Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Puffing induced in two model systems by microwave assisted drying under vacuum : an experimental and… Ressing, Mareike Johanne 2006

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

Item Metadata

Download

Media
831-ubc_2006-130360.pdf [ 12.63MB ]
Metadata
JSON: 831-1.0092881.json
JSON-LD: 831-1.0092881-ld.json
RDF/XML (Pretty): 831-1.0092881-rdf.xml
RDF/JSON: 831-1.0092881-rdf.json
Turtle: 831-1.0092881-turtle.txt
N-Triples: 831-1.0092881-rdf-ntriples.txt
Original Record: 831-1.0092881-source.json
Full Text
831-1.0092881-fulltext.txt
Citation
831-1.0092881.ris

Full Text

PUFFING INDUCED IN TWO MODEL SYSTEMS B Y MICROWAVE ASSISTED DRYING UNDER V A C U U M -A N EXPERIMENTAL AND NUMERICAL ANALYSIS by MAREIKE JOHANNE RESSING B.S., Clemson University, Clemson, USA 1995 M.S., The University of Hawaii, Honolulu, USA 1998 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES Food Science THE UNIVERSITY OF BRITISH COLUMBIA December 2005 © Mareike Ressing, 2005 ABSTRACT Microwave heating under vacuum (VM) removes water at reduced temperatures while minimally altering desirable biochemical characteristics of foods compared to traditional air drying methods. Concurrently, it achieves much greater drying rates than other vacuum drying methods. Puffing has been observed in V M dried products and reportedly contributes to improved rehydration and eating quality of V M processed foods. However, the underlying mechanisms this phenomenon are poorly understood. Wheat dough and gelatinized potato starch were selected as model systems to investigate V M puffing since their respective rheologies have previously been characterized. The thesis objective was to advance our understanding of the mechanism behind V M puffing, especially with regard to the rheological and dielectric properties of the selected model systems, and use these data, namely wheat dough results, to develop and validate a numerical model of V M puffing. One model system consisted of dough balls prepared with different quality flour, water and salt (0.0-1.5%) and dehydrated at 700 and 1300 W microwave power at 28 Torr absolute pressure. Cylinders of gelatinized potato starch (30-50% starch d.b., 0-2% NaCl, respectively) which served as the second model,were dehydrated at 350, 700, and 1300 W. For both model systems, volume, dielectric properties, Young's modulus and/or fracture strength were determined. A video recorder was used to observe the dough and starch model systems during dehydration. ii Both model systems exhibited pronounced puffing which was influenced by applied power levels, salt content, flour type or starch concentration and dielectric properties. While dough already expanded upon dropping chamber pressure, starch gels needed prior heating to decrease resistance to expansion. Both systems exhibited salt dependence with respect to their rheology. Dough showed reduced volume increases as salt concentration and elasticity decreased, while salt containing starch gels exhibited increased brittleness and lack of cohesion that decreased the expansion potential. Finite Element (FE) modeling described the dough development during microwave drying under vacuum. It illustrated that the puffing is highly influenced by the rheological properties of the dough more so than microwave interaction. While the accuracy was lower than expected, it revealed a stronger pressure interaction with the rheological properties in addition to steam-induced puffing. Image analysis used to characterize porosity of the dried product was only moderately successful and pointed to the possiblities in the field rather than offering a solution to current questions. iii TABLE OF CONTENTS Abstract 1 1 Table of Contents iv List of Tables viii List of Figures ix List of Symbols and Abbreviations xi Acknowledgements xiii CHAPTER I Overview 1.1 General introduction 1 1.2 Drying 2 1.2.1 Microwave drying 3 1.2.2 Vacuum assisted microwave drying 5 1.3 Puffing V 1.4 Selection of a model system 9 1.5 Numerical modeling 10 CHAPTER II Microwave drying under vacuum using wheat dough balls as a model system 2.1 Introduction 12 2.1.1 Dough properties, chemistry and rheology 13 2.1.2 Hypotheses 20 2.2 Materials and Methods 21 2.2.1 Materials 21 2.2.2 Basic dough preparation 22 2.2.3 Microwave drying under vacuum 22 2.2.4 Drying and heating rate 23 2.2.5 Storage 24 2.2.6 Density 24 2.2.7 Moisture 24 2.2.8 Protein analysis 25 iv 2.2.9 Dielectric properties 25 2.2.10 Rheology 2 6 2.2.10.1 Uniaxial extension test 26 2.2.10.2 Biaxial extension test 31 2.2.11 Video monitoring .32 2.2.12 Statistical analysis 33 2.2.13 Image analysis 34 2.2.13.1 Sample preparation 34 2.2.13.2 Image analysis program 35 2.3 Results 38 2.3.1 Microwave drying under vacuum 38 2.3.2 Heating rate 42 2.3.3 Drying rate and moisture content 46 2.3.4 Video monitoring 47 2.3.5 Protein analysis • 50 2.3.6 Rheology 50 2.3.7 Dielectric properties '• 57 2.3.8 Image analysis 59 2.4 Discussion 62 2.5 Summary and Conclusions 74 CHAPTER III Physical modeling of vacuum microwave drying of dough balls using the Finite Element Method 3.1 Introduction 76 3.2 Model description 79 3.3 Results and discussion 85 3.4 Summary 94 v CHAPTER IV Microwave drying under vacuum using potato starch gels as a model system 4.1 Introduction • 95 4.1.1 Gel atinization of starch 96 4.1.2 Rheology of gelatinized starch 96 4.1.3 Other starch gel characteristics 98 4.1.4 Hypotheses 1°0 4.2 Materials and Methods 101 4.2.1 Gel preparation 101 4.2.2 Microwave drying under vacuum 101 4.2.3 Storage 1 ° 2 4.2.4 Density 1 ° 2 4.2.5 Moisture 1 ° 2 4.2.6 Heating and drying rate 103 4.2.7 Young's modulus 103 4.2.8 Dielectric properties 104 4.2.9 Video monitoring 105 4.2.10 Air inclusion 106 4.2.11 Statistical analysis 106 4.3 Results 106 4.3.1 Microwave drying under vacuum 106 4.3.2 Drying rate 108 4.3.3 Young's modulus HO 4.3.4 Heating rate • I l l 4.3.5 Dielectric properties 113 4.3.6 Video monitoring 114 4.4 Discussion 123 4.5 Summary 133 CHAPTER V Summary and Future Directions 135 vi BIBLIOGRAPHY APPENDIX LIST OF TABLES Table 2.1 Wheat protein contents of Canadian harvests in 1999-2004 18 Table 2.2 A N O V A results for volume change in HG at high and medium 4 5 power and LG dried at high power Table 2.3 Temperature development for 0.0% salt HG dough under vacuum microwave drying ^ Table 2.4 Water content, in percent of total mass, for HG and L G dough, 0.0% salt respectively, with regard to water loss in high and medium power (HP, MP) drying over time ^ Table 2.5 a A N O V A determined by G L M for alveograph results for the parameter P, the resistance to extension 56 b A N O V A determined by G L M for alveograph results for the parameter L, the dough extensibility 56 c A N O V A determined by G L M for alveograph results for the parameter W, the deformation energy ^ 7 Formulations for HG and LG dough with their respective salt content and the corresponding dielectric properties 58 Material properties of the FE model 82 A N O V A results for volume change determined by G L M A N O V A results for volume change determined by G L M , excluding the gel concentration of 50% i K j y Dielectric properties E ' and 8" with respect to the varying starch gel formulations Video recording times for starch gels at low, medium and higher power drying at various starch and salt levels * ^ a A N O V A results determined by G L M for parameter bulging start 1 1 7 b A N O V A results determined by G L M for parameter rounding start 1 1 7 c A N O V A results determined by G L M for parameter rounding end : 1 1 8 Table 4 6 Selected results of DUNCAN multiple range test for bulging start for all starch gel formulations ^ Table 4.7 D U N C A N multiple range test for volume change and the parameters starch and salt concentration with respect to their drying response ^ 4 Table 2.6 Table 3.1 Table 4.1 Table 4.2 Table 4.3 Table 4.4 Table 4.5 viii LIST OF FIGURES Fi gure 2.1 Vacuum microwave dryer, Enwave Corp. Vancouver, BC 22 Figure 2.2 Schematic for dielectric properties determination setup 26 Figure 2.3 Schematic of uniaxial extension tests for wheat dough 28 Figure 2.4 a. Equipment used in uniaxial extension tests 29 b. Texture Analyser piston being lowered into a sheet of dough, 29 top view c. Texture Analyser piston as it is lowered into a sheet of dough, 29 bottom view Figure 2.5 Sample stress-strain curve for a dough uniaxial extension test 30 Figure 2.6 Alveograph inflating a wheat dough sheet into a bubble 32 Figure 2.7 Schematic for videorecording setup 33 Figure 2.8 Image analysis sample image capture for cross sections 34 Figure 2.9 DB2 program selected commands a. Original cross-sectional picture of dried doughball 36 b. Selection of pores by 'find loop' command c. Segmentation d. Final calculation results upon 'pore stats' command 37 Figure 2.10 Changes for HG at high power drying for different salt concentrations 39 Figure 2.11 Changes for HG at medium power drying for different salt concentrations 40 Figure 2.12 Changes for L G at high power drying for different salt concentrations 41 Figure 2.13 Volume change (%) for HG and L G doughs at varying salt concentrations for high power drying ^ Figure 2.14 Volume change (%) of HG dough at MP and HP drying 44 Figure 2.15 Water loss (g) over time for HG and L G dough at high and medium power 4g Figure 2.16 Young's modulus E (MPa) for uniaxial extension of HP dough at different salt concentrations ^ 1 ix Figure 2.17 Fracture strength Rm (MPa) for uniaxial extension of HP dough at different salt concentrations 52 Figure 2.18 a. Alveograph tenacity P values for selected salt formulations for low and high gluten dough, LG and HG 53 b. Alveograph L (abcissa of rupture) for selected salt formulations for low and high gluten dough, L G and HG 54 c. Alveograph deformation energy W (1CT4 J) for selected salt formulations for low and high gluten dough, L G and HG 55 Figure 2.19 Total cross section surface area of dried HG dough in high power, estimated and measured compared to salt concentration 50 Figure 2.20 Average pores size of HP-dried HG dough ball cross sections compared to the number of pores 61 Figure 3.1 FE-model of a dough ball 80 Figure 3.2 Schematic view of the FE-modeling process 84 Figure 3.3 Resulting dough ball deformation at 28 Torr absolute pressure and 180 s of microwave application 86 Figure 3.4 Temperature distribution after application of high power microwave field for 1 s (a), 10 s (b), 90 s (c), and 180 s (d) 87 Figure 3.5 Stress distribution after application of high power microwave field for 1 s (a), 10 s (b), 90 s (c), and 180 s (d) 88 Figure 3.6 Average temperature increase for various salt concentrations 89 Figure 3.7 Volume increase for various salt concentrations 91 Figure 3.8 Volume increase for various salt concentrations under the assumption of identical heat generation 92 Figure 3.9 Volume increase for 0.0% salt content including the reduction in stiffness by 2%/°C and a loss of air of 10 % over the 180 s of microwave exposure 93 Figure 4.1 Volume changes of different starch, salt formulations of potato starch gels dried at high, medium or low power 107 Figure 4.2 Moisture loss (g/min) versus drying times for 40% starch gels without added salt at high, medium and low power 110 Figure 4.3 Volume change of high, medium and low power drying for all starch/salt formulations compared to their respective E values I l l Figure 4.4 Heating rate of 40% starch gel without salt, dried at high, medium and low power 112 Figure 4.5 End of bulging time for high, medium and low power drying for all starch/salt formulations 121 Figure 4.6 End of rounding time for high, medium and low power for all starch/salt formulations 122 x LIST OF SYMBOLS AND ABBREVIATIONS 6 strain 8' dielectric constant e" dielectric loss factor A heat conduction [j, Poisson's ratio p density a stress c specific heat capacity D diameter of the holder d diameter of the plunger dp penetration depth E Young's modulus F force applied h height of the curve h latent heat L length after deformation L dough extensibility L 0 original length of the dough Mdw moisture per dry weight P probability P tenacity or resistance to extension P/L ratio curve configuration ratio xi Rm fracture strength r drying rate S area under the alveograph curve t thickness of the dough t0 original thickness of the dough W deformation energy w0 initial weight w{ final weight V volume of air ANOVA analysis of variance FE Finite Element FEM Finite Element Method G L M general linear model HG high gluten HP high power LG low gluten LP low power MP medium power V M D vacuum microwave drying xii Acknowledgements A Ph.D. thesis can never be just the product of one. The valuable input of others make it what it is. For that reason I hope to sufficiently acknowledge all who assisted my endeavour. I would like to express my gratitude to my supervisor Dr. Timothy D. Durance who had the courage to get me involved in this topic and assisted me with countless pieces of advice and encouragement. Many thanks to my committee members Dr. Christine Seaman, whose prompt responses to questions and counterquestion were invaluable, Dr. Shahab Sokhansanj for his precious outside food science knowledge and Dr. David Kitts for his encouragement and topical advice. Thanks to Dr. Henning Ressing for introducing me to the field of FE analysis, an opportunity a food scientist does not receive often. Dr. Qing Wang requires a special reference for the custom written image analysis and the many hours of adjustments. Thanks to the staff at the Canadian Grain Research Laboratory in Winnipeg, MB for their technical assistance and 'short course' on baking sience. Many thank yous go to Adam Figiel and Dariusz Piotrowski for their endless efforts in testing and recording during their stay at our department. A special thanks to Sherman Yee and Dr. Pedro Aloise for their technical assistance and invaluable creative problem solving skills. And Dr. Jaya Sundaram needs distinctive acknowledgement since she provided the spontaneous helping hand no matter when needed. CHAPTER I Overview 1.1 General Introduction Microwave dehydration under vacuum (VMD) is a drying technique to remove water by different means than conventional air drying. In addition to enabling rapid, efficient and economical removal of water, it offers clear advantages to conventional drying. The vacuum results in a the lower temperature of water to vapour transition, allowing product chemical composition to be only minimally altered. This particular effect has placed the value of V M D between air drying and freeze drying, the most gentle way of drying. Though the efficiency and practicality of V M D have been studied, the mechanisms by which textural changes occur, i.e., softer tissue, puffing and increased porosity, are poorly understood. One peculiarity of V M D is that during the drying process materials often do not collapse or even expand beyond their original size. While for example V M D potato cubes retain a puffed appearance, other commodities such as berries and cut fruit, may show collapse before V M dehydration is completed. The intent of this thesis was to investigate the mechanism behind inhibition of collapse or actual puffing induced by vacuum microwave drying. Due to the nature of microwave heating, modification of dielectric properties was an important tool for this study. For this reason special focus was placed on salt content and microwave power levels to investigate the effects of dielectric properties with respect to puffing behaviour. Wheat dough and gelatinized potato starch were used as model systems to study parameter 1 influences and their interactions on volume and density change. Image analysis was intended to serve as an aid and possible substitute for standard physical measures to determine and quantify porosity of the dehydrated products using analysis of cut cross sections. Furthermore, Finite Element Method (FEM) modeling was employed to predict puffing behaviour in a V M D system. The experimental results were intended to validate the numerical model. 1.2. Drying Drying or dehydration is among the oldest preservation methods for foods. Removal of water makes food more shelf-stable and less perishable. Dehydration technologies have been categorized into four generations, the first being the hot air drying type such as cabinet and bed dryers, mostly dealing with solids dehydration. Dryers suitable for slurries and purees such as spray and drum dryers fall into the second generation while freeze and osmotic dryers are third generation. Vacuum microwave assisted drying, along with radio frequency, refractance window and hurdle approach dehydration are seen as the fourth generation and represent the most recent advances in drying (Vega-Mercado etal., 2001). Dehydration also induces physical changes that are specific to the method used. Air-drying still remains the most common method of drying foods, however, it induces noticeable changes such as case hardening, shrinkage, flavour changes, texture change and nutrient losses. Freeze and osmotic drying, recognized as the most gentle drying methods, better preserve chemical properties, including volatile compounds. Freeze 2 drying in particular induces textural changes like a puffy appearance and porosity that facilitates rehydration. 1.2.1 Microwave drying In drying, microwave energy has been used for a number of years. The nature of microwave drying creates a heating pattern from the inside out if the penetration depth is sufficient. In addition, conductive and convective heating from already warmed locations will enhance the heating process. Dipolar rotation is created by microwaves, which creates heat from within the foodstuff through intermolecular friction (Oliveira & Franca, 2002). Water molecules and their solutes are agitated by the constantly changing orientation of the electric field (Buffler, 1993). For absorbed power, the following equation is used: Pv=2nfe0e"E2. (1.1) where P v = power absorbed per unit volume (W/min ) / = frequency (Hz) E - electric field inside the load (V/m) £o =permittivity of free space e" = dielectric loss factor Thus the temperature rise results from the following ^V(L)-p(g/mL)-C(cal/g°C)-AT(°C) P(W) = PVV = 69.8 — — —— (1.2) r(min) where V = total sample volume 3 p = load density C p = load heat capacity (specific heat) t = time that microwave power is applied T - temperature. Microwave heating time is significantly shorter than air drying due to the quick penetration of the microwaves (Oliveira & Franca, 2002). Once heat is created within the food, it is distributed further by conduction (Drouzas & Schubert, 1996). But depending on the conductive properties of the multiple components of a food, this may cause uneven heating, leaving hot and cold spots (Ryynanen, 2002). Other contributing factors to uneven microwave heating are shape and size of the sample, microwave field within the cavity, geometry and material of the food container (Ryynanen, 1995; 2002). Macroscopic agitation of the heating food within the microwave field can mitigate this shortcoming. Microwave heating is influenced by dielectric properties that mirror the temperature, moisture content and composition of the food item. Even in similar systems such as isolated starches, namely corn, rice, tapioca, wheat, waxy maize and amylomaize, in the dry form and in solution, significant differences were observed at temperatures between 30 and 95°C at 2450 MHz, as described by Ndife and colleagues (1999). In general, increased moisture was reflected in an increased dielectric constant and loss factor while dry starches showed little change in dielectric properties with rising temperatures. However, while 30% solutions of amylomaize and rice starch hardly lowered in e' with increasing temperature, the £' values for wheat, tapioca, waxy maize and corn starch noticeably decreased. Al l 30% starch solutions exhibited a drop in E" 4 values over this temperature range. The largest decrease was recorded for waxy maize and amylomaize. When looking at same starches in dry form, e" values slightly increased with rising temperature or stayed the same (Ndife et al., 1999). The water content, as a reflection of progression in drying, will adversely affect the efficiency of the latter portion of drying. Solutes, more specifically, salts enhance microwave heating (Ryynanen, 2002). Agar gels with added NaCl and sucrose have been used successfully to characterize microwave heating (Sakai et al., 2005). However microwave drying applied by itself may induce more shrinkage and slower rehydration than hot air drying, as demonstrated on kiwi (Maskan, 2001a, b). The combination of air and microwave drying has been effectively used and investigated. For potato and apple, the accelerated drying rate of microwave-assisted air drying compared to air drying alone was well demonstrated (Khraisheh et al., 2000; Prothon et al., 2001). While in microwave drying alone both wave guide and resonant chamber heating are applied, in microwave heating assisted by vacuum, resonant chambers have been favoured (Drouzas & Schubert, 1996; Erie & Schubert, 2001; Gunasekaran, 1999; Kaensup et al., 2002; Kim et al., 2000; Mousa & Farid, 2002). 1.2.2 Vacuum assisted microwave drying In combination with vacuum, microwave drying is heating more efficiently and is more protective of heat labile compounds than air drying. Vacuum allows evaporation of water to occur at lower temperatures. It has been noted that once the desired vacuum has been reached, the temperature in the chamber reaches a plateau and internal object temperatures will remain constant until the last stage of drying. Furthermore, in this 5 method, the reduction of oxygen present minimizes the degradation of susceptible compounds (Sham et al., 2001). Unlike microwave drying at atmospheric pressure, which induced shrinkage (Maskan, 2001b), its combination with vacuum prevents shrinkage and may actually increase volume and improve rehydration (Sablani & Rahman, 2002). With the exception of freeze-drying, V M D shows distinct advantages over other drying methods due to its low impact on biochemical compounds, volatiles and texture in general, as mentioned above. It is considered a good balance of cost versus food quality and therefore offers a viable alternative to more traditional drying techniques (Drouzas et al., 1999). The increase in volume, described by several researchers as expansion or puffing, facilitates rehydration (Krokida et al., 1998; Sablani & Rahman, 2002) and flavour is enhanced by good volatile retention (Mui et al., 2002). To reduce uneven heating, some research has focused specifically on rotary drum V M dryers. To evaluate drum rotation speed and energy efficiency, chilli peppers were V M -dried (Kaensup et al., 2002). Despite findings that lower vacuum pressure decreased drying time, the researchers point out that rotational speed needs to be adjusted to avoid a dead zone specific to the product. Drying efficiency, defined as the ratio of the heat utilized for evaporating the sample water to the heat supplied by the microwave oven, taking latent heat of vapourization into consideration, equals 100% at the beginning of the process but rapidly decreases as the moisture content decreases (Mousa & Farid, 2002). In the latter part of the dehydration process, the influence of the vacuum on 6 drying efficiency becomes more relevant (Mousa & Farid, 2002). In banana, it was universially noted that final texture differed from convection dried banana. Also product volume increased, the latter being likened to puffiness of freeze dried or fried products (Drouzas & Schubert, 1996; Mousa & Farid, 2002; Mui et al., 2002). In CaCl 2 treated apples, high vacuum settings increased the degree of puffing and their crispness (Sham et al., 2001). Cranberries dried in a pulsed vacuum microwave dryer had a softer texture than the conventionally air-dried (Gunasekaran, 1999). Even in combination with osmotic dehydration, V M D yielded low density apples and strawberries, exhibiting good quality regarding colour, taste, and vitamin C retention. The total volume was less than the original due to the osmotic treatment, however V M D did induce porosity (Erie & Schubert, 2001). In a direct comparison of carrots dried by V M , air, and freeze drying, V M D carrot slices performed similar or even better than freeze dried samples as judged by a sensory panel. This applied to both dry and reconstituted samples regarding their colour, texture, flavour and overall preference. Air-dried quality was considered lowest for colour compared to all other techniques (Lin et al., 1998). V M drying of basil and oregano showed good quality results in flavour and colour retention and excellent rehydration properties. More volatiles were detected for V M D basil compared to fresh samples due to chemical reactions in the process (Yousif et al., 1999, 2000). Colour retention has been judged as superior to air-drying and equivalent to freeze drying as well (Lin et al., 1998). 7 1.3 Puffing Puffing has been defined as the increase in volume after treatment (Varnalis et al., 2001a, b) or as a general lowering of density (Sham et al., 2001). While it has been well described for frying of foods, such as potatoes and corn chips, in extrusion, and in popping of popcorn, the mechanisms behind it are multiple. Wu and Schwartzberg (1994) suggested pore size and structure as essential for vapour-induced puffing, influencing the resulting characteristics of texture, mechanical deformation properties, breaking and cutting strength, heat and mass transfer mechanisms. Build-up time is critical since steam needs to be generated within the product (Yamsaengsung & Moreira, 2002). For popcorn, water content determines popping volume, increasing until the optimum water content has been reached and decreasing beyond this value (Shimoni et al., 2002). Microwave popcorn puffing has been modeled and the influence of salt content on puffing was described (Huang et al., 2002). Puffing has been observed in drying techniques as well. Namely in freeze-drying, induced by the necessary vacuum exposure, it is judged as a desirable property since it facilitates rehydration as the process induces porosity (Krokida et al., 1998; Sablani & Rahman, 2002). Hussain and colleagues (2002) developed a predictive model by neural networks, naming temperature of drying, moisture content, initial porosity, and product type as the key determining factors for pore formation. To avoid collapse of the puffed structure, the drying temperature should be kept below the glass transition temperature (Varnalis et al., 2001a). By inducing puffing prior to air-8 drying potato cubes by high temperature exposure, quality attributes increased in comparison to conventional methods (Varnalis et al., 2001b). In addition puffing may increase the drying rate in an air dryer according to Tabeidie and coworkers (1992), which however was not validated by McLaughlin and Magee (1999). 1.4 Selection of a model system In order to continue the research on microwave drying under vacuum, a suitable model system is necessary. It is assumed that the existence of rheological and power thresholds are prerequisites for puffing to take place in this drying method. To vary both, a thorough knowledge of the model system is crucial since it will allow direct inference of the mechanisms if it is well chosen. Since wheat dough was selected for study since its rheological properties have been widely researched. It is well known that expansion is possible due to its use in the production of leavened breads. This unique character distinguishes wheat from other cereal grains, namely because of the protein mixture gluten that exhibits a distinct viscoelastic character. Specifically the capacity to entrap gases characterizes wheat doughs. Several testing methods have been employed to unveil chemical and mechanical interactions in bread making. Empirical as well as fundamental methods involving uniaxial and biaxial extensions have been developed. The latter is crucial to understand bubble formation of wheat doughs (van Vliet et al., 1992; Dobraszczyk & Morgenstern, 2003). 9 As the complexity of wheat flour dough model exhibited a possibly anomalous salt effect in microwave drying under vacuum, gelatinized potato starch was chosen as a simpler model system. Not only is potato starch different from wheat flour since it lacks protein, its starch type and composition also differ. In addition, water content could be varied to modify the dielectric response without sacrificing ease in handling compared to wheat doughs which become unmanageable beyond a limited range. Potato starch has also been characterized with regard to its rheological properties, and its salt response is known (Donald, 2000). Testing methods are established. Due to the fact that it is a simple system of mostly one compound, starch as the dominant polysaccharide, its rheological and mechanical response were expected to facilitate the illumination of the puffing mechanisms in microwave drying under vacuum. 1.5 Numerical modeling Numerical modeling is increasingly being used as a means to help explain mechanisms behind observed phenomena and to serve as a predictive tool. It has been applied to food processing in general, dehydration, dough elasticity, baking performance, and in particular, puffing as well as shrinkage. Research in food processing has made use of Finite Difference and Finite Element (FE) methods as numerical modeling tools. A comprehensive review was published by Puri and Anantheswaran (1993) on the use of FE modeling of food processes. The focus of this research has been mainly the simple mechanisms of transport of mass or energy. 10 Kemp and Oakley (2002) reviewed the advantages and disadvantages of various modeling approaches to particulate drying. Other FE modeling in drying looked at mass transfer and structural change such as shrinkage in particular (Wu & Irudayarj, 1996; Jia et al., 2000; Yang et al., 2001). Microwave heating in particular has received attention with regard to its heat generation and the developing temperature profile (Lin et al., 1995; Oliveira & Franca, 2000; Ratanadecho et al., 2001; Pandit & Prasad, 2003). Yamsaengsung and Moreira (2002) combined two phenomena observed in tortilla chip preparation in a FE model, some shrinkage as well as expansion, with a good match to experimental data. Puffing of popcorn was also modeled by Huang and colleagues (2002). Shrinkage in drying and in general was modeled with regard to water removal during drying (Perre & May, 2001; Jia et al., 2000). Though V M drying has been the focus of modeling efforts, so far no attempts have been made to explain structural changes due to puffing; rather heat and mass transfer have dominated the research. The combination of thermal analysis of the drying process with that of structural deformation still is lacking. As such, modeling of V M D combining these aspects has not has been published to date. One intention of this study was to enhance the knowledge in this field. 11 CHAPTER II Microwave drying under vacuum using wheat dough balls as a model system 2.1 Introduction To undertake the study of puffing in microwave drying under vacuum, a model system is necessary that has a measurable rheological character which itself has been studied comprehensively. Preferrably the model system can be modified in aspects considered relevant to the puffing mechanism. For these reasons, wheat dough seemed a good choice for a model system for a process in which the texture of the material prior to drying is deemed important. Due to its importance to the baking industry, wheat dough chemistry and rheology have been studied intensively since the 1930's to aid optimization in processing and resources (Bagley et al., 1998). It is among the most studied food systems. Yet the mechanisms that govern dough rheology have not been entirely elucidated (Faubion & Hoseney, 1989). Tests have improved but even now the complexicity of dough rheology is not entirely explained and does not satisfactorily correlate with baking performance (Dobraszczyk & Morgenstern, 2003). For the purposes of studying puffing, a model system which is able to sustain stable bubbles is essential, and for exactly this property wheat dough is considered unique (Bloksma & Bushuk, 1988; Faergestad et al., 2000). To vary dielectric properties, salt has been used in model systems to study the response to microwave heating. Thus to understand volume expansion under vacuum-assisted microwave drying, wheat doughs 12 were examined with respect to added salt and the resulting changes in rheological property, their impact upon stress and strain, as well as temperature and expansion. 2.1.1 Dough properties, chemistry and rheology Doughs made from wheat flour have viscoelastic properties that are distinct from other cereal grains (Pomeranz, 1988). These characteristics are mostly due to gluten's polymeric and monomeric proteins. The former compounds, namely glutenins, cross-link with disulfide bonds between molecules to allow for the typical viscosity (Bloksma, 1990). Gliadin monomers are responsible for the elasticity as they SS-bond intramolecularly (Bloksma, 1990; Gupta et al., 1993; Cornec et al., 1994; Khatkar et al., 2002a, b). Yet, wheat dough rheology is very complex. Wheat dough is considered isotropic, i.e., its rheological characteristics are the same in all directions when force is exerted. The absolute values of stress and strain response are highly dependent on the magnitude of deformation. During kneading, an expandable but delicate gluten network forms, its gas retention potential allowing the formation of the viscoelastic dough typical of bread (Pomeranz, 1988; Bloksma & Bushuk, 1988; Autio & Laurikainen, 1997; Callejo et al., 1999). Nevertheless it is important to optimize the processing method. While under-mixing will not develop the gluten network in full and leave protein and starch portions unevenly distributed, over-mixing destroys the elastic network (Bloksma, 1990; Autio & Laurikainen, 1997). Dough in which gluten is well hydrated exhibits non-linear elasticity that is subject to shear thinning and work or strain hardening. This means that dough shows increasing resistance to further extension and the simultaneous decrease in elasticity or, in other words, where the stress increases more than proportionally to the strain (van Vliet et al., 1992). Interplay of viscosity and 13 extensibility are crucial for optimum dough performance (Bloksma, 1990). Dough rise is a result of gas bubble formation originating from C 0 2 and steam release. The foam stability, i.e., the stability to keep air entrapped, depends on properties such as surface elasticity, viscosity, film net surface charge, intermolecular interaction and film thickness, thermal and mechanical shock, local thinning and rupture, gas solubility and permeability of the film (Dobraszczyk et al., 2001). Particularly, local thinning has been identified as important to foam stability. The expansion of entrapped gases without coalescence can only happen when the thinning material serving as separator between adjacent gas cells is stronger in thin rather than thick form or at relatively high strain levels (van Vliet et al., 1992). The native starch in wheat flour dough provides sufficient viscosity to disrupt the strained dough films as their granules size determines the extension potential (van Vliet et al., 1992). Non-polar protein fractions such as prolamines weaken the elasticity of the polar proteins (Belitz et al., 1986). For this to happen, the thinning dough has to remain thicker than the local starch granules. The resistance to extension is largest at the thinnest locus. However, in good quality wheat doughs, rupture at these locations does not occur. Rather strain hardening is observed, which is considered essential to bubble stability (Dobraszczyk & Morgenstern, 2003; Kokelaar et al, 1996; van Vliet et al., 1992). For quick quality assessment and performance evaluation, several instruments have been developed to specifically assess dough characteristics. These empirical measurements of prepared dough have been used for decades since they reflect the processing conditions of dough and breadmaking well. However, these instruments such as mixograph, Chopin 14 alveograph, and Brabender farinograph cannot provide fundamental rheological properties (Martinant et al., 1998; Letang et al., 1999). To counteract this problem, fundamental measurements have found more application in wheat research in recent years. Some groups have concentrated on matching new methods with empirical methods while others have started to develop entirely new concepts. Numerical modeling has entered the science as well. Letang and colleagues (1999) matched Environmental Scanning Electron Microscope and dynamic rheometrical measurements with Brabender farinograph results. The same research group evaluated ultrasonic measurements for testing wheat dough (Letang et al., 2001). For non-linear viscoelastic properties of dough, a comparison of Brabender farinograph with mechanical spectrometry results and steady shear measurements via capillary rheometer (Instron) yielded a good model to predict steady shear viscosity and small oscillatory properties (Wang & Kokini, 1995). Developmental rheology was modeled with extensional tests of induced mixing by mixograph, investigating the interaction between addition of water and constant agitation itself (Gras et al., 2000). In a similar fashion, kneading behaviour was successfully described by using the Taylor-Galerkin algorithm (Binding et al., 2003). An extensograph allows characterization of the uniaxial extension response, yielding measurements of elastic moduli and viscosity parameters (Kieffer et al., 1998). Oscillatory rheometry was matched with laser scanning confocal microscopy results to examine the extent of protein network development due to under and over-mixing (Lee et al., 2001). This best described the resistance of the dough in rolling and relaxation. In a multidimensional study, oscillatory and steady shear, stress 15 relaxation and extensional viscosity were matched using extensographs, Bohlin and stress controlled rheometers and tension testers (Keentok et al., 2002). However, in all these approaches one needs to distinguish between small and large deformations in the dough-to-bread making process since the rheological response may vary greatly (Sliwinski et al., 2004c, d; Uthayakumaran et al., 2002). Gas bubble development is an important aspect of dough and bread development. Unlike the aforementioned procedures, which provide uniaxial extension measurements, the Chopin alveograph offers a measure characterized by biaxial extension of the dough in the two perpendicular planes, expanding a sheet of dough by blowing a stream of air to yield an air-filled dough bubble (Faridi & Rasper, 1987). This expansion closely resembles the deformation in the bread making process (Rasper & Danihelkova, 1986; Faridi & Rasper, 1987). The downside to the alveograph measurements is that under normal operations, the data generated cannot be converted to absolute stress and strain values (Bagley et al., 1998). To avoid this drawback, the alveograph was further modified to the Dobraszczyk/Robert (D/R) dough inflation system, enabling the determination of stress, strain and apparent viscosity (Dobraszczyk & Schofield, 2000; Dobraszczyk et al., 2001; Charalambides et al., 2002a, b). In an experimental study and subsequent modeling, Fan and colleagues (1999) found that in the early stages of baking bread, the bubble formation is only dependent on temperature and neither on heating rate nor viscosity when the internal pressure is equivalent to atmospheric pressure. However, with increasing viscosity of the dough, viscosity role becomes more instrumental in bubble formation, retarding the rise of the loaf. Further, Fan and colleagues (1999) found 16 that the degree of rise is dependent on C 0 2 level early on whereas water vapour dominates the later baking phase. Surface tension becomes a minor factor compared to forces applied in mixing as gases are incorporated (Mills et al., 2003). Bubble development itself has been studied and simulated with regard to proofing behaviour and bubble expansion and maturation (Shah et al., 1998; Chiotelli & Campbell, 2003; Mills et al., 2003). Larger gas concentrations as well as temperature increase the final bubble size while high surface tension decreases it (Shah et al., 1998). Strain hardening has been described in uniaxial extensions of dough (Janssen et al., 1996a; Uthayakumaran et al., 2002) as well as in biaxial extensions (Dobraszczyk & Roberts, 1994; Janssen et al., 1996a; Wikstrom & Bohlin, 1999). It is essential to bubble expansion since it enables the stretching of the enclosing dough 'membranes' to very thin layers without rupture. If strain hardening would not occur, the ratio of bubble to dough would be significantly lower. The foam/sponge characteristic observed in leavened wheat breads would not be as pronounced. In doughs that do not strain-harden, strain pulls the dividing walls apart and releases the entrapped air. In weak flour doughs, gas encapsulations can be achieved by additives such as lipids or other non-cereal proteins as is done in sweet breads (Calderon-Domfnguez et al., 2004). Wheat flour protein can be separated into non-gluten components, albumins and globulins, the minor portions of gluten proteins, and gliadins as well as glutenins, soluble and insoluble alike (Osborne Solubility Fractions, Osborne, 1907). Hard wheats, generally providing strong bread flours, tend to exhibit higher protein concentrations than soft wheats or durum wheats (Belitz et al., 1986). However, protein content alone does 17 not determine the rheological behaviour of a wheat flour. It is more important to look at the quality of the protein, especially the gluten components. Though protein contents for two tested flours were close, 13 vs. 12.5%, the former, categorized as strong flour, surpassed the other in biaxial extension tests (Tronsmo et al., 2000). Even more extreme, the other strong flour at 11.2% protein showed higher biaxial extension values than the high protein/low strength flour selected for this trial. This was confirmed in the standard baking tests (Tronsmo et al., 2000). General wheat protein contents of the Canadian harvests in 1999-2004 (Canadian Grain Commission, 2005) are given in Table 2.1. Table 2.1 Wheat protein contents of Canadian harvests in 1999-2004 (Canadian Grain Commission, 2005). Canada Western Red Winter (CWRW) wheat Canada Prairie Spring Red (CPSR) wheat Canada Prairie Spring White (CPSW) wheat Canada Western Extra Strong (CWES) wheat Canada Western Soft White Spring (CWSWS) wheat Protein content (%) 10.0-12.3 11.2-14.4 10.9-13.0 10.9-14.6 10.7-11.3 Quality aspects for wheat protein have been characterized by the glutenin and gliadin ratio as well as their respective sulfur bond counts and locations and the general amino acid composition. The presence of ionic bonds in addition to sulfur bridges strengthen the gluten and give each wheat variety its characteristic rheological properties. Bipolar ions in the form of amino acids, glycine in particular, contribute to stronger binding and increased elasticity. The subunits prolamine and glutelin in gluten act 18 counterproductively. While glutelins increase elasticity due to their polarity, prolamine in wheat specifically acts as a solvent/dilution factor due to its nonpolar character (Belitz etal., 1986). Unlike strong flours, weak flours typically exhibit a low mixing optimum, which is the time required to optimally develop the functional gluten network (Janssen et al., 1996a; Keentok et al., 2002; Rao et al., 2000; Hayta & Schofield, 2004). Strong flours show a higher viscosity in biaxial extension and apparent strain hardening, generally not recognized in weak flours. In weak flours, the resistance to deformation is relatively low (Janssen et al., 1996a, b). It is the protein phase that enables the entrapment of gas in dough since wheat proteins have film forming ability (Bloksma & Bushuk, 1988). In mixing, the gluten network forms by folding and unfolding a beta-sheet structure, eventually providing the optimum rheology for the flour. If this agitation continues, the beta-sheets unfold unrepearably. At the same time, disulfide as well as hydrogen bonds form. Al l contribute to the stability of the network (Belton, 2005). Starch provides the largest segment within the carbohydrate portion of wheat flour and it does take part in the rheology of wheat dough as its increasing content decreases the linearity of the elasticity (Hibberd, 1970). Pentosan adds a major water holding capacity to flour up to ten times their own weight though small in quantity (Bloksma and Bushuk, 1988). Non-flour ingredients play a role in dough properties as well. Not only does flour composition influence the baking quality (Magnus et al., 2000), other additives change rheology as well. It is known that gluten is responsive to the presence of salt as for 19 example, salt is added not only to modify taste but also for functional properties of dough, modification in dough cohesion and manageability in the baking industry (Mujumdar, 1987). With increasing salt content, water-binding increases, stickiness is reduced and elasticity decreased (Faubion & Hoseney, 1989; Chiotelli et al., 2004). In addition, incorporation of air is important for bubble formation and the rheology as well (Faubion & Hoseney, 1989). Air incorporation decreases specifically the density of the dough. Oxidation, known to make dough more elastic by attacking SS bonds, increases due to incorporated air (Faubion & Hoseney, 1989). Through mixing, gases are entrapped within the developing dough structure and bubble sizes normally range from 10 to 100 \im (Bloksma & Bushuk, 1988; Bloksma, 1990). It is estimated that 0.1 of the total volume are due to an increase through gas incorporation during mixing (Bloksma, 1990), which translates into the number of entrapped gas cells in the range of 10'2 to 1014 per square metre (van Vliet et al., 1992). Their shape is assumed to be nonspherical due to lack of space (van Vliet et al., 1992). 2.1.2 Hypotheses There are three hypotheses for this study: 1. Viscoelasticity of the dough has upper and lower thresholds between which microwave/vacuum exposure can result in puffing. The critical levels allow expansion through vacuum and steam response while avoiding collapse before drying is completed. 2. Expansion is dependent on microwave power supplied. There is a threshold above which the puffing response occurs. 20 3. Salt addition to the dough increases the microwave response by raising the dielectric loss factor and therefore increasing heat generation, which in turn will support more steam development and more puffing. Testing of the hypotheses was done by a series of experiments with doughs at different salt levels as well as different flour strength, drying trials at two power levels, raw dough rheology characterization, temperature development during drying, "dielectric properties measurement, and videorecording of the drying process. Dough samples were prepared from two different wheat flour types at varying salt levels to test the first hypothesis. The rheological properties were determined by uniaxial and biaxial extension tests. Dough samples were dried at different power levels to test the second hypothesis. Salt concentration in the dough formulation was varied from 0.0 to 1.5% NaCl to address hypothesis three. 2.2 Materials and Methods 2.2.1 Materials Commercial flours were used for dough. Best For Bread Homestyle White Flour (Robin Hood Multifoods Corporation, Markham, ON) served as high dough strength flour, for simplicity called high gluten flour (HG). Monarch Fancy Pastry Enriched Flour from A D M Milling Company A D M Agri-Industries Ltd (Windsor, ON) provided low dough strength flour (LG). The flour was purchased over a time period of 9 months, relying on quality consistency by the respective manufacturers. Windsor Iodized Table Salt was 21 food grade (Canadian Salt Company Limited, Pointe-Claire, QC). Distilled, deionized water was used for all experiments. 2.2.2 Basic dough preparation For 1750 g batches of dough, flour was weighed into a mixing bowl (20-25°C) and water (20-25°C) was weighed separately. Doughs consisted of 64.2-65.7% flour, 34.3% water and varying salt levels from 0.0% NaCl to 1.5%, in 0.1% -increments. For low concentrations, salt was dissolved in water and then added to the dry flour. For dough with > 0.6% salt, the dry ingredients were blended for 5 min, water was added and all quickly stirred with a dough hook. A l l ingredients were mixed in a 20 L-mixer (John Hunt Inc., Madison, WI) at 60 rpm for 10 min. The covered dough then rested for 30 min at room temperature (18-22°C). Halves of the dough were rolled out on a lightly oiled surface and rested for 10 min at room temperature. To control thickness, the wooden rolling pin was guided by two rods of 10 mm diameter. Uniform dough pieces were cut with a cork borer (diameter =10 mm). L G dough was cut with a 8 mm corkborer due to its brittleness. The dough pieces were then manually rolled into spheres and placed on a floured surface and covered until the batch was ready for drying. 2.2.3 Microwave drying under vacuum Before each drying run, the magnetron of the vacuum microwave dryer (Enwave Corp., Vancouver, BC, Figure 2.1) was warmed up by drying a sample batch of approximately 600 g of unused dough at 1300 W and 28 Torr chamber pressure for 7 min. Each of the 600 g-batches of dough balls was weighed into a polyethylene drum with a solid bottom 22 and a netted lid to provide good visibility. The drying proceeded at either medium (700 W) or high (1300 W) power, MP and HP respectively. Absolute chamber pressure was set at 28 Torr. Rotational speed of the drum was 4 rpm. A l l experiments were performed in triplicate. Figure 2.1 Vacuum microwave dryer, Enwave Corp. Vancouver, BC. On the left is the vacuum chamber holding the polyethylene drum on two bars of which one turns and rotates the drum. The wave guide leads the microwaves generated by the magnetron (top right) into the vacuum chamber. 2.2.4 Drying and heating rate Dough samples were prepared as previously described. Samples were taken at 1.5, 3, 4.5 and 6 min for HP and 3.5, 7.5 and 12 min for MP. Total batch weights, as well as 23 moisture content for individual samples, were determined by a standard method (AOAC, 1995). Immediately after full release of the vacuum and upon opening of the chamber, the temperature of the dough balls was measured using an infrared thermometer model 39650-04 (Cole Parmer Instruments, Co., Chicago Illinois, USA). 2.2.5 Storage Dried dough balls, once cooled to room temperature, were sealed Ziploc freezer bags (SC Johnson & Son Inc., Canada, Brantford, ON) and stored at room temperature. 2.2.6 Density Volume of dough balls was determined by displacement measurement with loose flax seeds before and after drying. A total of 25 dough balls were weighed into a graduated cylinder and covered adequately by loose flax seeds. Then the flax seeds' volume was determined. The difference between volume of the balls and flax seeds and flax seed volume alone yielded the total volume of the dough balls. Apparent density was determined as mass divided by volume of the dough balls. 2.2.7 Moisture Flour, dough and dried dough balls were weighed (wo) in pre-dried and pre-weighed aluminum weighing dishes and placed in a vacuum oven for 24 hours at 70°C and weighed after reaching room temperature in a desiccator (wf) (AACC, 1994). It was expressed as percent moisture per dry mass: 24 %Mdm ~ ' 100% (2.1) 2.2.8 Protein analysis Protein content of flour was determined by analysis using a LECO FP-528 Nitrogen/ Protein Determinator (LECO Corporation, St. Joseph, MI, USA), using the A A C C method No. 46-30/Dumas (combustion) method (1994). In this method the sample is flushed with oxygen for combustion at 850°C. Hot copper removes oxygen from the sample's gas, changing NOx to N 2 which in turn is measured by thermal conductivity. Protein was calculated using the factor 5.7 for wheat flour (Anonymous, 2003): % protein = N x 5.7 % d.b. (2.2) Samples were taken in triplicate from three random batches of flour. 2.2.9 Dielectric properties For all samples, relative dielectric constant E ' and loss factor E " were determined prior to drying for all NaCl concentrations at 24.5±1.0°C, using an open-ended coaxial probe and a network analyser (Hewlett Packard, Houston TX, USA, model HP 85070, software 85070 B Re B 0105). Relative dielectricities are unitless ratios of constants in test materials relative to those of empty space. Calibration was performed by measuring surrounding air, a metal short supplied by the manufacturer, and distilled water at 25°C. The coaxial probe was placed onto the dough.samples (width = 30 mm, length = 30 mm, depth = 15 mm) and fixed to the metal test stand while assuring close contact of the 25 sample and excluding air enclosures (Figure 2.2). Measurements at 2450 MHz were taken in triplicate at 25°C and £' and E " recorded. This was a modification of a method by Colpitis, Pelletier & Cogswell (1992). To verify the appropriate sample thickness, dough pieces were measured with and without aluminium foil placed under them. If the two measurements were equal, the thickness of the sample was deemed sufficient. v. j j J ° i 1 o data retrieval Figure 2.2 Schematic for dielectric properties determination setup. 2.2.10 Rheology 2.2.10.1 Uniaxial extension test Basic doughs of all formulations were prepared in advance as previously described. The rolling surface was dabbed with mineral oil to cover the rolling surface evenly, the dough sample of 35.5±0.2 g was rolled evenly out with a metal rod, and cut to fit two wooden plates (each 8.9 x 10.0 x 0.6 cm3). Excess dough was weighed to determine the weight of the sample to be tested. As well, the thickness of the enclosed dough was measured on all four corners by caliper and then averaged. To ensure proper measurements, the rolled out dough was weighed before and excess dough was weighed after. The difference in weight was devided by the average density and surface area to yield the thickness. The 26 s t a n d samp le ne twork analyser dough sample was secured between the two plates, each with a hole of 52 mm diameter and held in place on the platform of the TA.XT2i Texture Analyser™ (Stable Microsystems Ltd., Surrey, UK). The cylindrical probe (diameter = 50 mm) was lowered at a speed of 0.1 mm s"1 20 mm into the sample or until rupture occurred. The apparatus and method are shown as a schematic (Figure 2.3) and as images (Figure 2.4a-c). Strain and stress were determined from the measured force vs. displacement curves by (Beitz etal., 1994) L 2L E = L0 D-d (2.3) a = -f-. (2.4) d nt Under the assumption that the total volume of the deformed dough remains constant, the instantaneous dough thickness t can be written in terms of the original thickness to as t = - ^ , (2-5) 1 + £ thus equation (2.4) can be rewritten as a = - ^ — (1 + e). (2.6) d Jttg The rupture strength is defined as the maximum applied stress before rupture occurred, i.e., /?m=max(a). (2.7) 27 Figure 2.3 Schematic of uniaxial extension tests for wheat dough. Depicted is half the cross section of a Texture Analyser on the left, including the dough holder and the dough (red lines). On the right is the view from the top of the setup. L 0 = original length of the dough L = length after deformation t0 = original thickness of the dough t = thickness of the dough F = force applied d = diameter of the plunger D = diameter of the holder 28 Figure 2.4 a. Equipment used in uniaxial b. Texture Analyser piston being extension tests (from the top): lowered into a sheet of dough, oil, thermometer and hygrometer, rolling top view pin, blade, holders, infrared thermometer, sponge, sheeting c. Texture Analyser piston as it is lowered into a sheet of dough, bottom view 29 Young's modulus was determined by considering the center 2/3 portion of the stress-strain curve, i.e., 1/6 R m < a < 5/6 R m . From this, the slope of the curve was determined by fitting a straight line using a least-square approach (Holman, 2001), hence Young's modulus can be calculated by E = i i i (2.8) where n denotes the total number of recorded data points considered and i= 1, ..., n (Figure 2.5). 500 Fracture Strength (Rm) Figure 2.5 Sample stress-strain curve for a dough uniaxial extension test. The fracture strength (Rm) is defined as the maximum stress reached during the compression test. Young's modulus (E) is determined by fitting a straight line to the center 2/3 of the curve. E is defined as the slope of the straight line. 30 2.2.10.2 Biaxial extension test A Chopin alveograph, M A 82 (Chopin SA, Paris, France) was used to measure biaxial extension (Figure 2.6). Dough samples were prepared from 250 g flour, and the percentages of the recipe were kept in accordance with the original formulation. Total dough weight was 380.7 g. Doughs tested contained 0.0, 0.3, 0.5, 0.8, 1.0 and 1.5% of added salt for both HG and L G flour. Both types of flour were mixed for eight minutes, extruded and cut into small pieces and allowed to rest at 25°C in the tempering chamber. A dough sample was then placed in the 'vice', flattened, surface tension was released, and the bubble expanded through air flow, exerting constant resistance (Faridi & Rasper, 1987; Rasper & Danihelkova, 1986). Curves were recorded on original Chopin alveograph plotting paper and the following parameters were derived according to standard procedures (AACC, 1994): tenacity or resistance to extension P in millimetres P = / r l . l (2.9) with h being the maximum height of the curve, dough extensibility L (mm) equal to the abcissa at point of rupture, curve configuration ratio (P/L ratio), and deformation energy Win 10"4 J W =l.32V-^-S (2.10) with V being the volume of air and S the area under the curve. Each salt formulation and flour type was tested in triplicate, with individual dough samples tested five times. 31 Figure 2.6 Alveograph inflating a wheat dough sheet into a bubble 2.2.11 Video monitoring A small surveillance video camera (generic brand, 525 TV line resolution) was connected via an accelerator card (ATI Radeon 9200 vivo 128Mb DDR) to a PC. The camera supplied lighting in the form of LED lamps installed circularly around the lens. The video camera was inserted into a metal pipe, welded to the door (diameter = 40 mm), at the entrance of the vacuum chamber. In initial recordings, the camera was placed in the vacuum duct. Video images were recorded by Windows Movie Maker in x.WMV file format. For high and medium power drying, the highest resolution was chosen at 32 230 kilobites per second (kbps); due to lack of sufficient computer memory, low power drying recordings were saved in medium resolution (103 kbps) (Figure 2.7). Figure 2.7 Schematic for videorecording setup. 2.2.12 Statistical analysis Differences between treatments and their interactions were statistically analysed by analysis of variance A N O V A using SAS software (Version 6.12 TS 040, SAS Institute Inc., Cary, NC, USA). The General Linear Model G L M was used for the ANOVAs to manage the partially unbalanced nature of the data sets. Due to the nature of this program, variables are listed in the result tables even if they are not relevant to the model, however only with degrees of freedom equal to zero. Duncan's multiple range test served as a tool to distinguish means. 33 2.2.13 Image analysis 2.2.13.1 Sample preparation Dried dough balls were cut with a fine-toothed saw (12 teeth per centimetre) into even cross sections, placed on a black holder where the cross section's surface was positioned parallel to the lens. Four incandescent lamps were placed evenly around the sample holder with light illuminating the sample's surface from all sides (Figure 2.8). Images were taken using a Scalar Proscope (Bodelin Technologies, Lake Oswego, OR, USA) at a magnification lens 1-10. The objects were placed at 48 mm distance to the lens, and then saved as x.jpeg files, at the setting of 530 ppi. Using GraphicConverter V 4.6 (Lemke Software GmbH, Peine, Germany), the images' brightness was minimized and the contrast maximized. For each salt formulation of HG 10x2 cross sections were imaged. Pro-Scope stand sample holder Figure 2.8 Image analysis sample image capture for cross sections 34 2.2.13.2 Image analysis program Images were analysed by use of the image analysis program DB2, developed by Ph.D. candidate Qing Wang (Department of Computer Science, UBC). Once converted to grey scale, the program searched for areas below an adjustable darkness threshold. A closed loop was formed around the edges of these areas, which it equated to pores. From this pore, average circumference, average size as well as the largest pore's average were computed (Figure 2.9 a-d), based on the pixel count and its defined size of the pixel in the image analyzed. Due to frequent non-detection of pores within the cross sectional area of the cut dough balls, partially due to the lighting gradient created by the experimental setup, partially due to constraints in the program algorithms, the total number of pores needed to be estimated manually. Images were individually evaluated and an accuracy rating was assessed. The DB2 results was multiplied by this accuracy rating yielding the estimated pore measurements. 35 Figure 2.9 DB2 program selected commands a. Original cross-sectional picture of dried doughball. Ready d. Final calculation results upon 'pore stats' command. 37 2.3 Results 2.3.1 Microwave drying under vacuum In microwave dehydration tests assisted by vacuum, dough from HG flour dough was tested at both medium and high power levels (MP, HP) whereas tests for L G flour dough were completed only at HP. Tumbling of the dough balls within the rotating drum was assumed to move the balls through the microwave field sufficiently to obtain equivalent microwave exposure for each sphere. This was confirmed by the appearance of the dough balls, relatively evenly shaped round spheres and devoid of burn or browning marks. Drying times were arbitrarily chosen based on the finding that on average in the total drying process 33 g of water was removed per minute for HP and 16 g per minute for MP, approximately doubling the drying time from 1300 W to 700 W power settings. Starting weight and volume were the same for all drying runs (Figure 2.10, 2.11, 2.12). Final weights were the same due to choosing the same drying times, which consistently yielded the same moisture contents. For HP and MP combined, final moisture content was 6.51±0.81% and 6.55±1.53% for HG and L G dough balls respectively. Volume developments were noteworthy. For HP and MP dried HG samples, volume decreases were very similar in their linear response to increasing salt content. Yet, HP induced lower volume changes than did MP. This was confirmed statistically as HP and MP-dried HG dough balls did differ in their response slightly, and power level was a significant factor for volume change. In contrast, HP-drying of L G dough induced rather small expansion of dough balls compared to HG doughs. 38 500 450 400 + 350 , 300 5- 250 200 150 100 50 A 2 A £ • initial volume © final volume A initial weight A final weight Linear (final volume) =F 200 180 60 + 40 4- 20 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Salt (%) 1.1 1.2 1.3 1.4 1.5 Figure 2.10 Changes for HG at high power drying for different salt concentrations where the left y-axis show volume of 25 dough balls pre and post drying. The right y-axis depicts the weight of 25 dough balls in grams pre and post drying. Initial volume = closed diamonds; final volume = open diamonds; initial weight = closed triangles; final weight = open triangles. Error bars, when not obscured by the data point markers, indicate +/- one standard deviation. 39 500 200 450 • 400 350 + - 300 + •a % o Q- 250 £ Q. a> E 200 180 160 140 120 o 100 fi-ll) a D) 80 150 + * * * * * * * * * * * * * i • • 60 100 + 50 + _l 1 1 1 1 1 1 1 1 1 (-4- 40 • initial volume © final volume A initial weight A final weight Linear (final volume) 20 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 Salt (%) Figure 2.11 Changes for HG at medium power drying for different salt concentrations where the left y-axis show volume of 25 dough balls pre and post drying. The right y-axis depicts the weight of 25 dough balls in grams pre and post drying. Initial volume = closed diamonds; final volume = open diamonds; initial weight = closed triangles; final weight = open triangles. Error bars, when not obscured by the data point markers, indicate +/- one standard deviation. 40 500.0 450.0 -f 400.0 • Initial volume O Final volume A Initial weight A Final weight Linear (Final volume) 200.0 180.0 + 160.0 350.0 4- 140.0 300.0 120.0 | 250.0 + o > 200.0 A A A A A A A 100.0 80.0 150.0 + + 60.0 100.0 4-50.0 A A • « A ffi & 0. A A A A A A A A t * • -5 S- £ i . * i * I • i 1 40.0 20.0 0.0 -I H H 1 1 1 1 1 1- H 1- 0.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 Salt content (%) Figure 2.12 Changes for L G at high power drying for different salt concentrations where the left y-axis show volume of 25 dough balls pre and post drying. The right y-axis depicts the weight of 25 dough balls in grams pre and post drying. Initial volume = closed diamonds; final volume = open diamonds; initial weight = closed triangles; final weight = open triangles. Error bars, when not obscured by the data point markers, indicate +/- one standard deviation. 41 Nevertheless puffing was observed in all dough types and under both power levels. For ease of comparison, volume change, expressed in percent, was graphed directly for HG and LG flour doughs at HP and MP and HP for HG flour, as well (Figure 2.13, 2.14). HG exhibited the greatest volume change, the largest average expansion measured approximately 180% for 0.0% salt and the lowest about 80% for 1.5% NaCl. In contrast, LG exhibited only limited volume increases for low salt, showing the same trend of decreased expansion with increasing salt levels, between 50% and below 20% increase (0.0-1.5% NaCl, respectively). Applied power levels, salt content, flour type and 8", all significantly influenced the volume change (P = 0.0001), (Table 2.2). The day on which the experiments were run (DAY) added to variability. Though all experimental results exhibited large standard deviations, LG flour dough experiments showed particularly variable results (Figure 2.13). 2.3.2 Heating rate Initial heating behaviour was measured by taking internal temperatures every 30 s with a needle-tip thermocouple. After 20 s, microwave radiation was added to the vacuum chamber. The temperature rose an average 9.6±2.7°C to 37.07±2.6°C at 120 s. At 180 s, the temperature then observed appeared to decrease to an average 29.27±0.4°C (Table 2.3). Al l measurements were taken for HG dough without added salt. 42 Figure 2.13 Volume change (%) for HG and LG doughs at varying salt concentrations for high power drying. HG = open squares, L G = open diamonds. The error bars, if not obscured by the point markers, indicate +/- one standard deviation of 3 drying runs. 43 250 200 E a 100 O HG Medium power • HG High power ] < ) i L ( ] i > [ ( < > c i . [ 1 EDM c ] 1 ; ; > < r > I i ' ) L [ ] [ E t 1 * f J i • 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 Sal t concentrat ion (%) Figure 2.14 Volume change (%) of HG dough at MP and HP drying. HG, HP = open squares, HG, MP = open circles. The error bars, when not obscured by point markers, show +/- one standard deviation of 3 drying runs. 44 Table 2.2 A N O V A results for volume change in HG at high and medium power and L G dried at high power General Linear Models Procedure Sum of Mean Source DF Squares Square F Value Pr > F Model 105 1267043.36 12067.08 65.06 0.0001 Error 22 4080.22 185.46 Corrected Total 127 1271123.59 R-Square C.V. Root MSE VCAVE Mean 0.996790 11.47828 13.6185 118.646 Source DF Type I SS Mean Square F Value Pr > F SALT 15 52880 182 3525 345 19 01 0 0001 FLOUR 1 916244 682 916244 682 4940 27 0 0001 SALT*FLOUR 15 16723 475 1114 898 6 01 0 0001 POWER 1 237282 307 237282 307 1279 39 0 0001 SALT*P0WER 15 8267 700 551 180 2 97 0 0101 POWER*FLOUR 0 0 000 SALT*P0WER*FL0UR 0 0 000 DAY 58 35645 017 614 569 3 31 0 0015 Table 2.3 Temperature development for 0.0% salt HG dough under vacuum microwave drying. Drying time 0 s 40 s 60s 90s 120 s 180 s Temperature ^ 5 5 9 g j g l Q J 6 g 5 5 ? 9 Q increase (LJ 0 ^ ^ ^ ^ ^ ^ ^ J 2 ) ^ ^ Numbers are averages from three drying runs, one standard deviation is listed in brackets. 45 2.3.3 Drying rate and moisture content The total loss of moisture in percent during the drying process was very similar for HG and LG dough at high power, steadily removing moisture from the system (Table 2.4). Moisture loss for HG dough MP-dried at half this rate. When converting drying rate into grams of water removed per minute considering the batch weight of approximately 600 g, the following rates were obtained. Though at high power for HG, the 90 s interval removed 19.9 g water per minute, 27.7 g/min, 25.7 g/min and in the last interval 17.1 g/min, this appears to be random scatter. When taking the final moisture contents into consideration, the last 90 s interval 66.0 g of water per Table 2.4 Water content, in percent of total mass, for HG and L G dough, 0.0% salt respectively, with regard to water loss in high and medium power (HP, MP) drying over time Drying time (min) Power Testing Flour level time 1.5 3 4.5 6 7.6 9 12 15 HG HP pre 38.88 38.87 38.51 38.71 38.61 - - -HG HP post 33.9 26.95 20.17 16.11 6.47 - - -HG MP pre - 38.18 - 38.63 - 38.64 38.89 38.55 HG MP post 33.25 - 27.31 - 19.49 10.11 5.43 LG HP pre 38.93 38.68 38.63 39.11 38.82 - - -LG HP post 33.01 28.13 21.33 14.37 6.55 - - -minute were removed while the average rate overall was 31.3 g/min. The other experiments showed similar drying behaviour. For LG at HP, water removal was 23.7 g/min, 18.6 g/min, 27.0 g/min and 29.8 g/min. However, the final 90 second interval, until drying was completed, yielded a rate of 50.7 g/min. The overall average rate was 29.9 g of removed water per minute. For MP, moisture loss in HG dough was 46 9.8 g/min for the first 3 min, 12.8 g/min, 15.6 g/min and 19.3 g of water per minute in the following intervals. Only here the final three minutes showed a similar rate of 18.8 g/min. The overall drying rate was 15.3 g/min. The overall drying rates reasonably matched previously determined drying rates of 33 g and 16 g of water removed per minute based on total loss versus total drying time for high and medium power respectively. When looking at total water loss over time, the drying rate appeared linear as drying continued (Figure 2.15). The points graphed include the final measurement at the drying end point. However, since the rate increased dramatically in the last 1.5 and 3 min of high and medium power drying respectively, the trend lines did not include these data points to illustrate the linearity leading up to this period in drying time. 2.3.4 Video monitoring The majority of the drying runs were recorded for later viewing. Due to initial technical difficulties caused by the position of the video camera in the air evacuation duct, viewing quality was poor for many of the recordings because loose flour turbulently exited the chamber and obscured the view. Thus not all of the recordings could be used for evaluation and analysis. However, each formulation of L G and HG was recorded at good quality at least once. Due to the limited amount of useable drying runs, statistical analysis could not be applied to evaluate the drying event times. 47 Figure 2.15 Water loss (g) over time for HG and L G dough at high and medium power (HG, HP= open squares; HG, MP=open circles; LG, HP=open triangles). Final points for each dough type were taken at the very end of the drying process. At HP, HG doughs expanded as soon as air expulsion from the chamber was initiated. At low salt concentrations, dough balls expanded rapidly until the final pressure of 28 Torr was reached. Then, as microwave energy was introduced at approximately 20 s into the run, the expansion continued to a lesser degree, lasting until about 100 s into the drying process. During the interval of 0 to 100s, all HG formulations exhibited expansion. The extent of this was not determined due to the difficulty in distinguishing clearly the outline of the individual dough balls. For low salt HG formulations, the expansion was so extensive that the load of dough balls filled the entire drum cavity. Since all balls were 48 coloured the same, distinction between individual dough balls was very difficult. The higher the salt concentration, the lower the expansion and easier it was to separate the sillouettes of individual balls. Nevertheless, there was a distinct difference between HG and LG response. HG dough with low salt concentrations expanded greatly upon the start of air evacuation, approximately doubling in volume. Once the microwaves were channeled into the chamber, the additional expansion amounted to over four-fold the volume compared to the original size within 40 s of microwave exposure. Further in the drying process, the volume decreased slowly to about double the original volume at approximately 100 s of microwave power exposure, at which it then remained constant to the end. In this shrinking process, the shape of the dough balls became increasingly irregular, likely due to the balls' bouncing and constant contact. This ratio of immediate expansion under lack of pressure and further heating was particularly remarkable in no-or low-salt formulations. The timing for the expansion was the same for all runs, across formulations, power levels and flour type. LG dough at high power expanded slightly upon air evacuation. The degree of expansion was much lower than the HG formulations, indicated by the fact that the field of vision of the camera was never entirely blocked and the balls continued to tumble unobstructed. There was no perceived difference in size between the phase of reduced pressure and the following heating/drying phase under microwave exposure. Likewise, the different salt concentrations did not seem to have an impact on the amount of expansion under vacuum or microwave exposure. This is also reflected in a similar fashion in the final volumes throughout the L G dough formulations, as mentioned in the section above. 49 In MP drying runs, the HG dough expansion also immediately responded during the time necessary to reach 28 Torr absolute pressure. The behaviour was similar to HG dough dried at HP when comparing the different salt concentrations. After 100 s, further expansion was not evident. 2.3.5 Protein analysis LECO analysis yielded a protein content for HG flour of 13.57±0.04 g/100 g and 10.13±0.05 g/lOOg for L G flour. While HG flour was close to typical results for hard wheats of 14.9 g/100 g, L G was higher than reference values for soft wheat (Anonymous, 2003). 2.3.6 Rheology Young's modulus derived from uniaxial extension tests did not exhibit a clear salt response. Though the means suggest an increase of E with increasing salt concentrations, depicted in solid circles in Figure 2.16, statistically this could not be confirmed, likely due to very large standard deviations throughout the testing trials. The lowest E was determined for 0.1% salt dough at 0.05 MPa, the highest at 1.5% at approximately 0.09 MPa. For this reason, fracture strength was consulted as these measurements were much more consistent throughout the testing. Fracture strength results exhibited a positive linear relationship with increasing salt concentrations (Figure 2.17), (P = 0.001). The averages of fracture strength ranged from approximately 0.3 MPa to 0.55 MPa for 0.0% to 1.5% salt, respectively, the concentrations in between showing proportional results. 50 0:;14 0.12 0.1 10 5 0.08 o > o.o6: 0.04 002 u means within subsamples of replicates; :#, means of.all.replicates 0.1 0.2 0.3 04 0.5 0 6 0.7 0.8 0.9 1 1.1 1.2 13 1.4 1.5 Salt Content [%] Figure 2.16 Young's modulus E (MPa) for uniaxial extension of HP dough at different salt concentrations. Open squares symbolize means within subsamples of replicates, the solid circles represent means of all replicates. The error bars depict +/-one standard'deviation for subsamples of replicates. Biaxial extension of LG and HG yielded similar trends for the parameters examined, namely L, P, W, A, and P/L ratio. Al l responses for HG were several fold higher than LG. Surprisingly, dough viscosity P, also refered to as the resistance to deformation and considered essential for gas retention, was not a statistically significant factor for volume expansion (Figure 2.18 a). However, when viewing the results, the data for HG shows a poor match in response to added salt as there appears to be no change as salt increases. The LG response to salt is consistent with the expectations that with increasing salt concentration, P increases. Flour type and salt interaction with flour significantly influenced P values (P = 0.0001, Table 2.5a). For the extensibility value L, HG and LG 51 0 8: 0.7-;0;6 • means v/ithin subsamples of replicates ;<j means of ail replicates • to a. 0 5 0.4 :-o:3 0.-2 0.11 4 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 Salt Content [%] Figure 2.17 Fracture strength Rm (MPa) for uniaxial extension of HP dough at different salt concentrations. Open squares symbolize means within subsamples of replicates, the solid circles represent means of all replicates. The error bars depict +/-one standard deviation for subsamples of replicates. curves were the same with regard to salt response, increasing as salt increased (P = 0.001); HG responses were three-times the values of L G (Figure 2.18b). The deformation energy W, a flour strength indicator with respect to baked volume performance, exhibited a clear salt response for both L G and HG though the L G response was much lower (Figure 2.18c). Though HG's values are overall higher, LG's salt response toward W increases several fold. Thus salt exhibit more of dough strengthening power for L G dough, likely due to the fact that this flour has poor dough strength to begin with. Both parameters L and W were significant dependents of salt, flour type as well as their combination for the raw dough (P = 0.0001, respectively), (Table 2.5b-c). 52 160 140 120 100 Q_ 1 80 CO m 137 60 40 20 $ 46 m 146 Ltl 128.5 ?E 57 | 145 0 64 Ltl 130.5 Ltl 128 y = 11.138X + 50.473 R 2 = 0.7407 O LG • HG Linear (LG) $ 64 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 Salt % Figure 2.18 a Alveograph tenacity P values for selected salt formulations for low and high gluten dough, L G and HG. The values are the averages from 5 runs per trial, all performed in triplicate, error bars represent +/- one standard deviation. 53 Figure 2.18 b Alveograph L (abcissa of rupture) for selected salt formulations for low and high gluten dough, L G and HG. The values are the averages from 5 runs per trial, all performed in triplicate, error bars represent +/- one standard deviation. 54 700 600 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 Salt % Figure 2.18 c Alveograph deformation energy WOO"4 J) for selected salt formulations for low and high gluten dough, LG and HG. The values are the average from 5 runs per trial, all performed in triplicate, error bars represent +/- one standard deviation. 55 Table 2.5 a A N O V A determined by G L M for alveograph results for the parameter P, the resistance to extension Source Model Error Corrected Total DF 13 4 17 R-Square 0.978416 Sum of Squares 24542.7025 541.4225 25084.1250 C.V. 5.871282 Mean Square F Value 6135.6756 147.32 41.6479 Root MSE 6.45352 Pr > F 0.0001 PALV Mean 109.917 Source FLOUR SALT FLOUR*SALT REP DF 1 5 5 2 Type I SS Mean Square F Value Pr > F 24180.2500 5.0241 319.8162 37.6122 24180.2500 5.0241 319.8162 37.6122 580.59 0.12 7.68 0.90 0.0001 0.7339 0.0159 0.3593 Table 2.5 b A N O V A determined by G L M for alveograph results for the parameter L, the dough extensibility Source DF Sum of Squares Mean Square F Value Pr > F Model 13 15726.4199 3931.6050 156.95 0.0001 Error 4 325.6495 25.0500 Corrected Total 17 16052.0694 R-Square C.V. Root MSE LALV Mean 0.979713 5.674954 5.00499 88.1944 Source DF Type I SS Mean Square F Value Pr > F FLOUR 1 13301.7778 13301.7778 531.01 0.0001 SALT 5 2305.1684 2305.1684 92.02 0.0001 FLOUR*SALT 5 112.6514 112.6514 4.50 0.0538 REP 2 6.8223 6.8223 0.27 0.6105 56 Table 2.5 c A N O V A determined by G L M for alveograph results for the parameter W, the deformation energy Sum of Mean Source DF Squares Square F Value Pr > F Model 13 711899.320 177974.830 513.72 0.0001 Error 4 4503.805 346.447 Corrected Total 17 716403.125 R-Square C.V. Root MSE WALV Mean 0.993713 5.033962 18.6131 369.750 Source DF Type I SS Mean Square F Value Pr > F FLOUR 1 685584.000 . 685584.000 1978.90 0.0001 SALT 5 23458.993 23458.993 67.71 0.0001 FLOUR*SALT 5 196.106 196.106 0.57 0.4652 REP 2 2660.221 2660.221 7.68 0.0159 2.3.7 Dielectric properties While E ' hardly changed with addition of salt to the doughs, £" did increase steadily, as expected (Table 2.6). The variability was quite noticeable. Salt may not have been uniformly distributed throughout the different formulations. Though caution was exercised, the mixing took place at a temperature range of 22±1.5°C. This may have contributed to incorporation of more or less air into the dough, dependent on the rheological properties of the formulation, which would be reflected in the dielectric properties as well. Some data for L G is missing due to technical difficulties. 57 Table 2.6 Formulations for HG and L G dough with their respective salt content and the corresponding dielectric properties. salt% HG L G 0.0 25.22 (3.28) 8.54(1.98) 24.39 (2.28) 12.93 (0.90) 0.1 26.35 (4.34) 9.01 (4.12) 28.51 (0.97) 11.70 (0.58) 0.2 26.03 (2.88) 8.96(1.84) 23.73 (4.81) 9.61 (4.81) 0.3 25.47 (2.99) 8.48 (2.02) 23.19(4.24) 9.90 (2.02) 0.4 28.93 (6.76) 10.84 (3.01) 21.74 (7.40) 9.68 (3.50) 0.5 25.76(1.99) 10.95 (2.72) 25.28 (1.84) 9.73 (0.87) 0.6 25.90 (2.58) 10.56(1.40) 23.64 (5.78) 10.81 (3.16) 0.7 26.65 (3.59) 10.51 (2.71) 39.47 (9.08) 18.00 (4.05) 0.8 25.45 (2.62) 11.36(2.31) n.d. n.d. 0.9 25.74(1.92) 10.66(1.11) 24.49 (2.44) 10.89 (1.25) 1.0 26.45 (6.29) 11.52 (3.75) 26.44(1.19) 11.74 (0.53) 1.1 26.53 (2.36) 11.84(1.72) n.d. n.d. 1.2 26.25 (4.58) 12.80 (2.85) 32.12(0.35) 12.14(0.17) 1.3 26.09 (3.53) 12.76 (3.10) 23.71 (2.31 11.58 (1.27) 1.4 31.06 (7.15) 15.30(4.87) 25.40(1.78) 12.62 (0.85) 1.5 24.82 (3.33) 12.93 (2.33) 24.69 (2.57) 13.07(1.68) Numbers are averages from three tests, the standard deviation listed in brackets 58 2.3.8 Image analysis The cross sectional area of HG dough balls dried at high power showed a decreasing volume with increasing salt concentrations. The estimated cross sectional area consistently overestimated the actually measured area by about one third but did so following the identical linear declining trend with increasing salt concentration (Figure 2.19). For dough without added salt, the measured area ranged from 328.6 mm2; at 1.5% salt dough, the area was determined at an average 244.8 mm2. The average pore size enlarged as salt concentrations increased. While at low salt concentrations the average size did not differ, dough at above 0.6% salt exhibited a slight increase. The number of pores at low salt concentrations did not differ. In contrast, for dough at above 0.5% salt, the total number of pores apparently decreased (Figure 2.20). Neither of these results could be confirmed statistically. Even after removing seemingly aberrant points such as 0.3 and 0.6% salt samples, the r2 for the A N O V A G L M were below 0.55 and thus cannot be considered relevant. 59 600.00 500.00 400.00 rM < E E re <u u re m 300. 200.00 100.00 0.00 y = -53.108X + 308.34 0.4769 O Cross sectional area, estimated • Cross sectional area, measured Linear (Cross sectional area, estimated) Linear (Cross sectional area, measured) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 S a l t ( % ) Figure 2.19 Total cross sectional area of dried HG dough balls at high power, estimated (open diamonds) and measured (solid squares) compared to salt concentration. 60 1.60 4-y = -3.2456x + 171.5 R2 = 0.4211 y = 0.0157x + 0.5722 R2 = 0.365 • • average pores size O extrapolated # pores Linear (extrapolated # pores) Linear (average pores size) 250.00 200.00 150.00 + 100.00 50.00 0.00 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Salt (%) 0.9 1.1 1.2 1.3 1.4 1.5 Figure 2.20 Average pores size of HP-dried HG dough ball cross sections (open squares) compared to the number of pores (open diamonds). 61 2.4 Discussion The hypothesis for the volume expansions induced in V M drying was that different microwave power levels and formulations, i.e., varying protein content and type of flour as well as salt concentrations, would exhibit different degrees of expansion in the dried samples. This was based on the fact that higher dielectric loss factors would exhibit more rapid microwave heating. While HG flour showed fairly typical protein content for hard wheats of 14.9 g/100 g, L G protein content of 10.1 g was higher than the average for soft wheat, reported as 8.9 g/100 g (Anonymous, 2003). It matched the protein content of soft wheat and cracker wheat flour used to examine mixing development in dough (Lee et al., 2001). Though the amount of protein alone is not enough to evaluate the flour quality and its rheological properties, both flours were considered typical in their viscoelasticity for poor and good bread making flours. This is based on the comparison of mixing optima expressed as Mixograph peaktimes, although neither data are provided here (Magnus et al., 2000; Keentok et al., 2002). The possibility that one of the flour doughs was not properly mixed has to be considered. The Mixograph peaks could not be used for evaluation of over or undermixing because the mixing process in this type of testing apparatus did not match that of the mixer used in this setup. Over-mixing can destroy the viscoelastic properties of dough (Faubion & Hoseney, 1989; Lee et al., 2001). It is possible that L G doughs were over-mixed since their response to expansion was manyfold lower than the ones of HG. Lower baking quality flours need shorter mixing times for optimum development (Keentok et al., 2000; Rao et al., 2000; Hayta & Schofield, 2004). When weak flour dough was mixed at the optimum time to develop a 62 comparable strong one, it exhibited a loss of 62% of the original resistance (Janssen et al., 1996a). On the other hand, HG doughs may have been underdeveloped. In this case, the generally continuous network would not be as extensive as typical for this normally viscoelastical material. Protein patches would exist, interrupted by hydrated starch islet (Janssen et al., 1996a, b; Lee et al., 2001). This would explain the higher standard deviations for HG, also observed in the uniaxial extension trials where the identical dough preparation was used. In contrast, biaxial extension results were not as variable. It is to note that here, the doughs were prepared using the alveograph mixer, a device with a much higher force output than the mixer used in these experiments. In addition, one needs to take into account that not only protein content differed between the types of flour but also the protein composition and starch content. The composition and content both influence the quality (Pomeranz, 1988; Gupta et al., 1993; Janssen et al., 1996a; Magnus et al., 2000). Starch has more than filler qualities in dough rheology since its addition emphasizes the non-linear behaviour of dough in stress and strain applications (Hibberd, 1970). The initial expansion under vacuum exposure for MP is equivalent or at least similar to HP. Yet, the absolute volume at the end of drying was greater. Though the temperatures rose at a similar rate as in HP drying, the drying rate was lower and drying time accordingly longer. In HG doughs, water removal appeared to occur too rapidly to allow 63 stabilization of the puffed structure. The optimum energy level for expansion to occur needs to be determined and may lie between the two power levels chosen or even below MP. As water is removed, the rheological properties of the dough change. Janssen and colleagues (1996a) suggested that the higher resistance as the water content is lowered is due to a concentration effect rather than the lack of water as a plasticizer. The protein properties are highly dependent on water to exhibit the viscoelastic characteristics since they rely on charge. Hydrogen bonding will not take place in the same manner when water is limited. The hygroscopic nature of salt is more pronounced as water content becomes limited, thus affecting the local hydration levels of the gluten network. The 8" values matched the results of starch/water 1:2 mixtures measured at 2.4 GHz by Ndife and colleagues (1998). Zuercher and colleagues (1990) investigated commercial ready-bread dough of unknown composition and ingredients and sampled it prior to and throughout baking. At 2 GHz, the standard dough measured e' equaling 22, increasing with added water and decreasing accordingly when flour was added. The loss factor of 12 was determined at the same frequency. For samples taken during baking, 8' at 2.4 GHz dropped to 12, 9.5 and 4.5 at the 10, 20 and 30 min intervals, respectively. Meanwhile, 8" decreased from 4.5 at 10 min to 1.2 at 30 min, at which time baking was completed. Therefore, with decreasing water content, both s' and 8" dropped significantly, which is to be expected in this drying process as well. Literature values for wheat flour at 14% moisture at 1 GHz was determined to be an e'of 4.5 and 8" of 3.2 (Meredith, 1998). For 2.4 GHz, no data was available but it was presumed to be lower as 64 E ' and £" values drop with rising frequency (Zuercher et al., 1990). No useable values for dried dough balls were measured in this study due to the inability to maintain proper contact between the probe and the sample surface. The penetration depth of microwaves into dough is considered low at 8-10 mm compared to 10-15 mm in other foods (Mellgren et al., 1988). When measuring dielectric properties of eight different flours with varying water contents during heating, the increase in £' before reaching 100°C, for dry flour, or with 10% water content, shows that the water interacts with the polysaccharide matrix and thus is bound. When water evaporates, £' decreases to stable levels in a dry product (Seras et al., 1987). Although £' varies with water content, decreasing as water is lost, temperature has only a minor effect (Kent & Kress-Rogers, 1987; Venkatesh & Raghavan, 2004). If water content is low, E ' and £" are equally low for flours at microwave frequencies (Kent & Kress-Rogers, 1987). The measured density is in line with previous findings: for 42-46% moisture content ranging from 1.03 to 1.1 g/cm3 (Sablani & Rahman, 2002) and for 55% moisture content approximately 1.2 g/cm (Chin et al., 2005). However it is important to note that with increasing salt content, density increased as well (P = 0.02), which is related to the decreased incorporation of gases into the dough (Chin et al., 2005). The observed differences between HG and LG are reflected in the literature as well. LG flour doughs disperse elastic strain faster than HG doughs (Tschoegl et al., 1970; Janssen et al., 1996a, b). This is mainly due to the less developed gluten network necessary to withstand the strain in processing. The nature of the proteins do not allow the extensive 65 cross linking by disulfide bonds in weak flour dough. Rather, the stability of the dough would rely on noncovalent bonds that can rearrange during a stress response (Lee et al., 2001). This would explain why LG dough does not expand as much during V M drying. Because softer doughs disperse strain faster, they also occlude more air (Campbell et al., 1993; Chin, 2003; Chin et al., 2005). Dough density increases with increasing levels of salt (Chin et al., 2005), suggesting that less gases are being folded into the dough. The general size of entrapped gas bubbles ranges between 10 and 100 Ltm (Bloksma, 1990). However, low speed mixing, as applied in these experiments, may have incorporated more air than in other processing methods, as suggested by Autio & Laurikainen (1997). It has been theorized that oxygen is physically adsorbed by starch granules instead of chemically reacting with dough components. On the other hand, only considering adsorption does not explain the fact that dough contains five-fold the oxygen of water alone (Xu, 2001). Though all dough formulations and applied power levels yielded volume increases, it is evident that volume change was dependent on salt content, likely due to the incurred change in rheological properties. Dough for all formulations had similar E values at 21.0±1.0°C taking the range into consideration. In extensional experiments by Keentok and colleagues (2002), curves for different commercially strong and weak quality wheat flours were very similar to the current results. The E value alone did not show the expected sharp decrease in elasticity in higher salt doughs. Nevertheless, the maximum stress value, meaning the dough's fracture strength, showed a positive relationship with increasing salt concentrations. In comparison of weak and strong wheat flours, the peak 66 stresses were not consistently higher or lower for either quality of flour and could not be correlated with the protein content alone. However, the strongest had the highest peak stress, the weakest the lowest. More significantly, the peak strain clearly distinguished between strong and weak flour facilitating discrimination of the flours by this measure (Keentoketal.,2002). Stress and strain in dough decrease as temperature increases until protein denaturation occurs (Sliwinski et al., 2004a, b, Angioloni & Dalla Rossa, 2005). This has been explained by the fact that at temperatures up to 45°C, sulfhydryl bonds dominate and non-covalent bonds become less important (Sliwinski et al., 2004b). Below 45°C, changes induced by heating are considered reversible (Faubion & Hoseney, 1989). SH-content showed little change up to 30°C but a steady decline to about half the original amount in high gluten flour and a decrease of 25% of the starting amount whereas SS-content hardly changed for both types of flour (Hayta & Schofield, 2004). Thus, it can be hypothesized that at a higher temperature, salt formulations would cause a stiffer, less elastic dough due to the hydrophilic nature of salt. Therefore salt would bind the water that otherwise would make the gluten structure more pliable. This temperature range was achieved in the current experiments and a lower stress and strain rate has been suggested as part of the theory to explain the induced expansion. Furthermore, the fact that salt containing doughs did not expand as much can be explained by these findings, in particular at concentration above 0.8% hardly expanded at all. In oscillatory measurements including 0, 1.5 and 3% salt formulations, higher salt doughs showed a higher storage modulus G ' above 55°C which then leveled off at or above 80°C. The 67 same was evident for a series of 0, 2.5, 3.5 and 4.5% salt doughs made of weak flour (Chiotelli et al., 2004; Angioloni & Dalla Rossa, 2005). From the graphs supplied by Chiotelli and colleagues (2004) it is difficult to determine whether the differences below 55°C were significant while higher salt formulations showed slightly higher G ' values. However, data published by Angioloni and Dalla Rossa (2005) seems to support this. Though oscillatory experiments were not conducted, these results (Chiotelli et al., 2004; Angioloni & Dalla Rossa, 2005) suggest that salt in the ranges added to the dough in the current experiments contributed to stiffening as well. Statistical analysis did reveal that salt, flour and their interaction influenced the outcome of the volume expansion. While no salt or low salt formulations for HG doughs expanded greatly at 28 Torr absolute pressure, the degree to which higher salt formulations expanded initially was reduced. The further expansion with continued heating occurred only in lower salt formulations. As reviewed previously, polarity is the key to understanding dough rheology. Salt will reduce hydration of protein. In the process of drying, this requirement for elastic behaviour to occur is reduced. Thus, the high salt-gluten interaction sufficiently stiffens the dough to prevent expansion. It appears that puffing is less dependent on steam buildup than expansion of incorporated gases and temperature as shown in previous studies. In the temperature range comparable to this study, bubble formation and expansion were influenced by temperature as the single most important parameter due to the low viscosity (Fan et al., 1999). 68 Initial temperature development was especially important to determine the change of rheological properties of the dough since rising temperature makes dough softer and more expandable. In alveograph experiments conducted at temperatures of 10, 15, 20 and 30°C, the curves at the highest temperature clearly exhibited the lowest stress/strain response, meaning doughs softened with increasing temperature (Rasper & Danihelkova, 1986). The same is true for sweet dough extensibility when measured at <19, 19-22, 22-25, and 25-29°C temperature ranges with uniaxial extensions by an extensograph (Calderon-Dominguez et al., 2004). This temperature range was reached prior to the sample time at which the initial temperature measurement was made in the current experiments. Notably in this work, microwave radiation did not commence until the desired pressure condition had been reached after approximately 20 s. Starting temperatures were on average 24°C. After 180 s, the dough in the chamber reached 35-41°C. There is no evidence that the temperature increased further immediately after this point in time, confirming earlier findings (Sham et al., 2001). However at the end of drying, temperatures were measured in the range of 65 to 80°C. It is noted that this temperature increase occured late in the drying process. This was also observed by Sham and colleagues (2001) and Yaghmaee and Durance (2005). Organic compounds with e' values below 3 and e" below 0.1 are dielectrically irrelevant and do not react to microwaves at average moisture levels (Mudgett, 1985). Only when water content is so low that it is considered mostly bound and does not react with microwave radiation do the components with a low loss factor become the major factors in heating (Mudgett, 1985). At this time, drying efficiency decreases (Kaensup et al., 2002). 69 The irregular heat development at the end of microwave drying is repeatedly cited as a drawback for microwave drying as the temperature fluctuation for this type of drying method is known (van der Veen et al., 2004). This was also reflected in the increased drying rate in the final minutes of the process as the rate almost doubled for high power dried HG and L G doughs. Air-drying at a constant temperature to reach the desired moisture content in the last stages of drying could circumvent this problem. Under atmospheric pressure, the dough balls had the measured initial volume, i.e., in a compressed state under normal pressure. During gas evacuation from the vacuum chamber, the atmospheric pressure dropped to the absolute pressure of 28 Torr (3.7 kPa) This lack of pressure allowed decompression of the dough and therefore a volume expansion was observed in the first 20 s, just as the gas evacuation begins. Microwaves were added as soon as the desired pressure had been reached and heating could begin. Within a short time, the temperature of the dough was approximately 35-41 °C. The continuing expansion needs to be explained. However, at the time of expansion, it is highly unlikely that the steam build up within the dough balls would be able to create a pressure differential. Consulting with steam tables yields the steam pressure of 10 kPa, which is still considerably lower than the original atmospheric pressure removed of 100 kPa (Anonymous, 1982). When dough expands initially, the findings suggest that it is due to the induced pressure differential releasing the expansion potential of the dough itself as well as the expansion and drive outward of entrapped gas already found in the dough due to incorporation during mixing. The introduction of microwave energy will enable the water-to-steam transition. Furthermore, the function of microwaves is to 70 supply the necessary energy to warm the dough to the point where its rheological properties offer a softer, more extensible rather than brittle texture. This trend increases steadily until temperatures reach 70-80°C, manifesting itself in increased gluten elasticity, both for gliadin and glutenin in a similar fashion (Kokini et al., 1994). Beyond this temperature range, tensile strength increases as elongation potential decreases (Cuq et al., 2000), therefore stopping further expansion. At the time where these maximum temperatures are reached within this drying system, the dough is sufficiently dry and would not allow further expansion. The continuing air evacuation allows further expansion, though the pressure differential from within the dough ball toward the outside is still the same but the dough would be softer. The texture of HG dough allows this expansion due to sufficient resistance whereas LG does not contain the necessary gluten network to supply the visco-elasticity required. This was already suspected as during dough ball forming the paste-like properties of LG dough were noted. Though generally it is recognized that softer doughs trap more gases, it is possible that L G dough in these experiments did not incorporate as many gas bubbles or did so but could not retain them as well as HG doughs possibly due to higher permeability. The shrinkage observed in the drying process can be attributed to the release of entrapped gases as well as vapour when the dough is permeable enough to release coalesced bubbles. Puffing was expected to be greater for increasing salt contents due to the higher E " values, as determined experimentally based on the assumption that it would increase the heating rate and steam generation. Since £" values decrease with rising temperature (Ndife et al., 1998), the heating process was expected to slow and with it the temperature 71 development. However, as it appears, the pressure differential affected the dough expansion much more than any other effect. Therefore, the larger volume increase was induced under the influence of initial gas evacuation. Yet not contradictory to the original thought is the fact that higher-salt-doughs heat up more rapidly as the loss factor almost doubles when comparing 0.0% to 1.5% salt formulations. However neither the final temperature nor the final moisture content after drying differed between the different dough formulations. In addition, the overall drying rate was the same for all. How the drying rate developed stays unanswered since the initial, more detailed drying rate was only determined for 0.0% salt dough. It is possible that moisture removal in the beginning is more rapid for higher salt-doughs due to the higher loss factor. Then it would level off as the moisture content decreases, yielding the same overall drying rate as for the other formulations. The temperature would not continue to rise since less moisture would be available for microwave induced heating. However, there is no evidence for cooling capacity either. As there is less water available, less energy is necessary for latent heat to enable liquid to vapour conversion. As with the observed temperature rise in the final moments of drying for all drying runs, the higher salt samples would be expected to attain a higher temperature at the end of drying. Yet this was not measured as final temperatures for all formulations were in the same range. However, it is to note that the method to determine temperatures of dough balls was flawed as the measurements were not taken in the vacuum chamber instantly at the sampling time. The drying process needed to be interrupted and gases reintroduced in order to open the chamber for sampling. Thus valuable time was lost to accurately determine the dough ball temperature. 72 The intent of employing image analysis was to quantify porosity of VM-dried product by looking at cross sectional images of cut dried dough balls. The results from the image analysis were promising but also showed limitations. As expected, the number of pores decreased as their size increased with rising salt concentrations. The image analysis results consistently overestimated the actually measured total cross-sectional area of the cut balls but the trend was seen as the area decreased with rising salt concentrations. This matches the results of the actual volume development. The custom written program was set up to recognize the cross sectional areas for open pores using the contrast in shading compared to the surrounding dough, also refered to as membranes. Due to this simplification, the drawbacks were that pore depth could not be used to assess actual pore size with regard to its volume. Dough membranes were not uniform shape, shading and smoothness throughout. If the image analysed showed gaps in pixels, the programs would have not necessarily recognized the complete circumference of the pore and therefore discarded this area as a non-pore. The way images were taken, differential shading was avoided by supplying lighting from four corners of the setup. However, this induced an over-illumination of the dough ball edges and thereby eliminating the edge as an area for analysis. When large air bubbles were trapped in the dough, either through direct entrapment or coalescence compared to the general occlusion of gases during mixing, the average pore cross sectional area would be off. As there would be one or several larger cavities, the average area would be based on their cross section and would therefore increase the 73 average pore size even when all other pores are very small in comparison. Thus this process would give misleading results. Likewise, the occurrence of long and shallow pores cut either on the long section or the narrowest section would distort the average pore opening and distribution as well. However if recognition is good, a quick analysis and comparison between samples would be a great asset to evaluation of porosity. Since the size of the actual image is not relevant, just its clarity and focus of the picture, any size specimen could be analyzed. In all, the image analysis program showed promise but did not give solid results in this study. If the quality of the images and the recognition of pores could be improved, the threshold adjustment for determination of pore versus non-pore areas might be facilitated. The potential is there but thus far, the program does not allow batch mode processing. Since it requires interactive steps, the analysis is very labour intensive and time consuming. But once improvements are implemented, image analysis could facilitate porosity evaluation for manufacture of foams or quality checks for dried or puffed goods. 2.5 Summary and Conclusions Puffing under vacuum in microwave drying was achieved and the mechanisms were revealed to a certain extent. As expected, volume changes in V M drying were due to low pressure as gases were removed from the vacuum chamber. This occurred to a much larger degree than anticipated as became obvious through video documentation. Initial expansion appeared possible due to entrapped gases as well as decompression of the dough. In addition, the rheology of wheat dough was an important factor in determining 74 how much expansion due to this pressure differential could occur. This was made obvious by the contrasting properties of high and low quality flour HG and LG. Likely, strain hardening facilitated HG dough expansion and holding of the newly found volume to yield large puffed balls with even pore distribution, both at high and medium power. This model system suggests possibilities for adjusting rheological properties of product to be dried to where it can be puffed. There clearly is a threshold below which Young's modulus or fracture strength are insufficient to withhold expansion without subsequent collapse. The addition of salt did not enhance the puffing, rather it proved counterproductive due to the interaction between flour and salt. More specifically, salt and gluten reduced gluten elasticity and therefore prevented the expansion instead of the quickly rising dough ball temperature facilitating volume increase. However, image analysis may provide the opportunity to analyze the porosity in terms of pore size and distribution as a response to microwave drying under vacuum with much less effort and increased precision than current manual methods. 75 CHAPTER III Physical modeling of vacuum microwave drying of dough balls using the Finite Element Method 3.1. Introduction Numerical modeling has become of interest, to understand mechanisms behind observed phenomena but more importantly, as a predictive tool. If a model fits well, computation may replace experimental assessment of processes. However, when models describe specific parameters only, their versatility is limited. Numerical modelling has been applied to food processing in general, and more specifically, dehydration (air, vacuum, microwave and freeze-drying), dough elasticity as well as baking performance and puffing. Both, Finite Difference and Finite Element (FE) Method have been used as numerical modeling tools in food processing by various researchers. Here, the FE Method has a clear advantage in its flexibility towards modeling structures of complicated geometry by dividing them into subdomains or small elements. This allows the specification of properties or conditions for each individual element instead of creating complicated equations which must be valid over the entire model. Furthermore, FE modeling allows the combined modeling of multi-physics problems such as structural deformation, heat flow, fluid flow and electro-magnetic problems. Here, FE modeling represents a readily available and established tool to find a solution even for complicated inputs such as 76 nonlinear material properties, state changes of materials, or transient conditions throughout the analysis. FE modeling of fluid flow was employed by Connelly and Kokini (2004) and Feigl and colleagues (2003). Martins and Silva (2004a, b) modeled the quality loss upon thawing of green beans. Models of air drying have focused on solving equations for mass and heat transfer problems. Wu and Irudayarj (1996) analyzed the heat, mass and pressure transfer in air-drying of a starch based food systems using a two-dimensional FE model, while Yang and colleagues (2001) simulated non-isotropic shrinkage deformation of potato spheres. Heat generation due to microwave power absorption has been simulated using FE by Ratanadecho and colleagues (2001), Oliveira and Franca (2000) and Lin and colleagues (1995). A review of the advantages and disadvantages of various modeling approaches to particulate drying is given by Kemp and Oakley (2002) while Puri and Anantheswaran (1993) published a comprehensive review of the usage of FE modeling of food processing. Microwave heating has been modeled using FE analysis for heating per se or heat and mass transfer in food stuff. Heat generation due to microwave power absorption has been simulated using two-dimensional FE analysis to compute the non-uniform temperature distribution throughout the heating process (Oliveira & Franca, 2000). Further, the effects of sample size and shape, salt concentration, and power operation mode were investigated in this study. A two-dimensional FE analysis of starch based food systems investigated the relation between heat, mass and pressure transfer for heating in general (Wu & 77 Irudayaraj, 1996). It was shown that the inter-dependence of the different models has a great influence on the results. For this reason coupled models that combine the dependence of one factor on another, i.e., temperature versus pressure and heating versus rheological properties, achieve much better solutions than uncoupled ones that look at one factor individually. A 3D FE model to predict temperature and moisture distribution in a cylindrically and slab shaped potato specimen was developed by Zhou and colleagues (1995) for microwave heating. The model included heat and mass transfer effects and results of temperature and absolute moisture were within 15% and 2.5% of experimental values, respectively. Pandit and Prasad (2003) computed the transient temperature profile of potato during microwave heating. The puffing mechanism was numerically investigated by Yamsaengsung and Moreira (2002). They developed a two-dimensional FE model simulating the frying and cooling of tortilla chips. Input parameters included temperature as well as water and oil saturation. Resulting changes in structure, i.e., shrinkage or expansion, compared well to experimental data. Al l FE models were based on time-integration from initial condition and incorporated non-linear material properties. Thus development of accurate material models are a key component to a successful modeling approach. Modeling of vacuum microwave drying considering heat and mass transfer while analysing the resulting structural change due to puffing has been thus far neglected. While many studies have focused on temperature distribution and moisture content during the drying process, only few studies have combined the thermal analysis of the 78 drying process with that of structural deformation. Perre and May (2001) modeled the shrinkage of potato due to water removal during drying. Thermal stresses in corn kernels during drying were analyzed by Jia and colleagues (2000). However, no modeling of the puffing process during V M D has been published to date. The intention of this study was to describe the mechanisms involved in the physical dough expansion in vacuum microwave drying, including the progressing change in temperature, water loss and volume change. The parameters used in building the model were partly derived from experimental findings of microwave drying under vacuum of wheat dough balls as well as literature values of general commodities like water, flour, pressure differentials and steam table findings where no data was retrievable from the current experiments. Where assumptions were made, these are stated as well. 3.2 Model description The FE model was developed to provide a better understanding of the thermodynamical and structural behaviour of dough balls when microwave-assisted drying under vacuum is used. The FE Method is a numerical tool which divides the complex domain into substructures (elements) of simple shape for which the respective physical equations can be solved. The FE model of the dough balls is shown in Figure 3.1. A detailed description of the model is provided by Ressing and colleagues (2005), which can also be viewed in theAppendix. Dough balls were modeled as a 2-D circular geometry consisting of approximately 300 quadrilateral elements. Dimensions were 20 mm for the diameter and 79 10 mm for the thickness. For simplicity only one quarter of the dough ball was actually modeled. The dough ball model was subjected to two type of loads: first, application of vacuum , i.e. the reduction of the external pressure in the chamber, resulting in an internal over pressure of the trapped air inside the dough, and second, microwave heating as heat generation within each element due to the applied microwave electric field. Here, a penetration depth for the microwave electric field of dp = 20 mm was considered (Buffer, 1993). Figure 3.1: FE-model of a dough ball. Yellow elements symbolize dry dough, blue elements symbolize wet dough. Wet dough elements are distributed randomly. For symmetry only one quarter is modeled. Material properties of dough ball elements were either those of dry or wet dough. Thermodynamically dry dough elements represent the flour mass of the system while wet dough elements represent the water mass. The ratio of dry to wet dough elements was set 80 to the water mass content of 42% of the dough mix throughout the experiment. Using the density of water and wheat flour this yields a water content of 35.6% in terms of volume. Therefore, 35.6% of all elements were randomly selected to be wet dough elements with the remainder being dry dough elements. Material properties of wet and dry dough elements are listed in Table 3.1. Differences exist primarily in their thermal behaviour. Due to their much higher loss factor, wet dough elements will generate most of the heat supplied by the microwave field. Wet dough elements were randomly distributed because it allowed the occurrence of local hot spots for water heating in the microwave field. This was done in order to simulate the effect of uneven heating due to unevenly distributed water in the dough balls. If no difference in thermal properties between wet and dry dough elements were to be made or if wet dough elements were simply distributed evenly, heat generation due to microwave application would be completely uniform. And the influence of local hot spots would not be observed. Additionally, e" values for wet dough were dependent on salt concentration of the dough mix. The structural behaviour of wet and dry dough elements were deduced from measured stress-strain curves for various salt concentrations. Here, the nonlinear portion of the measured stress-strain curve for low strains was also considered. Since stress-strain curves are a global function of the dough mix, no distinction can be made between Young's modulus of wet dough and dry wheat flour. Therefore, the structural properties of wet and dry 81 Table 3.1 Material properties of the FE model. Properties Dry Dough Wet Dough Thermal Heat conduction A [W/(m K)] 0.33 a 0.55 a Thermal Specific heat capacity c [J/(kg K)] 908 a 4200 a Thermal Dielectric loss factor e" 0.1 8.1 - 13.4b Thermal Drying rate r [%/s] 0 0.175 Thermal Latent heat h [MJ/kg] 0 a 2.26 a Structural Density p [kg/m3] 740 1000 Structural Young's modulus E [kPa] Salt concentration (%): 0.0 117.8 Structural 0.1 121.8 Structural 0.2 125.7 Structural 0.3 129.6 Structural 0.4 133.5 Structural 0.5 137.5 Structural 0.6 141.4 Structural • 0.7 145.3 Structural 0.8 149.2 Structural 0.9 153.2 Structural 1.0 157.1 Structural 1.1 161.0 Structural 1.2 165.0 Structural 1.3 168.9 Structural 1.4 172.8 Structural 1.5 176.7 Structural Poisson's ratio \x 0.48 "Anonymous, 1982 b see Table 2.3 for details. 82 dough elements were set to be identical, with the exception of the possibility of internal pressure buildup due to steam generation in wet dough elements. A schematic of the modeling procedure is shown in Figure 3.2. The initialization step creates the element geometry and assigns material properties to all elements. The first load step applies the internal over pressure of trapped air due to vacuum to an estimated air content volume of 10 % (Bloksma & Bushuk, 1988; Bloksma, 1990). The structural solution, i.e., deformation and resulting increase in volume, due to the applied vacuum is computed. Step two applies the microwave field for the specified time step size and computes the heat generation in each element with respect to its specified loss factor. Temperature distribution is then computed based on specified thermal properties of heat conduction and heat capacity. The average temperature within each element is evaluated and yields the respective vapour pressure for a wet dough element. At the same time, the current element volume is computed and yields the ratio of water liquid to vapour. The required latent heat is removed from the system- Water vapour is removed according to the specified drying rate, considering the loss of mass and energy. Step three re-assigns the structural material properties for each element based on the current thermal and structural solution, i.e., the temperature and strain distributions within the FE model. For wet dough elements the computed internal pressure is applied as internal forces on the nodes of the element, simulating the vapour pressure of the 83 CN CL Read Parameters and FE mesh, time = 0 I Randomly select wet dough elements Assign material properties , ST | Apply vacuum | | Compute deformation | I t ime^ao s I I Assign thermal material properties Solve FE equations for temperature distribution I Compute liquid-ratio in wet dough elements and remove latent heat for each element I Remove water vapour according to drying rate and remove latent heat for each element I Compute heat generation to MW Set time = time + time step size Compute vapour pressure according to temperature distribution for each element and apply vapour pressure as internal forces on element nodes C5 Q . OJ -4—» V) I Assign structural material properties according to current value of temperature and strain I Solve FE equations for structural solution, i.e. deformation, stresses, strains and volume increase I | Check if time > time max Yes^ I Stop~| No Figure 3.2 Schematic view of the FE-modelling process 84 steam. The new deformation of the dough ball is solved for and the resulting stresses and strains for each element are computed. After this, steps two and three are repeated until the maximum simulation time is reached. Modeling was only done on high gluten flour dough with the range of 0.0 to 1.5% salt. The power level applied was only high power. Thus comparisons can only be made carefully toward the composition and power response with respect to a threshold of puffing. The dough was assumed to be properly mixed, existing only of water, flour and salt and 10% air. Modeling was scheduled for the initial 200 s only, including the 20 s duration of initial gas evacuation to reach the desired reduced pressure. This was based on the observation that expansion during the drying process was terminated before or at 3 min. Though all calculations were done taking all parameters into account the images only display one factor at once. This was done for clarity and better overview because superimposition of expansion and temperature change or stresses would be difficult to discern for either parameter. 3.3 Results and discussion Figure 3.3 depicts a simple reflection of the dough ball deformation, showing the initial and final volume superimposed. Figures 3.4 and 3.5 show the deformation and temperature and stress distribution of a dough ball with 0.0% salt concentration for times t = 1, 10, 90 and 180 s of microwave exposure, respectively. In the early stages, the uneven heating due to the non-uniform water distribution can be observed. As the heat penetrates through the dough ball at later points in time, the temperature distribution is 85 primarily dependent on the penetration depth and the resulting lower heat generation in the centre of the dough ball. The resulting stress distribution shows the increased tensile stresses from higher vapour pressure in regions of higher temperature. Increases in temperature as a function of time for various salt concentrations are shown in Figures 3.6. For 0.0 % salt concentration a temperature of 55°C is reached. This compares reasonably to experimentally observed values of 45°C. For greater salt concentrations the temperature rise is unreasonably high and greatly exceeds experimentally observed values. This is due to the greater dielectric loss factor as input parameter to the FE model. 4 Figure 3.3: Dough ball deformation under vacuum pressure and 180 s of microwave application. The blue grid shows the original shape, the red grid depicts the expanded dough ball in its final form. 86 (a) time = 1 s (b) time = 10 s 20.5 20.6 20.7 20.B 20.9 21 21.1 21.2 temperature (°C) 22 22.5 23 23.5 24 24.5 25 temperature (°C) (c) time = 90 s (d) time = 180 s 38 40 42 temperature (°C) 52 53 54 temperature (°C) Figure 3.4 Temperature distribution after application of high power microwave field for 1 s (a), 10 s (b), 90 s (c), and 180 s (d). Initially hot spots due to non-uniform wet dough element distribution can be recognized. At later times the temperature distribution is dominated by the penetration depth. 87 (a) time = 1 s (b)time= 10 s 10 12 14 16 18 20 22 tensile stress (kPa) 10 12 14 IE 18 20 22 tensile stress (kPa) (c) time = 90 s (d) time = 180 s 10 12 14 16 tensile stress (kPa) 10 12 14 16 tensile stress (kPa) Figure 3.5 Stress distribution after application of high power microwave field for 1 s (a), 10 s (b), 90 s (c), and 180 s (d). As the temperature rises, steam pockets create an internal pressure stretching the surrounding material. 88 0 20 40 60 80 100 120 140 160 1 8 0 2 0 0 Time (s) Figure 3.6 Average temperature increase for various salt concentrations. Symbols are markers to distinguish the curves, not individual data points. Higher salt concentrations show unrealistically high increase due to their greater dielectric loss factor. However, there is no unambiguous explanation for this loss of energy. It is possible that, while the overall drying rate for higher salt concentrations is the same as for no salt, the drying rate is initially greater, thus more water is removed in the initial stages of the drying process, causing less heat generation at later points in time. It can also not be excluded that the microwave efficiency, the rate of absorbed energy by the dough compared to the absorbed energy by the standard 2 L of water, is significantly lower than the assumed 100%, in particular as more water is removed. In addition, no data is available for residence chamber vacuum microwave dryers. The literature values for parameters used in this model were taken from research conducted in wave guide dryers 89 (e.g., Ratanadecho et a., 2001) that expose the product only once to microwaves. The efficiency of microwave energy is increased in residence chamber dryers since the microwaves can be 'recycled' due to bouncing of the wave of the chamber walls The likelihood that the assumption is true 100% of the microwave energy were used in this experimental setup would increase compared to a wave guide dryer whose energy usage is significantly lower due to the, transient exposure. Buffler (1993) estimates an energy efficiency of 44% for microwave ovens. When observing the increase in volume (Figure 3.7) for various salt concentrations, it can be seen that the increase in volume due to vacuum application of 60 % for 0.0 % salt concentration makes up the major portion of the overall increase in volume and also reflects the manner in which salt concentration affected the experimentally measured final volume as the highest volume was obtained for the lowest salt concentration. The increase due to steam generation, i.e., internally applied vapour pressure, contributes only an additional increase of 15% of the total volume increase. For higher salt concentrations the calculated volume increase initially is lower than that of lower salt concentrations. This is due to the greater Young's modulus which was observed in dough formulations containing higher levels of salt. However, due to the increased heat generation caused by the significantly higher dielectric loss factor for higher salt concentrations the temperature increases over time to significantly higher levels compared to low salt concentrations. After 40-45 s microwave exposure, the FE results yield a reversal of this trend where high salt formulations dominate the additional volume increase. This 90 unreasonably great increase in volume is a direct result of the high temperature increase for high salt concentrations discussed above and was not observed in experiments. Time (s) Figure 3.7: Volume increase for various salt concentrations. Symbols are markers to distinguish the curves, not individual data points. Initial increase for high salt concentrations are smaller than for low salt concentrations, because of their higher E values. Later high salt concentrations increase dramatically because of their unrealistic high temperatures. When ignoring the increased heat generation due to higher salt concentrations, it becomes clear that at similar temperature increases dough formulations with lower salt concentrations exhibit a greater increase in volume due to their lower stiffness (Figure 3.8). This result matches the results from the experiments qualitatively. The overall increase in volume of 60 % for high salt concentrations and 88 % for 0.0 % salt match the experimental values reasonably well. From the video recordings for 0.0 % salt, 91 it can be observed that the volume increase between 90 and 120 s of microwave exposure of 240 % is much greater than the volume increase after 180 s. However, this behaviour was not observed for greater salt concentrations. The model does not accommodate for this expansion and subsequent shrinkage. 20 40 60 80 100 120 140 160 180 200 Time(s). Figure 3.8: Volume increase for various salt concentrations under the assumption of identical heat generation. Symbols are markers to distinguish the curves, not individual data points. Higher salt concentrations increase less, because of their higher E values. However, volume increases due to vapour pressure are less than 5% of calculated total volume increase. Results of volume increase due to a temperature dependent dough stiffness and a removal of air are shown in Figure 3.9. The reduction of dough stiffness with an increase in temperature has been noted in the literature by Rasper and Danihelkova (1986) and Fan et 92 al. (1999). Here, the dough stiffness was reduced by 2 %/°C temperature increase in each element yielding a 50 % reduction in Young's modulus for an average dough temperature of 45°C. At the same time, the loss of air, and thus the loss of internal pressure, was assumed to be 10 % of the original trapped air mass over the 180 s of microwave exposure. As observed in the video recordings, the increase in volume rises more rapidly initially, due to the reduced dough stiffness corresponding to the rise in temperature. However, as more and more air mass is lost, the internal pressure decreases and the volume increase is reduced towards the end of the drying time. 140 k 0 20 40 60 80 1 0 0 1 2 0 140 160 180 200 Time (s) Figure 3.9 Volume increase for 0.0% salt concentration including the reduction in stiffness by 2 %/°C and a loss of air of 10 % over the 180 s of microwave exposure. 93 3.4 Summary In summary it can be concluded that in principle the FE model does show a volume increase of the dough balls under vacuum assisted microwave drying based on the specified physical principles. Quantitatively the results only show a reasonable match for dough balls with low salt concentrations. For higher salt concentrations the thermal analysis results yield an unrealistic temperature profile. The loss of heat from the system that is apparent from the experimental values of temperature increase for higher salt concentrations cannot be explained unambigously. Furthermore, the FE model suggests that the increase in steam pressure due to the rise in temperature is not the only driving force behind the increase in volume. Additionally, the rise in pressure of the internally trapped air as well as the reduced dough stiffness due to the rise in temperature must be considered as well. 94 CHAPTER IV Microwave drying under vacuum using potato starch gels as a model system 4.1 Introduction As the wheat flour dough model exhibited a dominating salt effect on gluten elasticity, counterproductive to the expected increase in dielectric response in microwave drying under vacuum, gelatinized potato starch was chosen as a simpler model system. Not only is potato starch different from wheat flour since it lacks protein, but its starch type and composition also differs. Potato starch shows a hexagonal B-crystalline structure, dominated by amylopectin of longer chain length, compared to the A-crystallinity typical of cereals such wheat starch (Waigh et al., 2000; Tester et al., 2004). By choosing gelatinized starch, the fragility and pronounced temperature dependence of dough was circumvented. Potato starch and its gelling properties, including rheology of the raw and gelatinized starch and its interaction with water and solutes such as salt, have been studied extensively (Donald, 2004; Tester et al., 2004). Compared to starch from other sources, potato starch has unique properties such as high consistency in gelatinization and film forming characteristics, high binding potential and low gelatinization temperature, particularly well suited for pasting and gelling into a highly viscous entity (Bergthaller, 2004). In addition, native potato starch compared to for example cereal starches contains few impurities such as lipids, protein and salts (Conde-Petit & Escher, 1994; Bergthaller, 2004; Tester et al., 2004). The lack of lipid particles prevents certain retrogradation phenomena (Howling, 1980; Bergthaller, 2004; Tester et al., 2004). 95 4.1.1 Gelatinization of starch The mechanism of gelatinization is not entirely understood but good models have been proposed (Tester et al., 2004). The irreversible process of gelatinization, induced by heating, causes swelling of the starch granules to where the mostly crystalline structure of amylopectin is converted into a wholly amorphous structure (Parker & Ring, 2001; Karim et al., 2000). The amylose portion of starch solubilizes while the higher molecular amylopectin does not (Parker & Ring, 2001). Amylose forms a perpetual matrix in which the swollen, tightly packed and amylopectin-rich starch granules are dispersed (Langton & Hermansson, 1989; Svegmark & Hermansson, 1991). The gelation process is aided by complex formation between amylose molecules (Conde-Petit & Escher, 1994). Granule swelling behaviour is dependent on the close packing concentration, the swelling ratio and the amylopectin packing fraction (Evans & Lips, 1991; Donald, 2004). In contrast, during retrogradation, crystallites form, initially from amylose, long-term from amylopectin, and cause increased rigidity and eventually syneresis (Conde-Petit & Escher, 1994; Karim et al., 2000). 4.1.2 Rheology of gelatinized starch Starch suspensions and gels are considered viscoelastic systems (Ellis et al., 1989). Large and small strain tests have been employed to study the rheology of gels. Large strain tests were used to simulate sensory experience in the form of puncture, uniaxial compression, torsion and folding tests (Tabilo-Munizaga & Barbosa-Canovas, 2005). In addition, retrogradation effects were studied using large deformation tests for prepared gelatinized starch (Jankowski & Rha, 1986; Keetels et al., 1996a-c). Small strain tests 96 were employed to simulate processing, i.e., oscillatory and stress relaxation test, yield stress determination, and rheological characterization of time dependent fluids (Tabilo-Munizaga & Barbosa-Canovas, 2005). Compression tests were successfully employed to study rheology of various starch types (Inaba et al., 1994; Kortstee et al., 1998; Choi & Kerr, 2003). More specifically, simple starch gels as well as baked bread samples were exposed to uniaxial extension test to study retrogradation and sensory attributes (Jankowski & Rha, 1986; Keetels et al., 1996a-c; Seow & Teo, 1996). Instron compression curves yielded matching results with pulsed NMR measurements regarding increased firmness for retrograding corn starch gels, wheat bread and rice cupcakes. These tests reflect the molecular change as they relay their effect to the macroscopic level (Seow & Teo, 1996). By testing compression work, force, resilience and compressibility, differences in starch origin were discovered based on their rheological properties and gelling temperatures (Inaba et al., 1994). Young's modulus was determined for 40% corn starch gels, wheat bread or rice cupcakes using Hencky strain calculations from data generated by a Zwick.Universal Testing Machine (Conde-Petit & Escher, 1994; Keetels et al., 1996a-d). Apparent Young's modulus was determined in compression experiments with Instron for, among other items, corn starch gels and bread using the slopes of force deformation curves for 5-10% compression results (Conde-Petit & Escher, 1994; Bagley, 1987; Keetels et al., 1996d) and cooked wheat kernels (Jankowski & Rha, 1986). 97 To simulate sensory response, texture profile analysis (TPA) has been employed as well (Mandala et al., 2002; Choi & Kerr, 2003). These experiment more closely resemble eating responses since they create larger deformation stresses. However the stresses were low enough to ensure that immediate destruction of the samples was avoided (Pons & Fiszman, 1996; Mandala et al., 2002). Dynamic and oscillatory rheological testing has focused on studies for starch pastes and suspensions (Muhrback & Eliasson, 1987; Evans & Lips, 1991; Aguilera & Rojas, 1998; Svensson et al., 1998; Chen & Ramaswamy, 1999; Chang et al., 2004). 4.1.3 Other starch gel characteristics Gelatinized potato starch, either in suspension at low concentrations or in self-contained gels up to medium concentrations, is known to be transparent (Roberts & Cameron, 2002; Bergthaller, 2004), making possible air inclusions visible. It is not susceptible to short-term retrogradation due to the higher content of branched amylopectin that packs less tightly and lower amylose content, which in general are known to contribute to retrogradation/syneresis by leaching (Visser et al., 1997). Thus potato starch gels were ideal to handle in a larger scale pilot series of experiment such as this one. When gelatinization temperatures are reached, viscosity rapidly rises (Bergthaller, 2004). Potato starch's swelling capacity surpasses most starches (Parker & Ring, 2001). To ensure complete gelation, a distinct time at which gelatinization is completed facilitates filling and packaging of experimental samples. In addition, the rheological properties for this type of starch in native form have shown clear linearity in its concentration response (Evans & Lips, 1991; Roberts & Cameron, 2002), making potato starch suitable to 98 observe rheological responses and possibly creating a standard response curve regarding the rheology and drying and puffing behaviour as well. Inaba and colleagues (1994) noted an increase in gel strength when allowing gelation of starch to take place at 4°C over 24 hours. Potato starch stands out in its properties as it has been reported to exhibit a higher phase angle and elasticity even at higher concentrations such as 20 and 25% starch (Evans & Lips, 1991; Inaba et al., 1994). The rupture force, compression work and resilience was highest compared to wheat, corn and tapioca starch at high concentrations. Compressibility was recorded as lower than tapioca but still higher than wheat starch gels (Inaba et al., 1994). Interactions with other compounds are known for potato starch. Apart from pH sensitivity, salts notably influence gelation of the starch. Cations enhance rapid viscosity increase by binding water and withholding it from rehydration of the starch molecule. This has been attributed to the phosphate monoesters typical for potato that leave the starch molecules negatively charged causing the strong response to salts such as NaCl (Howling, 1980; Muhrbeck & Eliasson, 1987; Chiotelli et al., 2002). This polyelectrolyte character influences its high swelling properties in that it binds water to the charged starch granules (Parker & Ring, 2001). If the ionic strength in processing is increased, the swelling capacity of potato starch is decreased (Parker & Ring, 2001). In contrast, calcium and potassium chloride have demonstrated the opposite effect. In general, the Hofmeister series reflects the influence of salts on starch (Howling, 1980). 99 4.1.4 Hypotheses There are three hypotheses in this study: 1. Rheology of the gelatinized starch formulations such as its elasticity has upper and lower thresholds between which microwave/vacuum exposure can result in puffing. The critical levels allow expansion through vacuum and steam response while avoiding collapse before drying is completed. 2. Expansion is dependent on microwave power supplied. There is a threshold above which the puffing response occurs. 3. Salt addition to the starch gels increases the microwave response by raising the dielectric loss factor and therefore increasing heat generation, which in turn will support more steam development and more puffing. Testing of the hypotheses was done by a series of experiments with 30 to 50% starch concentrations with 0 to 2% added salt, drying trials at three power levels, starch gel rheology characterization, temperature development during drying, dielectric properties measurement, and videorecording of the drying process. Part one of the hypothesis was to be tested by comparing rheological measurements, namely determination of Young's modulus, to volume expansion results. Due to the range of formulations, a rheological threshold was expected to be detectable. The second part of the hypothesis was to be tested by drying at three different levels of microwave power, 350, 700 and 1300 W. The last part of the hypothesis was to be addressed by the addition of salt to the different starch concentrations. The dielectric loss factor was determined and results were compared to volume expansion. 100 4.2 Materials and Methods 4.2.1 Gel preparation Potato starch was supplied by K M C (Brande, Denmark). The 'Superior potato starch' (CN#1108.13.00) was exclusively used from one batch. Gels were prepared at 30, 35, 40, 45 and 50% starch concentration with 0, 1 and 2% added salt (w/w), by stirring the starch/water/salt mixture at 20°C into a 1 L-beaker before placing it on a magnetic stir/heating plate in batches of 500 mL. Then the slurry was heated under constant stirring at medium heat and medium high speed until thickened, and the vortex slowed. The semi-gelled mass was then filled into Betan-B2 artificial sausage casing, diameter = 20 mm, (Naturin Canada, Ville St. Laurent, Qc, Viscofan USA Inc. Montgomery, AL). The ends were tied and the strands steamed at 95°C for 20 min in a steam kettle to complete gelation. The ready gel strands were removed, photographed and placed in a cooler at 5°C overnight. Prior to drying, strands of gels were allowed to reach room temperature, the casing was removed, and gels were cut into even 20 mm pieces by a six-bladed aluminium gel cutter. 4.2.2 Microwave drying under vacuum The vacuum microwave dryer (Enwave Corp., Vancouver, BC) was warmed up by placing two litres of water in a three-litre glass beaker into the vacuum chamber at high power for 9 min and chamber pressure at 28 Torr. Starch gels were weighed into a polyethylene drum with a solid bottom and a netted lid to provide good visibility. Drying proceeded in batches of approximately 600 g at either low (350 W), LP, medium (700 W), MP, or high (1300 W), HP, power. Absolute chamber pressure was set at 101 28 Torr while the drum rotated at 4 rpm. A l l experiments were performed in triplicate. The details of the setup can be viewed in Figure 2.1. 4.2.3 Storage Dried starch gels, once cooled to room temperature, were sealed into Ziplock™ freezer bags (SC Johnson & Son Inc., Canada, Brantford, ON), and stored at room temperature. 4.2.4 Density Volume was determined by displacement measurement with loose flax seeds. A total of five starch gel cylinders were weighed and covered adequately by loose flax seeds in a graduated cylinder. Then the flax seeds' volume was determined. The difference between volume of the starch cylinders and flax seeds and flax seed volume alone yielded the total volume of the cylinders. Apparent density was determined as mass divided by volume of the gel cylinders, measured before and after drying. 4.2.5 Moisture Starch gels, dried starch gels and potato starch were weighed on aluminium weighing dishes and placed in a vacuum oven for 24 hours at 70°C and weighed after reaching room temperature in a desiccator (wf) (AACC, 1994). Moisture was expressed as percent moisture per dry mass: %Mdm ~ ' 100% (4.1) 102 4.2.6 Heating and drying rate Gels of 40% starch and 0% salt were prepared and placed in the vacuum microwave dryer for drying as previously described. Sampling intervals were chosen as follows: High power: 0, 1.5, 3, 4.5, 6, 11 and 13.5 min; Medium power: 0, 5, 7.5, 12, 19, 22.5, 29 and 30.5 min; Low power: 0, 10, 22, 35, 45, 48, 59 min. For each sampling, surface temperature was measured by infrared thermometer (Cole Parmer Instruments, model 39650-04, Chicago IL USA) and moisture was determined (see 4.2.5). A l l trials were done in triplicate with three repeats each. 4.2.7 Young's modulus Gel cylinders (diameter = 20 mm, thickness = 20 mm) were placed on the holder of the Texture Analyser TA.XT2i (Stable Micro Systems, Surrey, UK) where the plunger (diameter = 50 mm) was lowered at a speed of 0.1 mm s"1 5 mm into the gel. The resulting slope of the curve of this compression test was multiplied by the penetration depth, and divided by the surface area of the sample to yield the modulus of elasticity or Young's modulus E. From the measured force vs. displacement curves, strain 8 and stress a were determined by L 2L and (4.2) D-d 103 (4.3) Calculations were done following the same approach as for wheat dough, using the least-square approach (Holman, 2001). See Chapter 2.2.10 for details. For the harder 45 and 50% starch gel samples, core and peripheral samples were taken with a cork borer (diameter = 8 mm) and treated accordingly. For these samples, calculations for E were done in the same manner as for the other samples. Calculations followed previous research (Jankowski & Rha, 1986; Conde-Petit & Escher, 1994; Keetels et al., 1996d). 4.2.8 Dielectric properties For all samples, dielectric constant and loss factor, e' and e" respectively, were determined prior to drying for all formulations at 24±1°C, using an open-ended coaxial probe and a Hewlett Packard network analyser (Houston TX, USA, model HP 85070, software 85070 B Re B 0105). Calibration was performed by measuring surrounding air, a metal short supplied by the manufacturer and distilled water at 25°C. For 30-45% starch gels, the probe was placed onto the original gel samples (cylinder height = 20 mm, diameter = 20 mm) and kept in close contact during the time the measurement was taken (Figure 2.2). For 45 and 50% starch gels, the readings were taken from the cut surface for consistent readings. Measurements at 2450 MHz were taken in triplicate at 25°C and 8' and 8" recorded. This was a modification of a method by Colpitts, Pelletier & Cogswell (1992). To verify the appropriate sample thickness, starch gel cylinders were measured with and without aluminium foil placed under them. If the two measurements were equal, the thickness of the sample was deemed sufficient. 4.2.9 Video monitoring Drying was monitored by a video camera (generic brand, 525 TV line resolution) connected via an accelerator card (ATI Radeon 9200 vivo 128Mb DDR) to a PC. The video camera was inserted to the edge of the vacuum chamber supported by a metal pipe welded to the door (Figure 2.6). Videos were recorded with Windows Movie Maker in x.WMV file format. For HP and MP drying, the quality was chosen at the highest resolution, however, LP drying recordings were saved in medium quality to manage limited computer memory. Recordings were viewed and events recorded according to time of occurrence of start and end of bulging, start and end of rounding, puffing and cluster formation. Start of bulging was defined as the point in time when the gel cylinders exhibited changes on the surface, mostly an upward bending of the cut surfaces. End of bulging was the time when these changes no longer occurred. Start of rounding was defined as the point in time where the gel samples lost their cylindrical shapes and rounded into spheres. End of rounding was defined as the time at which no more change in the spherical shape could be detected. Puffing signified the occurrence of sudden expansions in no defined location on the samples. Cluster formation was the observation of several samples aggregating and permanently forming a cluster of more than two gel cylinders. 105 4.2.10 Air inclusion After steaming, starch gel strands were photographed when still warm and translucent and air inclusions were visible throughout. Images were categorized into low, medium and high levels of air inclusions. 4.2.11 Statistical analysis Differences between treatments and their interactions were statistically analysed by analysis of variance A N O V A using SAS software (Version 6.12 TS 040, SAS Institute Inc., Cary, N.C., USA). The General Linear Model G L M was used for analysis of variance to manage the partially unbalanced nature of the data sets. Duncan's multiple range test served as the tool to distinguish means. 4.3 Results 4.3.1 Microwave drying under vacuum Most starch formulations under low, medium and high power exposure exhibited volume increases. The exception were the 30% starch series with added salt where shrinkage was recorded for formulations at MP and LP and 35% starch, 1% salt for LP as well as minor shrinkage for 45% starch, 2% salt formulations at MP and LP (Figure 4.1). The higher starch formulations showed the most increase at all power levels with a few exceptions. The most consistent results were HP dried samples. The maximum volume increases were recorded at HP drying for 50% starch, 0% salt gels were recorded as 75%, for 45% gels approximately 60% and for 30% gels approximately 15%. For no-salt formulations 106 70 60 50 40 e 10 a | 20 o > 10 -10 -20 35,0 40,0 45,0 50,0 35 40,1 45,1 50,1 30 35,2 40,2 4§§2 50,2 HHP • L P Starch, salt (%) Figure 4.1 Volume changes of different starch, salt formulation of potato starch gels dried at high, medium or low power (HP=fading, MP=striated, LP=dotted). The sample formulations were expressed as 30,0, signifying 30% starch, 0.0% salt. at HP, this relationship response appears linear. Al l salt-added formulations exhibited lower volume increases compared to their equivalents without salt and were more inconsistent in the concentration to volume expansion response where a linear relationship cannot be confirmed. Change in volume was dependent on starch content, salt (P = 0.0001, respectively), their interaction (P = 0.021), power level, and starch/power level interaction (P = 0.0001, respectively), and the interaction of starch, power level and salt (P = 0.015, Table 4.1). Due to the irregularities in gelatinization of 107 50% starch gels, these results were excluded from further analysis (Table 4.2). Without these data, results showed a similar relationship between starch, salt and power, most of these parameters yielded highly significant influence on volume changes. Interaction between starch and salt as well as starch and power showed lower P values (0.0167 and 0.0018, respectively) while salt/power and starch, salt and power interaction were not significant. Replicates showed a slightly higher influence on the results than when 50% starch gels were included. When comparing means, the volume change averages of 30 and 35% starch gels were different from each other as well as all other formulations. Analysis of air inclusions did not reveal any relevance with regard to their extended influence on puffing. Al l trials and formulations exhibited a similar range of high, medium and low amounts of air inclusions. 4.3.2 Drying rate The moisture loss over time, determined for 40% starch without salt added, increased linearly for HP and was relatively constant for MP and LP drying after a slight delay in the beginning stages of the drying process (Figure 4.2). For HP, initial moisture removal was below three grams per minute for the first 90 s but increased in the next intervals up to 32.7 g/min at the 11 min-sampling. For MP, the rate of water removed per minute stayed relatively steady throughout. With the exception of the first five minutes of drying, the averages were 10.7±2.0 g/min. LP drying proceeded at a level rate as well. While in the first 10 min, an average of 2.2 g of moisture was removed, continued drying yielded 5.1±1.5 g/min. 108 Table 4.1 A N O V A results for volume change determined by G L M Sum of Mean Source DF Squares Square F Value Pr > F Model 48 59117.0664 1231.6055 10.49 0.0001 Error 81 9510.2291 117.4102 Corrected Total 129 68627.2955 R-Square C.V. Root MSE VOLCH Mean 0.861422 66.82369 10.8356 16.2152 Source DF Type I SS Mean Square F Value Pr > F STARCH 4 17978.6861 4494.6715 38.28 0.0001 SALT 2 2586.4142 1293.2071 11.01 0.0001 STARCH*SALT 8 2277.9986 284.7498 2.43 0.0211 POWER 2 24565.2971 12282.6485 104.61 0.0001 STARCH*POWER 8 4376.0082 547.0010 4.66 0.0001 SALT*P0WER 4 2110.8632 527.7158 4.49 0.0025 STARCH*SALT*P0WER 16 3984.7867 249.0492 2.12 0.0148 REP 4 1237.0123 309.2531 2.63 0.0400 Table 4.2 A N O V A results for volume change determined by G L M , excluding the gel concentration of 50% Sum of Mean Source DF Squares Square F Value Pr > F Model 37 36548.5636 1015.2379 10.66 0.0001 Error 65 6284.6403 95.2218 Corrected Total 102 42833.2039 R-Square C.V. Root MSE VOLCH Mean 0.853276 77.61322 9.75817 12.5728 Source DF Type I SS Mean Square F Value Pr > F STARCH 3 11895.7612 3965.2537 41.64 0.0001 SALT 2 2083.9967 1041.9983 10.94 0.0001 STARCH*SALT 6 1611.2484 268.5414 2.82 0.0167 POWER 2 15465.0420 7732.5210 81.21 0.0001 STARCH*POWER 6 2286.3302 381.0550 4.00 0.0018 SALT*POWER 4 839.2380 209.8095 2.20 0.0781 STARCH*SALT*POWER 12 1878.5874 156.5489 1.64 0.1008 REP 2 . 488.3597 488.3597 5.13 0.0268 109 35.00 30.00 y = 3.0634X - 0.3145 R2 = 0.983 • HP o MP A LP Linear (HP) 15 20 25 30 35 Drying time (min) 40 45 50 55 60 Figure 4.2 Moisture loss (g/min) versus drying times for 40% starch gels without added salt at high, medium and low power. HP (closed diamonds), MP (open circles), LP (closed triangles). 4.3.3 Young's modulus Young's modulus generally increased with increasing starch concentration when looking at each salt concentration individually. It proved to be highly influenced by the initial moisture, salt, as well as the interaction of both parameters (P = 0.0001, respectively). Salt contents of one and two percent yielded the same Young's modulus values yet different from 0% salt. Volume change was dependent on the E values of all starch gels when 50% starch formulations were excluded (P = 0.0013). There was a general trend (Figure 4.3) that suggests that with increasing Young's modulus, the potential for more pronounced volume expansion is realized for high power drying. This relationship appeared to be linear in nature. For low power, the volume increases plateau for 40 and 110 45% at 0 and 1% salt. At this power level, 45% starch and 2% salt, the volume increase is lower than the 35 and 40% formulations. Medium power does not exhibit a consistent trend. 70 GO 50 40 oj <?. 30 CO U | 20 ,>: 10 or -10 h -20 • HP 0 MP A LP 6 • A O o A O • o o • A o A • 6 o A 30,0 35.0 40;0 45,0 30.1 35,1 40,1 45,1 30.2 starch content (%). salt content (%) 35.2 40,2 45,2 Figure 4.3 Volume change of high, medium and low power drying for all starch/salt formulations compared to their respective E values (histogram). HP (open squares), MP (open circles), LP (open triangles). The sample formulations were expressed as 30,0, signifying 30% starch, 0.0% salt. 4.3.4 Heating rate As the 40% starch gel specimens were exposed to microwave power, temperatures reached 40-50°C very quickly (Figure 4.4). Even so, the temperature increases varied widely between repeats. However, after a quick rise in temperature in the first minute for high power, reaching 40°C, the temperature only rose just above 45°C, which appeared to 111 be the main drying temperature. Medium power samples reached 42°C within the first sampling interval and continued to rise in temperature until 50°C. Prior to the end of drying time, temperatures neared 55°C. For low power drying, the initial reading was 45.5°C and continued to increase steadily until the last sampling time of 45 min, close to the end of drying completion where the temperature surpassed 55°C. Low power drying, which required the longest drying time, also exhibited the highest temperatures within the parameters applied. The heating rate at different salt concentrations was not determined. 60.0 55.0 50.0 ~45.0 S 35.0 S • o 30.0 25.0 20.0 • HP OMP ALP 10 15 20 25 30 Drying time (min) 35 40 45 50 Figure 4.4 Heating rate of 40% starch gel without salt, dried at high, medium and low power (HP=closed diamonds, MP=open circles, LP=closed triangles). 112 4.3.5 Dielectric properties Though 8' values varied throughout the sample formulations, there appears to be a trend of decreasing 8' values with increasing solids content for 30 and 50% starch gels. In contrast, 8" increased for all formulations in similar fashion with rising salt content (Table 4.3). Thus, the increase in salt and decrease in moisture content influenced the change in 8" (P = 0.002). Table 4.3 Dielectric properties e' and e" with respect to the varying starch gel formulations. Values in the same column followed by a different letter are significantly different. Starch (%) NaCl (%) e' s" 30 0 56.80 ± 0.89 a 11.31 ± 0.29 c 30 1 54.62 ± 1.71 a 17.99 ±0.81 b 30 2 52.88 ±2.32 a 25.08 ± 1.89 a 35 0 50.88 ±0.81 a 12.17 ± 0.55 c 35 1 51.06±2.16a 17.84 ± 1.32 6 35 2 50.21 ± 1.47 a 23.48 ±0.59 a 40 0 49.53 ± 0.89 Z> 11.94 ± 0.32 c 40 1 51.74 ± 1.83 a 17.39 ±2.69 b 40 2 45.89 ±3.05 b 26.07 ± 1.09 a 45 0 ' 46.00 ± 4.32 b 12.19 ± 0.56 c 45 1 51.93 ± 1.38a 19.44 ±0.99 b 45 2 . 35.02 ± 3.99 c 16.11 ±2.55 b 50 0 50.22 ±3.18 a 11.90 ± 1.00 c 50 1 41.52 ± 2.45 c 17.89 ± 1.39 6 50 2 30.36 ± 4.48 d 15.83 ±2.95 b 113 4.3.6 Video monitoring All recorded events are listed in Table 4.4. There were some factors that significantly affected the start of bulging, namely salt, starch concentration and power levels individually (P = 0.0286; P = 0.0098; P = 0.0001, respectively) (Table 4.5a). End of bulging showed no significance for any parameters. Start of rounding only exhibited significant changes for power levels (P = 0.0001) while end of rounding showed significance for starch and power (P = 0.0242 and 0.0001, respectively) (Table 4.5b). Sudden expansions and cluster formation did not show significance with respect to starch/salt formulations or power levels (Table 4.5c). It should be noted that the extent of the volume changes was not measured in detail from the video images, particularly because at low power the viewing quality was too poor. The final volume changes and their respective response to formulations and power were explained earlier in section 4.3.1. The only significant and consistent recorded events were bulging and rounding of the gel samples. Therefore, they are highlighted in the following. Figure 4.5 depicts the time at which bulging ends for HP, MP and LP drying of various starch/salt formulations. Initial bulging, defined as the upward curving of the cut surfaces, was visible at 1.5-2 min for no salt formulations for HP, for MP at 4 min and for LP at 6 min. For one and two percent salt formulations, these times were reduced to 1.5 min and 1 min respectively for HP. For MP and LP, the latter effect was not noted. The difference between power levels becomes obvious when comparing means, with HP exhibiting the lower bulging start times, LP the highest. This appears to be related to salt concentrations as almost all 114 Table 4.4 Event times during drying for starch gels at low, medium and higher power at varying starch and salt concentrations. Times listed are averages of three runs, all indicated in minutes. Starch, salt Power Bulging Bulging Rounding Rounding Puffing content start end start end 30,0 High 1.3 1.7 1.8 2.8 5.5 35,0 High 1.2 1.8 1.6 2.7 1.8 40,0 High 1.5 n.d. 1.5 2.5 2.1 45,0 High 1.3 n.d. 1.5 2.8 1.9 50,0 High 1.3 1.8 n.d. 2.8 1.7 30,0 Medium 2.9 4.4 6.3 7.7 8.0 35,0 Medium 3.3 4.9 5.3 6.8 n.d. 40,0 Medium 2.7 3.7 4.4 5.8 n.d. 45,0 Medium 4.0 n.d. 4.9 6.2 6.5 50,0 Medium 3.4 3.6 2.1 4.9 n.d. 30,0 Low 6.7 n.d. 10.8 10.2 8.0 35,0 Low 5.4 6.3 7.7 n.d. 10.3 40,0 Low 3.9 3.5 5.8 6.7 n.d. 45,0 Low 5.8 n.d. 6.9 8.0 n.d. 50,0 Low 6.0 5.8 7.7 7.5 8.0 30,1 High 1.6 2.7 n.d. 6.0 2.8 35,1 High 1.3 1.5 n.d. n.d. 3.0 40,1 High 1.3 n.d. n.d. 2.9 2.0 45,1 High 1.25 n.d. 1.4 3.4 1.7 50,1 High 1.3 n.d. n.d. 2.6 1.6 30,1 Medium 3.7 4.5 5.5 6.0 10.0 35,1 Medium 3.4 n.d. 5.8 3.4 6.9 n.d. = not detected 115 Table 4.4 continued Starch, salt Power Bulging Bulging Rounding Rounding Puffing content start end start end 40,1 Medium 2.9 4.2 6.4 6.4 6.0 45,1 Medium 3.2 3.7 4.3 5.8 5.8 50,1 Medium 3.1 3.8 4.6 6.3 6.0 30,1 Low 6.3 7.7 6.5 9.6 n.d. 35,1 Low 5.3 7.0 n.d. 8.5 12.9 40,1 Low 6.4 n.d. 7.5 n.d. 9.5 45,1 Low 5.7 n.d. 8.0 n.d. 10.7 50,1 Low 6.5 n.d. 6.6 8.0 11.7 30,2 High 1.3 n.d. 1.7 2.3 n.d. 35,2 High 1.3 n.d. n.d. 2.4 2.1 40,2 High 1.5 n.d. n.d. 2.0 2.0 45,2 High 1.5 n.d. 1.8 2.0 2.2 50,2 High 1.3 1.7 1.8 2.0 2.5 30,2 Medium 4.0 6.5 5.8 5.8 n.d. 35,2 Medium 3.9 n.d. 6.0 5.9 7.7 40,2 Medium 3.4 n.d. 4.1 5.1 n.d. 45,2 Medium 4.3 n.d. 5.3 8.6 5.9 50,2 Medium 3.3 4.2 4.2 5.0 6.0 30,2 Low 5.8 7.0 8.0 13.0 n.d. 35,2 Low 8.3 10.0 7.5 9.3 11.3 40,2 Low 5.4 7.5 7.0 10.1 n.d. 45,2 Low 7.7 n.d. 9.8 n.d. n.d. 50,2 Low 5.6 n.d. 8.1 8.9 n.d. n.d. = not detected Table 4.5a A N O V A results determined by G L M for parameter bulging start Sum o f Mean Source DF Squares Square F V a l u e P r > F Model 83 413.726712 4 .984659 12 .25 0 .0001 E r r o r 25 10.173090 0 .406924 C o r r e c t e d T o t a l 108 423.899802 R-Square C . V . Root MSE BULG1 Mean 0 .976001 18.60023 0 .63791 3 .42956 Source DF Type I SS Mean Square F V a l u e P r > F SALT 2 3 .347679 1.673839 4 . 1 1 0 .0286 STARCH 4 6 .823495 1.705874 4 . 1 9 0 .0098 PWR 2 358.657274 179.328637 4 4 0 . 6 9 0 .0001 SALT*STARCH 8 7.153295 0 .894162 2 . 2 0 0 .0633 STARCH*PWR 8 3.886213 0 .485777 1.19 0 .3420 SALT*STARCH*PWR 16 15.714876 0 .785744 1.93 0 .0600 DAY 39 18.143879 0 .465228 1.14 0 .3676 Table 4.5b A N O V A results determined by G L M for parameter rounding start Sum o f Mean Source DF Squares Square F V a l u e P r > F Model 57 340.343084 5 .970931 3 .85 0 .0331 E r r o r 7 10.846718 1.549531 C o r r e c t e d T o t a l 64 351.189802 R-Square C . V . Root MSE R0UND1 Mean 0 .969114 22.71168 1.24480 5 .48089 Source DF Type I SS Mean Square F V a l u e P r > F SALT 2 4 .105863 2 .052932 1.32 0 .3251 STARCH 4 17.277069 4 .319267 2 . 7 9 0 .1122 PWR 2 237.024457 118.512228 76 .48 0 .0001 SALT*STARCH 8 11.869607 1.483701 0 . 9 6 0 .5295 STARCH*PWR 8 9 .547408 1.193426 0 . 7 7 0 .6414 SALT*STARCH*PWR 16 17.455688 1.454641 0 . 9 4 0 .5602 DAY 21 43.062992 2 .050619 1.32 0 .3706 Table 4.5c A N O V A results determined by G L M for parameter rounding end Sum o f Mean Source DF Squares Square F V a l u e P r > F Model 67 565.555824 8 .441132 7 . 3 4 0 .0089 E r r o r 6 6 .900662 1.150110 C o r r e c t e d T o t a l 73 572.456486 R-Square C . V . Root MSE R0UND2 Mean 0 .987946 19.70288 1.07243 5 .44302 Source DF Type I SS Mean Square F V a l u e P r > F SALT 2 3 .475328 1.737664 1.51 0 .2942 STARCH 4 29.056724 7 .264181 6 . 3 2 0 .0242 PWR 2 404.210174 202.105087 175 .73 0 .0001 SALT*STARCH 8 10.115112 1.264389 1.10 0 .4672 STARCH*PWR 8 9 .334472 1.166809 1.01 0 .5075 SALT*STARCH*PWR 16 37 .944133 2 .232008 1.94 0 .2113 DAY 26 71.419881 2 .746919 2 . 3 9 0 .1402 formulations show the same order, the lower salt concentration and the lower the power level, the higher the bulging time (Table 4.6). Rounding of the gels to spheres occurred after bulging was completed. It distinguished itself from bulging in that the cylinder walls curved outward to make the whole sphere evenly rounded. The time at which rounding ended was also comparable throughout the formulations and only differed between power levels, occurring later with decreasing power supplied (Figure 4.6). 118 Table 4.6 Selected results of DUNCAN multiple range test for bulging start for all starch gel formulations Alpha= 0 . 0 5 df= 25 MSE= 0 .406924 WARNING: C e l l s i z e s a r e not e q u a l . Harmonic Mean o f c e l l s i z e s = 35 .6422 Number o f Means 2 3 C r i t i c a l Range .3112 .3269 Means w i t h t h e same l e t t e r a r e not s i g n i f i c a n t l y d i f f e r e n t . Duncan Group ing Mean N PWR A 6 0137 30 1 B 3 4243 42 2 C 1 3403 37 3 119 Table 4.6 continued DUNCAN multiple range test for bulging start times for all starch gel formulations Level of Level of Level of BULG1-SALT STARCH PWR N Mean SD 0 30 1 3 6 66666667 1. 15470054 0 30 2 3 2 86100000 0. 47403903 0 30 3 3 1 27766667 0. 50229108 0 35 1 2 5 35000000 0 14142136 0 35 2 3 3 25333333 0 82802979 0 35 3 1 0 75000000 0 40 1 .1 3 93000000 0 40 2 3 2 72223333 0 54218351 0 40 3 2 1 50000000 0 00000000 0 45 1 3 5 41666667 0 62915287 0 45 2 3 4 02233333 0 59303991 0 45 3 3 1 27780000 0 12729741 0 50 1 2 5 97500000 2 08596500 0 50 2 3 3 37666667 0 42027769 0 50 3 3 1 32233333 0 33406037 1 30 1 2 6 31500000 0 26162951 1 30 2 3 3 72223333 1 18241827 1 30 3 3 1 61100000 0 25475849 1 35 1 2 5 33350000 0 94257334 1 35 2 3 3 41666667 1 01036297 1 35 3 2 1 24850000 0 11525841 1 40 1 2 6 41650000 0 11808683 1 40 2 3 2 88900000 0 34707780 1 40 3 2 1 34168500 0 10604480 1 45 1 2 6 00000000 0 47093312 1 45 2 2 3 20000000 0 42426407 1 45 3 3 1 25000000 0 25000000 1 50 1 1 6 50000000 1 50 2 3 3 08333333 0 14433757 1 50 3 3 1 30456667 0 42961921 2 30 1 2 5 75000000 1 06066017 2 30 2 2 4 04000000 1 00409163 2 30 3 3 1 33223333 0 33335561 2 35 1 2 8 25000000 1 06066017 2 35 2 3 3 91666667 0 80363756 2 35 3 2 1 33350000 0 47164022 2 40 1 3 5 36110000 1 21431241 2 40 2 2 3 41650000 0 11808683 2 40 3 2 1 54185000 0 17698883 2 45 1 1 7 66700000 2 45 2 3 4 32233333 0 38935374 2 45 3 2 1 50000000 0 23617366 2 50 1 2 5 60000000 0 84852814 2 50 2 3 3 25000000 0 00000000 2 50 3 3 1 26110000 0 16189388 120 Figure 4.5 End of bulging time for high, medium and low power drying for all starch/salt formulations (HP=fading, MP=striated, LP=dotted). The sample formulations were expressed as 30,0, signifying 30% starch, 0.0% salt. 121 12 10 H high power El medium power Fl O low power 30,0 35,0 40,0 45,0 50,0 30,1 35,1 40,1 45,1 50,1 30,2 35,2 40,2 45,2 50,2 Starch, salt (%) Figure 4.6 End of rounding time for high, medium and low power for all starch/salt formulations (HP=fading, MP=striated, LP=dotted). The sample formulations were expressed as 30,0, signifying 30% starch, 0.0% salt. 4.4 Discussion To explain the puffing mechanism of microwave drying under vacuum, gelatinized potato starch/salt preparations were investigated. Unlike the wheat dough experiments, here starch, water content and salt were varied as well as the power levels generated by the vacuum microwave dryer. Starch/salt formulations were gelatinized, cut into cylinders of uniform size and dried under similar conditions as wheat dough balls. Microwave power levels were chosen to be low, medium and high, all applied at 28 Torr absolute pressure. 122 Volume increases were observed for almost all preparations, particularly for HP and MP-dried samples. However, in certain instances, shrinkage was observed, particularly for the lowest chosen starch concentrations and at low power. At HP, a higher volume increase relative to the initial starch concentration was observed, resulting in higher expansion for higher starch content. With increasing salt levels, the recorded volume increases were lower. The combined effect of starch and salt was significant. The highest volume increase at above 80% was recorded for HP-dried 50% starch without added salt compared to the lowest for 20% for 30% starch, 2% salt gels dried at HP. At LP, volume change was recorded in the range of -10% for 30% starch to 5-20% for gel with the highest starch concentrations. However, results for MP and LP levels were statistically indistinguishable. For 40 and 50% starch gels, Duncan multiple range test revealed volume increases to be the same (data not shown). As reflected in large standard deviations for dielectric properties and visual inspections, 50% starch gels were unevenly and incompletely gelatinized. This particular range of concentration had been noted to cause gradients within the samples after gelatinization is completed (Lui et al., 1991), likely due to one granular portion taking away the opportunity for swelling of other granules (Parker & Ring, 2001), thus gelatinizing unevenly. Therefore, segments of these gels are likely more similar to a lower starch concentration than the intended 50% and the equivalent texture. Nevertheless, the average pre-drying moisture of these pieces was indeed 51.2±1.3%, 123 Table 4.7 DUNCAN multiple range test for volume change and the parameters starch and salt concentration with respect to their drying response. D u n c a n ' s M u l t i p l e Range T e s t f o r v a r i a b l e : VOLCH A lpha= 0 . 0 5 df= 66 MSE= 95 .22182 Harmonic Mean o f c e l l s i z e s = 25 .70252 Number o f Means 2 3 4 C r i t i c a l Range 5 .435 5 .718 5 .905 Means w i t h t h e same l e t t e r a r e not s i g n i f i c a n t l y d i f f e r e n t . Duncan Group ing Mean N STARCH A 22 .538 26 40 A 20 .769 26 45 B 11 .708 24 35 C - 4 . 1 4 8 27 30 Duncan ' s M u l t i p l e Range T e s t f o r v a r i a b l e : VOLCH Number o f Means 2 3 C r i t i c a l Range 4 .708 4 . 9 5 3 Means w i t h t h e same l e t t e r a r e not s i g n i f i c a n t l y d i f f e r e n t . Duncan Group ing Mean N SALT A 17 571 35 0 B 12 611 36 2 C 7 063 32 1 L e v e l o f L e v e l o f VOLCH-STARCH SALT N Mean SD 30 0 10 -3 .4000000 15 4430855 30 1 9 -8 .4444444 8 1103500 30 2 8 -0 .2500000 8 4978989 35 0 9 17.1111111 13 6879915 35 1 7 3 .1428571 12 3885508 35 2 8 13.1250000 13 8092878 40 0 8 26.7500000 16 3423202 40 1 7 20 .1428571 17 6958429 40 2 11 21.0000000 15 1525575 45 0 8 35.1250000 27 8333690 45 1 9 I5.4444444 21 0778979 45 2 9 13.3333333 25 7099203 124 equaling the intended concentration. In a sufficient watenstarch ratio, first starch granules rapidly absorb water, expanding quickly. Further, the crystallinity is lost while cross-links dissipate. As the temperature is raised, the remaining crystallites undergo a conventional melting transition (Waigh et al., 2000). However, when water availability is limited, this uncoiling and unraveling to an amorphous structure does not fully occur (Jenkins et al., 1999). The process of disrupting the crystalline portions with hydrogen bonds to separate mixed amylose/amylopectin regions into their starch sub-components only happens incompletely (Keetels et al., 1996a-c). In this case, no specific tests were performed to examine the degree of completion in starch gelatinization. The uneven distribution of water throughout the gel yielded a texture unlike any other of the gel formulations. The degree of gradation in gelling found in 50% starch gels was the same for 0, 1, and 2% salt formulations. If absorption stays incomplete, the texture is likened to brittleness rather than elasticity since the crystallinity would be still prominent compared to well-hydrated portions of plasticized starch (Choi & Kerr, 2003). Furthermore, there was an observable hard core of non-hydrated starch surrounded by a hard gel, yielding toward the outside edge into a softer and wetter gel and water pockets at selected areas in the casing. In all likelihood, starch rapidly absorbed water at a faster rate in some portions of the casing strands, depriving the centre parts from full hydration as mentioned above. Since no agitation occurred at this stage of sample preparation, water pockets were trapped toward the outside edges along the casing. It can be hypothesized that this constituted syneresis water from retrogradation. However, apart from the fact that potato starch is known for its low degree of retrogradation, this liquid only pooled at the tied ends and the bent portions of the filled casing observed immediately after steam processing was terminated. 125 Retrogradation would have occurred to a certain extent in the overnight storage of the gels. It has been described as a hardening effect upon retrogradation for amylose within a short time and for amylopectin occurring over the period of several weeks of storage (Miles et al., 1985a, b; Conde-Petit & Escher, 1994). But this type would have taken place in all gels strands to the same extent since all gels were stored overnight under refrigeration temperatures. Likely, due to the partially crystalline regions within the gel strands, the rigidity and subsequently E increased overproportionally compared to the lower starch concentrations. Though large variations were noted in previous experiments as well (Kortstee et al., 1998), it was decided to exclude 50% starch gels of all formulations from the final analysis to have consistent, reliable data. In addition, dielectric properties measurements reflected the inconsistencies in preparation very clearly as well (Table 4.3) and justified the omission. This yielded a clear relationship in concentration dependence of the Young's modulus, as well as volume change dependence on starch, salt, Young's modulus and power level. When comparing means according to salt concentrations, 0% salt samples across all formulations showed the highest volume increase, and were significantly different from the other formulations. HP had the highest impact on volume increase while MP and LP yielded statistically the same lower outcome. As expected, the dielectric loss factor increased with rising salt levels for all starch gels while the dielectric constant remained the same for all (Table 4.3). 126 As these results confirmed, Young's modulus was dependent on starch and salt content. When increasing starch in the formulation, Young's modulus rose. The results were similar to previous results exhibiting a linear relationship between concentration of starch and stiffness (Evans & Lips, 1991). Choi and Kerr (2003) found an increase in hardness with increasing starch concentrations. However, their tests showed the same degree of cohesiveness and springiness for 25, 33, and 40% wheat starch gels, unlike in the current results where hardness increased beyond the 40% starch concentration,. It has been suggested that hardness of the gels is a product of lower mobility of water attributed to granule remnants (Choi & Kerr, 2003). For interaction with salt, Bircan and Barringer (1998) hypothesized that salt binds the water and makes it unavailable for hydration for starch granules. This would confirm the perceived but not measured brittleness of salt-containing formulation gels due to the limited plasticizing effect of water on starch as NaCl would draw the water away from the granules. Muhrbeck and Eliasson (1987) recorded decreased gelatinization viscosity upon addition of only slightest amounts of salt. Monophosphate esters have been suggested to be responsible for this effect by increasing the negative charge of potato starch molecules (Chiotelli et al., 2002). Therefore salt changes the gelatinization behaviour and, in turn, it would affect the rheology of the starch gels in these experiments. The results for the higher starch formulation with respect to Young's modulus stands in contrast to the others, especially when looking at the effect of these formulations for medium and low power runs. The formulations of 45 and 50% starch show a much reduced volume gain. These results may suggest that the determination of Young's modulus is flawed. Since the cross sections used for 45% starch gels were significantly 127 smaller, an error is possible due to the creation of artifacts by size reduction. Furthermore, the values may not be equated to the type of stress to which the gels are exposed during the drying process. Young's modulus was determined by a compression test compared to the expansion which took place at 28 Torr absolute pressure. Since the direction of the stress is opposite and multidimensional, inconsistencies between these results could be expected. It is possible that tensile stresses are not linear compared to those in compression. Though this applies to all starch concentrations, it may be amplified for the higher ones such as 45 and 50%. The initial Young's modulus E is presumably different from the E at which expansion occurs. The increased temperature would have lowered E values. In contrast, as drying continued, it is not known what effect the lower moisture content would have had on the concommittant E. Lowering the moisture content by drying may have done more than just remove a certain amount of moisture. As known from other drying methods, textural changes, for example case hardening, occur as well as chemical changes as energy is continually absorbed to accelerate chemical reactions, according to Arrhenius rate law. Though the starch is already gelatinized at this point in time, it is possible that the starch network changes. When the initial Young's modulus was evaluated in combination with power, it yielded a significant influence on volume change. With increasing E, the potential for more pronounced volume expansion was realized for all HP- and most MP and LP- dried samples with the exception of 45% starch and 2% salt formulations (Figure 4.3). This was most apparent in the high power trials and to a lesser degree in medium and low 128 power drying. In addition, the destabilizing effect of salt on the gel strength was reflected in the limited volume increases for salt-added formulations. No-salt-gels made of 30 and 35% starch were the same volume, 40% gels were different from the lower as well as the higher 45%. The 45% gels were different from all the other in that they exhibited the highest increase in Young's modulus and volume for 0% salt dried at HP. For salt concentrations of 1 and 2%, E values for 45% starch gels remained the highest, however the LP and MP drying runs exhibited the same or lower volume increases than equivalent 40% starch gels. Observation of the drying process clarified that the initial expansion recorded during wheat dough drying, induced by the immediate change pressure, does not occur in the starch gels. The starch gels remained unchanged for most of the drying time; only edges smoothened gradually, and bulging and rounding were observed later in the drying process. This is in stark contrast to the increased Young's modulus for wheat dough where the degree of puffing decreased. Considering that initial temperatures measured in the heating process did not exceed 45-50°C, it is unlikely that steam pressure within the gels alone is responsible. Since all gel samples contained entrapped gases, their expansion would cause the quickest expansion of the gels. The bulging and rounding can possibly be attributed to this. As the samples continue to heat, liquid is converted into vapour and is able to build up where the Young's modulus values are above a certain threshold and the gel lacks brittleness. 129 Notably gel samples taken to determine temperature development and moisture loss were characterized as forming a rubbery casing in the initial stages of drying while the centres of the gels cylinders exhibited a rather soft gel structure. Rheological data for these states were not determined. Possibly this stage in drying enhances expansion similar to the gelatinization effect on the outer surface of blanched potato cubes that increased puffing (Varnalis et al., 2001a, b). The mechanism suggested by this research group was reduced water vapour permeability. However, the initial shrinkage and subsequent expansion during drying of potato was not observed in these starch gel experiments. It is very likely that the delay in expansion is due to the heating behaviour as well as steam buildup. Addressing the first, the gelatinized potato starch is stiffer than raw wheat dough and depends on the increase in temperature to lower the E characteristics to the degree where the gels are expandable. The critical temperature range appeared to be 40-45°C. Before these temperatures were reached, no changes were noted during the drying process, as documented in the video recordings. This phenomenon had been recorded for gels prepared from other starches when reheated (Miles et al., 1985a, b). In this case, by the energy supplied, the starch structures shift from a smectic phase, a layer structure with perpendicular molecules, into a nematic phase, a structure of parallel molecules (Donald, 2004). The same linear relationship of a higher E with higher starch concentrations between the formulations is to be expected at higher temperatures since the intercept of the curve would be expected to shift not its slope or direction. Young's modulus of starch gels behaves proportional to the concentration (Keetels et al. 1996c; Roberts & Cameron, 2002). There is no indication that this relationship should be 130 inverted. The salt response on rheological properties was opposite; where dough balls became stiffer upon salt addition, starch gels with added salt became weaker. Though degree of brittleness was not determined by TPA, the samples with 2%-salt were judged to be more brittle and notably did not expand as much. This suggests that the gel structure became more susceptible to quick water loss. On the other hand, the differences between HP and MP drying were likely caused by steam generation and buildup. While gels expanding to both HP and MP did not reach 43°C until 5 min drying time, samples dried under HP already exhibited bulging and rounding at 1.5-2 min and 2-3 min respectively. For MP dried starch gels, these events did not occur until 3.8±0.4 min for bulging and 5.7±1.9 min for rounding. Due to the increased energy input in HP drying, conversion to steam is quicker and allows the starch gels to expand more noticeably. For LP, the exact temperature at which bulging starts is not known as it is before 6 min,. At 22.5 min, the temperatures just surpassed 45°C. However, the temperature development up to this point is not known and can only be estimated. There is an interesting aspect to the microwave response of gelatinized potato starch formulations. An interaction between microwave power and salt can be suggested. Though not evident in the final volume outcome, the video recordings revealed shortened bulging times with increasing salt concentration (Table 4.6). This suggests that NaCl does increase the heating rate lowering the gel cylinders' E earlier and can respond to the bulging pressure. This confirmed the anticipated heating response by salt induced increased dielectric loss factor. 131 While puffing in raw potato cubes requires casing formation for pressure buildup (Varnalis et al., 2001 a, b) or water content at critical point for semi-dry corn kernels causing a different pressure buildup due to a drop in temperature (Shimoni et al., 2002), it appears that microwave dehydration under vacuum relies on the rheological properties of the sample and gas inclusion as well as steam buildup to enable expansion. In this case, the gas inclusion could not be identified as a significant factor since all samples contained equal amounts of gas inclusions based on the evaluation of images as the were cooling in the gelatinization process. Yet, there remains the possibility of case hardening in the case of starch gels that allows the expansion since no immediate effect of expansion upon pressure reduction could be observed. However, Young's modulus mostly corresponded to the expansion potential. 4.5 Summary For elucidating the mechanism of volume expansion in microwave, drying potato starch gels served as a model system. This system enabled the detection of puffing within the parameters selected. Videorecording of the drying process facilitated understanding of the interaction between the expanding gases induced by the gas evacuation and steam generation and the rheological properties of the gelled samples. Gels heated till they reached the critical temperature at which Young's modulus was small enough to yield to the pressure differential and to expand. Simultaneously, the Young's modulus had to be large enough to resist collapse. Young's modulus in this case was dependent on starch and salt content, both parameters working counterproductively 132 to each other in terms of strength of the gels. The volume expansion was possible under high energy supplied by the magnetron. While evacuation of gases from the chamber sets the stage for pressure loss, the expansion is not possible without microwave heating that enables the expansion. Once Young's modulus for the starch gels is lowered sufficiently and entrapped gas as well as the generated steam accumulate and can push the substance outward and increase the volume. The reduced pressure in the chamber allows the temperature to stay low enough to ensure low quality losses. Thus, the critical parameters for microwave drying under reduced absolute pressure for potato starch gels are the texture characteristics, suitable energy supplied by the microwaves and the pressure differential induced in the chamber. 133 CHAPTER V Summary and Future Directions In order to study and describe puffing induced in microwave drying under vacuum, wheat and starch preparations were used as models. Both systems were purposely modified in their rheology with salt. Additionally, dough balls only were modified with different flour composition, specifically comparing high to low gluten content. In contrast, gelatinized potato starch varied in starch concentration, i.e., the water content varied in addition to the salt concentration. Furthermore, drying conditions were modified by using high, medium and low power settings for microwave energy. Dough balls were dried at 700 and 1300 W while starch gels were exposed to 350, 700 and 1300 W in the drying trials, both model series were dried at 28 Torr absolute pressure. Both of these drying setups were intended to elucidate the drying mechanisms. Their respective responses differed at first glance. Namely salt interactions for both systems were opposite in their rheological response. While wheat doughs lost their elasticity and increased fracture strength, salt containing starch gel formulations exhibited a lowering of Young's modulus. However, the two systems complemented each other and actually helped explain better and more solidly which mechanisms were at play. To aid the process, video recording, FE modeling as well as image analysis were employed to characterize the degree of the expansion and nature of porosity induced in the drying process for dough samples. The results provide a much more comprehensive picture than anticipated, though the three-part hypothesis was not confirmed in its entirety. 135 As expected, volume increases were recorded for both model systems. Wheat dough responded very strongly to microwave drying under vacuum in the extent to which expansion occurred. The volume changes in V M D occurred due to the reduced absolute pressure but to a much larger degree than anticipated as became obvious through video documentation and FE modeling. Though the expectations were not quantified, it was anticipated that largest portion of the expansion would be due to steam buildup and sudden release of gases. The reduction of pressure and the induced decompression of the dough as well as the expansion of entrapped gases were responsible for this initial volume increase in dough samples. Thus the degree to which entrapped gases already expanded upon reduction of pressure was underestimated. Rheology of wheat dough was the deciding factor that allowed expansion to occur due to the pressure differential. This was clarified by the divergent properties of high and low quality flour HG and L G as well as salt concentration. Likely the mechanism behind HG dough expansion was strain hardening, a phenomenon typical of good baking quality wheat flour. It facilitated expansion and holding of the newly reached volume to yield large puffed balls with even pore distribution. Gelatinized starch provided much less expansion upon drying. Nevertheless it was measurable and responsive to the parameters of starch concentration, salt and power levels. The higher the initial starch content measured, the more pronounced expansion was. Salt concentration modified rheological properties such that the E values were lowered and those samples became less cohesive. This in turn reflected in the lower expansion recorded. 136 The first hypothesis, stressing the importance of upper and lower thresholds for rheological properties between which microwave/vacuum exposure can result in puffing, was clearly confirmed. Though not possible to distinctly define the upper and lower limits, the fracture strength of wheat doughs clearly showed the contrast in flour strength and the resulting expansion potential. Salt addition however modified the rheological properties such that the fracture strength increased and impaired the expansion of both flour systems. The upper limit must lie above 0.4 MPa for strong flours, or rather the equivalent value at the corresponding higher temperature. As fracture strength was not determined for L G dough, a statement cannot be made. For dough, the difference lies in the disparity between low and high gluten flours, more specifically in this case the quality of the flour for dough making as well as its salt content and the resulting salt interaction with regard to the fracture strength and their ability to exhibit strain hardening. The thresholds themselves could not be established, though the intent of the study was rather the start of staking the upper and lower limits than quantifying them. However it was established that the pressure and generated steam response allow expansion while avoiding collapse before drying is completed. For gelatinized starch samples, the rheological properties demonstrated clearly a threshold dependence of starch concentration and salt as well. The minimum E appeared to be approximately 0.15 MPa, the maximum 1 MPa or the equivalent values at higher temperatures. Yet, the expansion potential was lower and therefore the possibility of collapse decreased. Nevertheless, since shrinkage was recorded for low starch concentrations at 30% and 0-2% salt and for medium and low power and 35% starch and 137 1% salt at low power, this model system shows limits to the gels' stability and confirms the hypothesis as well. The second hypothesis stated the microwave power dependence of the expansion as well as the existence of a theshold. For the dough system it was obvious that the lower limit has not been found. The two levels chosen, 700 W and 1300 W were both successful in triggering volume expansion, shown in HG doughs while L G was only tested at high power as it poorly expanded even under these condition of ample energy supply. This threshold must lie lower than the tested 700 W per 600 g sample load. Starch experiments on the other hand narrowed the range at which puffing for gels in a vacuum microwave are possible since high power drying runs yielded clearly puffed gels whereas medium and low power showed similarly low expansions. However, the lower limit of power was not yet reached in these experiments though 350 W drying runs did yield a significantly lower puffing outcome. The third hypothesis entailed salt addition to the formulations in order to increase the microwave response by raising the dielectric loss factor and therefore increasing heat generation, which in turn would support more steam development and more puffing. The proof of this hypothesis segment is still outstanding since the interaction between salt and the remaining formulation was stronger than the effect modification in heating behaviour might have had. The dielectric loss factor was raised, as demonstrated. However, it did not establish itself as a puffing enhancing factor. In the contrary, the opposite was true. Volume varied inversely with salt concentration for both model systems. However the 138 data obtained raised a new question: does a higher dielectric loss factor raise the temperature within the model system and how does temperature develop throughout the drying process and in turn influence the drying rate? As with other puffed products, just one factor usually is not sufficient to define the mechanism for its expansion. Generally it is an interplay between energy input into the system and physical and chemical product characteristics, as summarized by previous authors (Wu & Schwartzberg, 1994; Shimoni et al., 2002; Yamsaengsung & Moreira, 2002). What distinguishes V M dehydration from other drying methods is the mode of heat transfer. Since microwaves do not primarily supply direct heating but provide heat by generating it in dipolar agitation, choosing and modifying the product to be dried becomes more challenging. As well, a suitable model system needs to be found where low and high threshold levels for dielectric properties modification do not affect other parameter qualities. What this research was unable to provide is the distinct characterization of the dielectric response. Obviously the chosen model systems had dielectric properties in the range where they responded to microwave energy input. But as stated prior, the change in dielectric properties did not manifest itself in an enhanced puffing. When looking specifically at the response to salt, neither model provided the benefit of increased dielectric heating response that translated into different puffing response. The individual heating response could not be judged since only 0.0% salt doughs and starch gels were directly measured. Instead, both showed a different salt effect with regard to texture changes. 139 Wheat protein became less elastic and did not allow expansion beyond a certain volume limit. Thus the addition of salt proved counterproductive as the results of likely interaction between flour and salt previously observed in baking research. When salt is added to wheat dough, the hydration of the protein changes. The plasticizing effect of water is reduced due to the binding of to salt, thus changing the viscoelasticity of the dough. However, the two levels of microwave power chosen for dough did not explain differences. The lower limit of the puffing threshold for HG dough was clearly below the chosen experimental energy levels. For starch experiments, the responses were different to varying levels of power. In contrast to flour doughs, starch gels, though they also did not expand more rather less under salt addition, appeared to change the texture through lower cohesion of the gels. The elasticity of the gels mostly came from the network of entangled swollen granules, in which all crystallinity has been lost (Keetels et al., 1996a-c), and where the double helices are unwound and amylopectin molecules in hydrogen bond based structures (Waigh et al., 2000) move freely. Once the critical temperature of 40-45°C and the resulting E were reached, the pressure differential could enhanced the expansion of the enclosed gas bubbles as well as the steam generated in the heating process. Thus, the critical parameters for microwave drying under vacuum for potato starch gels are the rheological characteristics, namely Young's modulus and possibly cohesiveness, suitable power supplied by the microwaves and the pressure differential induced in the chamber. Young's modulus showed dependence on starch and salt content, two parameters that work counterproductively in terms of strength of the gels. The destabilizing effect of salt 140 was obvious: However, using Young's modulusas the sole texture parameter to explain the puffing, did not yield completely satisfying results because not all high E values yielded the most expansion recorded upon drying. It could be useful to expand the research of rheological properties by looking at texture profile analysis (TPA) and the cohesiveness value in particular because differences in cohesiveness in the samples were observed but not quantified. However, at the same time the proper rheological properties, namely Young's modulus and fracture strength as investigated here, are necessary so that the steam and/or entrapped gases do not immediately escape the system. Rather the permeability has to be below a limit that would prevent expansion. In the case of wheat doughs, the occurrence of strain hardening enhanced the phenomenon. Strain hardening is specific to local stresses and strains and generally occurs at high stresses. It is difficult to predict the local strain within the dough exactly, as has been attempted in baking science research, since the dimensions in question are very small and strain gages may not be applied in satisfactory manner. With correlations or empirical values from a series of test runs with some samples, it would be possible to predict the puffing potential. Clearly, there are factors that would prohibit puffing potential and thus it could be predicted whether or not a product can be puffed or modified to the extent that puffing is possible. The establishment of suitable rheological property measurements to define puffing capacity are crucial to the success of designing puffable products. Puffing under vacuum in microwave assisted drying was achieved and the mechanisms were revealed to a certain extent. From these model systems, a threshold is evident below which Young's 141 modulus or fracture strength are insufficient to withhold expansion without subsequent collapse. More experiments would be needed to determine the limit more precisely. The application of FE modeling in this particular case was a good beginning for understanding microwave assisted drying under vacuum as FEM combined heat and mass transfer with the volume expansion phenomena observed in the process. On the other hand, it also clarifies how complex food processing is. Though it reflected the degree of volume expansion reasonably well, it failed to predict the increase and subsequent partial shrinkage of low salt wheat dough formulations during the drying process, observed in the video recordings. By estimating a certain degree of fracture strength reduction, this shortcoming could be corrected. However, this was not based on specific knowledge about the dough rheology but rather an educated guess. Furthermore, the model did not calculate the temperature development the way it was recorded experimentally. This in turn yielded estimates for volume changes that were unrealistically high for high salt concentrations. The loss of heat from the system, assumed to be present by the temperature differential to the experimental temperature values, for higher salt contents cannot be explained. Though the final temperatures recorded in the experimental portion were the same for all salt concentrations, one would expect the temperature development for higher salt formulations to differ from no salt added due to the dielectric heating effect. Possibly the fact that the model incorporated data from wave guide dryers instead of resonant chamber skewed the results. A resonant chamber shifts the emphasis of the dielectric loss factor on the heating behaviour to where its impact would be lower compared to the generally used wave guide dryer considered in the literature and used as 142 the model. Additionally, energy was funneled into chemically changing the dough structure. Since the dried dough balls were not analysed for degree of starch gelatinization or protein denaturation, it can be speculated that chemical change took place. Regardless, these contrasting results do, however, point out the discrepancy between experimental and numerical results. Since one part of the hypothesis regarding the enhanced puffing with an increased loss factor was reflected in the FE results, it becomes clear that there is a knowledge gap waiting to be filled. More experiments are needed to confirm or discard these questions raised. Detailed temperature studies regarding different salt concentrations would be useful to determine the heating behaviour during the drying process. Furthermore, determination of fracture strength with rising temperatures would reduce the guess work regarding the rheological behaviour throughout the drying process. As the modeling time was restricted to 180 s microwave exposure based on the observation that expansion is completed by 3 min into the drying procedure, the final temperature achieved in these experiments was not modeled. The model is clearly lacking parameters that are important to the drying development. Furthermore, the FE model confirms that the increase in vapour pressure due to the rise in temperature is not the only driving force behind the increase in volume. Entrapped gases as well as the effects of decompression of the dough do influence the initial expansion with lowering of the chamber pressure. The FE model matched in principle the expected mechanism of vacuum and microwave effects. 143 However, when more parameters are known a better model can be implemented and more precise predictions can be made. In this case, several factors such as heating behaviour of salt containing formulations, rheological changes upon heating and degree of gas entrapment were not known or ignored for simplicity. Through the combination of videorecording and FEM, it was possible to clearly pinpoint the crucial parameters in this process. Now, a better FE model describing drying curves over a longer time can be built, taking the following into consideration: compressibility factors and heating behaviour of dough at different salt levels. With these modifications, an improved understanding of rheology and puffing for V M drying can be achieved. The intent of employing image analysis was to quantify porosity of VM-dried product by looking at cross sectional images of dried dough balls. The custom written program was set up to recognize the cross sectional areas of open pores according to contrast in shading compared to the surrounding dough, also refered to as membranes in dough processing. Therefore, image analysis opened the opportunity to analyze the porosity regarding the innate pore sizes and their distribution as a response to microwave drying under vacuum with much less effort and increased precision. Image analysis has a lot of potential in product evaluation. The generally non-destructive and safe nature of this method makes it a good alternative to chemical methods to evaluate porosity. If recognition is good, a quick analysis and comparison between samples would be a great asset to evaluation of porosity. Since the size of the actual image is not relevant but only the clarity and focus, any size specimen could be analyzed. 144 This series of experiments was able to document puffing in vacuum microwave drying for wheat dough and potato starch gels in varying formulations. It was able to characterize some of the important parameters necessary for puffing to occur. In combination with FE modeling, good points were raised regarding possible mechanisms and interactions that warrant further study. Especially FEM offers a suitable tool to further explore the field of food science, or drying in this particular case. Though standardization of the formulation regarding a puffable product in V M D is still far, this study showed the possibilities for future product development in this particular area. 145 BIBLIOGRAPHY A A C C . 1994. Method 46-30. Protein Determination by LECO Analysis. Approved Methods (8 th Ed.). American Association of Cereal Chemists. St. Paul, M N , USA. AO AC. 1995. Moisture determination. Official Methods of Analysis. Association of Official Analytical Chemists. Gaithersburg, MD, USA. A A C C . 1994. Approved Methods (8 th Ed.). American Association of Cereal Chemists. St. Paul, M N , USA. Aguilera, J.M. and Rojas, G.V. 1997. Determination of Kinetics of Gelation of Whey Protein and Cassava Starch by Oscillatory Rheometry. Food Research International. 30(5):349-357. Angioloni, A. and Dalla Rossa, M . 2005. Dough Thermo-Mechanical Properties: Influence of Sodium Chloride, Mixing Time and Equipment. Journal of Cereal Science. 41:327-331. Anonymous, 1982. Handbook of Chemistry and Physics. 63 r d Edition. CRC Press. Boca Raton, USA. Anonymous, 2003. Organic Application Notes. Form No. 203-821-176. LECO Corporation. St. Joseph, Michigan, USA. Autio, K. and Laurikainen, T. 1997. Relationship between Flour/Dough Microstructure and Dough Handling and Baking Properties. Trends in Food Science and Technology. 8:181-185. Bagley, E.B., Dintzis, F.R., and Chakrabarti, S. 1998. Experimental and Conceptual Problems in the Rheological Characterization of Wheat Flour Dough. Rheologica Acta. 37:556-565. Beitz, W., Kuttner, K. -H. and Davies, B.J. 1994. Dubbel Handbook of Mechanical Engineering. Springer, Berlin, Germany. Belton, P.S. 2005. New Approaches to Study the Molecular Basis of the Mechanical Properties of Gluten. Journal of Cereal Science. 41 (2):203-211. Belitz, H.-D., Kieffer, R., Seilmeier, W. and Wiser, H. 1986. Structure and Function of Gluten Proteins. Cereal Chemistry. 63(4):336-341. Bergthaller, W. 2004. Development of Starches. In: Eliasson, A.-C. (Ed.) Starch in Food. CRC Press, pp. 241-257. 146 Binding, D.M., Couch, M.A. , Sujatha, K.S., and Webster, M.F. 2003. Experimental and Numerical Simulation of Dough Kneading in Filled Geometries. Journal of Food Engineering. 58:111-123. Bircan, C. and Barringer, S. A. 1998. Effect of Viscosity on the Dielectric Properties of a Starch Solution when Salt is Present. Journal of Food Science. 63(6):983-986. Bloksma, A .H. and Bushuk, W. 1988. Rheology and Chemistry of Dough. In: Pomeranz, Y . (Ed.) Wheat Chemistry and Technology, Vol. 2. American Association of Cereal Chemists. St. Paul, M N . pp. 131-217. Bloksma, A .H. 1990. Dough Structure, Dough Rheology, and Baking Quality. Cereal Foods World. 35(2):237-244. Buffler, C.R. 1993. Microwave Cooking and Processing. Engineering Fundamentals for the Food Scientists. Van Nostrand Reinhold, New York, USA. Calderon-Dominguez, G., Vera-Dominguez, M . , Farrera-Rebollo, R., Arana-Errasquin, R. and Mora-Escobedo, R. 2004. Rheological Changes of Dough and Bread Quality Prepared from a Sweet Dough: Effect of Temperature and Mixing Time. International Journal of Food Properties. 7(2): 165-174. Callejo, M.J., Gil , M.J., Rodriguez, G. and Ruiz, M.V. 1999. Effect of Gluten Addition and Storage Time on White Pan Bread Quality: Instrumental Evaluation. Zeitschriftfuer Lebensmittel Untersuchung undForschung A. 208:27-32. Campbell, G.M., Rielly, C D . , Fryer, P.J. and Sadd, P.A. 1993. Measurement and Interpretation of Dough Densities. Cereal Chemistry. 70(5):517-521. Canadian Grain Commission. 2005. Harvest Quality. Protein Content. http://www.grainscanada.gc.ca/Quality/protein/proteinmenu-e.htm Accessed March 14, 2005. Chang, Y.H. , Lim, S.T. and Yoo, B. 2004. Dynamic Rheology of Corn Starch-Sugar Composites. Journal of Food Engineering. 64(4):521-527. Charalambides, M.N. , Wanigasooriya, L., Williams, J.G. and Chakrabarti, S. 2002a. Biaxial Deformation of Dough using the Bubble Inflation Technique. I. Experimental. Rheologica Acta. 41:532-540. Charalambides, M.N. , Wanigasooriya, L. and Williams, J.G. 2002b. Biaxial Deformation of Dough using the Bubble Inflation Technique. II. Numerical Modeling. Rheologica Acta. 41:541-548. Chen, C.R. and Ramaswamy, H.S. 1999. Rheology of Tapioca Starch. Food Research International. 32:319-325. 147 Chin, N . L. 2003. Bread dough aeration and rheology during mixing. Ph.D. thesis. University of Manchester. Abstract. Chin, N.L, Campbell, G.M. and Thompson, F. 2005. Characterisation of Bread Doughs with different Densities, Salt Contents and Water Levels using Microwave Power Transmission Measurements. Journal of Food Engineering. 70(2) :211-217. Chiotelli, E. and Campbell, G.M. 2003. Proving of Bread Dough I: Modelling the Evolution of the Bubble Size Distribution. Food and Bioproducts Processing: Transactions of the Institution of Chemical Engineers, Part C. 81(C3):194-206. Chiotelli, E., Pilosio, G. and Le Meste, M . 2002. Effect of Sodium Chloride on the Gelatinization of Starch: a Multi-measurement Study. Biopolymers. 63(1):41—58. Chiotelli E., Rolee, A. and Le Meste, M . 2004. Rheological Properties of Soft Wheat Flour Doughs: Effect of Salt and Triglycerides. Cereal Chemistry. 81(4):459-468. Choi, S.G. and Kerr, W.L. 2003. Water Mobility and Textural Properties of Native and Hydroxypropylated Wheat Starch Gels. Carbohydrate Polymers. 51(1): 1 -8. Colpitts, B., Pelletier, Y . and Cogswell, S. 1992. Complex Permittivity Measurements of the Colorado Potato Beetle using Coaxial Probes Techniques. Journal of Microwave Power and Electromagnetic Energy. 27(3): 175-182. Connelly, R.K. and Kokini, J.L. 2004. The Effect of Shear Thinning and Differential Viscoelasticity on Mixing in a Model 2D Mixer as Determined using F E M with Particle Tracking. Journal of Non-Newtonian Fluid Mechanics. 123:1 -17. Conde-Petit, B. and Escher, F. 1994. Influence of Starch-Lipid Complexation on the Ageing Behaviour of High Concentration Starch Gels. Starch/Staerke 46(5): 172-177. Cornec, M . , Popineau, Y. and Lefebvre, J. 1994. Characterisation of Gluten Subfractions by SE-HPLC and Dynamic Rheological Analysis in Shear. Journal of Cereal Science. 19(2): 131-139. Cuq, B., Boutrot, F. Redl, A. and Lullien-Pellerin, V. 2000. Study of the Temperature Effect on the Formation of Wheat Gluten Network: Influence on Mechanical Properties and Protein Solubility. Journal of Agriculture and Food Chemistry. 48(7):2954-2959. Dobraszczyk, B.J., Campbell, G.M., and Gan, Z. 2001. Bread: a Unique Food. In: Dendy, D.A.V. and Dobraszczyk, B.J. (Eds.) Cereal and Cereal Products, Chemistry and Technology. Aspen Publishers, Gaithersburg, MD, USA. pp. 182-232. 148 Dobraszczyk, B.J. and Morgenstern, M.P. 2003. Rheology and the Breadmaking Process. Journal of Cereal Science. 38:229-245. Dobraszczyk, B.J. and Roberts, C.A. 1994. Strain Hardening and Dough Gas Cell-wall Failure in Biaxial Extension. Journal of Cereal Science. 20(3):265-274. Dobraszczyk, B.J. and Schofield, J.D. 2000. Measurement of Biaxial Extensional Rheological Properties using Bubble Inflation and the Stability of Bubble Expansion in Bread Doughs. Proceedings of the XHIth International Congress on Rheology, Cambridge, UK. Donald, M.A. 2004. Understanding Starch Structure and Functionality. In: Eliasson, A.-C. (Ed.). Starch in Food. CRC Press, pp. 128-155. Drouzas, A.E. and Schubert, H. 1996. Microwave Application in Vacuum Drying of Fruits. Journal of Food Engineering. 29(2):203-209. Drouzas, A.E., Tsami, E. and Saravacos, G.D. 1999. Microwave/Vacuum Drying of Model Fruit Gels. Journal of Food Engineering. 39:117-122. Durance, T.D. and Wang, J.H. 2001. Energy Consumption, Density, and Rehydration Rate of Vacuum Microwave- and Hot-Air Convection-Dehydrated Tomatoes. Journal of Food Science. 67(6) :2212-2216. Ellis, H.S., Ring, S.G. and Whittam, M.A. 1989. A Comparison of the Viscous Behavior of Wheat and Maize Starch Pastes. Journal Of Cereal Science. 10:33^14. Erie, U . and Schubert, H. 2001. Combined Osmotic and Microwave-Vacuum Dehydration of Apples and Strawberries. Journal of Food Engineering. 49(2/3): 193-199. Evans, I.D. and Lips, A . 1991. Viscoelasticity of Gelatinized Starch Dispersions. Journal of Texture Studies. 23:69-86. Faridi, H. and Rasper, V.F. 1987. The Alveograph Handbook. American Association of Cereal Chemists. St. Paul, M N , USA. Faergestad, E.M., Molteberg, E.L. and Magnus, E .M. 2000. Interrelationships of Protein Composition, Protein Level, Baking Process and the Characteristics of Hearth Bread and Pan Bread. Journal of Cereal Science. 31:309—320. Fan, J., Mitchell, J.R. and Blanshard, J.M.V. 1999. A Model for the Oven Rise of Dough during Baking. Journal of Food Engineering. 41:69-77. 149 Faubion, J.M. and Hoseney. 1989. Viscoelastic Properties of Wheat Flour Dough. In: Faridi, H.A. and Faubion, J.M. (Ed.). Dough Rheology and Baked Texture. Van Nostrand Reinhold, New York (1989), pp. 29-66. Feigl, K., Kaufmann, S.F.M., Fischer, P. and Windhab, E.J. 2003. A Numerical Procedure for Calculating Droplet Deformation in Dispersing Flows and Experimental Verification. Chemical Engineering Science, 58:2351 -2363. Goedeken, D.L., Shah, K . K . and Tong, C.H. 1998. True Thermal Conductivity Determination of Moist Porous Food Materials at Elevated Temperatures. Journal of Food Science. 63(6): 1062-1066. Gras, P.W., Carpenter, H.C. and Anderssen, R.S. 2000. Modelling the Developmental Rheology of Wheat-Flour Dough using Extension Tests. Journal of Cereal Science. 31(1):1-13. Gunasekaran, S. 1999. Pulsed Micro wave-Vacuum Drying of Food Materials. Drying Technology. 17(3):395-412. Gunasekaran, N . 2002. Effect of Fat Content and Food Type on Heat Transfer during Microwave Heating. Master Thesis. Virginia Polytechnic Institute and State University, Blacksburg, Va. Gupta, R.B., Khan, K. and Macritchie, F. 1993. Biochemical Basis of Flour Properties in bread Wheats. I. Effect of Variation in the Quantity and Size Distribution of Polymeric Protein. Journal of Cereal Science. 18:23-41. Hayta, M . and Schofield, J.D. 2004. Heat and Additive induced Biochemical transitions in Gluten from Good and Poor Breadmaking Quality Wheats. Journal of Cereal Science. 40:245-256. Hibberd, G.E. 1970. Dynamic Viscoelastic Behaviour of Wheat Flour Doughs. Part II. Effects of Water Content in the Linear Region. Rheologica Acta. 9(4):497-500. Holmann, J.P. 2001. Experimental Methods for Engineers. McGraw-Hill, New York, NY, USA. Howling, D. 1980. The Influence of the Structure of Starch on its Rheological Properties. Food Chemistry. 6:51-61. Hussain, M.A., Rahman, M.S. and Ng, C.W. 2002. Prediction of Pores Formation (Porosity) in Foods during Drying: Generic Models by the Use of Hybrid Neural Network. Journal of Food Engineering. 51(3):239-248. 150 Inaba, H., Hoshiziwa, M . , Adachi, T., Matsurhura, Y. and Mori, T. 1994. Characterization of Texture and Mechanical Properties of Starch Gels. Food Hydrocolloids. 8(l):33-44. Jankowski, T. and Rha, C.K. 1986. Retrogradation of Starch in Cooked Wheat. Starch/Staerke. 38(l):6-9. Janssen, A . M . , van Vliet, T. and Vereijken, J.M. 1996a. Rheological Behaviour of Wheat Glutens at Small and Large Deformations. Comparison of Two Glutens Differing in Bread Making Potential. Journal of Cereal Science. 23(1): 19—31. Janssen, A . M . , van Vliet, T. and Vereijken, J.M. 1996b. Rheological Behaviour of Wheat Glutens at Small and Large Deformations. Effect of Gluten Composition. Journal of Cereal Science. 23(1) 22-A2. Janssen, A . M . , van Vliet, T. and Vereijken, J.M. 1996c. Fundamental and Empirical Rheological Behaviour of Wheat Flour Doughs and Comparison with Bread Making Performance. Journal of Cereal Science. 23(l):43-54. Jenkins, P.J. and Donald, A . M . 1998. Gelatinisation of Starch: a combined SAXS/WAXS/DSC Study. Carbohydrate Research 308:699. Jia, C. -C, Sun, D.-W. and Cao, C.-W. 2000. Mathematical Simulation of Stresses within a Corn Kernel during Drying. Drying Technology, 18(4, 5):887-906. Kaensup, W., Chutima, S. and Wongwises, S. 2002. Experimental Study on Drying of Chilli in a Combined Microwave-Vacuum-Rotary Drum Dryer. Drying Technology. 20(10):2067-2079. Karim, A.A. , Norziah, M.H. and Seow, C.C. 2000. Methods for the Study of Starch Retrogradation. Food Chemistry. 71(l):9-36. Keentok, M . , Newberry, M.P., Gras, P., Bekes, F. and Tanner, R. 2002. The Rheology of Bread Dough made from four Commercial Flours. Rheologica Acta. 41:173-179. Keetels, C.J.A.M., van Vliet, T. and Walstra, P. 1996a. Gelation and Retrogradation of Concentrated Starch Systems: 1 Gelation. Food Hydrocolloids. 10(3):343—353. Keetels, C.J.A.M., van Vliet, T. and Walstra, P. 1996b. Gelation and Retrogradation of Concentrated Starch Systems. 2. Retrogradation. Food Hydrocolloids. 10(3):355-362. Keetels, C.J.A.M., van Vliet, T. and Walstra, P. 1996c. Gelation and Retrogradation of Concentrated Starch Systems. 3. Effect of Concentration and Heating Temperature. Food Hydrocolloids. 10(3):363-368. 151 Keetels, C.J.A.M., van Vliet, T. and Walstra, P. 1996d. Relationship between the Sponge Structure of Starch Bread and its Mechanical Properties. Journal of Cereal Science. 24(1):27-31. Kemp, I.C. and Oakley, D.E. 2002. Modeling of Particulate Drying in Theory and Practice. Drying Technology. 20(9):1699-1750. Kent, M . and Kress-Rogers, E. 1987. The COST90is Collaborative Work on the Dielectric Properties of Foods. In: Jowitt, R., Escher, F., Kent, M . , McKenna, B. & Roques, M . (Eds.). Physical Properties of Foods. 2. COST 90bis Final Seminar Proceedings. Elsevier Applied Science, London, pp. 171-197. Khatkar, B.S., Fido, R.J., Tatham, A.S. and Schofield, J.D. 2002a. Functional Properties of Wheat Gliadins. I. Effects on Mixing Characteristics and Bread Making Quality. Journal of Cereal Science. 35:299-306. Khatkar, B.S., Fido, R.J., Tatham, A.S. and Schofield, J.D. 2002b. Functional Properties of Wheat Gliadins. II. Effects on Dynamic Rheological Properties of Wheat Gluten. Journal of Cereal Science. 35:307-313. Khraisheh, M . A . M . , McMinn, W.A.M. and Magee, T.R.A. 2000. A Multiple Regression Approach to the Combined Microwave and Air Drying Process. Journal of Food Engineering; 43(4):243-250. Kieffer, R., Wieser, H., Henderson, M.H. and Graveland, A. 1998. Correlations of the Breadmaking Performance of Wheat Flour with Rheological Measurements on a Micro-scale. Journal of Cereal Science. 27:53-60. Kim, H.O., Durance, T.D., Seaman, C.H. and Kitts, D.D. 2000. Retention of Alkamides in Dried Echinacea purpurea. Journal of Agricultural and Food Chemistry. 48(9):4187-4192. Kokelaar, J.J., van Vliet, T. and Prins, A. 1996. Strain Hardening Properties and Extensibility of Flour and Gluten Doughs in Relation to Breadmaking Performance. Journal of Cereal Science. 24(3): 199—214. Kokini, J.L., Cocero, A . M . , Madeka, H. and de Graaf, E. 1994. The Development of State Diagrams for Cereal Proteins. Trends in Food Science and Technology. 5:281-288. Kortstee, A.J., Suurs, L.C.J.M, Vermeesch, A,M.G. , Keetels, C.J.A.M., Jacobsen, E. and Visser , R.G.F. 1998. The Influence of an Increased Degree of Branching on the Physico-Chemical Properties of Starch from Genetically Modified Potato. Carbohydrate Polymers. 37(2): 173-184. 152 Krokida, M.K., Karathanos, V.T. and Maroulis, Z.B. 1998. Effect of Freeze-Drying Conditions on Shrinkage and Porosity of Dehydrated Agricultural Products. Journal of Food Engineering. 35(4):369-380. Langton, M . and Ffermansson, A . - M . 1989. Microstructural Changes in Wheat Starch Dispersions during Heating and Cooling, FoodMicrostructure. 8:29-39. Lee, L., Ng, P.K.W., Whallon, J.H. and Steffe, J.F. 2001. Relationship between Rheological Properties and Microstructural Characteristics of Nondeveloped, Partially Developed, and Developed Doughs. Cereal Chemistry. 78(4):447-452. Letang, C , Piau, M . and Verdier, C. 1999. Characterization of Wheat Flour-Water Doughs. Part I: Rheometry and Microstructure. Journal of Food Engineering. 41(2):121-132. Lin, Y.E. , Anantheswaran, R.C. and Puri, V . M . 1995. Finite Element Analysis of Microwave Heating of Solid Foods. Journal of Food Engineering, 25, 85-112. Lin, T.M., Durance, T.D. and Seaman, C.H. 1998. Characterization of Vacuum Microwave, Air and Freeze dried carrot slices. Food Research International. 31(2):111-117. Liu, Q., Yada, R. and Arul, J. 2002. Characterization of Thermal Properties of Potato Dry Matter-Water and Starch-Water Systems. Journal of Food Science. 67(2):560-566. Liu, H., Lelievre, J. and Ayoungchee, W. 1991. A Study of Starch Gelatinisation using Differential Scanning Calorimetry, X-ray and Birefringence Measurements. Carbohydrate Research 210:79-87. Mandala, I.G., Palogou, E.D. and Kostaropoulos, A.E. 2002. Influence of Preparation and Storage Conditions on Texture of Xanthan-Starch Mixtures. Journal of Food Engineering. 53(l):27-38. Magnus, E.M. , Brathen, E., Sahlstram, S., Vogt, G. and Fa;rgestad, E .M. 2000. Effects of Flour Composition, Physical Dough Properties and Baking Process on Hearth Loaf Properties Studied by Multivariate Statistical Methods. Journal of Cereal Science. 32:199-212. Martinant, J.-P., Nicolas, Y. , Bouguennec, A., Popineau, Y. , Saulnier, L. and Branlard, G. 1998. Relationships Between Mixograph Parameters and Indices of Wheat Grain Quality. Journal of Cereal Science. 27:179-189. Martins, R.C. and Silva, C .L .M. 2004a. Inverse Problem Methodology for Thermal-Physical Properties Estimation of Frozen Green Beans. Journal of Food Engineering 63(4):383-392. 153 Martins, R.C. and Silva, C .L .M. 2004b. Green Beans (Phaseolus vulgaris, L.) Quality Loss upon Thawing. Journal of Food Engineering. 65(l):37-48. Maskan, M . 2001a. Kinetics of Colour Change of Kiwifruits during Hot Air and Microwave Drying. Journal of Food Engineering. 48(2): 169-175. Maskan, M . 2001b. Drying, Shrinkage and Rehydration Characteristics of Kiwifruits during Hot Air and Microwave Drying. Journal of Food Engineering. 48(2): 177-182. McLaughlin, C P . and Magee, T.R.A. 1999. The Effect of Air Temperature, Sphere Diameter and Puffing with CO2 on the Drying of Potato Spheres. Drying Technology. 17(1/2):119-136. Mellgren, E., Ohlsson, T., Risman, P.O., Skjoldebrand, C. and Wass, B. 1988. Dielectric Properties of Wheat Bread Dough. In: Asp, N-G. (Ed.) Cereal Science and Technology in Sweden, Proceedings from an International Symposium pp. 322-324 Ystad, Sweden. Meredith, R.J. 1998. Microwave Interaction with Dielectric Materials. Engineers' Handbook of Industrial Microwave Heating. IEE power series: 25. Institution of Electrical Engineers, London, UK. pp. 19-51. Miles, M.J., Morris, V.J. and Ring, S.G. 1985a. Gelation of Amylose. Carbohydrate Research. 135(2):257-269. Miles, M.J., Morris, V.J. and Ring, S.G. 1985b. The Roles of Amylose and Amylopectin in the Gelation and Retrogradation of Starch. Carbohydrate Research. 135(2):271-281. Mills, E .N.C, Wilde, P.J., Salt, L.J. and Skeggs, P. 2003. Bubble Formation and Stabilization in Bread Dough. Food and Bioproducts Processing: Transactions of the Institution of Chemical Engineers, Part C. 81 (C3): 189-193. Mousa, N . and Farid, M . 2002. Microwave Vacuum Drying of Banana Slices. Drying Technology. 20(10):2055-2066. Mudgett, R.E. 1985. Dielectric Properties of Food. In: Decareau, R.V. (Ed.) Microwaves in the Food Processing Industry, Academic Press, Orlando, FL pp. 15-37. Muhrbeck, P. and Eliasson, A.C. 1987. Influence of pH and Ionic Strength on the Viscoelastic Properties of Starch Gels. A Potato and Cassava Starches. Carbohydrate Polymers. 7:291-300. 154 Mui, W. W.Y., Durance, T.D. and Seaman, C.H. 2002. Flavor and Texture of Banana Chips Dried by Combinations of Hot Air, Vacuum, and Microwave Processing. Journal of Agricultural and Food Chemistry. 50 (7):1883-1889. Mujumdar, A., Beris, A . N . and Metzner, A.B. 2002. Transient Phenomena in Thixotropic Systems. Journal of Non-Newtonian Fluid Mechanics. 102:157-178. Ndife, M.K., Sumnu, G. and Bayindirli, L. 1999. Dielectric Properties of Six Different Species of Starch at 2450 MHz. Food Research International. 31(l):43-52. Oliveira, M.E.C. and Franca, A.S. 2000. Finite Element Analysis of Microwave Heating of Solid Products. International Communications of Heat and Mass Transfer. 27(4):527-536. Oliveira, M.E.C. and Franca, A.S. 2002. Microwave Heating of Foodstuffs. Journal of Food Engineering. 53(4):347-359. Osborne, T.B. 1907. The Proteins of the Wheat Kernel. Carnegie Institute, Washington. Publication No. 84. Pandit, R.B. and Prasad, S. 2003. Finite Element Analysis of Microwave Heating of Potato - Transient Temperature Profiles. Journal of Food Engineering, 60:193-202. Parker R. and Ring, S.G. 2001. Aspects of the Physical Chemistry of Starch. Journal of Cereal Science. 34(1):1-17. Perre, P. and May, B.K. 2001. A Numerical Drying Model that Accounts for the Coupling between Transfers and Solid Mechanics. Case of Highly Deformable Products. Drying Technology, 19(8): 1629-1643. Pomeranz, Y. 1988. Composition and Functionality of Wheat Flour Components. In: Pomeranz, Y . (Ed.) Wheat Chemistry and Technology, Vol. 2. American Association of Cereal Chemists. St. Paul, M N . pp. 219-370. Pons, M . and Fiszman, S.M. 1996. Instrumental Texture Profile Analysis with particular Reference to Gelled Systems. Journal of Texture Studies. 27(6):597-624. Prothon, F., Ahrne, L . M . , Funebo, T., Kidman, S., Langton, M . and Sjoholm, I. 2001. Effects of Combined Osmotic and Microwave Dehydration of Apple on Texture, Microstructure and Rehydration Characteristics. Lebensmittel-Wissenschaft und Technologic 34(2):95-101. Puri, V. M . and Anantheswaran, R. C. 1993. Finite Element Method in Food Processing -a Review. Journal of Food Engineering. 19:247. 155 Rao, V.K., Mulvaney, S. J. and Dexter, J.E. 2000. Rheological Characterisation of Long- and Short-Mixing Flours Based on Stress-Relaxation. Journal of Cereal Science. 31:159-171. Rasper, V.F. and Danihelkova, H. 1986. Alveography in Fundamental Dough Rheology. In: Faridi, H. and Faubion, J.M. (Eds.) Fundamentals of Dough Rheology. American Association of Cereal Chemists. St. Paul, M N . pp. 169-180. Ratanadecho, P., Aoki, K. and Akahori, M . 2001. A Numerical and Experimental Study of Microwave Drying using a Rectangular Wave Guide. Drying Technology. 19(9):2209-2234. Ressing, H., Ressing, M . and Durance, T.D. 2005. Modeling the Mechanisms of Dough Puffing during Vacuum Microwave Drying using the Finite Element Method. Submitted to Journal of Food Engineering. Roberts, S.A. and Cameron, R.E. 2002. The Effect of Concentration and Sodium Hydroxide on the Rheological Properties of Potato Starch Gelatinisation. Carbohydrate Polymers. 50:133-143. Ryynanen, S. 1995. The Electromagnetic Properties of Food Materials: A Review of the Basic Principles. Journal of Food Engineering. 26:409-429. Ryynanen, S. 2002. Microwave Heating Uniformity of Multicomponent Prepared Foods. EKT series 1260. Dissertation. Sablani, S.S. and Rahman, M.S. 2002. Pore Formation in Selected Foods as a Function of Shelf Temperature during Freeze Drying. Drying Technology. 20(7): 1379-1391. Sakai, N . , Mao, W. Koshima, Y. and Watanabe, M . 2005. A Method for Developing Model Food System in Microwave Heating Studies. Journal of Food Engineering. 66(4):525-531. Seow, C.C. and Teo, C.H. 1996. Staling of Starch Based Products: a Comparative Study by Firmness and Pulsed N M R Measurements. Starch/Staerke. 48(3):90-93. Seras, M . , Courtois, B., Quinquenet, S. and Ollivon, M . 1987. Measurement of the Complex Permittivity of Dielectrics during Microwave Heating: Study of Flours and Starches. In: Jowitt, R., Escher, F., Kent, M . , McKenna, B. and Roques, M . (Eds.) Physical Properties of Foods, 2. COST 90bis Final Seminar Proceedings, Elsevier Applied Science, London, pp. 217-223. Shah, P., Campbell, G.M., McKee, S.L. and Rielly, C D . 1998. Proving of Bread Dough: Modelling the Growth of Individual Bubbles. Food and Bioproducts Processing. 76(C2):73-79. 156 Sham, P.W.Y., Seaman, C.H. and Durance, T.D. 2001. Texture of Vacuum Microwave Dehydrated Apple Chips as Affected by Calcium Pretreatments, Vacuum Level, and Apple Variety. Journal of Food Science. 66(9): 1341-1347. Shimoni, E., Dirks, E .M. and Labuza, T.P. 2002. The Relation Between Final Popped Volume of Popcorn and Thermal -Physical Parameters. Lebensmittel-Wissenschaft und Technologie. 35:93-98. Sliwinski, E.L., van der Hoef, F., Kolster, P. and van Vliet, T. 2004a. Large-Deformation Properties of Wheat Dough in Uni- and Biaxial Extension. Part II. Gluten dough. Rheologica Acta. 43(4):321-332. Sliwinski, E.L., Kolster, P. and van Vliet, T. 2004b. On the Relationship between Large-Deformation Properties of Wheat Flour Dough and Baking Quality. Journal of Cereal Science. 39(2):231-245. Sliwinski, E.L., Kolster, P. and van Vliet, T. 2004c. Large Deformation Properties of Wheat Dough in Uni and Biaxial Extension. Part I. Flour Dough. Rheological Acta. 43(4):306-320. Sliwinski, E.L., Kolster, P., Prins, A. and van Vliet, T. 2004d. On the Relationship between Gluten Protein Composition of Wheat Flours and Large-Deformation Properties of Their Doughs. Journal of Cereal Science. 39(2): 247-264. Svegmark, K. and Hermansson, A . M . 1993. Microstructure and Rheological Properties of Composites of Potato Starch Granules and Amylose: A Comparison of Observed and Predicted structure, Food Structure. 12 :81—193. Svensson, E., Autio, K. and Eliasson, A.-C. 1998. The Effect of Sodium Dodecylsulfate on Gelatinization and Gelation Properties of Wheat and Potato Starches. Food Hydrocolloids. 12:151-158. Tabeidie, Z., Payne, F.A. and Cornelius, P.L. 1992. Puffing Potato pieces with CO2. Transactions of the ASAE 35:1935-1940. Tabilo-Munizaga, G. and Barbosa-Canova, G.V. 2005. Rheology for the Food Industry. Journal of Food Engineering. 67(1-2): 147-156. Tester, R.F, Karkalas, J. and Qi, X . 2004. Starch - Composition, Fine Structure and Architecture. Journal of Cereal Science. 39(2):151-165. Tronsmo, K . M . , Faergestad, E.M. , Magnus, E.M. and Schofield, J.D. 2000. Gluten Quality vs. Quantity: Rheology as the Arbiter. Wheat Gluten: Proceedings of the 7th International Workshop Gluten. The Royal Society of Chemistry, Cambridge, UK. pp. 387-390. 157 Tschoegl, N.W., Rinde, J.A. and Smith, T.L. 1970. Rheological Properties of Wheat Flour Doughs. Rheologica Acta. 9(2):223-238. Uthayakumaran, S., Newberry, M . , Phan-Tien, N . and Tanner, R. 2002. Small and Large Strain Rheology of Wheat Gluten. Rheologica Acta. 41:162-172. van der Veen, M.E., van der Goot, A.J., Vriezinga, C.A., de Meester, J.W.G. and Boom, R.M. 2004. On the Potential of Uneven Heating in Heterogeneous Food Media with Dielectric Heating. Journal of Food Engineering. 63(4):403-412. van Vliet, T., Janssen, A . M . , Bloksma, A .H. and Walstra, P. 1992. Strain Hardening of Dough as a Requirement for Gas Retention. Journal of Texture Studies. 23(4):439-460. Varnalis, A.J., Brennan, J.G. and MacDougall, D.B. 2001a. A Proposed Mechanism for High Temperature Puffing of Potato. Part I. The Influence of Blanching and Drying on the Volume of the Puffed Cubes. Journal of Food Engineering. 48:361-367. Varnalis, A.J., Brennan, J.G. and MacDougall, D.B. 2001b. A Proposed Mechanism for High Temperature Puffing of Potato. Part II. Influence of Blanching and Initial Drying on the Permeability of the Partially Dried Layer to Water Vapour. Journal of Food Engineering. 48:369-378. Vega-Mercado, H., Gongora-Nieto, M . M . and Barbosa-Canovas, G.V. 2001. Advances in Dehydration of Foods. Journal of Food Engineering. 49(4):271-289. Venkatesh, M.S. and Raghavan, G.S.V. 2004. An Overview of Microwave Processing and Dielectric Properties of Agri-food Materials. Biosystems Engineering. 88(1):1-18. Visser, R.G.F., Suur, L.C.J.M., Steeneken, P.A.M. and Jacobsen, E. 1997. Comparison between Amylose-Free and Amylose Containing Potato Starches. Starch/Staerke. 49(ll):438-443. Waigh, T.A., Gidley, M.J., Komanshek, B.U. and Donald, A . M . 2000. The Phase Transformations in Starch during Gelatinisation: a Liquid Crystalline Approach. Carbohydrate Research. 328(2): 165-176. Wang, C.F. and Kokini, J.L. 1995. Prediction of Nonlinear Viscoelastic Properties of Gluten Doughs. Journal of Food Engineering. 25:297-309. Wikstrom, K. and Bohlin, L. 1999. Extensional Flow Studies of Wheat Flour Dough. II. Experimental Method for Measurements in Constant Extension Rate Squeezing Flow and Application to Flours Varying in Breadmaking Performance. Journal of Cereal Science. 29:227-234. 158 Wu, Y. and Irudayaraj, J. 1996. Analysis of Heat, Mass and Pressure Transfer in Starch Based Food Systems. Journal of Food Engineering. 29:399-414. Wu, P.J. and Schwartzberg, H.G. 1994. Determination of Vapor Pressure in Vapor-Induced Puffing. AIChE Journal 40(\):160-165. Xu, F. 2001. Adsorption of Oxygen Gas by Hydrated Wheat Flour. Lebensmittel-Wissenschaft und Technologie. 34(2):66-70. Yaghmaee, P. and Durance, T.D. 2005. Destruction and Injury of Escherichia coli during Microwave Heating under Vacuum. Journal of Applied Microbiology. 98(2):498-506. Yamsaengsung, R. and Moreira, R.G. 2002. Modeling the Transport Phenomena and Structural Changes during Deep Fat Frying Part I: Model Development. Journal of Food Engineering. 53:1-10. Yang, H., Sakai, N . and Watanabe, M . 2001. Drying Model with Non-Isotropic Shrinkage Deformation undergoing Simultaneous Heat and Mass Transfer. Drying Technology. 19(7):1441-1460. Yousif, A.N. , Seaman, C.H., Durance, T.D. and Girard, B. 1999. Flavor Volatiles and Physical Properties of Vacuum-Microwave- and Air-Dried Sweet Basil (Ocimum basilicum L.) Journal of Agricultural and Food Chemistry. 41:A111-A1%\. Yousif, A.N. , Durance, T.D., Seaman, C.H. and Girard, B. 2000. Headspace Volatiles and Physical Characteristics of Vacuum-Microwave, Air, and Freeze-Dried Oregano (Lippia berlandieri Schauer). Journal of Food Science. 65(6):926-930. Zheng, H., Morgenstern, M.P., Campanella, O.H. and Larsen, N.G. 2000. Rheological Properties of Dough During Mechanical Dough Development. Journal of Cereal Science. 32:293-306. Zhou, L., Puri, V. M . , Anantheswaran, R.C. and Yeh, G. 1995. Finite Element Modeling of Heat and Mass Transfer in Food Materials during Microwave Heating — Model Development and Validation. Journal of Food Engineering. 25(4):509-529. Zuercher, J., Hoppie, L., Lade, R., Srinivasan, S. and Misra, D. 1990. Measurement of the Complex Permittivity of Bread Dough by an Open-Ended Coaxial Line Method at Ultra-High Frequencies. Journal of Microwave Power Electromagnetic Energy. 25(3):161-167. 159 APPENDIX Modeling the Mechanisms of Dough Puffing during Vacuum Microwave Drying using the Finite Element Method H. Ressing, M . Ressing and T. Durance As submitted to the Journal of Food Engineering, Elsevier, The Netherlands, 2005. Abstract A two-dimensional finite element (2-D FE) model was developed in order to simulate the puffing of a dough ball being dehydrated under vacuum with microwave energy. The model enabled the coupling of thermal and solid mechanics effects providing an insight into the mechanisms of puffing of food stuffs during vacuum microwave dehydration. The model predicted the rise in temperature reasonably well for low salt contents. The model suggests puffing is due to two mechanisms: first, the difference in pressure between air trapped in the dough and the chamber pressure of the vacuum microwave dehydrator (VMD) ; second, the generation of vapour due to the temperature rise in the dough. The temperature distribution is primarily determined by the penetration depth of the microwaves into the material. It was further shown that the partial collapse of dough balls in the later stages of the drying process can be attributed to the softening of dough at higher temperatures and air loss throughout the drying process. 160 Keywords: vacuum microwave drying, puffing, Finite Element Method, coupled modeling, dough modeling 1 Introduction Vacuum microwave drying (VMD) enables rapid, efficient and economical removal of water at lower temperatures while only minimally altering the composition. This particular effect has placed the value of V M D between air drying and freeze drying, the most gentle way of drying. Though the efficiency and practicality of V M D has been studied, the mechanisms by which textural changes occur are poorly understood. It has been noted that during the V M D process, items may inflate. However, it is not clear why certain food stuffs such as potato cubes retain the puffed appearance while others such as fruits collapse before dehydration is complete. It is difficult to determine the mechanisms of food puffing under V M D based on experimental studies alone. Temperature, pressure and moisture measurements can only provide a global measure of the puffing process but cannot give the local insight necessary to understand the mechanism of puffing. In addition, the installation of measurement equipment in a vacuum microwave environment is quite difficult, especially when the products are agitated duringdrying. On the other hand, physics-based 161 modeling provides a tool to easily study the influence of various parameters and mechanisms of the puffing process. Numerical modeling of food processing has been done by various researchers using the Finite Difference or Finite Element (FE) Method. FE modeling of fluid flow was used by Connelly & Kokini (2004) and Feigl, Kaufmann, Fischer & Windhab (2003). Martins & Silva (2004b) modeled the quality loss upon thawing of green beans. Models of air drying have focused on solving equations for mass and heat transfer problems. Wu & Irudayarj (1996) analyzed the heat, mass and pressure transfer in starch based food systems using a 2-D FE model, while Yang, Sakai & Watanabe (2001) simulated shrinkage deformation of potato spheres. Heat generation due to microwave power absorption has been simulated by Ratanadecho, Aoki & Akahori (2001), Oliveira & Franca (2000) and Lin, Anantheswaran & Puri (1995). A 3-D FE model to predict temperature and moisture distribution in a cylindrically and slab shaped potato specimen was developed by Zhou, Puri & Anantheswaran (1995). Pandit & Prasad (2003) computed the transient temperature profile of potato under microwave heating. A comprehensive review of the usage of FE modeling of food processing has been published by Puri & Anantheswaran (1993). While many studies have focused on temperature distribution and moisture content during the drying process, only few studies have combined the thermal analysis of the drying process with that of structural deformation. Perre & May (2001) modeled the shrinkage of potato due to water removal during drying. Thermal stresses in corn kernels during drying were analyzed by Jia, Sun & Cao (2000). No modeling of the puffing process during V M D has been published to 162 date. This research aims to combine the equations of heat transfer and solid mechanics using the FE Method to study the mechanisms of puffing during V M D on a local level. In our case, dough balls were assumed to be tumbled within the chamber in such a way as to average hot and cold spots within the microwave field of the V M D . 2 Materials and Methods A 2D FE model was developed to simulate the puffing of dough balls under V M D application. The modeling procedure solved the coupled heat transfer and solid mechanics equations in order to determine the puffing due to the externally applied vacuum and internal pressure build-up from water vapour. Dough balls were modeled as a 2-D circular geometry consisting of approximately 300 quadrilateral elements. For symmetry reasons, only a quarter of the dough ball was actually modeled. The FE mesh is depicted in Figure 3.1. The details of the FE modeling process are described in the following sections. 2.1 Thermal modeling Thermal modeling of microwave drying was based on Fourier's equations of heat conduction (Bathe, 1996) _ d 86. 8 86. 8 36. ... Pc0- — (k—)---(k—)- — (k—) = qh (1) dx dx 8y 8y 8z az 163 where 6=6 (x,y,z) denotes the temperature, pthe density, c the specific heat capacity and k the thermal conductivity with respect to the principle axis x, y, and z. qb represents the rate of heat generated per unit volume. Equation (1) must further satisfy the boundary conditions dL = 0s (2) = qs (3) dn where 6s is the known surface temperature on the surface Sm k„ is the body thermal conductivity, n denotes the direction of the unit normal vector to the surface Sq and qs is the prescribed heat flux input on the surface Sq of the body. In the 2D case, i.e., with no heat flow in the z-direction, equation (1) simplifies to • d 36. d .. dO. ( A , pc 6- — (kx —)-—{k —) - qb • (4) dx dx dy dy In order to solve this partial differential equation the principle of virtual temperatures is employed in which the equation (1) is rewritten in an integral sense (Bathe, 1996) /86 (pc) 6dV + f 86,T k &dV = f 66 qbdV + /86 qs dS + ^ 86, Q{ (5) V V V S i 164 where 8' denotes the gradient of 6 and Q, are concentrated heat flow inputs. 86 denotes the virtual temperature which can be arbitrary, but must be zero on S0. When employing the FE method, discretization of the domain into elements of finite size leads to c t h e + Kthe~Q or 2c«H^ + 2K^ = E e e <6> e e e where 8 denotes the vector of nodal temperatures, C,h and C t h e are the global and element thermal capacitance matrices Ceth= fHT(pc)KdV, (7) v K t h and K t h e are the global and element conductivity matrices K ; h = J*B Tk B dV, (8) v and Q and Qe are the global and element nodal point heat flow input vectors Qe = fHTqbdV+ fHsTqsdS (9) v S' 165 Here, H and B are the common linear shape functions and their derivatives for isoparametric elements defined over the element volume V, respectively. The rate of heat generated per unit volume equals the absorbed power from the microwave, thus qb = PV =2nfe0e"E2 (10) where/is the microwave frequency, E is the electric field inside the load, e0 is the permitivity of free space and E" is the dielectric loss factor (Buffler, 1992). In general, the electric field inside the microwave cavity is not constant, but depends on the shape and size of the cavity and subsequently the dominating modes of the microwaves. This results in hot and cold spots inside the cavity. Based on the assumption that the dough balls are continuously moved around in the microwave cavity, it can also be assumed that their exposure to the electric field is a constant at all times and equal to the average electric field strength of the microwave oven. This field strength was determined in experiments with a known load of 2 L of water being exposed to microwave irradiation for 1 minute and by measuring the temperature increase of the water. The total absorbed power P was then determined as P = (11) 166 where mw denotes the water mass, cw the specific heat capacity of water, AT the measured temperature difference and At the duration of microwave exposure. Hence, the average electric field strength <X>0 can be calculated from ^ P dough (12) 2jtfe0e"mdough The electric field strength acting inside a solid is further diminished with increased penetration into the material. A measure for this is the penetration depth dp which is defined as the depth at which the microwave power has decreased to Me or 36.8 % of its original power (Buffler, 1992). Hence, the electric field strength acting on each element in the model is computed as where r is the distance of the element center to the outer edge of the model geometry. Combining equations (10), (12) and (13) yields O 2 = O 0 2 e (13) qh = P (14) 167 Finding the transient solution was achieved using the Euler backward implicit time integration method which ensures solution stability for an arbitrary time step size. In this method 6 is approximated by 6 = d,+Al Q' (15) t\t where 1st represents the time step size throughout the integration. Equation (6) is then written at time t+At % + K »K*-^ + ^ <I6) and can be solved, if the solution at time t is known. The initial conditions at time t - 0 are set to 6=TR (17) for all nodes, where TR denotes the reference temperature, here 25°C. 2.2 Solid mechanics modeling The principle of virtual displacement for modeling solids at static equilibrium (Bathe, 1996) is well known and generally written as 168 f6eT D £ dV = / 8uTfB dV + f 8uT fs dS V V s (18) where, in the 2-D case of plane strain, e denotes the strain vector [e„, e>y and y^,] T , / B represents the body force applied to the body,/ 5 the traction on the surface, 8e and 5u the virtual strain and displacements, respectively, and D is the material matrix which can be written as D = l-fl fi 0 0 0 0 1-2/t 2(1 + ^ )1 (19) with E and [i denoting Young's modulus and Poisson's ratio, respectively. By discretizing the domain the FE equations are obtained and written as KstU-F or 2K : t c/=5> (20) e e where U denotes the vector of nodal displacements, K,., and K,.,6 are the global and element stiffness matrices = / B T D B a V , (21) 169 and F and F° are the global and element nodal point forces vectors F< = fHrfbdV + JnsTfsds. (22) V S* Here, H and B are the common linear shape functions and strain-displacement matrices for isoparametric elements, respectively. It must be noted that the rate of displacement during puffing is considered small, thus the equations can be solved quasi-statically. For a solid under vacuum fB can be ignored and/ 5 is equal to the internal over pressure of trapped air due to vacuum with an estimated air content of 10% (Bloksma & Bushuk, 1988; Bloksma, 1990) yielding 2.3 Distinction between "wet" and "dry" dough elements A unique feature of the developed FE model was the implementation of "wet" and "dry" dough elements. Differences between wet and dry dough elements lie primarily in their thermal behavior, while their structural properties are identical. During the thermal analysis process, wet dough elements represent the water content in the dough mass. Wet dough elements are randomly distributed according to the mass ratio of water in the dough. This ensures that the FE model contains the right amount of water and flour mass that will be heated throughout the analysis. At the same time, the concentration of /,=0.1-/7V. (23) 170 thermal properties of water in only certain elements and the random distribution of these elements allow the simulation of uneven heating of the entire dough mass due to uneven distribution of water in the dough. This is possible, because the thermal properties, such as thermal conductivity and dielectric loss factor, of water and flour are well distinguished. In addition, the dielectric loss factor of wet dough elements depends on the dough's salt content. The salt content was varied between 0 and 1.5 % with low salt content exhibiting lower loss factors than higher salt contents. On the other hand, structurally no distinction between wet and dry dough elements is feasible. The structural stiffness of dough is simply not the result of a random distribution of the material properties of flour and water, but is determined from global rheological experiments on the flour water mix. Hence, the structural material properties of dough were applied equally to wet and dry dough elements. In order to simulate puffing of the dough balls due to the generation of water vapour inside the dough mass, the state of water contained in wet dough elements was checked at every time step of the analysis. For this, the pressure inside the wet dough elements was determined and compared to the vapour pressure of water at the element temperature. If the element pressure was below the vapour pressure, part of the water mass was converted into steam filling the available element volume completely. At the same time, energy in the amount of latent heat for the converted water mass was removed from the system. Once a wet dough element was determined to contain water vapour, the vapour pressure was applied as internal pressure to the element edges during the structural analysis process to simulate the expansion of the dough ball due to vapour generation. 171 Drying was simulated by water mass removal from wet dough elements containing water vapour. The water mass removal rate was determined in experiments and, for simplicity, is applied as a function of time ignoring the effects of temperature. 2.4 Analysis procedure Due to the unique requirements of the wet dough elements, behaving thermally as water while structurally as dough, the water mass removal due to drying, removal of latent heat, and the application of internal pressure due to steam generation it was not easily possible to employ a standard, commercially available FE code to solve this simulation problem. Therefore, an existing 2-D thermal and solid mechanics FE code written in Matlab (The Mathworks, 2004) was used and adapted to these specific requirements. The program allowed the analysis of the coupled thermal and solid mechanics simulation by sequentially solving each problem for a small time step before using the results to update the input data set for the subsequent analysis. A schematic of the modeling procedure is shown in Figure 3.2. The initialization step created the element geometry and assigned material properties to all elements. The first load step applied the internal over pressure of trapped air relative to the chamber vacuum. The structural solution, i.e., deformation and resulting increase in volume, due to the applied vacuum was computed. 172 Step two applieds the microwave field for the specified time step size and computed the heat generation in each element with respect to its specified loss factor. Temperature distribution was then computed based on specified thermal properties of heat conduction and heat capacity. The average temperature within each element was evaluated and yielded the respective vapour pressure for a wet dough element. At the same time, the current element volume was computed and yielded the ratio of water liquid to vapour. The required latent heat was removed from the system. Water vapour was removed according to the specified drying rate, considering the loss of mass and energy. Step three re-assigned the structural material properties for each element based on the current thermal and structural solution, i.e., the temperature and strain distributions within the FE model. For wet dough elements the computed internal pressure was applied as internal forces on the nodes of the element, simulating the vapour pressure of the water vapour. The new deformation of the dough ball was solved for and the resulting stresses and strains for each element were computed. After this, steps two and three are repeated until the maximum simulation time was reached. 2.5 Model parameters and material properties The applied power level of the microwave field was 1300 W, the applied vacuum pressure was set to 35 mbar in absolute pressure or 978 mbar below atmospheric pressure. Modeling was only done on high gluten flour dough with the range of 0.0 to 1.5% salt. The dough was assumed to be properly mixed, consisting only of water, flour and salt and an air content of 10% by volume. Material properties used for wet and dry 173 dough elements are listed in Table 3.1. Values were either taken from the literature or determined experimentally. In particular the structural behavior of wet and dry dough elements were deduced from measured stress-strain curves for various salt contents (Ressing, 2005). Figure 2.5 gives an example of the measured non-linear stress-strain curve for dough with 0% salt content. From this, Young's modulus was determined by fitting the linear portion of the curve, between 1/6 and 5/6 of the rupture stress, to a straight line and finding its slope. However, for the implementation in the FE model, the full non-linear curve was considered in order to describe the solid mechanics behavior of wet and dry dough elements. The listed values in Table 1 served only as comparison for the increased overall stiffness of the dough with the increase in salt content. The simulated time frame was set to 200 s, including an initial duration of 20 s to apply the vacuum and during which no microwave exposure was taking place.. This was based on the experimental observation that expansion during the drying process was terminated before or at 3 min (Ressing, 2005). The time step size during steps 2 and 3 of the analysis procedure was set to 1 s. The drying rate was set to 30% of the total water content over the entire drying time of 180 s, a value consistent with experimental results . 3 Results and discussion Figure 3.3 depicts a simple reflection of the dough ball deformation, showing the initial and final volume superimposed. Figures 3.4 and 3.5 show the deformation, temperature and stress distribution of a dough ball with 0.0% salt concentration for times tdrying =1,10, 174 90 and 180 s of microwave exposure, respectively. In the early stages, the uneven heating due to the non-uniform water distribution can be observed. As the heat penetrates through the dough ball at later points in time, the temperature distribution is primarily dependent on the penetration depth and the resulting lower heat generation in the centre of the dough ball. The resulting stress distribution shows the increased tensile stresses from higher vapour pressure in regions of higher temperature. Increases in temperature as a function of time for various salt concentrations are shown in Figure 3.6. For 0.0 % salt concentration a temperature of 55°C was reached. This compared reasonably to experimentally observed values of 45°C (Ressing, 2005). For greater salt concentrations the temperature rise was unrealistically high and greatly exceeded experimental results. However, the higher temperatures are a direct result of the greater dielectric loss factor as input parameter to the FE model. Hence, there appears to be a mechanism for loss of energy which the FE model is not accounting for. It is possible that, while the overall drying rate for higher salt concentrations is the same as for no salt, the drying rate is initially greater, thus more water is removed in the initial stages of the drying process, causing less heat generation at later points in time. It can also not be excluded that the microwave efficiency, the rate of absorbed energy by the dough compared to the absorbed energy by the standard 2 L of water, is significantly lower than the assumed 100%, in particular as more water is removed. When observing the increase in volume (Figure 3.7) for various salt concentrations, it can be seen that the increase in volume due to vacuum application of 60% for 0.0% salt 175 concentration makes up the major portion of the overall increase in volume. The increase due to vapour generation, i.e., internally applied vapour pressure, contributes only an additional increase of 15% of the total volume increase. For higher salt concentrations the calculated volume increase initially is lower than that of lower salt doughs. This is due to the greater stiffness which was observed in dough formulations containing higher levels of salt. However, due to the increased heat generation caused by the significantly higher dielectric loss factor for higher salt concentrations the temperature increases over time to significantly higher levels compared to low salt concentrations. Hence, after 40-45 s of microwave exposure, the FE results yield a reversal of this trend where high salt formulations dominate the additional volume increase. This unreasonably great increase in volume is a direct result of the high temperature increase for high salt concentrations discussed above. If the increased heat generation due to higher salt concentrations is ignored, it becomes clear that at similar temperature increases dough formulations with lower salt concentrations exhibit a greater increase in volume due to their lower stiffness (Figure 3.8). An overall increase in volume of more than 60% for high salt concentrations and 88% for 0% salt can be observed. However, the volume increase due to vapour pressure is less than 40% of the overall volume increase. Results of volume increase due to a temperature dependent dough stiffness and a removal of air are shown in Figure 3.9. The reduction of dough stiffness with an increase in temperature has been noted in the literature by Rasper & Danihelkova (1986) and Fan, 176 Mitchell & Blanshard (1999). Here, the dough stiffness was reduced by 2%/°C temperature increase in each element yielding a 50% reduction in Young's modulus for an average dough temperature of 45°C. At the same time, the loss of air, and thus the loss of internal pressure, was assumed to be 10% of the original trapped air mass over the 180 s of microwave exposure. As a result, the volume rises more rapidly initially, due to the reduced dough stiffness corresponding to the rise in temperature. However, as more and more air mass is lost, the internal pressure decreases and the volume increase is reduced towards the end of the drying time. A similar behaviour has been observed in experimental studies (Ressing, 2005). 4 Conclusions A 2-D FE model was developed in order to simulate the puffing of a dough ball applied to VMD. The model enabled the coupling of thermal effects, e.g., uneven heat generation due to microwave exposure, heat conduction and drying, with solid mechanics effects, e.g., external vacuum application, internal pressure due to vapour generation and the non-linear stiffness of dough. The combination of these effects allowed insights into the mechanisms of puffing of food stuffs during V M D . The model predicted the rise in temperature reasonably well for low salt contents. For higher salt contents an unrealistic rise in temperature due to the higher dielectric loss factor was observed. Thus, there appears to be a mechanism for loosing energy, which has not been accounted for in the model and is not yet explained unambiguously. 177 Howeveiyit can be shown that puffing is due to two mechanisms: first, the application of external vacuum and thus an internal over pressure of the trapped air; second, the generation of vapour due to the temperature rise in the dough. Here it wasshown, that the ratio of volume increase due to externally applied vacuum compared to volume increase due to vapour generation is approximately 3:2. Temperature gradients at the local level due to uneven heating if a non-uniformly distributed water mass played only a minor role. The temperature distribution is primarily determined by the penetration depth of the material. It was further shown that the partial collapse of dough balls at later stages in the drying process can be attributed to the softening of dough at higher temperatures and air loss throughout the drying process. 5 References Anonymous. 1982. Handbook of Chemistry and Physics 63 r d Edition, CRC Press, Boca Raton, USA. Bathe, K.-J. (1996). Finite Element Procedures, Prentice-Hall, Upper Saddle River, USA. Bloksma, A. H. & Bushuk, W. (1988). Rheology and Chemistry of Dough. In: Pomeranz, Y. (Ed.) Wheat Chemistry and Technology, Vol. 2. American Association of Cereal Chemists. St. Paul, USA. 131-217. Bloksma, A. H. (1990). Dough Structure, Dough Rheology, and Baking Quality. Cereal Foods World. 35(2), 237-244. Buffler, C. R. (1992). Microwave cooking and processing, AVI , New York, USA. 178 Connelly, R. K. & Kokini, J. L. (2004). The effect of shear thinning and differential viscoelasticity on mixing in a model 2D mixer as determined using FEM with particle tracking. Journal of Non-Newtonian Fluid Mechanics, 123, 1-17. Fan, J., Mitchell, J. R. & Blanshard, J. M . V. (1999). A Model for the Oven Rise of Dough during Baking. Journal of Food Engineering, 41, 69-77. Feigl, K., Kaufmann, S. F. M . , Fischer, P. & Windhab, E. J. (2003). A numerical procedure for calculating droplet deformation in dispersing flows and experimental verification. Chemical Engineering Science, 58, 2351-2363. Jia, C . -C, Sun, D.-W. & Cao, C.-W. (2000). Mathematical simulation of stresses within a corn kernel during drying. Drying Technology, 18 (4&5), 887-906. Lin, Y. E., Anantheswaran, R. C. & Puri, V. M . (1995). Finite Element Analysis of Microwave Heating of Solid Foods. Journal of Food Engineering, 25, 85-112. Martins R. C. & Silva C. L. M . (2004). Green beans (Phaseolus vulgaris, L.) quality loss upon thawing, Journal of Food Engineering, 65, 37-48. The Mathworks (2004). Matlab Release 14. Natick, USA. Oliveira, M . E. C. & Franca, A. S. (2000). Finite element analysis of microwave heating of solid products. Int. Comm. Heat Mass Transfer, 27 (4), 527-536. Pandit, R. B. & Prasad, S. (2003). Finite element analysis of microwave heating of potato - transient temperature profiles. Journal of Food Engineering, 60, 193-202. Perre, P. & May, B. K. (2001). A numerical drying model that accounts for the coupling between transfers and solid mechanics. Case of highly deformable products. Drying Technology, 19 (8), 1629-1643. 179 Puri, V. M . & Anantheswaran, R. C. (1993). Finite element method in food processing -a review. Journal of Food Engineering, 19, 247. Rahman, S. 1995. "Food Properties Handbook" p. 242. CRC Press, Boca Raton, USA. Rasper, V. F. & Danihlkova, H. (1986). Alveography in Fundamental Dough Rheology. In: Faridi, H. & Faubion, J. M . (Eds.) Fundamentals of Dough Rheology. American Association of Cereal Chemists. St. Paul, USA, 169-180. Ratanadecho, P., Aoki, K. & Akahori, M . (2001). A numerical and experimental study of microwave drying using a rectangular wave guide. Drying Technology, 19 (9), 2209-2234. Ressing, M . (2005). Puffing Induced in two Model Systems by Microwave Assisted Drying under Vacuum - an Experimental and Numerical Analysis. Ph.D. Thesis, The University of British Columbia, Vancouver, Canada. Wu, Y. & Irudayaraj, J. (1996). Analysis of Heat, Mass and Pressure Transfer in Starch Based Food Systems. Journal of Food Engineering, 29, 399-414. Yang, H., Sakai, N . & Watanabe, M . (2001). Drying model with non-isotropic shrinkage deformation undergoing simultaneous heat and mass transfer. Drying Technology, 19 (7), 1441-1460. Zhou, L., Puri, V. M . , Anantheswaran, R. C. & Yeh, G. (1995). Finite Element Modeling of Heat and Mass Transfer in Food Materials During Microwave Heating - Model Development and Validation. Journal of Food Engineering, 25, 509-529. 180 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0092881/manifest

Comment

Related Items