 16 electrodes according to the equation n e (n e -3)/2, where n e is the number of electrodes per plane. Eventually, the DAS communicates the quantitative data to the host image reconstruction computer, where data are processed using a suitable image reconstruction algorithm. An image reconstruction algorithm is applied to determine the interior distribution of the resistance in the process vessel after gaining the measurements from a set of electrodes sitting on the border of a vessel (Pinheiro et al., 1999;Mann et al., 1996). There are two types of algorithms: non-iterative and iterative. A non-iterative image reconstruction algorithm (Linear Back Projection-LBP) is used to convert a voltage measurement to conductivity values of each plane (Barber and Brown, 1984). The linear back projection algorithm, compared to an iterative image reconstruction algorithm, has a low computational requirement (Wang, 2002). The iterative algorithm has high computational requirement and is too slow for real-time image reconstruction compared to the non-iterative algorithm (Madupu et al., 2005).
The P2000 system (Industrial Tomography Systems-ITS, Manchester, UK), used in this study, comes with a qualitative, non-iterative algorithm on linear back-projection.

Material and Methods
In this study, we assessed the mixing of aqueous wheat straw slurries of 5 wt%, 7 wt%, and 10 wt% concentration. For each test, the required amount of the wheat straw was loaded into the mixing vessel and the tap water was added to reach the desired slurry height (400 mm) inside the tank.
In this study, experiments were conducted using three impellers at many impeller speeds, and for two fiber sizes.   The mixing tank was equipped with a top-entering impeller driven by a 2-hp motor. The impeller rotational speed was varied using a variable frequency drive. The impeller torque and speed were measured using a rotary torque meter (Staiger Mohilo, Germany) and a tachometer, respectively. In this study three impellers were used (A100, A200, A310), each with a 180 mm diameter (D). Each impeller was located on plane P3 with an offbottom clearance of 160 mm.  This impeller generates an axial-flow pattern. Normally this impeller is used for low to medium viscosity flow.

Power Measurements
The power input to the impeller (P) was obtained from torque (M) and impeller rotational speed (N) measurements using: Impeller torque and speed were measured using a torque sensor (Staiger Mohilo, Germany). The bearing friction was measured by operating the system with an empty vessel. This friction torque was subtracted from all subsequent measured torques.

Using tomography to measure the cavern size
In this study, electrical resistance tomography (ERT) was utilized to measure the cavern size. To achieve this goal, 30 mL of 20% saline solution (as a tracer) was injected into the wheat straw slurry near the impeller blade using a plastic syringe after the mixing system reached steady-state conditions. The injection took about 2-3 s for all experiments. A conductivity meter (PC10 portable meter, Oakton Instruments, USA) was employed to take an individual conductivity measurement before each experiment run to make sure that the conductivity of the wheat straw slurries were the same for all experiments. The size of the cavern was measured after the injection once the cavern size remained unchanged. The cavern boundary was determined as a position at which the tracer concentration was zero. Each tomography test was repeated three times. The standard deviation was found to be 0.56 % for cavern dimensions.
Colors in tomograms display the dispersion of the tracer in the vessel (Fig 3.5). The dark blue color in these tomograms demonstrates low-conductivity zones, which represents lower tracer concentration. The red color in the tomograms shows the high-conductivity regions, which indicates the higher tracer concentration in those zones. At the boundary of the cavern, the concentration of the tracer was zero. This means that the conductivity of the slurry at the cavern boundary was equal to the conductivity of the slurry before the injection of the tracer.
The 2D tomogram shows that the tracer injected near the impeller blade remained within the cavern and no tracer was found in the surrounding.

Xanthan Gum as a Reference to Check the Accuracy of the Tomography Measurements
In this study, an aqueous 0.5 wt% xanthan gum solution was used as a reference to check the accuracy of the ERT measurements. To prepare xanthan gum solutions, 2.5 kg of the xanthan gum powder was added slowly to water in the tank while the A310 impeller was working.
ERT was able to image the cavern formed around the impeller at 0.5 wt% xanthan gum solution once 30 ml of 10% saline solution tracer was injected near the impeller blade.
Measurements of cavern were collected from the four planes of electrodes until the cavern size remained unchanged. These tomography images were then used to measure the cavern diameter (D c ) and cavern height (H c ), using 2D and 3D images respectively. The cavern dimensions obtained from the tomography measurements were then substituted into the cylindrical model proposed by Elson (Equation 1.11), and the yield stress for 0.5 wt% xanthan gum solution was calculated. As shown in Table 3.2, the yield stress estimated using the tomography images was in good agreement with that measured using a Bohlin CVOR Rheometer 150 (Malvern Instruments, USA) as reported by Pakzad (2007). The relative error between the estimated values obtained from the rheometry and the tomography techniques was 0.61 %.

RESULTS AND DISCUSSION
In this chapter, the experimental results are presented and discussed for the impeller torque, power consumption, yield stress, and cavern formation in the mixing of wheat straw slurries as a function of the impeller type, impeller speed, slurry concentration, and fiber size.

Evaluation of the Torque Sensor Precision
The impeller torque was measured using a rotary torque meter (Staiger Mohilo, Germany) as a function of the impeller rotational speed. Each test was repeated 3 times and the standard deviation for the torque measurements was calculated using the following equation: where N is the number of measurements, M i is the torque measurement for sample i and is the mean value of the torque measurements.    As shown in Figure 4.3, for the same size and concentration of wheat straw at 80 rpm, impeller A100 shows highest torque compared to impellers A310 and A200. Impeller A200 has the lowest torque. A100,2mm,7wt% A200,2mm,7wt% A310,2mm,7wt%

Cavern Formation in Mixing Wheat Straw Slurries
Several experimental methods have been employed to investigate the formation of cavern as discussed in literature review; each of these techniques has its own restrictions. In order to prevail over these limitations, electrical resistance tomography, which is an emerging technology, can be the best alternative to measure the cavern size and analyze fluid flow in 2D and 3D in this study.

Using ERT to Measure the Cavern Diameter and Height
In this study, the size of cavern formed around the impeller in the mixing of the wheat straw slurries was measured as a function of the impeller type and speed for two sizes of wheat straw fibres (≤ 2mm and 8 ± 0.014 mm) with 5, 7, and 10 wt% fiber through the tomography images. agitated by A310 impeller using 3D tomogram at 60 rpm.

Dimensionless Cavern Diameter (Dc/D)
Various models have been recommended in the literature (Solomon et al., 1981;Elson et al., 1986 and1988) to estimate the dimensionless cavern diameter (Dc/D) versus the product of two dimensionless groups of P o and Re y = ρN 2 D 2 /τ y . As mentioned before, Dc is the cavern diameter, D is the impeller diameter, P o is the impeller power number, and  shows the cavern diameter of 32 cm, 28 cm and 21 cm for wheat straw concentration of 5 wt%, 7 wt%, and 10 wt% respectively.

Effect of Wheat Straw Concentration on the Cavern Dimensions
The largest cavern diameter was measured for 5 wt% wheat straw slurries. This finding is in agreement with those reported in the literature. For instance, Pakzad et al. (2008) showed that by increasing the yield stress, the cavern diameter decreased. In fact, the presence of the yield stress leads to the formation of stagnant regions outside the cavern in which the shear stress falls below the slurry yield stress.

Effect of Impeller Speed on Cavern Diameter
In order to see the effect of impeller speed on the cavern diameter, two illustrative tomograms (as a sample of obtained tomograms) are shown in Figure 4.16. This figure shows the effect of impeller speed when it was varied from 70 rpm to 95 rpm, on the diameter of the cavern for 10 wt% wheat straw slurry (≤ 2 mm) agitated by A100 impeller. These images were obtained from tomography plane 3, which was located at the impeller position. These tomograms clearly show that the diameter of the cavern increased when the impeller speed increased to 95 rpm. Similar trends were also observed for other cases.

Effect of Impeller Type on Cavern Diameter
Figure 4.17 shows tomograms obtained from plane 3 for the A310, A200, and A100 at the rotational speed of 40 rpm and 7 wt% wheat straw slurry (≤ 2 mm). The results show the cavern diameter of 18.5 cm, 21 cm, and 21.5 cm for A310, A200, and A100, respectively.
The larger cavern diameter created by the A100 impeller and A310 resulted in the smallest cavern.

Tomography Data
The yield stress of wheat straw suspensions for concentrations of 5, 7, 10 wt% and two fiber sizes were estimated using the cylindrical model of Elson (Equation 1.11). The term N 2 D 2 ρ/τ y on the right hand side of Equation (1.11) refers to the yield stress Reynolds number (Re y ) .
As explained in the previous section, the diameter and height of the cavern were determined through ERT. These dimensions as well as the impeller speed and the power number were substituted into Equation (4.2) to obtain the slurry yield stress. The yield stresses calculated using three impellers at three concentrations (5, 7, and 10 wt%) for two fiber sizes (≤ 2 mm and 8 mm) are listed in Tables 4.2 and 4.3. It can be seen that the average yield stress for the fiber size of ≤ 2 mm at 5, 7, and 10 wt% were 1.3, 4.2, and 14.8 Pa, respectively. For the fiber size of 8 mm, the average yield stress was 3.4, 6.8, and 16.7 Pa for 5, 7, and 10 wt% concentrations, respectively.

Effect of Fiber Length on Yield Stress
As shown in Tables 4.5 and 4.6 , the yield stress of the wheat straw slurry with a fiber size of 8 mm was more than that for a fibre size of ≤ 2 mm, at all three concentrations.
When the fiber size increases the fibrous network will be stronger, thus the yield stress will increase. This result is in agreement with those reported in the literature. Samaniuk et al. (2011) observed that the yield stress of concentrated lignocellulosic biomass (corn stover) increased with particle length. In another study, Viamajala et al. (2009)  hydrolyzed corn stover slurries behaved like yield stress fluids, at various concentrations, and that with increasing particle size, the yield stress increased. Bashir (2008) measured the rheological properties of wheat straw suspension at concentrations between 5.0-20.0 wt%. The yield stress was found to increase with the size of the wheat straw fibers. Rosgaard et al. (2007) investigated the effect of solids content and enzymatic hydrolysis on the apparent viscosity of barley straw biomass slurries, with solids fraction varying from 5 wt% to 15 wt%. They showed that the apparent viscosity increased with solids fraction. Hanley (2003 and2004) measured the apparent rheological properties of corn stover suspensions of 5, 10, 20, and 30 wt% and showed that the viscosity and the yield stress of the suspensions enhanced as the fiber size increased.

Effect of Fiber Mass Concentration on Yield Stress
Several studies have been conducted to relate the yield stress (τ y ) as a function of the mass concentration (C m ) in both pulp and biomass suspensions (Bennington et al., 1990;Dalpke and Kerekes, 2005;Knutsen and Liberatore, 2009;Stickel et al., 2009;Hue et al., 2009).
As was shown in Tables 4.5 and 4.6, by increasing fiber concentration, due to the fiber network strength the amount of yield stress is increased, which is in good agreement with Chaussy et al. (2011), Derakhshandeh et al. (2010b, and Bashir (2008) studies.

Power Number versus Yield Stress Reynolds Number
Typically, the power consumption of an impeller is presented using the power curve: power number versus Reynolds number. Since, in this study, we measured the yield stress of the wheat straw slurry, the yield stress Reynolds number (Equation 4.3), which is a function of the yield stress, was employed instead of Reynolds number.

Conclusion
Electrical resistance tomography (ERT) was used to measure the cavern diameter (2D) and cavern height (3D) in the mixing of non-Newtonian wheat straw slurries with yield stress, for the A100, A200, and A310 impellers.
In this study for the first time wheat straw slurry yield stress was estimated from the cavern size obtained from ERT tomograms for three concentrations of wheat straw slurry (5,7, and 10 wt%) and two sizes of fiber (≤ 2 mm 8 ± 0.014 mm) with three types of axial impellers (A100, A200, and A310) . The average yield stress of wheat straw slurries at 5, 7, and 10 wt% were 1.31 Pa, 4.2 Pa, and 14.8 Pa, respectively for the fiber size of ≤ 2 mm and were 3.4 Pa, 6.8 Pa, and 16.7 Pa, respectively for the fiber size of 8 mm. As expected, these values were irrespective of the type of impeller. As the slurry concentration and fiber size increased, the yield stress of the slurry also increased due to a stronger fibrous network structure.
Analysis using the 2D images at different slurry concentrations showed that the diameter of the cavern increased when the impeller speed increased for the three impellers, and that the highest ratio of the cavern height to the cavern diameter was achieved by using the A310 impeller.

Recommendations for Future Work
The results from experimental work in this study drew attention to areas for future considerations, as follows:  Using computational fluid dynamics (CFD) to study the mixing of wheat straw slurries.
 Studying the performance of other mixing systems (e.g. the combination of a radial and an axial flow impellers, and the combination of a close clearance impeller and a central impeller) in the mixing of wheat straw slurries.
 Studying the rheological behavior of wheat straw slurry using the rheometry techniques.
 Exploring the mixing of other types of agricultural wastes used for the production of bioethanol.
 Assessing the effect of pre-treatment on the mixing of agricultural wastes.

Sample of Yield Stress Calculations for a 5 wt% Wheat Straw Slurry
Cavern dimensions were measured using ERT. For each impeller speed, the experiment was repeated 3 times.  Step by step calculations for N = 30 rpm: 1. Value of "N" (impeller speed) was measured using a tachometer; the value was read in revolutions per minute (rpm) and then converted to revolutions per second (rps). For N = 30 rpm, N = 30/60 = 0.5 rps 2. Value of "M" (impeller torque) was measured using a torque meter in N.m. At