C O M P A R I S O N OF CONSTANT RETORT T E M P E R A T U R E AND VARIABLE RETORT T E M P E R A T U R E THERMAL P R O C E S S E S FOR QUALITY IMPROVEMENT OR C O S T REDUCTION OF CONDUCTION-HEATED C A N N E D FOODS By BOB Y O N G S H E N G XIANG B. Sc. in Ag., Southwest Agricultural University, P. R. China, 1987 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE D E G R E E OF MASTER OF SCIENCE In THE FACULTY OF G R A D U A T E STUDIES (Food Science Program) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA N O V E M B E R 2003 © Bob Yongsheng Xiang, 2003 In presenting this thesis in partial fulfillment of the requirement for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or his representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Food, Nutrition and Health Program The University Of British Columbia 6650 NW Marine Drive Vancouver, B C V6T 1Z4 Date -ii-ABSTRACT Almost all commercial retort processes for canned foods use constant retort temperature (CRT) process. However, variable retort temperature (VRT) process, as one of the potential technologies to improve both the economy and quality of some canned foods, has been receiving increasing attention. The VRT process has been shown to be very promising in this regard, especially in improving food quality and reducing process time. The surface color is an important quality attribute of canned foods. Discoloration and browning of canned foods are the results of various reactions, including Maillard reaction. Heat treatment affects the surface color of canned foods. Surface color changes measured by HunterLab are used to predict both chemical and quality changes in canned foods. In this study I examined the surface color change characteristics of macaroni and cheese (MC). Surface color change of M C followed first order reactions and D values of the surface color change and z value of the surface color change were measured. This study evaluated the application of the "Retort" program and the random centroid optimization (RCO) program for modeling and optimization of VRT thermal processing for conduction-heated foods. This study tested whether canned macaroni and cheese (MC) surface quality would be improved or process times decreased by using the optimal V R T process as compared with the optimal CRT process. From this study, I concluded that the optimal VRT process was superior. It improved the surface quality (i.e., reduced the surface cook value by 8.9-11.2 %) or reduced the process time by 23.6-34.2 % compared with the optimal C R T process. -III-T A B L E O F C O N T E N T S Page Abstract ii Table of Contents iii List of Tables vi List of Figures ix Acknowledgements xi C H A P T E R I 1 C H A P T E R II L ITERATURE REVIEW 4 2. 1. Color Measurement 4 2 .2 . Color Change With Heat Treatment 5 A. Maillard Reaction 5 B. Color Change in Canned Foods 8 C. Thermal Kinetics of Color Change of Foods 9 2. 3. Thermal Processing of Canned Foods 10 A. Goals of Thermal Processing for Canned Foods 10 B. Processing Media of Canned Foods 11 C. Optimization Sterilization of Canned Foods 12 D. Temperature Measurement and Heat Penetration Tests 12 E. Process Determination 13 F. The Improved General Method and Sterilization Value (F0) 14 2. 4. C R T Process and VRT Process 17 A. Definition of the C R T and VRT processes 17 B. Retort Program 19 C. R C O Program 20 D. Computer Simulation of the C R T and VRT Processes 27 E. Estimation of Rho (Fraction of Sterilization Value) 27 2. 5 Quality of the Thermally Processed Canned Foods 28 A. Basic Consideration for Canned Foods 28 -iv-B. Effect of Canned Size 29 C. Effect of Processing Temperature 30 D. Surface Quality for Canned Foods 31 E. Goals of This Research Project 31 C H A P T E R III EXPERIMENTAL METHODS 33 3. 1. Sample Preparation 33 3. 2. Surface Color Change of MC with Heat Treatment 34 3 .3. Surface Color Measurement of MC 35 3. 4. Heat Penetration Test 35 3. 5. Determination of Sterilization Value (F0) 36 3. 6. Retort Program 36 3. 7. R C O Program 38 3. 8. Confirmation of the Results for the C R T and VRT Processes in an Actual Steam Retort 40 C H A P T E R IV R E S U L T S AND DISSCUSSION 42 4. 1. Surface Color Changes of MC 42 4 .2 . D Values and z Value of MC 50 4. 3. Heat Penetration Parameters 55 4. 4. Comparison of Can Center Temperatures by Retort Program and Retort Experiment 57 4. 5. Rho, Retort Temperature and Unaccomplished Temperature 59 4. 6. Surface Cook Values of the C R T and VRT Processes 62 4. 7. Process Times of the C R T and VRT Processes 81 4. 8. Compare the Results of the C R T and VRT Processes for MC 97 4. 9. Confirmation of the Optimum C R T and VRT Processes in an Actual Steam Retort 103 A. Confirmation of Sterilization Value (F0) 103 B. Confirmation of Surface Cook Values and Surface Color Parameters 105 -V-C. Confirmation of the Surface Cook Values of MC 107 C H A P T E R V CONCLUSIONS 109 APPENDIX A. Terminology and Abbreviations 111 APPENDIX B. Processing Conditions for Computer Simulation Model 114 R E F E R E N C E S 116 BIBLIOGRAPHY 124 -vi-LIST OF TABLES Page Table 1. Processing conditions of retort experiment for MC 307 x 409 cans 41 Table 2. D values at different heat temperatures (°C) 51 Table 3. The average heating rate index and the average cooling rate index for MC obtained during process determination work work in three process runs (12 cans) 56 Table 4. The C R T processes at different surface z values in term of surface cook value (Fs) with the same F0=6 min 63 Table 5. Optimization experiments to minimize Fs with Pt < 124.8 min and 5.9 < F 0 < 6.1 min (z= 28 C°), the best result was the bold value (F s = 56.2 min) 66 Table 6. Optimization experiments to minimize F s with P t < 148.1 min and 5.9 < F 0 < 6.1 min (z= 24 C°), the best result was the bold value (F s = 50.4 min) 68 Table 7. Optimization experiments to minimize F s with P t < 148.1 min and 5.9 < F 0 < 6.1 min (z= 26 C°), the best result was the bold value (F s = 53.6 min) 70 Table 8. Optimization experiments to minimize F s with P t < 124.8 min and 5.9 < F 0 < 6.1 min (z= 30 C°), the best result was the bold value (F s = 59.6 min) 72 Table 9. Optimization experiments to minimize F s with P t < 124.8 min and 5.9 < F 0 < 6.1 min (z= 32 C°), the best result was the bold value (F s = 61.2 min) 74 Table 10. Comparison of the optimum VRT processes with minimum F s and P t in term of different z values 77 -VII-Table 11. Comparison of F s of MC for the optimum C R T and VRT processes in terms of different z values (F S ) min) 80 Table 12. Optimization experiments for VRT processes to minimize P t with F s < 63.2 min and 5.9 < F 0 < 6.1 min (z=28 C°), the best result was the bold value (P t = 95.3 min) 82 Table 13. Optimization experiments for VRT processes to minimize P t with F s < 56.1 min and 5.9 < F 0 < 6.1 min (z=24 C°), the best result was the bold value (P t = 106.9 min) 84 Table 14. Optimization experiments for VRT processes to minimize P t with F s < 59.8 min and 5.9 < F 0 < 6.1 min (z=26 C°), the best result was the bold value (P t =97.5 min) 86 Table 15. Optimization experiments for VRT processes to minimize P t with F s < 66.1 min and 5.9 < F 0 < 6.1 min (z=30 C°), the best result was the bold value (P t = 88.2 min) 88 Table 16. Optimization experiments for VRT processes to minimize P t with F s < 68.9 min and 5.9 < F 0 < 6.1 min (z=32 C°), the the best result was the bold value (P t = 87.5 min) 90 Table 17. Comparison of the optimum VRT processes with minimum P t and F s in term of different z values 93 Table 18. Comparison of P t for the optimum C R T and VRT processes in terms of different surface z values 96 Table 19. The optimum C R T and VRT processes of M C (z=28 C°) In term of the minimum surface cook value and the minimum process time 102 Table 20. Sterilization values (F0) for MC with three process runs for each treatment and calculations done using improved general method 104 Table 21. The surface color parameters of MC in terms of the different C R T and VRT processes (confirming experimental -VIII-results) 106 Table 22. Comparison of surface cook values (Fs) of MC in terms of computer simulation and retort experiments (three process runs for 8-10 cans, based on the sterilization value F 0 of 6.0 min) 108 -ix-LIST OF FIGURES Page Figure 1. Reference TDT curve, where F=1 and z=18 Fo (Durance, 1995) 15 Figure 2. Comparison of retort temperature histories of Conduction-heated canned foods with the C R T and VRT processes (Durance, 1997) 18 Figure 3. The simplified flow diagram of R C O procedure 24 Figure 4. A comprehensive operation chart of R C O program (Nakai et al., 1999) 26 Figure 5. Simplified flow diagram of the Retort program procedure 37 Figure 6. Effect of heating time and heating temperature on the surface color L values 43 Figure 7. Effect of heating time and heating temperature on the surface color a values 44 Figure 8. Effect of heating time and heating temperature on the surface color b values 45 Figure 9. Surface color parameters (L, a and b) changes with the heating time (hr) at heating temperature 100 °C 48 Figure 10. Surface color difference versus heating time (hr) at heating temperature 100 °C 49 Figure 11. Effect of heating time on the log L value of MC at different heating temperatures (80, 100, 110, 120 and 125 °C) 53 Figure 12. Effect of heat temperature on the Log D values of MC 54 Figure 13. Comparison of the can center temperature histories of MC (retort experiment and Rretort Program) 58 Figure 14. The relationship of Rho and final unaccomplished temperature (g) 60 -X-Figure 15. The relationship of Rho and retort temperature 61 Figure 16. The C R T processes at different z values in terms of surface cook values (Fs) 64 Figure 17. The optimum VRT processes to yield the minimum F s of MC in terms of different z values 78 Figure 18. The optimum VRT processes to yield the minimum P t of MC in terms of different z values 94 Figure 19. The optimum C R T and VRT processes for the minimum surface cook values. RT and T c indicated retort temperature and can center temperature for the respective C R T and VRT computer simulations 98 Figure 20. The optimum CRT and VRT processes for the minimum process time. RT and T c indicated retort temperature and can center temperature for the respective C R T and VRT computer simulations 100 -xi-ACKNOWLEDGEMENTS The author wishes to express his gratitude to Dr. Tim Durance, research supervisor for his encouragement, support and guidance throughout the course of this research project. He wishes to thank the members of the research committee: Dr. Christine Seaman, Dr. Victor Lo and Dr. Gary Sandberg for their advice during the research phase of this project and in the review of this manuscript. Special thanks are extended to Mr. Jinglie Dou for his help with the Random Centroid Optimization (RCO) program of this project and the use of the R C O program that he, together with Dr. S. Nakai, had written; to Mr. Sherman Yee and Ms. Val Skura for their assistance with the surface color measurement and laboratory equipment; to Ms. Brenda Barker for her all assistance for my thesis writing, printing and committee meeting and more; to Ms. Parastoo Yaghmaee for her advice on the operation of retort. The help provided by several other students and staff within the Food, Nutrition and Health Program of the Faculty of Agricultural Sciences at UBC during the course of the research is greatly appreciated. In addition, I would like to thank my dear wife, Manna Ma. She gave me much more support, encouragement and love when I study at UBC. I also would like to thank my adorable, little daughter, Esther Xiang and she gave me more love, happiness and enjoyment. I cannot finish my study without her help, support and love. Finally, many thanks for my parents, my brothers and sisters, my father-in-law, mother-in-law for believing in me. God Bless them all. 1 CHAPTER I INTRODUCTION Food industry is pressed with the need to provide foods that are safe, nutritious and convenient at competitive prices. In the last decade, various studies have been carried out for quality optimization of thermally processed foods. Computer simulation has made this possible since the kinetics of microorganisms and quality factors, and physics of conduction heat transfer are very well understood and can be described with mathematical models (Sablani et al., 1995). Optimization of the sterilization process is based on the fact that thermal inactivation of microorganisms is much more temperature dependent than quality factors (Lund, 1977). Teixeira et al. (1969) were probably the first to use computer simulations for quality optimization. Now several researchers have used such models for predicting optimal conditions for thermal processing of foods (Ohlsson, 1980; Silva et al., 1992; Hendrickx et al., 1990, 1993; Durance et al., 1997). The first goal in designing a sterilization process is to achieve a reduction in the number of undesirable microorganisms, leading to a safe product with increased shelf life. Because of the applied heat treatment a concomitant decrease in the quality attributes (essential nutrients, color, flavor, texture and so on) is observed (Lund 1982). Conduction-heated foods have a slow rate of heat transfer. Very high temperatures will cause severe thermal degradation of the food near the surface long before the food at the center of the container has risen in temperature. On the other hand, a relatively low retort temperature will cause great quality losses because of the long time it will take to obtain commercial sterility. Consequently, there is an optimum time-temperature relationship that will minimize the quality losses while still providing a microbiologically safe food (Ohlsson, 1980). The optimum constant retort temperature (CRT) processes have been calculated for the case of optimization of surface quality of canned foods (Hendrickx et al., 1990; 2 Banga et al., 1991). The use of recent optimization techniques to solve the problem of finding the optimal retort profile for the optimization of surface retention (Banga et al., 1991) leads to the conclusion that the use of variable retort temperature (VRT) processes represents a valuable policy. An appreciable increase in the surface quality retention (20%), over the optimal CRT processes could be achieved. A considerable reduction in the process time could also be achieved using the VRT processes. These conclusions were based on a limited number of case studies (Noronha et al., 1993; 1996b, Durance et al., 1997). Most of the available work on optimization of thermal processes considers the calculation of the optimum CRT processes. Several authors have investigated the use of the VRT processes. When maximization of mass average quality was considered, there was no significant improvement in the use of optimum VRT processes compared with the optimum CRT processes (Banga et al., 1991). However when the optimization of surface quality retention is considered, substantial increases in quality retention and decreases in the process time could be achieved using the optimum VRT processes, as compared with the optimum C R T processes (Banga et al., 1991; Noronha et al., 1996a). Banga et al. (1991) indicated that surface quality was improved by up to 20% under the optimum VRT process and that the process time could be reduced by up to 16.5% compared with the optimum CRT process. Noronha et al. (1996b) demonstrated that the optimal VRT processes allowed a significant reduction in the surface cook value (22%) or the process time (26%) without reduction of the quality compared to the optimum C R T processes. Almonnacd et al. (1993) also obtained the conclusions that a change from the C R T processes to the VRT processes increased canning capacity by 20 to 50%. Conventional thermal processes, that is constant retort temperature (CRT) processes, have been widely studied for a variety of optimization purposes. Almost all commercial retort processes for canned foods use constant retort temperature (CRT) processes. However, variable retort temperature (VRT) process, as one of the potential technologies to improve both the economy and quality of some canned foods, has been receiving increasing attention (Durance, 1997). Some researchers 3 have focused on optimization of some objective functions, such as surface quality, process time and energy conservation. The VRT process has been shown to be very promising in this regard, especially in improving food quality and reducing process time (Chen and Ramaswamy, 2002b). The surface color is an important quality attribute of foods. This is due to the reactionship among color, flavor and aroma of food products. Discoloration and browning of canned foods are the results of various reactions, including Maillard reaction (Cornwell and Wrolstad, 1981). Heating temperature and heating time both affect the surface color of canned foods. Surface color changes measured by tristimulus reflectance colorimetry may be used to predict both chemical and quality changes in a food (Little, 1976). Durance et al. (1997) reported a study using a finite difference model program (Retort Program) and random-centroid optimization (RCO) program (Dou et al. 1993) to optimize the optimum VRT processes to treat canned salmon by specific small can size (307 x 115 cans) and got good results. Durance et al. (1997) concluded that the best VRT process decreased process time (16%) and the thiamine losses from 19.6 % to 16.8 % which maintained equal F 0 and surface quality compared with the best C R T process. Chen and Ramaswamy (2002) used the small cans (111 x 306 cans) to evaluate the optimum CRT and VRT processes to affect on the surface quality or process time by using coupled neural networks and genetic algorithms. But no one has used the Retort program and R C O program to select the VRT processes to the canned foods in bigger cans. Also no one reported research about surface color change of canned foods and used the surface color change index to decide the best sterilization process. The aim of this project was to study the surface color change of conduction-heated canned foods and use this knowledge to choose the optimum thermal sterilization processes. This study evaluated the application of the Retort program and R C O for modeling and optimization of the CRT and VRT thermal processes for conduction-heated foods. 4 CHAPTER II LITERATURE REVIEW 2 . 1 . Color Measurement Color measurement is a critical objective quality parameter that can be used for the following applications: as a quality index measurement of processed foods for use in quality control documentation and communication; for determination of conformity of food quality to specifications; and for analysis of quality changes as a result of food processing, storage and other factors (Giese, 2000). The color measurements can be used in an indirect way to estimate quality changes of foods, since they are simpler and faster than chemical analysis. HunterLab color parameters (L, a and b) have previously proven to be valuable in describing visual color deterioration and providing useful information for quality control of canned foods, such as pear puree (Ibarz et al., 1999). Of course, the measurement of brown color is one of the most common analytical methods used to study the effects of food composition, storage environment and packaging system on the non-enzymatic reaction of foods (Palombo et al., 1984). For objective color measurement of foods, color scales are used to measure color and color differences. Color is often defined using three-dimensional color scales that describe the different components of color. Light reflected from a colored object is composed of a light or dark component in addition to a red or green and a blue or yellow component. HunterLab measures the degree of lightness or blackness (L), the degree of redness or greenness (a), and the degree of yellowness or blueness (b). Sometimes only one specific dimension of color is needed to determine the quality of a product. For example, Lightness (L) was used to monitor the formation of the Maillard reaction products (MRPs)(Bates et al., 1998). As pH and temperature increased, the L value decreased and the samples became darker. In the tomato industry, the color red is the color by which the quality of the product is evaluated. A 5 set of indices has been derived to measure or score tomato ketchup, sauce, juice, paste, and puree for the degree of redness (Mabon, 1993). Color is one of the three major quality attributes of food along with flavor and texture. However, if the color is unattractive, a consumer may never get to judge the other two quality attributes (Francis, 1991). Color is among the most important quality attributes of canned foods or dehydrated foods for consumers (Driscoll and Madamba, 1994). Color change in canned foods during manufacturing and storage is of vital interest to the food industry, because the first quality judgment made by a consumer on a food at the point of sale is its visual appearance. Appearance analyses of foods, color, taste, odor and texture are used in the maintenance of food quality throughout and at the end of processing (Avila and Silva, 1999; Lopez at al., 1997; Maskan et al., 2002). The color of food products can be specified by three co-ordinates in the color space that can be obtained directly with a tristimulus colorimeter. A variety of color scales are used to describe color. Those most often used in the food industry include the HunterLab system, the CIELab system and the Munsell control solid (Giese, 2000). The HunterLab system is the most frequently used scale to measure the color of food products (Hutchings, 1994). The HunterLab systems decide the L, a, b color coordinates. The L coordinate measures the value or lightness of a color and ranges from black at 0 to white at 100. The a coordinate measures red when positive and green when negative. The b coordinate measures yellow when positive and blue when negative (Chen et al., 1999). 2. 2. Color Change with Heat Treatment A. Maillard Reaction The Maillard reaction is a type of non-enzymatic browning reaction that involves the reaction of carbonyl compounds, especially reducing sugars, with compounds that possess a free amino group, such as amino acids and proteins. The reaction products are significant in foods because they are responsible for flavor and color changes, which may be desirable or undesirable depending on the type of foods (Ames, 1990). 6 Non-enzymatic browning reactions between amino acids and reducing sugars are the basics of the Maillard reaction, which take place in thermally processed foods. The Maillard reaction results in the formation of complex mixtures of colored and colorless reaction products, which range from flavor volatiles to melanoidins, a series of brown pigments with high molecular weights. Brown pigment formation is desired during some types of food processing (baking, cocoa and coffee roasting, cooking of meat), while it is absolutely undesirable in other technologies (milk drying, thermal treatments for the stabilization of milk, fruit juices and tomatoes). The Maillard reaction often has negative consequences not only on the sensory characteristics of foods (color changes and volatile compound formations), but also on the nutritional value (amino acid and protein unavailability for human metabolism) (Lerici et al., 1990). When food is cooked, the Maillard reaction plays an important role in improving the appearance and taste of foods. Maillard reaction is related to aroma, taste and color, particularly in traditional processes such as roasting of coffee and coco beans, the baking of bread and cakes, the toasting of cereals, the cooking of meat, the sterilization of canned foods (Martins et al., 2001). The Maillard reaction also plays an important role in the production of undesirable flavor compounds, and in the development of browning color during thermal processing (Palombo et al., 1984). Various factors are responsible for changing the color during processing of food products. These include Maillard and enzymatic browning and process conditions, such as pH, acidity, packaging materials and duration and temperature of storage (Ahmed and Shivhare, 2001). The Maillard reaction is largely responsible for the roasted, toasted, or caramel-like aromas, as well as the development of browning color in protein and carbohydrate rich foods following a thermal treatment (Nursten, 1986). Because of the inherent complexity of many food systems, such as coffee, much of the work on the Maillard reaction has been accomplished in simple Maillard reaction model systems of individual amino acids and reducing sugars or lipids (Friedman, 1996; Namiki, 1988). The Maillard reaction occurs nonenzymatically in foods between reducing sugars and 7 available amino groups during thermal processing and home cooking operations. The Maillard reaction is influenced by many factors such as temperature, time, pH, water activity (aw), and reactants (Wijewickreme et al., 1997). Maillard reactions include those involving reducing sugars, aldehydes, and ketones with amines amino acids, peptides, and proteins. In food, the normal reactants are reducing sugars and amino acids. Reactions can be divided into three phases. The early phase consists of defined chemical reactions without browning. The second phase consists of many reactions involving the formation of volatile or soluble substances. The final phase consists of reactions leading to the production of insoluble browning polymers. Most chemical changes that occur during caramelization also occur in Maillard browning. Many reactions that take place in pure sugars only at very high temperatures occur at lower temperatures once they have reacted with amino acids (Mauron, 1981). Maillard browning can be found in three different areas of food manufacture. It has a traditional use in the development of aromas and flavors in roasting, baking and cooking; it is used deliberately to engineer flavors in non-traditional foods; and it occurs as an undesirable byproduct of food processing, affecting color or flavor, or both (Buckholz et al., 1980). The Maillard reaction is of considerable importance to food companies. In particular, pasta industries need more knowledge to control browning during processing; in fact, pasta color is generally considered as one of the major components of quality (Fogliano et al., 1999). The Maillard reaction produces a multitude of small molecular weight intermediates, collectively referred to as Maillard reaction products (MRPs), and high molecular weight polymeric compounds known as melanoidins. Melanoidins were isolated from different model systems consisting of a single amino acid and carbohydrate (Fogliano et al., 1999). The typical brown color formed by Maillard reaction is due to chromophores, which have been widely studied in different model systems. In a gluten-glucose model system, colored low molecular weight molecules became entrapped in the high molecular weight polymers formed by gluten proteins (Fogliano et al. 1999). In a 8 casein-sugar model system, it is established that color formation is mainly due to the formation of protein oligomers mediated by chromophoric substructures derived from carbohydrates. In different model systems, the Maillard reactions are different and they produce different Maillard reaction products (MRP). Thus the food product will have different color changes with different heat treatments (Hofmann, 1998). A number of kinetic studies have been carried out on the Maillard reaction. Two approaches with respect to Maillard reaction kinetic studies have been proposed in the literature. The first approach focuses on the rate of browning, and the other relates to the rate of loss of sugar and amino acids (Xing, 2002). Baisier and Labuza (1992) reported that although the overall kinetics of Maillard reaction are more complex than the individual loss of sugar or amino acids, the initial stage of the reaction follows pseudo-first order kinetics. After the initial first order period, the loss of reactants tapers off into a phase with little reactant disappearance (no loss period), which can be explained by means of steady state kinetics (Baisier and Labuza, 1990) B. Color Change in Canned Foods The time-temperature combinations used in canning have a substantial effect on most naturally occurring pigments in canned foods. For example, in meats the red oxymyoglobin pigment is converted to brown metmyoglobin and purplish myoglobin is converted to red-brown myohaemichromogen. Maillard browning and caramelisation also contribute to the color of sterilized meats. However, this is an acceptable change in cooked meats. In fruits and vegetables, chlorophyll is converted to pheophytin, carotenoids are isomerized from 5, 6-epoxides to less intensely colored 5, 8-epoxides, and anthocyanins are degraded to brown pigments. In sterilized milk slight color changes are due to caramelization, Maillard browning and changes in the reflectivity of casein micelles (Fellows, 1998). In pasta industry, pasta color is generally considered as one of the major components of quality. Consumers like an amber-yellow color while an intense brown tone causes a decrease of the commercial pasta value (Fogliano et al., 1999). 9 The heat treatment of foods rich in reducing sugars and free amino acids results in the production of MRPs . Therefore, heat treatments such as frying, baking, broiling, stewing and thermal processing have an integral role in the quality of browning which in turn will influence the sensory, color and nutritional compositions of the foods (Xing, 2002). In a macaroni and cheese system, it contains sugars, proteins and amino acids. Through heat treatment, MC could take place the Maillard reaction. The sugars and amino acids of the MC will change their compositions and new compounds are produced. MC will have different color changes with the Maillard reaction. C. Thermal Kinetics of Color Change of Foods Food color changes can be associated with its heat treatment history. Various reactions such as pigment destruction (carotenoids and chlorophylls) and non-enzymatic browning (Maillard) reactions affect the color of foods during blanching of fruits and vegetables and during the heat processing to canned foods (Cornwell and Wrolstad, 1981). The retention of total color can be used as a quality indicator to evaluate the extent of color deterioration during thermal processing (Shin and Bhowmik, 1995). Several researchers have published work on modeling of thermal degradation kinetics of color in the temperature range of sterilization conditions. The majority of the published work reported first order or zero order degradation reaction kinetics (Avila and Silva, 1999). The kinetics of color change in food products is a complex phenomenon, and dependent on models to predict experimental color change. Experimental studies and application of various simplified models to represent the behavior are required. Several authors studied the color kinetics of food materials during thermal processing in terms of changes in Hunter tristimulus color values L, a and b (Berry, 1998; Weemaes et al., 1999). To optimize the thermal process of a food, it is important to determine the kinetic parameters (reaction order, reaction rate constant, and activation energy) for color change (Weemaes et al., 1999). Hence, if the kinetics of color change are determined and the order of color change is established, the total 10 color can be used to evaluate quality of food product during thermal processing (Ahmed and Shivhare, 2001). Calculating and predicting a quality indicator in food systems involves development of a mathematical model during processing (Samaniego-Esguerra et al., 1991). A quality indicator such as color is usually modeled using a general reaction rate equation: dC/dt = k C n (1) Where C is the measured HunterLab color value (L, a, b) of the product, C 0 is the measured HunterLab color value at zero time, t is the heating time (min) and k is the reaction rate constant (per min). The order of a chemical reaction is generally zero or first order (Ozdemir and Devres, 2000). The Maillard reaction in foods is generally first-order or zero-order reactions (Driscoll and Madamba, 1994; Chen and Ramaswamy, 2002a). The results of Ahmed et al., (2000) and Shin and Bhowmik (1995) indicated that color degradation during thermal processing of chilli puree followed first order reaction kinetics. 2. 3. Thermal Processing of Canned Foods A. Goals of Thermal Processing for Canned Foods Heat sterilization of foods is a preservative technique that aims to obtain a safe product with a long shelf life and is based on the application of suitable time-temperature conditions to thermally inactivate microorganisms, spores and enzymes (Maesmans et al., 1990). The recommended sterilization processes are not designed to kill all microorganisms in canned foods. In canned food sterilization, the main concern of the canning industry is to prevent the growth of Clostridium botulinum, the food poisoning bacterium capable of producing a highly lethal toxin (Lopez, 1981). Where Forn>1 , C / C 0 = (1 + (n-1) kt) 1 / ( 1 " n ) For n = 1 (first-order), C = C 0 exp (-kt) For n = 0 (zero-order), C = C 0 - kt • •(2) (3) ..(4) 11 Clostridium botulinum is the most heat-resistant, anaerobic, spore-forming pathogen that can grow in low-acid canned foods, and consequently its destruction is the criterion for successful heat processing of this canned food (Lund, 1991). A sterilization process that assures the destruction of Clostridium botulinum usually also kills all other microorganisms capable of producing canned food spoilage under normal conditions of canned food handling and storage (Lopez, 1981). The thermal processing of canned foods is one of the most widely used methods of preservation in the twentieth century (Teixeira and Tucker, 1997). The concept of thermal processing is based on heating of canned foods for a certain length of time to obtain a safe product complying with public health standards. The thermal processing is based on established time-temperature profiles. Associated with thermal processing is always some degradation of heat-sensitive quality factors that is undesirable. Since much demand is on safe and shelf-stable food products along with a high quality attributes, processing schedules are designed to keep the process time to the required minimum (Afaghi and Ramaswamy, 2001). The differences in the temperature-sensitivity between the rate constants of destruction of microorganisms and those of quality factors, such as color, flavor, texture and nutrients, allow the choice of an appropriate heating process that minimizes the degradation of quality factors while still achieving the necessary destruction of undesirable microorganisms (Noronha et al., 1996b). B. Processing Media of Canned Foods The processor wishes to provide the consumer with a safe product, and within economic constraints, one exhibiting the maximum possible retention of quality attributes (Durance, 1995). Different heating mediums are used to optimize the retorting of different forms of food and types of packaging. In the food industry, there are three kinds of heating media, which have been used for processing of filled containers in retorts: steam, water immersion/overpressure systems and steam/air mixtures. In general, steam is used for cans and is the most popular heating medium and is used in many retort designs. Steam is easily manufactured, regulated and held for immediate use, the steam pressure within the retort helps to 12 counterbalance the pressure in the can. Steam produces large amounts of latent heat available to heat the food (Durance, 1995). C. Optimum Sterilization of Canned Foods In commercial heat sterilization of canned foods, the cans have been heated in a retort at certain conditions of temperature and time. Much attention has been given to maximizing quality retention for a specified reduction in undesirable microorganisms during sterilization (Terajima and Nonaka, 1996). Quality optimization is possible because the degradation kinetics of quality is much less temperature-sensitive than the kinetics of microorganism destruction (Lund, 1977). More researchers have optimized sterilization of canned foods in terms of quality retention (Lund, 1982, Holdsworth, 1985, Silva et al., 1993). Teixeira et al. (1969) calculated the optimum retort temperature for cylindrical cans using thiamine retention as optimization criteria. It is necessary to obtain an optimal compromise with regards to quality and consistency (Hildenbrand, 1980). Now more techniques such as computer simulation, expert systems, on-line monitoring and semi-automatic control systems are used in the food industry to optimize sterilization process and allow canned foods have a long shelf life with a minimum quality loss (Ramesh, 1995). D. Temperature Measurement and Heat Penetration Tests Data obtained from heat penetration tests conducted on containers of foods during processing can be used to calculate the process time required for that product. Temperature measurements are made at the slowest heating spot (cold spot) in the filled container. Procedures for conducting such heat penetration tests have been described by Bee and Park (1978). Obtaining accurate data regarding the heating and cooling of the food in a container is extremely important if an accurate time and temperature for product sterilization is to be determined. The results of a heat penetration test are experimentally derived 13 heating and cooling curves. The type of curve obtained is dependent on the kind of product involved. Parameters obtained from the data plot are dependent on the manner in which data are plotted. Generally, factors influencing rate of heat penetration are retort temperature, size and shape of a container, fill-in weight, thermal properties of the food, initial product temperature and heating medium (Downing, 1996). In cans, Ecklund Type-T rigid thermocouples (O. F. Ecklund Inc., Cape Coral, FL, USA) have been the primary means used to obtain temperature measurements for heat penetration work (Bee and Park, 1978). Thermocouples are preferable to thermometers in measuring temperature changes because of the physical properties of canned foods. To insert the thermocouple into a can, a hole is cut in the sidewall of the can. Thermocouples are placed in the cold spot of a can. A gasket receptacle is placed through the hole, and screwed in place. The thermocouple with the receptacle adaptor is inserted, and then the can filled and closed. For conduction-heated foods, the cold spot is the geometric center of the can. The thermocouples are also placed outside the cans to monitor the retort temperature (T r). The temperature of each thermocouple is measured at set intervals of time (every 60 seconds). These temperatures are collected with a data logger, and then presented in a standard manner (Durance, 1995). (Terminology and abbreviations in thermal processing are presented in Appendix A.) E. Process Determination The process time required to sterilize a canned food is influenced by the heat resistance of microorganisms or enzymes in the foods, the heating conditions, the pH of the food, the size of the container and the physical state of the food. It is also necessary to have information about both the heat resistance of microorganisms, particularly heat resistance spores, or enzymes that are likely to be present and the rate of heat penetration into the food (Fellows, 1998). The main objective of thermal process calculations is to determine the process time for achieving a pre-selected process lethality or making heat treatment sufficient to 14 destroy expected spoilage organisms or evaluating the lethality of a given process (Afaghi and Ramaswamy, 2001). The sterilization value of a process is generally expressed as the F 0 value which is equivalent to the number of minutes required to destroy a specified number of Clostridium botulinum spores at 121.1 °C (250 °F) when z value equals 10 C 0 (18 F°) (Downing, 1996). F. The Improved General Method and Sterilization Value (F0) The Improved General Method is the most accurate method for a given experimental condition, as it makes use of real time-temperature data for process calculations (Afaghi et al., 2001). However, this method provides little flexibility in allowing mathematical determination of process changes when variations in conditions occur. A general rule of thumb is that a process should have a total lethality three times F to insure a safe process for a low acid canned food (Durance, 1995). Lethality can be derived from the graph in Figure 1 in the following manner: AY/AX= Iogt-Iog1/T-250 (5) Log t - l og 1/(T-250) =-1/z (6) Log (1/t) = T-250/18 (7) 1 / t = 1 0 ( T - 2 5 0 ) / 1 8 ( 8 ) L = "lethal rate" = 1/t (9) L _ 1 Q (T-250)/18 _ 1 Q (T-121.1)/10 ^ Accumulated Lethality (F0)= £ Lx At (11) Where At = time interval over which L is considered constant. The lethality of the Improved General Method is a special case, based on the unit (the decimal death time (TDT)) curve where z = 10 C° (18 F°) and the reference temperature =121.1 °C (250 °F). It is given the symbol F 0 (with units of time). The reference TDT curve can be used to construct a "lethality" curve from any heat penetration curve. Thus, we are no longer dependent on the knowledge of the TDT of any organism. We can determine the F value of any process and compare it to the F value of any other process and thus tell which of the two is more effective. If F 0 is 6.0 min, then the entire thermal process is equivalent in terms of lethality to 6.0 minutes at 121.1 °C for any microorganisms with a z =10 C° (Durance, 1995). 15 1C0 10 (T.t) J2 V . « -u u> OS O Q (250°, 1) 0.1 2 2 0 Temperature (*F) - 18 F 3 250 l i l i l l i f l i Figure 1. Reference TDT curve, where F=1 and z = 18 F° (Durance, 1995) 16 After time and temperature data for a given product in a given can have been obtained by heat penetration studies, these data may be analyzed by the general method. The Improved General Method is used to measure the exact sterilization value of a process when such conditions as come-up time, cooling water temperature or the holding time after processing but before water-cooling are different from normal retort procedures. Time and temperature data during the cooling cycle as well as the heating cycle must be recorded in order to use the general method (Downing, 1996). The accumulative lethality method, in which the time-temperature data from heat penetration test is analyzed for determining process lethality, is the most accurate method possible (Stumbo, 1973). In developing a process schedule, a specific target lethality value for the product must be known and heat penetration tests performed with thermocouples installed in the center point of the cans. The test product must be adjusted to an initial temperature normally encountered during commercial production. The retort temperature used for the heat penetration test must be no higher than the retort temperature intended for use during commercial production. The process time can be increased by calculation over the test process time if additional process lethality is required (Spinak and Wiley, 1982). The target sterilization value F 0 depends on the expected number of spores and the medium where the spores are processed. For example, products higher in acid or salt will require a less severe heat process. The number of organisms is also important. Mushrooms and pet food have high concentrations of spores, while baby food spore counts are lower. The typical target F 0 for canned mushrooms and pet food ranges from 10 to 18 minutes, while baby food may have a F 0 of 3 to 7 minutes (Durance, 1995). However, the food composition of canned foods can dramatically influence the survival of spores, target F 0 should preferably be determined individually for each type of product. For a new product, the target F 0 is based on previously established processes for similar products (Durance, 1995). The sterilization value (F) at the coldest point in container assures a minimum sterility in all points of the foods; therefore this is the most adequate criterion (Silva et al., 1993) 17 2. 4. Constant Retort Temperature (CRT) Process and Variable Retort Temperature (VRT) Process A. Definition of the CRT and VRT processes C R T process is defined as the process which includes a come-up time (the time needed for the initial retort temperature to rise to the prescribed retort temperature, for example, 119 or 121 °C), a holding time at constant heating temperature, cooling time with cooling water. The come-up time includes the vent time plus the time for the retort to reach the prescribed retort temperature after the vent is closed. Process time (P t) of a C R T process is defined as holding time not including vent time or cooling time. A VRT process is defined as a process which includes a come-up time, Until retort temperature reaches 104 °C; a variable temperature period in which retort temperature changes with the heating time and cooling time. Like the CRT process, the process time (P t) of a VRT process does not include come-up time or cooling time (Durance et al., 1997). Durance (1997) compared the difference of the CRT and VRT processes. Figure 2 shows the difference of the typical CRT and VRT processes. This Figure shows that their retort temperatures are different. For the CRT process, the retort temperature is constant from vent time (about 6 minutes) until the steam turns off. For the VRT process, the retort temperature was variable from about 104 to 130 °C after vent time (from initial retort temperature to 104 °C). Often, the process time of the VRT process is shorter than that of CRT process with the same sterilization value. 18 140.00 120.00 O 100.00 3 80.00 re o. E 60.00 0) -2 40.00 20.00 o.oo n Retort temp. (CRT) Retort temp. (VRT) i r ~i r 0 20 40 60 80 100 120 140 160 Heating time (min) Figure 2. Comparison of retort temperature histories of conduction-heated canned foods with the C R T and VRT processes (Durance, 1997) 19 B. Retort Program Finite difference models of heat transfer into packaged food have been successfully applied in optimization and control (Teixeira et al, 1969; Teixeira and Tucker, 1997; Durance et al., 1997). The main feature of this model is the prediction of the temperature profile based on the governing heat transfer equations of packaged food products. A finite difference model requires several input data related to the food product and system such as thermal diffusivity of the food product, heat transfer coefficient of the heating and cooling medium, and processing conditions. When these conditions are known, time-temperature data at any specific location of the product can be obtained by solving the appropriate governing equations. Because of its ability to provide accurate time-temperature history, this model has largely replaced the need to carry out experiments for routine data gathering when the boundary conditions are well defined (Afaghi et al., 2001). However, actual heat penetration experiments are still a regulatory requirement for determination of commercial food sterilization process. A finite difference model is based on a numerical solution of unsteady state heat transfer, providing transient temperature distribution throughout the container. At the beginning of the process time, all the interior points of the cylinder are set to the initial temperature of the product, while the temperature at the surface is set at the retort temperature. With a known set of initial conditions, these equations are solved at each time interval. The new temperature distribution at the end of each time interval is used to set the initial conditions for the following time interval. This procedure is continued for a pre-determined process time, during which the temperature profile of product is computed. The same procedure is applied for cooling of the product by changing the ambient temperature to cooling water temperature and continuing the calculation process (Afaghi et al., 2001). The objective of any heat-transfer analysis is to predict heat flow or the temperature that results from a specified heat flow. During commercial sterilization, the heat transfer within the can was estimated with a two-dimensional finite difference model 20 (Sandberg, 1991; Sandberg et al. 1994). Average thermal diffusivity (a) of the food material was calculated as follows: a = (0.398) / [ fh (1 / r2 + 0.427 / b 2 ) ] (12) Where fh is the average heating rate index determined in a retort trial, r is the radius of the can and b is the half-height of the can. The thermal diffusivity of the heating side and that of cooling side may be different and the authors used a factor to adjust the thermal diffusivity of the cooling side of thermal process. This model controlled initial retort temperature, retort temperature, cooling water temperature, initial product temperature and final product temperature. Surface heat transfer coefficients for heating and cooling were 10,000 and 800 W/m 2 °K, respectively. Steam-off condition was based on the sterilization value (F 0) at the time of steam-off. Output included temperature histories at the surface of the can and the center-point of the can, as well as sterilization value (F 0) at the end of cooling and the accumulated surface cook value (F s) at the end of cooling and the process time (Durance et al., 1997). The product temperature was assumed to be uniform throughout the can at the beginning of the cook. Heat penetration measurements were used for comparison with the model only if the measured center point initial temperature was < 1 °C from the nominal initial temperature. A perfect thermal contact at the surface of the container was also assumed, in an attempt to simplify the model. Lastly, due to the large temperature difference between the interior of the container and the saturated steam environment of the retort, the convective boundary condition was ignored at the container surface and was set at retort temperature at the beginning of the process (Sandberg et al., 1994). C. RCO Program Since Morgan and Deming (1974) applied simplex optimization for selection of analytical conditions, this method has become one of the most popular optimization techniques in chemistry. This optimization technique can accommodate nonlinear equations to predict response values by including a subroutine; however, it is incapable of handling constraints with exception of a boundary constraint (Nakai, 21 1981). Vazquez-Arteaga (1990) modified the Complex (constrained simplex) technique of Box (1965) for application to meat formulation. The method (Forplex) searches for the best quality within an acceptable cost range, in contrast to least-cost formulation. Equations to predict quality parameters were derived as functions of the ingredient composition through small-scale experiments for frankfurter preparation. Forplex was superior to least-cost formulation as it could obtain quality parameter values, closer to the values for desirable product than those obtained by least-cost formulation (Dou et al., 1993). In addition to the incapability of handling constraints, simplex optimization suffers from the following shortcoming: 1. quick loss of efficiency during optimization and 2. difficulty in homing-in on the global optimization. Random centroid optimization was established in UBC food science (Nakai, 1990) to circumvent these shortcomings. It is possible to accommodate constraints through mapping by selecting new search scales to avoid trespassing the level values, which will violate the constraints (Dou et al., 1993). Now there are different programs to optimize the optimization VRT processes for getting the optimum result. Banga et al. (1991) proposed a new algorithm, ICRS/DS, for the solution of fixed terminal a combination of a robust parameterization of the control function and a computationally efficient non-linear programming algorithm. The objective was to calculate optimum VRT in order to maximize surface and overall retention or minimize process time (Silva et al., 1993). Noronha et al. (1996b) used the F O R T R A N program using a quasi-Newton multivariable optimization subroutine to calculate the VRT processes. The genetic algorithms (GAs) are a combinatorial optimization technique, which searches for an optimal value of a complex objective function by simulation of the biological evolutionary process based on crossover and mutation. Chen and Ramaswamy (2002) optimize the VRT processes by GAs and got good results. A random centroid optimization program (RCO) is used to search simulantously for optimal levels of many factors. R C O is an effective optimization programme while 22 allowing testing of several treatments at a time (Girard and Nakai, 1991). The R C O program consists of a random search, a centroid search and mapping, which together constitute a search cycle (Dou et al., 1993). R C O is also modified by introducing a penalty function to accommodate constraints in formula optimization. A new program of random centroid optimization (RCO) was written that is useful for graphical solutions of multimodel cases of optimization. The R C O repeats a cycle of random search—centroid search—mapping. The mapping defines the search spaces to be used in the random search of the succeeding cycle (Nakai et al., 1998). It is expected that broader application of R C O is feasible not only in food research and development but also a variety of optimization purposes in different fields of study (Nakai etal . , 1999). A deterministic rule was modified in order to obtain more uniform distribution of experimental points. Centroid search is conducted by altering the vertex to be excluded in the centroid computation from the worst to the second worst and then to the third and so on until the subsequent response becomes worse than the preceding response (Aishima and Nakai, 1986). Mapping is an approximation of the response surface. Mapping assists visualization of the true response surface steps of the simplex optimization (Dou et al., 1993). A mathematical model for 15 factors (xi - x -15) was formulated using the matrices of Bowman and Gerard (1976). R C O was applied to the 15-factor model. This model also was used to optimize computations for 3-15 factors by replacing unused factors with their optimal level values. To optimize these models, the mapping process was automated by selecting narrower search spaces for subsequent search cycles to be one-third the size of search spaces of the previous search cycle around the best response values (Dou et al., 1993). Dou et al. (1993) showed the number of experimental points for search convergency for mathematical models with different number of factors. Dou et al. (1993) also got results that in situations when the number of factors is less than eight, the number of experimental points required for optimization slowly increases up to 50. Normally, it needs about 30-50 experimental points when there are 5 - 6 factors for R C O program. Then the optimization result 23 from these 30 to 50 points is obtained. A potential risk of missing the global optimum exists in this strategy as a result of narrowing the search ranges of factors selected in the first series of optimization. Random search possesses high flexibility by freely extending its search spaces outside of the set ranges if required and finally homing-in on the global optimum in the case of models with local optima. Therefore, the global optimum may not be frequently overlooked (Dou et al., 1993). The R C O repeats a cycle of random search-centroid search-mapping. The centroid search, which computes averages of level values in better groups of response values, also contributed to improvement of the optimization efficiency. Continuation of searching around the best response that was found in the random search would more thoroughly utilize the information derived from the random search. The mapping defines the search spaces to be used in the random search of the succeeding cycle. The new search space for each factor should be sufficiently narrow near the global optimum. Therefore, this mapping step is highly critical in achieving the efficient optimization without being stalled at local optima. Success of the R C O for global optimization owes mainly to "a factor ignoring process" (to ignore factors during computation of trend lines). Mapping process for automating the intensified line-drawing process was included in the R C O program by eliminating one or two factors simultaneously in a search cycle by rotating the factors to be ignored. A S a result, by using model functions appeared in the global optimization papers reported in the literature, the R C O has found the global optima by running less than 50 experiments for most functions (Nakai et al., 1999). Figure 3 shows the simplified flowchart diagram of the R C O procedure. 24 Start ± Random search (Define upper and lower limit for each factor: list random combination) Experiment (Conduct the experiments on given combination) * Centroid (Enter the results: program narrows the ranges and lists combinations) Experiment (Conduct the experiments on given combination) t Mapping (Enter the results in the program as response: map the results) Figure 3. The simplified flow diagram of R C O procedure. 25 Figure 4 shows the operation chart for R C O program. By entering the search spaces of all factors narrowed by mapping in cycle 1, the random search design would be printed out or saved to files. After entering the response values obtained by experimenting, Centroid 22 would print the centroid design. Upon entering experimental results (response values), Sum/Map23 would print out or save the summary data of the cycles 1-2 combined and its mapping was then initiated. Then the procedure was continued to random 31 to random 41 until the optimal results were obtained (Nakai et al., 1999). MaxMin was the option button for selecting maximization or minimization. "Select cycle" contained four options for cycle 1-3 and Simult. Additional cycles 4 and 5 were for optimization involving a larger number of factors (the program can accommodate 3-30 factors). After one of these option buttons had been "clicked", the processes in each procedure list should be followed step-by-step for random search, centroid search, and summary/mapping, except Simult. The two digits after each step title were the identification numbers to show the step in use (Nakai et al., 1998). 26 rMaxMirv 0 Maximization 0 Minimization f Select cycle-0 1x\ cycle 0 2nd cycle 0 4th cycle 0 3rd cycle O Sth cycle 0 Simult Shift Procodura Open first Open first Open f irst RandomH Centtoid12 Sum/Map! 3 Random21 Centioid22 Sum/Map23 Random31 Centroid32 Sum/Map33 Shf(Gornb41 SeirShfM2 Sum/Map43 Figure 4. A comprehensive operation chart of R C O program (Nakai et al., 1999) 27 D. Computer Simulation of the CRT and VRT Processes Durance et al. (1997) defined the CRT and VRT processes through the Retort Program. C R T processes were defined by a 6-minute vent time, during which the retort temperature rose linearly to the nominal retort temperature, a period of constant retort temperature and a period of cooling. Retort temperature during cooling decreased over 7 minutes, from the final retort temperature of the heating cycle to the cooling water temperature of 10 °C, then remained constant until the can center-point reached 90 °C (Durance et al., 1997). VRT processes included a 6 min vent time to 104 °C, which was the vent time necessary to ensure a pure steam environment in the test retort. The shape of the subsequent retort temperature versus time profile was defined by four coordinate pairs on the profile, (0.25 tv, R T ^ , (0.50 tv, RT 2 ) , (0.75 tv, RT 3 ) , and (tv, RT 4 ) . Straight-line segments between such points can be made to approximately curvilinear temperature profiles. The five variables; total time of variable retort temperature heating (tv) and values of the four intermediate retort temperatures (RT-i, RT 2 , etc.) were adjusted in each computer simulation experiment as directed by the Random Centroid Optimization search procedure. The search was further constrained such that temperatures increased through the cook (i.e. RTi< RT 2< RT 3< RT 4 ) . If process time specified by R C O exceeded vent + tv then R T 4 was maintained until accumulated bacterial lethality equaled the target F 0 multiplied by fraction of sterilization value (Rho), at which point cooling was begun (Durance et al., 1997). E. Estimation of Rho (p) (Fraction of Sterilization Value) The fraction of bacterial lethality that occurs in the heating side of thermal processing or Rho was estimated, allowing the experimenter to end heating at the correct time and achieve the target F 0 at the end of the cooling time. Relationship of Rho (p) to retort temperature (retort temperature from 120 to 130 °C), final unaccomplished temperature (g = RT-T f ; 1< g <12; Tf = center-point temperature at time of steam-off), and thermal diffusivity (a) of the can contents (a from 1x 10 "7 to 2.2 x 10 "7 m 2/s) was estimated with repeated computer simulations (Durance etal . , 1997). 28 Accurate prediction of Rho (p), the fraction of total F 0 which occurs prior to steam-off, was desirable because Rho (p) would greatly reduce the number of experiments required for computer optimization of the VRT processes. If Rho (p) is unknown, many simulations of each VRT process must be performed in order to arrive at a suitable process time to give the target total F 0 , while one is sufficient if p is known since "Retort" can be set to begin cooling once a given interim F 0 is achieved in the heating side. Stumbo (1973) estimated p as a function of final unaccomplished temperature (g), the difference between maximum retort temperature and the can center-point temperature of the product at steam-off (Durance et al., 1997). Through the retort program (computer simulation), Durance et al. (1997) found Rho (p) was also a function of thermal diffusivity (a), retort temperature, container geometry and container size. 2. 5. Quality of the Thermally Processed Canned Foods A. Basic Consideration for Canned Foods Thermal resistance of food components of canned foods must be considered to develop strategies for maximizing retention of quality attributes. Examinations of these data indicate several important points. The temperature dependence for vulnerable quality attributes, both sensory and nutritional qualities are similar. Thus, optimization for one quality attribute will generally optimize the retention of all quality attributes (Rizvi and Acton, 1982). In thermal processing of low acid foods, the primary concern of the processor is to achieve a condition of total absence of microorganisms of public health significantly especially Clostridium botulinum and its spores as well as other nonpathogenic microorganisms that may be capable of growing and causing spoilage of the food under normal storage and distribution conditions. It is only after having assured the safety of the food that the canner then chooses adequate temperature-time combinations that would optimize nutrient and organoleptic quality retentions (Ariahu and Ogunsua, 1999). 29 B. Effect of Container Size Each point within the container must receive a heat treatment sufficient to destroy the microbial population of concern in order to produce a safe product. In conduction-heated products, the rate of temperature response within the product is limited by the distance within the food through which the heat must penetrate and by the thermal diffusivity of the product. The thermal diffusivity is a material property for a particular product, but the thickness of material through which the heat must penetrate can be changed by altering the container geometry (Teixeira et al., 1975a). By reducing the distance required for heat penetration, process times required to achieve a safe product can be reduced and retention of quality attributes improved (Teixeira et al., 1975a). Teixeira et al. (1975a) used a finite difference computer model to calculate temperature histories at many locations within containers, coupled with microbial spore and thiamine degradation kinetics to predict thiamine retention in conduction-heated foods processed at 121.1 °C (250 °F) in different cylindrical can sizes receiving the same sterilization effect. Ohlesson (1980) did the same research about different can sizes that provided different volumes. The integrated effect on quality was expressed as the cook value. Her results showed that improved quality could be obtained by using cans that provided a minimum distance for heat penetration to the center (Young, 1984). The concern for producing high quality products has led to investigations in which different processes, that accomplish the major objective of safety, have been compared on the basis of quality retention. High temperature short time processes have been used to achieve these objectives with convection-heating products and in aseptic processing (Lund, 1977). Variations in container geometry provide greater promise for improved quality retention. A significant increase in the nutritional value of a thermally processed food is possible with the use of container geometries, which allow more rapid heat penetration compared to conventional cans (Texieira et al., 1975b). A change in container geometry offers the possibility of improving retention of quality attributes. For such improvements to be observed, it would appear that 30 careful control of processing conditions must be maintained to ensure that overprocessing does not occur (Young, 1984) C. Effect of Processing Temperature Teixeira et al. (1969) used a finite difference model to predict thiamine retention in a conduction-heated product processed in a cylindrical container. Thiamine retention may decrease with increasing process temperature. When the product receives a relatively severe heat treatment at the outer surfaces in order to heat the food sufficiently at the center, this results in lower thiamine retention overall. It was demonstrated that the optimum temperature would vary depending on the conditions under study. A heat labile factor with a relatively low z value showed optimum retention with a relatively low process temperature compared to a high z value quality factor for which retention was favored by a higher temperature process. Ohlsson (1980) used a similar type of study to predict the integrated effect on quality (cook value) in conduction-heated foods in cylindrical cans. Their results showed the same trends as did those of Teixeira et al. (1969). Also tested were the effects of changing can size, process lethality (F 0) and initial temperature on the optimum process temperature required for the minimum cook value. Increasing the can size shifted the optimum temperature to lower values. Many researchers suggested that optimal retort temperatures were in the range of 113 to 119 °C for normal can sizes. Of course, some substantially smaller can diameters or heights would be advantageous for thermal sterilization at a higher temperature (Young, 1984). Thus, there is potential to improve quality retention in thermally processed foods by altering the container geometry and /or retort temperature. Of the two, changes in container geometry can provide a larger improvement (Teixeira et al., 1975b). The magnitude of the differences and the optimum retort temperature will depend on the product and container tested, as well as the thermal degradation kinetics of the quality attribute under investigation (Young, 1984). 31 D. Surface Quality for Canned Foods The maximization of the final canned food quality can be considered in terms of surface quality retention (Banga et al., 1991; Hendrickx et al., 1993; Silva et al., 1994) or volume average quality retention (Banga et al., 1991). Average quality is important for nutrient retention, texture characteristics, etc., while optimum surface quality is necessary for quality attributes such as appearance, color and aroma (Silva et al., 1992). The experimental determination of optimum retort temperature to minimizing the surface cook value of canned foods is an important procedure for evaluation the effect of thermal processing on product surface quality. Rate of the surface quality loss (Q) was defined as Q = 10 /z At (Durance et al., 1997). When conduction products are processed to adequate center lethality and they inevitably received excessive surface cooks. A C R T process, which yielded the minimum surface cook value, existed for each combination of containers, product and surface z (Durance et al., 1997) The comparison of the optimum CRT's with the optimum VRT 's for the same process time showed that it is possible to get improvement in the quality retention at the surface up to 20 % by using the VRT process. From the case studies of the VRT process, there was no straightforward relationship between the achieved improvements and the z value or target sterilization value (Noronha et al., 1993) E. Goals of this research project The overall objective of this project was to evaluate the optimum CRT and VRT processes to decrease the process time or improve surface quality for macaroni and cheese (MC) by using the Retort program and R C O program. A limited number of studies have been conducted to compare surface quality or process time of MC in 32 307 x 409 cans. As previously discussed, surface quality and process time of the optimum CRT and VRT processes have been evaluated in conduction-heated canned foods in small cans (301 x 115 cans or 111 x 305 cans) by several researchers (Noronha et al., 1993; Durance et al., 1997; Chen and Ramaswamy, 2002). But no one used the big cans to evaluate the surface quality or process time for conduction-heated canned foods by using the optimum C R T and VRT processes. In this project, the big cans (307 x 409 cans) were used to evaluate surface quality or process time for MC by using the optimum CRT and VRT processes. The use of a canned MC product as the test product provided a conduction-heated material that was relatively homogeneous and susceptible to Maillard browning. Conduction-heated foods would be expected to show quality attribute benefits of the cans because of the very slow heat penetration rate through such products. In this project, the effect of heating time and heating temperature on surface color change of MC was evaluated. The objective was to determine whether the surface quality of MC would be improved or process times decreased by using the optimum CRT and VRT processes. The objectives for this research project were: 1) . to consider if first order reaction kinetics could be used to describe the thermally induced surface color changes in MC product processed in cans by using different temperatures. 2) . to use the Retort Program and R C O program to select the optimum CRT and VRT processes for MC. 3) . to compare the surface cook values and process times for MC by using the optimum C R T and VRT processes with the same sterilization value (F 0). 4) . to confirm the results for surface color parameters, surface cook values for MC by using an actual steam retort. 5) . to determine whether the optimum VRT processes would decrease the surface cook value or improve surface quality or decrease process time compared to the optimum C R T processes. 33 CHAPTER III EXPERIMENTAL METHODS 3.1. Sample Preparation In this study, macaroni and cheese (MC) were used as a sample of a food sensitive to color change and studied its surface color change. This product also was used to study the CRT and VRT processes. According to the instructions of Kraft Dinner (The Original, MACARONI & C H E E S E , Kraft Canada, Don Mills, ON, Canada), macaroni was cooked in salted, boiling water for 9-10 minutes. Water and cheese powder were added to the macaroni in the prescribed amounts. The product was stirred about 3-5 minutes and then used a blender to make MC to a paste. The MC paste was hand-filled into 301 x 106 cans to study the surface color change or 307 x 409 cans for heat penetration tests or other retort tests. The cans were filled to maximum capacity, sealed in a hand-operated sealing machine. The cans were stored in the cooling room until processed. To study the surface color change, for each test, four cans were fitted with thermocouples and connected to the data logger (Model DT 505, Sydney, Australia) to measure the surface temperature of the cans during heat treatment. The surface color changes were studied using MC samples heated at different temperatures between 80 and 125 °C. Temperatures of 80 and 100 °C were achieved in a water bath, and temperatures of 110, 120 and 125 °C were achieved in a saturated steam retort. The processing times varied from 0 to 9 hours. For the test temperatures of 80 or 100 °C, four cans were prepared. A wire was soldered to the surface of each can and then connected with a data-logger, connected to a laptop equipped with the software program Decipher. The data-logger was used to monitor the changes of the surface temperature of the can. Then 34 the cans were put in the water bath at 80 or 100 °C separately and the test began to show the surface temperature changes of the cans according to different heat treatment time. For the test temperatures of 110, 120 or 125 °C, four cans were prepared and put in a horizontal steam retort (FMC Corporation Central Engineering Laboratories, Santa Clara, CA, USA). Then the cans were soldered with wires to monitor the surface temperature of the cans. The wires were connected with data-logger and laptop to obtain the surface temperature changes of the cans and the retort temperature change. 3. 2. Surface Color Changes of MC with Heat Treatment The surface color changes of MC were analyzed as first order reactions with respect to time such that t = D(log L 1 - log L 2 ) . . . . (14) Where t was time at a particular temperature, Li and L 2 were the surface color index (lightness) at time 0 and time t respectively, and D was the time associated with a tenfold surface color change of MC. Surface D value was defined as the time in hours at a specified temperature required to have a 90% of surface color changes. From the plot of the log L versus heating time, the D value was calculated from the negative inverse of the slope. Then D values of different temperatures were calculated from surface color L value changes. The surface z value was the temperature required for a one-log reduction in the log D value. The z value was a measure of the sensitivity of the surface color to change in temperature. The surface color change curve could be described using D and z to predict the effect of the temperature history of the surface color change. Using any two points on a surface color change curve, surface z value and D value at any temperature could be determined. The Minitab software (Version 13, Minitab Inc., State College, PA, USA) was used to evaluate the heating time and heating temperature to affect on the surface color 35 changes. Two-way analysis of variance (ANOVA) was performed to find out the effects of temperature and time on the surface color of MC. 3. 3. Surface Color Measurement of MC MC samples were treated at different temperatures and different times in a water bath or a retort. Cans were taken from the water bath or the retort and cooled for 4 -6 hours. The sample was stored at 4 °C and the surface color measurements were performed within 24 - 48 hours. Following processing, four cans from each run were opened and the surface color of MC was measured using the HunterLab (Hunter Associates Laboratory Inc., Reston, Virginia, USA), with a 1.0 cm diameter aperture. A HunterLab standard black tile and white tile were used to calibrate the colorimeter. Each test used four cans to measure the surface color parameters. Each can (240 ± 5g in weight) was opened and then the MC put into two plastic Petri dishes (Fisher Brand, 100 x 15 mm) to measure the surface color parameters. The sample was measured on a plastic Petri dish covered with a black plastic box as a light-shield. Each dish was turned 90 degree after each measurement. It was determined that four individual readings for each plastic dish were sufficient to produce repeatable results with acceptable standard deviations. Then the average surface color parameter values for one sample (L, a and b value) were calculated by 16-32 readings from the HunterLab. 3. 4. Heat Penetration Test MC sample was packed into 307 x 409 cans, fitted with Ecklund nonprojecting thermocouple fittings, steam retorted and center temperature histories were collected as previously described (Durance and Collins, 1991). All thermocouples needed for experiments were calibrated against the retort thermometer at the same conditions used during processing runs. In each trial, the vent time of 6 min was used and temperatures at the can center and in the retort were recorded every minute using a data-logger. When the can center temperature was 121 °C and it was the same with retort temperature (121 °C), the steam was turned off. The cans were cooled to about 50 °C with cooling water at the end of the process. During processing the 36 retort temperature was maintained to the prescribed temperature (121 °C) within 1 °C. There were three replicate retort runs carried out for each of the processes. The average heating rate index (fh) and the average cooling rate index (fc) of cans were obtained from can center point temperature histories of 12 cans. After correction for heat conduction along the thermocouple fitting (Ecklund, 1956) temperature histories of cans were compared with a computer simulation of the same process (Durance et al., 1997). 3. 5. Determination of Sterilization Value (F0) Sterilization value of the thermal processes was calculated from the time-temperature data by the Improved General Method with the reference temperature of 121.1 °C and a z value of 10 C°. Accumulated bacterial lethality or sterilization value (F 0) at the center of the can was determined by the following equation: Fo =1 L A t = E (10 ( ( T c - 1 2 1 - 1 ) / 1 0 ) )A t , (15) Where T c is the can center temperature. In this project, the sterilization value (F 0) of the thermal process of 6.0 min was used. That is to say that any sterilization process for MC was based on the sterilization value of 6.0 min which determined process time, retort temperature and so on. 3. 6. Retort Program According to the instruction of Retort Program (Durance et al., 1997), the CRT and VRT processes were defined. Parameters as required were entered and the results of the C R T and VRT processes were obtained by Retort Program. Appendix B showed the parameters for the Retort program for the C R T and VRT processes. Figure 5 showed the simplified flow diagram of the Retort Program procedure. Start 1 r Input (Product parameters menu) Cylindrical geometric diameter and height, heating rate index, cool diffusivity modification, initial product temperature Input (Retort heating control menu) Retort temperature, initial retort temperature, surface color z value, reference temperature for surface color, number of ramps, ramp parameters, heat off lethality t Input (Retort cooling control menu) Cooling water temperature, temperature at the can center (for terminating the process) Output Sterilization value, surface cook value, process time Retort program calculation Figure 5. Simplified flow diagram of the Retort program procedure 38 3. 7. RCO Program Random centroid optimization (RCO) program was used to find the optimum VRT process (Dou et al. 1993). For the VRT processes, different ramp times and total times were chosen to obtain the results through the Retort program and R C O program. Here five variable factors were defined. They were the total ramp time (tv) and four variable retort temperature values (RT-i, RT 2 , R T 3 and RT 4 ) . The five factors were optimized by using R C O program. Values for parameters in each experiment were suggested by the R C O program, and then the retort program was used for simulation experiments. The objective here was to determine the values for five factors which minimize the process time or the surface cook value for the VRT process. Then the five variables were adjusted in each Retort Program experiment. Total ramp time was 90 to 160 minutes and the variable retort temperature ranges were 104 to 130 °C. Retort temperature ramps were further constrained such that RTi< RT 2< RT 3< RT 4 . Each ramp was linear with time and extends one quarter of the total ramp time. The first cycle of optimization included 10 random design experiments and 4 centroid experiments, after which the results were mapped. The mapping process is automated by selecting narrower search spaces for subsequent search cycles to be one-third the size of search spaces of the previous search cycle around the best response values (Dou et al., 1993). Then the second and third cycles were continued. Then all results were mapped. The mapping process aids in visualization of the experimental response surface, indicating the result (Nakai et al., 1984). According to Dou et al. (1993), about 30-50 experimental points are needed for optimization if there are five factors for a project. R C O is usually repeated until one gets a response considered adequate. After the first cycle of the R C O program, the approximate position of the optimum was clear. Then the second cycle and third cycle was continued to conduct random search and centroid search. Because of the random search, the different search spaces were selected and this will result in different numbers of experiment. Sometime more experiments and sometime fewer experiments would be needed for the same factors for a project. Mapping of results, revealed points which best approach the optimum point and the R C O program would 39 be stopped. Therefore, about 30 -50 experiments were performed to determine an optimum VRT process for each case. Because R C O is a random search program, every search is different. As well as the main optimization objective, constraints or conditions were included in the experiments. For example, a VRT process that resulted in a minimum process time to achieve the sterilization value F 0 , with the constraint that surface cook value must not exceed a certain value of F s . After first cycle of the R C O program, maps were drawn of the experimental results. The R C O program generates arrows, which appeared at the bottom of maps, showing the assigned locations of the optimum, which were used for computing the approximated slope curves of the response surface. The maps were drawn according to the methods of Nakai (1990), by assuming that the optimum is located at an x value marked by a large shaded arrow at the bottom of each map. A pair of small arrows on maps indicates the boundaries on the x scale between which the optimum may be located (Dou et al., 1993). Through this method, the optimum VRT process would be obtained by the R C O program and the retort program. While R C O maps require experience to interpret correctly, they are very useful for narrowing the boundaries of search areas for each factor between which the optimum is likely to be found. The narrowed field may then be searched more intensely in the next cycle. In my project, sometime about 31 experiments (i.e. 3 cycles) were needed for the optimum result and sometime about 41 or 49 experiments (i.e. 4 cycles) were conducted. The search experiments are mapped and the lowest value of P t or F s was selected from all search experiments. In my project, I would decide to stop the R C O program if the objective values had differences of less than 3 minutes for the last 6 experiments when the three search cycles were finished. Otherwise I should go on the fourth cycle to continue to search the optimum. In no case did I continue beyond the fourth cycle which corresponded to about 41 to 49 experiments. 40 3. 8. Confirmation of the Results for the CRT and VRT Processes in an Actual Steam Retort To confirm the results of the computer simulation of the optimum C R T and VRT processes, retort experiments were done by using the parameters obtained with the Retort Program and R C O program. Heat penetration data were obtained for MC packed and processed under the specified conditions. The data obtained were used to determine process times that were calculated on the same basis for each treatment. In this study, the C R T and VRT processes were chosen to confirm the experiments. An experimental design was employed with four processing temperatures ((113 °C (it had the minimum surface cook value), 121 °C (normal RT for canned foods) and two variable retort temperatures (one had the minimum surface cook value, the other had the minimum process time)) for 12 experimental runs. The cans were connected to thermocouples to monitor the center point temperature of the cans and monitor the surface temperature of the cans (that is the retort temperature). Process times were established based on the sterilization value (F 0) of 6 minutes for MC according to the data from the Retort Program, based on the fh and fc and the dimension of the cans. Temperature readings were recorded at 60 s intervals. During the processing, the retort temperature was maintained to the prescribed temperature within 1 °C. Each experimental run consisted of a 6-min vent time, a fixed constant retort temperature time or a fixed variable RT time and then by 15-20 minutes' cooling time until the center point temperature of the cans reached to about 90 °C after the established process time. Then cans were taken from the retort and went on cooling until 4 - 6 hours later. Following processing, four cans from each run were opened and then the surface color of macaroni and cheese was measured by using HunterLab. The processing conditions, container dimensions and heat treatment used for macaroni and cheese were outlined in Table 1. The 307 x 409 cans (87.3 mm diameter by 116 mm high) were packed with 625 ± 5 g of macaroni and cheese. The cans were processed together in each run. Cans were placed in a metal crate. 41 Table 1. Processing conditions of retort experiment for MC of 307 x 409 cans Process abbreviation CRT =113°C CRT =121 °C The VRT processes Container 307 x 409 307 x 409 307 x 409 Amount of product per can (g) 625 +5 625 ±5 625 ±5 Process temperature (°C) 113 121 104-130 Heating medium Steam Steam Steam Sterilization value (F 0 , min) 6.0 ±0.1 6.0 ±0.1 6.0 ±0.1 Product initial temperature (°C) 20 20 20 Cooling water temperature (°C) 10 10 10 Initial retort temperature (°C) 22.5 22.5 22.5 42 CHAPTER IV RESULTS AND DISCUSSION 4.1. Surface Color Changes of MC The surface color of MC changed from a light yellow color to dark yellow color with an increase in heating time, which corresponded to a decrease in the surface color parameters L, a and b values of MC. As expected, the surface color change of MC was more rapid at higher temperatures as evidenced by steady changes in L, a and b values. The surface color parameters L, a and b values changed at different rates with different temperatures. During heat treatment, L values change from 66.88 to 46.73, a values from 8.64 to 7.13 and b values from 24.72 to 18.93. However, it was observed (Figure 6, 7 and 8) that there were clear differences in the time dependency of these surface color parameters. A N O V A was performed to determine the factors affecting the surface color changes of MC. For MC, two-way A N O V A indicated that L, a and b values of MC surface color were significantly changed by heating temperature (p<0.05). From the results of A N O V A analysis, both heating time and heating temperature had significant effects on the surface color of MC as measured by L, a and b values. Figure 6. Effect of heating time and heating temperature on the surface color Lvalues of MC 4 5 6 7 Heating time (hr) Figure 7. Effect of heating time and heating temperature on the surface color a values of MC 45 4 5 6 7 Heating time (hr) - 0 - b value-80 b value-100 — A — b value-110 —•— b value-120 b value-125 Figure 8. Effect of heating time and heating temperature on the surface color b values of MC 46 The average L value of the uncooked sample was 66.88 ± 0.13, whereas for the other samples, after the different heat treatments, the average L values were between 46.73 and 66.57. Thus the uncooked sample and the cooked samples were all relatively bright samples but brightness decreased as a result of heat treatment, especially with higher heating temperatures. An increase in temperatures from 80 to 125 °C markedly decreased the brightness of MC (figure 6). Two-way A N O V A analysis indicated that L value was significantly affected (p<0.05) by heating time and heating temperature. From figure 7, the final a values, indicative of the redness of MC, varied between 7.13 and 8.61 for the heated samples, whereas the uncooked sample had an average a value of 8.64 ± 0.12. The low magnitude of a value indicated that the development of the red color was small during heat treatment for MC. Two-way A N O V A analysis indicated that a value was also significantly associated with (p<0.05) heating time and heating temperature. From figure 8, the positive b values indicated the yellowness of MC. The uncooked sample had the average b value of 24.72 ± 0.15, which showed the prominence of the yellow color in the sample due to the presence of the cheese powder. The heated samples had final b values in the range of 18.93 and 24.58, showing that these samples had a marked decrease in yellow color. Two-way A N O V A analysis indicated that b value was also significantly affected (p< 0.05) by heating time and heating temperature. From Figure 9 and Figure 10, one can see that L values decreased more than the a and b values with the same conditions. Figure 9 shows the surface color overtime at 100 °C, while figure 10 shows the surface color differences with heating time at temperature 100 °C. MC darkened to a brown color with increased heating. The surface color changes of MC are related to the formation of browning pigments in the MC probably due to 47 Maillard browning. On the basis of high correlations, HunterLab L values were found to be the best predictor of surface color change. L values were chosen as the main surface quality indicator and used to study the surface color change of MC. In this study, L values were considered to determine the surface color changes of MC and a and b values were not considered. Figure 9. Surface color parameters (L, a and b) changes with the heating time (hr) at heating temperature 100 °C Figure 10. Surface color difference versus heating time (hr) at heating temperature 100 °C 50 4. 2. D Values and z Value of MC The surface color changes of MC followed first order reactions. The D value was calculated by the following equation: t = D(log U - l o g L 2) (16) Where t was time at a particular temperature, l_i and L 2 were the surface color index at time 0 and time t respectively, and D was the time associated with a tenfold surface color change of the MC. Table 2 showed the D values at different temperatures. Table 2. D values at different heating temperature (°C) Temperature (°C) Linear equation D values (hr) R ' 80 °C y = -0.0013x+ 1.8245 769.2 R 2 = 0.9902 100 °C y = -0.0071x+ 1.8187 140.8 R2 = 0.9828 110°C y = -0.0172x+ 1.8204 58.1 R 2 = 0.9749 120 °C y = -0.035x + 1.8159 28.6 Rz = 0.9777 125 °C y = -0.0525x+ 1.8196 19.0 R2 = 0.988 52 Figure 11 showed the effect of heating time on the log L of MC at 80, 100, 110, 120 and 125 °C. The D values with different temperatures were obtained from Figure 11. We found that the correlation coefficients (R 2) of all regression models were larger than 0.97, meaning there were good agreements between the model-predicted values and experimental values. Thus, the kinetic models of first order reactions were assumed to adequately describe the surface color changes of MC during heat treatment. Figure 11 showed that higher temperature had the lower D values. That was to say that the higher temperature had more effect on the surface color. The surface z value was the temperature required for a one-log reduction in the log D value. Figure 12 showed that the surface z value of MC through the log D values at different heating temperatures. From the linear equation y=-0.036x+5.7611, z value was calculated and the surface z was 28 C° for MC. 53 O) O 2 3 4 5 6 7 8 9 Heating time (hr) R2 = 0.9902 R2 = 0.9837 A Log L (80 C) o Log L (100 C) oLogL(110C) x Log L (120 C) • Log L (125 C) Figure 11. Effect of heating time on the log L of MC at different heating temperatures (80, 100, 110, 120 and 125 °C). Figure 12. Effect of heating temperature on the log D values of MC 55 4. 3. Heat Penetration Parameters The average heating rate index (fh) and the average cooling rate index (fc) were determined from the heat penetration data for MC in 307 x 409 cans by retort experiments in three process runs (12 cans). The calculation of the heating rate index (fh) and the cooling rate index (fc) was performed using the heat penetration curves. The linear portion of the log (T r-T c) versus time curve was chosen and the heating rate index (fh) was calculated by linear regression. Similarly, the linear portion of the log (T c-T w) versus time curve was chosen and the cooling rate index (fc) was calculated. The average heating rate index (fh) and the average cooling rate index (fc) of MC through heat penetration test were presented in Table 3. Note that the fc was greater than fh, indicating that the MC heated faster than it cooled. 56 Table 3. The average heating rate index and the average cooling rate index for MC obtained from heat penetration tests in three process runs (12 cans). CRT= f h fh fh Mean fc fc fc Mean 121 °C (4 (4 (4 value (4 (4 (4 value cans) cans) cans) f h cans) cans) cans) fc 307x409 59 58 57 58.0 ± 76.9 77.5 77.5 77.3 ± cans 1.0 0.4 57 4. 4. Comparison of Can Center Temperatures for the CRT Processes by Retort Program and Retort Experiment The finite difference model for cylindrical containers (Retort Program, Durance et al., 1997) was tested against experimental data obtained by measuring the can center temperature. In Figure 13, the can center temperature predictions of the finite difference model were compared with the can center temperature measured during the processing of 307 x 409 cans for MC by using retort experiment. A good agreement between predicted and experimental temperatures was observed in the heating phase of the process. There were small differences observed in the cooling phase of the process. There were some difficulties in controlling the conditions during the cooling phase of the phase. This was because water used for cooling was not controlled could change significantly between different processes. Figure 13 compared the two kinds of can center temperature histories; the results were very similar to each other. From this figure it was concluded that the Retort program was sufficiently accurate for process optimization purposes. 58 125 o 100 (0 o. E a> +•» c a> O 75 50 25 20 40 60 80 100 Heating time (min) o Can #1 - Model 120 Figure 13. Comparison of the can center temperature histories of MC (retort experiment and Retort Program), C R T =121 °C. 59 4. 5. Rho, Retort Temperature and Unaccomplished Temperature Rho is the fraction of sterilization value (F 0), which occurs during the heating side of thermal processing. Prior knowledge of Rho greatly reduces the number of experiments required for computer optimization of the VRT processes. When Rho is known, the experimenter knows when to end heating in the simulation and achieve the target sterilization value (F 0) at the end of the cooling phase (Durance et al., 1997). In this project, one can size and one product were used and only the effect of unaccomplished temperature and retort temperature on Rho was considered. Through the Retort program, Rho values were obtained from retort temperature and unaccomplished temperature. Figure 14 and Figure 15 showed the relationships among Rho, retort temperature (111 to 121 °C) and final unaccomplished temperature (3 < g < 15). From the Retort Program, the results were concluded that Rho decreased with increasing final unaccomplished temperature, but Rho was only slightly changed with increasing retort temperature. Therefore for simplicity, Rho was predicted from unaccomplished temperature alone. In addition, one-way A N O V A analysis indicated that the unaccomplished temperature was a significant factor affecting the Rho (p<0.001) and retort temperature was not a significant factor affecting the Rho (p>0.001). 60 0.55 CRT=111 oC y = 0.679e° 1 0 6 2 x = 0.996 3 4 5 6 7 8 9 10 11 12 13 14 15 g values (C) CRT=116oC y = 0.6896e° 1 1 0 8 x R2 = 0.996 CRT=121 oC y = 0.654e 0 0 9 9 3 x R2 = 0.994 o Tr=111 C • Tr=116 C X Tr=121 C •Expon. (Tr=111 C) Expon. (Tr=116C) Expon. (Tr=121 C) Figure 14. The relationship of Rho and final unaccomplished temperature (g) 61 o .c 0.55 0.45 0.35 0.25 0.15 0.05 111 113 115 117 Retort temperature (C) 119 121 -—g=l5 C ^^g=14C _+_g=13 c _^g=12C -«-g=11 C g=10 C - » - g = 9 C - « - g = 8 C -x-g=7 C —x-g=6 C ^-g=5C ^-g=4C -o-g=3 C Figure 15. The relationship of Rho and retort temperature. 62 4. 6. Surface Cook Values (Fs) of the CRT and VRT processes Surface cook values (F s) for MC were estimated from the equation F s = Z (10 ( T s " 1 2 1 1 ) / z ) At. Here T s was the surface temperature of the can. The surface temperature of MC was assumed to equal to the can surface temperature since the thermal conductivity of the steel can was so large as not to provide any significant insulation of MC from the steam. Table 4 and Figure 16 compared surface cook values of MC for C R T processes at different surface z values. The optimum RT varied from 111 °C to 113 °C, depending on different surface z values. The bold values of table 3 were the minimum surface cook value for the optimum C R T process. In each case, the CRT process had the same sterilization value (F 0) of 6.0 min. 63 Table 4. The C R T processes at different surface z values in terms of surface cook value (Fs) with the same F 0 =6 min CRT, Tr (°C) Fo (min) Pt (min) Fs (z=24) Fs (z=26) Fs (z=28) Fs (z=30) Fs (z=32) 111 5.91 148.1 55.3 59.7 63.7 67.5 71.0 113 5.99 124.8 56.3 59.9 63.2 66.3 68.9 115 5.95 108.0 58.8 61.7 64.3 66.7 69.0 117 5.97 96.0 63.2 65.4 67.3 69.0 70.6 118 5.94 91.0 65.9 67.7 69.2 70.6 71.9 119 5.96 87.2 68.8 70.1 71.4 72.5 73.4 1.21 6.01 80.0 76.3 76.7 77.1 77.4 77.7 123 5.95 74.0 85.3 84.5 83.9 83.3 82.9 125 5.95 69.1 96.3 94.0 92.1 90.6 89.3 64 Figure 16. The C R T processes at different z values in terms of surface cook values (F s) with the same F0=6 min 65 To get the minimum surface cook value of MC in 307 x 409 cans, the optimum CRT process of 113 °C was found through Retort Program when z value of MC equaled to 28 C°. When the F s of CRT process of 113 °C was 63.2 min and F 0 was 5.99 min, its process time (P t) was 124.8 min plus vent time. R C O was employed to determine the optimum VRT processes for MC that minimized Fs based on the different z values of MC. The optimum VRT process reduced the surface cook value (F s) while maintaining the F 0 value very close to 6.0 min and maintaining the P t no higher than the P t of the CRT process. When z value was 28 C°, 49 computer simulation experiments were used to complete the research for the optimum VRT process and the results were summarized in Table 5. By using the same methods, the optimum VRT processes were determined when z was 24, 26, 30, 32 C° (Table 6, 7, 8, 9). 66 Table 5. Optimization experiments for VRT processes to minimize F s with Pt < 124.8 min and 5.9 < F 0 < 6.1 min (z= 28 C°), the best result was the bold value (F s = 56.2 min) Trial Ramp time RT! R T 2 R T 3 R T 4 F 0 Fs(min) Pt(min) 1 153.7 108.9 112.9 121.7 128.1 5.96 59.0 117.1 2 135.3 105.1 11.7 123.2 125.1 6.03 59.3 103.3 3 101.9 107.4 110.6 123.6 127.9 6.05 64.0 96.3 4 132.0 109.6 119.2 123.8 128.3 6.04 61.2 96.7 5 148.9 109.0 116.2 123.0 129.6 5.93 58.8 107.6 6 98.2 107.8 112.7 120.4 129.3 5.99 64.2 95.7 7 143.2 105.4 113.1 121.9 127.7 5.93 58.3 114.5 8 96.7 106.5 117.1 121.2 128.9 6.02 64.3 91.9 9 147.7 106.5 112.6 124.7 126.4 6.08 60.3 113.9 10 153.2 106.3 113.4 122.9 126.4 6.06 58.9 116.8 11 146.9 106.9 114.7 122.5 127.4 5.93 58.4 111.7 12 149.4 107.2 113.6 122.8 127.6 5.98 58.9 114.2 13 145.7 106.5 114.6 123.1 127.0 6.01 58.9 111.3 14 145.8 107.0 114.5 122.9 127.4 5.97 58.8 111.2 15 121.8 104.5 115.7 122.9 125.9 5.98 59.8 101.8 16 149.3 104.3 110.0 124.7 125.7 5.97 59.1 119.5 17 124.5 107.7 113.5 123.5 129.8 5.92 60.9 103.0 18 121.7 105.7 115.7 125 127.7 6.07 61.9 99.1 19 127.6 106.6 112.2 120.8 129.0 5.98 59.1 109.8 20 146.1 107.0 114.6 122.7 127.8 6.01 58.8 111.5 21 146.8 107.1 114.4 122.6 127.9 5.92 58.5 111.9 22 146.2 106.6 114.1 122.7 127.4 6.02 58.8 112.6 23 146.8 107.0 114.4 122.7 127.9 5.96 58.6 112.0 24 136.2 104.7 115.9 121.5 129.4 6.03 58.3 108.4 25 140.2 106.5 116.4 121.3 125.3 6.09 58.4 108.2 26 133.5 107.4 112.9 123.0 129.8 6.01 60.2 108.5 27 133.4 104.5 111.3 120.7 127.7 6.02 58.7 115.0 28 148.5 107.1 111.0 122.6 129.2 6.02 59.9 118.5 29 143.2 104.4 110.1 123.0 126.0 5.96 58.2 118.4 30 141.9 105.6 114.0 122.1 127.2 5.93 58.2 112.2 31 141.9 105.6 114.0 122.1 127.3 5.93 58.2 112.2 32 143.3 105.7 113.6 122.3 127.7 5.94 58.4 113.1 33 142.7 105.9 114.3 122.2 127.2 5.91 58.2 111.7 34 130.9 106.6 110.0 120.5 125.3 6.06 59.3 115.1 35 146.5 104.4 117.6 122.3 126.9 5.97 56.2 109.8 36 134.4 105.0 116.7 124.0 127.0 6.04 59.8 103.9 37 143.3 105.2 114.0 122.3 126.9 5.94 58.2 112.7 38 141.8 105.3 115.2 122.0 127.6 5.94 58.0 110.6 39 142.1 105.0 114.4 122.2 127.4 5.98 58.2 111.9 40 142.1 105.0 114.4 122.2 127.3 5.97 58.2 111.9 41 145.6 106.3 116.3 120.9 125.7 5.99 57.6 110.5 42 149.9 106.1 117.2 122.3 125.1 6.04 58.1 109.2 43 142.3 104.4 116.7 123.8 126.9 6.03 57.5 107.6 44 145.1 105.7 119.6 121.3 127.4 5.96 58.0 103.9 45 144.3 105.2 117.1 122.1 126.9 5.98 57.9 108.2 46 145.9 105.4 117.5 122.1 126.4 5.99 57.9 107.8 47 145.2 105.3 116.6 122.3 126.4 5.96 57.9 109.0 48 145.1 105.2 117.3 122.4 126.8 5.97 57.9 109.1 49 145.8 105.5 117.2 121.7 126.5 5.93 57.6 108.5 68 Table 6. Optimization experiments for VRT processes to minimize F s with Pt < 148.1 min and 5.9 < F 0 < 6.1 min (z = 24 C°), the result was the bold value (F s = 50.4 min) Trial Ramp time RTi R T 2 R T 3 R T 4 F 0 (min) F s (min) Pt (min) 1 107.7 108.5 111.6 124.1 128.2 6.07 61.4 97.1 2 159.3 107.1 115.2 122.7 125.6 5.93 52.5 114.8 3 125.9 106.8 119.1 122.0 127.0 6.02 \ 56.4 98.4 4 156.0 106.1 112.8 122.7 126.3 6.07 53.3 121.2 5 102.9 108.2 111.7 120.6 129.8 5.95 60.8 98.1 6 122.2 107.9 119.7 121.9 128.1 5.92 57.2 95.6 7 97.2 104.5 115.8 121.8 128.3 5.98 61.6 93.1 8 120.7 10.07 117.5 123.4 125.3 6.06 57.8 97.7 9 101.7 109.9 117.6 123.7 128.0 6.07 63.1 89.6 10 107.2 104.5 117.3 124.0 128.4 6.04 61.2 93.5 11 136.8 107.0 116.9 122.5 126.4 5.93 54.5 104.3 12 133.2 107.2 115.7 122.0 127.4 5.91 54.4 105.3 13 132.9 107.0 115.3 122.3 126.8 6.00 54.9 105.9 14 132.2 107.3 115.4 122.2 127.0 5.99 54.9 105.5 15 130.0 104.7 117.4 121.6 127.7 5.93 54.4 103.6 16 126.7 104.6 111.4 121.3 125.6 5.96 54.2 111.0 17 157.9 109.2 118.0 123.5 126.8 6.04 54.2 106.5 18 126.3 104.7 114.2 122.9 127.4 6.01 55.9 105.3 19 159.2 107.3 112.1 123.5 125.0 5.94 53.8 120.1 20 129.8 108.8 111.7 123.1 127.7 5.92 56.4 107.6 21 137.8 104.4 119.6 124.2 126.4 6.04 55.9 102.6 22 151.8 106.9 113.9 122.7 125.9 5.97 53.4 114.8 23 152.5 106.9 115.1 122.8 126.3 5.91 53.4 111.3 24 146.2 106.0 113.8 122.3 126.0 6.04 53.6 114.2 25 146.0 106.3 115.0 122.3 126.4 6.06 53.6 111.8 26 131.0 104.2 115.4 120.4 126.8 5.93 52.1 109.9 27 158.6 105.4 115.8 120.2 127.4 5.96 51.2 117.1 28 151.9 109.4 112.8 121.9 127.1 5.95 53.8 11.6.1 29 133.8 109.4 117.5 123.6 125.8 6.09 56.8 100.3 30 158.2 107.9 117.2 121.5 126.7 5.94 52.5 100.4 31 152.9 106.3 116.5 121.9 127.8 5.99 52.7 111.5 32 152.0 106.2 116.0 121.3 126.8 5.91 52.1 112.6 33 152.6 106.1 115.3 121.5 126.6 5.92 52.1 114.2 34 151.3 106.0 115.5 121.3 127.0 5.94 52.2 113.6 35 151.6 105.8 115.1 121.6 126.8 5.98 52.4 114.5 36 151.5 104.2 112.1 120.1 126.0 5.95 50.4 123.8 37 158.5 104.6 116.6 122.8 128.9 6.06 52.7 114.0 38 152.4 105.4 115.0 123.9 128.4 6.01 53.8 113.2 39 149.1 105.2 114.9 120.7 126.7 5.93 51.8 115.3 40 153.2 105.6 115.0 120.9 126.8 5.94 51.7 116.0 41 149.0 105.2 114.8 120.7 126.7 6.02 52.1 115.8 42 148.9 105.2 115.0 120.7 126.8 6.04 52.2 115.3 43 150.9 106.1 113.8 122.6 128.4 6.02 53.4 115.5 44 148.2 104.6 116.3 111.1 126.7 6.06 53.0 111.7 45 152.3 105.1 114.5 120.5 126.7 5.93 51.5 117.6 46 153.0 105.3 114.6 120.7 126.7 5.95 51.7 117.3 47 153.0 105.3 114.6 120.7 126.7 5.96 51.7 117.2 48 152.2 105.2 114.6 120.6 126.7 5.96 51.7 117.1 70 Table 7. Optimization experiments for VRT processes to minimize F s with Pt < 148.1 min and 5.9 < F 0 < 6.1 min (z = 26 C°), the result was the bold value (F s = 53.6 min) Trial Ramp time R T T R T 2 R T 3 R T 4 F 0 (min) F s (min) Pt (min) 1 106.7 108.4 111.5 124.1 128.1 5.97 62.4 96.7 2 158.3 107.0 115.1 122.6 125.5 5.93 55.3 115.0 3 125.0 106.7 119.0 121.9 127.0 6.08 58.4 98.5 4 155.1 106.0 112.7 122.6 126.3 6.06 56.3 119.5 5 102.0 108.1 111.6 120.6 129.7 6.03 62.5 98.0 6 121.2 107.8 119.6 121.8 12.08 6.00 59.1 95.9 7 96.3 104.4 115.6 121.7 128.2 6.01 62.4 93.8 8 119.7 106.9 117.4 123.3 125.2 6.09 59.6 97.7 9 100.8 110.0 117.5 123.6 127.9 6.05 64.1 89.6 10 106.3 104.5 117.2 123.9 128.4 5.97 62.1 93.3 11 135.9 106.9 116.7 122.5 126.4 5.94 56.8 104.5 12 133.2 106.4 116.7 122.6 127.0 6.03 57.3 104.0 13 132.9 106.2 116.3 122.9 126.5 6.04 57.4 104.3 14 132.1 106.4 116.4 122.9 126.7 6.04 57.6 103.7 15 130.5 104.1 115.9 122.4 125.3 6.01 55.7 106.8 16 147.3 109.7 118.1 124.2 125.2 6.05 57.9 102.5 17 133.7 106.3 113.4 121.1 125.5 5.95 56.1 110.7 18 137.7 108.9 113.7 122.9 128.4 6.04 57.9 108.4 19 112.6 105.7 115.1 122.4 127.8 6.04 59.8 98.9 20 151.2 106.4 11.6.6 121.4 126.7 5.91 54.9 110.9 21 148.7 105.9 115.8 124.4 126.7 6.09 57.0 109.9 22 142.8 107.3 112.4 120.3 127.7 5.92 55.8 116.4 23 143.3 106.2 114.7 121.6 126.5 6.00 55.8 111.9 24 147.6 106.1 114.5 121.9 126.6 5.91 55.4 113.4 25 145.8 106.0 114.7 122.0 126.2 6.04 55.9 112.6 26 148.2 106.6 114.0 121.6 126.7 5.98 55.6 114.8 27 141.8 105.4 115.7 123.1 129.8 6.06 56.7 108.9 28 147.8 107.5 115.2 124.0 125.8 5.97 56.9 109.4 29 137.9 108.7 118.7 120.2 125.3 5.94 56.3 102.2 30 137.7 105.1 114.1 121.4 128.1 6.05 56.1 111.7 31 130.3 108.3 117.9 123.0 129.6 6.03 58.5 99.7 32 152.4 105.0 118.8 123.1 126.4 6.02 56.2 107.1 33 147.2 106.0 115.2 122.0 126.5 5.98 55.6 112.2 34 149.7 106.5 115.0 121.8 126.7 5.92 55.3 113.1 35 146.2 106.0 115.4 122.0 126.5 5.96 55.6 111.3 36 146.3 106.1 115.3 121.9 126.5 5.96 55.5 111.7 37 154.7 107.5 118.8 121.2 125.2 6.03 55.8 106.8 38 153.9 106.8 114.2 123.5 129.9 5.93 56.1 114.4 39 150.6 106.4 115.3 121.9 126.8 5.98 55.4 123.1 40 150.6 106.4 115.3 121.9 126.7 5.98 55.4 113.1 41 150.3 106.4 115.5 121.9 126.7 6.01 55.5 112.6 42 148.2 106.2 115.4 121.8 126.9 6.07 55.7 112.7 43 145.1 104.6 117.3 120.9 129.3 5.97 54.9 109.6 44 153.3 104.4 112.3 120.8 126.5 5.97 53.6 123.0 45 151.5 105.8 115.3 121.5 127.3 5.91 54.8 114.0 46 150.0 105.7 115.3 121.4 127.5 5.94 55.0 113.7 47 151.7 105.8 115.3 121.5 127.3 5.91 54.8 114.0 48 151.4 105.8 115.0 121.6 127.0 5.95 55.1 114.6 Table 8. Optimization experiments for VRT processes to minimize F s with P t < 124.8 min and 5.9 < F 0 < 6.1 min (z = 30 C°), the result was the bold value (Fs = 59.6 min) Trial Ramp time RTi R T 2 R T 3 R T 4 F 0 (min) Fs(min) Pt (min) 1 147.9 108.4 114.0 12.01 127.6 5.93 60.8 113.9 2 129.5 104.6 111.1 123.4 128.6 6.00 62.1 109.7 3 96.1 106.9 119.8 124.9 125.7 6.05 67.1 86.3 4 126.2 109.1 111.2 124.2 126.9 6.00 63.6 105.5 5 143.1 108.5 117.0 122.7 125.7 6.08 61.3 105.3 6 92.4 107.3 119.5 120.9 129.4 6.03 67.0 88.4 7 137.4 104.9 119.3 121.1 127.0 5.93 59.8 103.1 8 90.9 106.0 112.4 123.1 129.2 5.97 67.1 90.4 9 141.9 106.0 113.1 120.9 129.1 5.94 60.6 114.9 10 147.4 105.8 113.7 121.3 126.8 5.91 60.2 115.8 11 143.5 106.7 115.4 121.4 127.2 6.08 60.6 110.9 12 140.8 106.0 114.3 121.6 127.8 6.06 60.7 111.9 13 139.9 106.0 114.9 121.9 127.4 6.05 60.7 110.3 14 141.1 106.4 115.0 121.9 127.2 5.97 60.5 110.1 15 115.9 105.0 118.8 124.3 127.3 5.98 63.3 94.5 16 127.3 106.6 116.7 124.0 128.7 5.94 62.0 100.2 17 144.9 107.2 119.4 123.3 129.6 6.00 61.1 101.9 18 112.2 106.4 116.7 120.7 125.5 5.95 61.6 99.0 19 119.7 108.8 117.2 124.9 127.8 6.07 64.1 95.5 20 138.4 104.6 114.0 120.7 125.1 5.97 59.9 113.3 21 141.6 105.7 115.5 121.3 126.7 5.93 59.9 110.4 22 141.2 105.6 115.0 121.2 127.0 5.95 60.0 111.4 23 141.7 105.6 115.1 121.1 127.0 5.96 60.0 111.5 24 140.5 105.7 115.4 121.2 127.1 6.02 60.2 110.5 25 145.7 106.9 119.5 120.9 126.7 6.04 60.2 103.7 26 121.6 106.6 119.2 121.8 127.5 6.09 61.6 98.6 27 141.5 106.7 118.3 123.2 125.3 6.05 61.0 103.4 28 126.2 106.6 115.8 121.0 128.0 5.97 60.9 104.4 29 140.1 105.3 115.8 121.1 126.6 5.92 59.7 109.8 30 140.9 105.5 116.7 12110 126.5 5.91 59.6 108.4 31 141.5 105.7 116.9 121.1 126.9 5.94 59.7 108.1 32 149.9 106.8 115.1 122.3 125.1 5.98 60.5 1.12.5 33 134.8 105.5 114.4 123.9 126.9 6.01 61.6 107.0 34 140.2 105.4 117.1 121.1 126.7 6.07 60.1 107.8 35 141.0 105.6 116.3 121.1 126.6 5.98 59.9 109.2 36 140.5 105.5 117.0 121.1 126.7 6.05 60.0 108.0 37 140.2 105.4 116.8 121.1 126.7 6.04 60.0 108.3 38 148.4 107.4 117.8 123.2 127.7 6.04 60.8 105.9 39 140.2 105.4 117.1 121.1 126.7 6.07 60.1 107.8 40 140.9 105.5 116.5 121.1 126.6 6.00 59.9 108.9 41 140.3 105.5 117.2 121.1 126.7 6.08 60.1 107.7 42 140.1 105.4 117.0 121.1 126.7 6.07 60.1 108.0 Table 9. Optimization experiments for VRT processes to minimize F s with P t < 124.8 min and 5.9 < F 0 < 6.1 min (z = 32 C°), the result was the bold value (F s = 61.2 min) Trial Ramp time RT-, R T 2 R T 3 R T 4 F 0 (min) F s (min) Pt (min) 1 94.7 106.0 114.9 120.7 126.4 6.05 65.4 94.2 2 148.7 107.4 118.5 124.3 127.4 6.06 63.0 104.1 3 145.2 107.7 112.3 123.6 129.4 5.91 64.0 113.3 4 137.0 109.4 116.0 124.3 125.7 5.94 63.9 102.7 5 91.7 105.9 114.9 122.2 129.5 5.96 67.5 90.0 6 157.6 106.1 112.9 123.5 128.1 5.94 63.5 118.8 7 118.3 107.2 119.0 123.4 125.8 6.03 64.3 94.9 8 147.7 108.3 110.7 125.0 127.9 6.05 65.8 115.4 9 110.8 107.7 110.8 120.3 127.8 6.01 64.6 103.9 10 118.8 105.1 110.5 120.6 125.5 5.98 63.0 109.1 11 141.5 107.1 114.1 123.3 127.2 5.94 63.3 109.8 12 136.1 107.0 115.4 123.2 126.5 6.03 63.3 106.0 13 137.7 106.7 114.6 123.1 127.2 5.95 63.1 107.9 14 133.6 107.4 115.3 123.3 126.7 6.00 63.4 104.9 15 125.9 107.6 111.9 123.6 125.4 5.97 64.2 105.9 1 6 1 4 4 . 3 1 0 4 . 4 1 1 1 . 7 1 2 0 . 9 1 2 6 . 5 5 . 9 4 6 1 . 2 1 1 9 . 7 17 133.8 108.2 110.5 121.4 128.4 5.94 64.3 113.1 18 106.2 107.6 115.7 120.7 126.5 5.98 64.3 97.3 19 93.6 106.6 110.5 122.7 128.2 6.03 66.9 93.5 20 90.9 104.6 114.0 123.8 128.1 5.99 67.2 89.6 21 137.1 106.1 114.1 122.4 126.6 5.95 62.8 109.6 22 138.2 106.1 113.9 122.4 126.8 5.91 62.7 110.1 23 137.9 106.2 114.0 122.5 126.6 5.95 62.9 109.9 24 135.7 106.1 113.3 122.2 126.6 5.97 63.0 110.4 25 155.2 108.0 112.5 122.8 126.8 5.93 64.0 117.9 26 121.4 105.9 113.0 120.5 127.3 5.95 63.0 106.9 27 142.0 104.9 114.9 120.5 126.4 5.98 61.9 112.2 28 141.7 106.0 113.8 122.2 128.6 6.05 63.1 112.3 29 139.9 105.5 113.7 121.7 126.6 5.93 62.5 112.5 30 136.6 105.5 113.5 121.4 126.7 5.99 62.7 111.8 31 136.7 105.5 113.5 121.4 126.7 5.97 62.6 111.9 32 136.5 105.5 113.5 121.4 126.7 5.99 62.7 111.8 76 Table 10 compared the results obtained for the optimization of the VRT processes, considering the minimum surface cook values for each surface z value. Table 10 showed that when the surface z values increased, the minimum surface cook value of its optimum VRT process also increased. Figure 17 showed the best VRT processes to yield minimum surface cook values at different surface z values. 77 Table 10. Comparison of the optimum VRT processes with minimum F s and Pt in term of different z values z value Ramp time RTi R T 2 R T 3 R T 4 F 0 (min) F s (min) P t (min) 24 151.5 104.2 112.1 120.1 126.0 5.95 50.4 123.8 26 153.3 104.4 112.3 120.8 126.5 5.97 53.6 123.0 28 146.5 104.4 117.6 122.3 126.9 5.97 56.2 109.8 30 140.9 105.5 116.7 121.0 126.5 5.91 59.6 108.4 32 144.3 104.4 111.7 120.9 126.5 5.94 61.2 119.7 Figure 17. The optimum VRT processes to yield the minimum F s of MC in terms of different z values. 79 Table 11 compared the surface cook values of MC for the CRT and VRT processes at different z values. Here it was found that the optimum VRT processes had smaller surface cook values than the optimum C R T processes. For example, when surface z value was 28 C°and C R T 113 °C was chosen as the optimum C R T and its minimum surface cook value was 63.2 min, but the surface cook value of the optimum VRT process is only 56.2 min. At the same time, comparison of the optimum CRT and VRT processes showed that the process times of the VRT processes were smaller than those of the C R T processes. The optimum VRT process decreased surface cook value 11.1 % relative to the optimum CRT process with the same sterilization value (F 0 equaled to 6.0 min). Of course, different surface z values resulted in different surface cook values for the C R T and VRT processes. From Table 11, it was found that when the surface z value increased, its surface cook values (F s) of the CRT and VRT processes all increased. For the C R T processes, when z value increased from 24 to 32, their optimum RT increased from 111 to 113 °C and their surface cook values increased from 55.3 to 68.9 min. The surface cook values of the CRT processes increased 13.6 min with an increase of z value from 24 to 32. For the VRT process, when z value increased from 24 to 32, its surface cook value increased from 50.4 to 61.2 min. The surface cook values of the VRT processes increased 10.8 min with an increase of z value from 24 to 32. Table 11 showed that when the z value was higher, the surface cook values of the VRT processes decreased more from 4.9 to 7.7 min and their surface cook values decreased from 8.9 to 11.2 % compared with the surface cook values of the CRT processes. From Table 11, it was found the process times of the C R T processes were from 148.1 to 124.8 min but the process times of the VRT processes were from 123.8 to 108.4 min. The process times of the VRT processes decreased from 25.1 to 5.1 min. Then the process times of the VRT processes all were shorter than those of the C R T processes. When z values decreased, the process times of the VRT processes decreased more and when z values increased, the process times of the VRT processes decreased less (Table 11). Table 11. Comparison of F S of MC for the optimum C R T and VRT processes in terms of different z values ( F S , min) z = z = z = z = z = / 24 C° 26 C° 28 C° 30 C° 32 C° 55.3 59.7 63.2 66.3 68.9 (CRT (CRT (CRT (CRT (CRT F S for the best C R T process =111) =111) =113) =113) =113) F S for the best VRT process 50.4 53.6 56.2 59.6 61.2 Decrease surface cook value (min, F S , C R T - F S , V R T ) 4.9 6.1 7.0 6.7 7.7 Decrease surface cook value (%) 8.9 10.2 11.1 10.1 11.2 148.1 148.1 124.8 124.8 124.8 (CRT (CRT (CRT (CRT (CRT P t for the best CRT process =111) =111) =113) =113) =113) P t for the best VRT process 123.8 123.0 109.8 108.4 119.7 Decrease P t(min, Pt, C R T -P t , V R T ) 24.3 25.1 15.0 16.4 5.1 Decrease P t (%) 16.4 17.0 12.0 13.1 4.1 81 4. 7. Process Time of the CRT and VRT processes A significant advantage of the VRT processes was the reduction in process time while maintaining product surface quality similar to that of the C R T processes. MC in 307 x 409 cans would typically be processed at 113 °C since this gave the best surface quality for a CRT process when z value was 28 C°. When a CRT process of 113 °C was chosen, its sterilization value (F 0) of 5.99 min and F s, z=28c°of 63.2 min, the process time (P t) of 124.8 min plus vent time was obtained. R C O program was applied to find the optimum VRT process that reduced the process time (P t), maintained the F 0 value very close to 6.0 min and maintained the F s no higher than the F s of the best CRT process. 33 computer simulation experiments were used to search for the best VRT process and the results were summarized in Table 12. By using the same methods, the optimum VRT processes were determined when z was 24, 26, 30, 32 (Table 13, 14, 15, 16). 82 Table 12. Optimization experiments for VRT processes to minimize P t with F s < 63.2 min and 5.9 < F 0 < 6.1 min (z=28 C°), the best result was the bold value (P t = 95.3 min) Trial Ramp time RTi RT 2 ) R T 3 R T 4 Fo(min) Fs (min) Pt (min) 1 145.8 107.1 118.2 122.2 125.2 6.04 58.5 105.3 2 142.4 107.5 116.8 124.2 128.8 6.06 59.7 104.9 3 134.1 109.2 118.2 120.5 129.2 5.93 58.8 101.5 4 158.8 105.7 114.1 124.3 126.7 6.00 58.7 116.8 5 154.8 105.8 116.6 122.9 129.2 6.07 58.1 111.8 6 115.4 106.9 116.4 120.6 125.3 5.98 59.6 100.5 7 144.9 108.1 119.5 122.7 129.7 6.09 59.5 101.5 8 108.0 107.5 110.2 122.6 126.5 6.04 62.0 100.8 9 115.9 104.9 110.8 120.3 126.2 5.96 59.4 107.7 10 118.8 107.5 115.1 121.5 127.9 6.03 60.3 101.6 11 120.5 107.1 114.8 122.2 127.8 5.96 60.2 101.8 12 118.3 107.4 114.6 121.8 127.7 5.92 60.2 101.5 13 125.7 107.5 116.4 121.8 128.3 5.97 59.6 101.9 14 119.1 105.8 1.10.3 122.6 129.2 5.99 61.3 105.9 15 133.3 107.2 110.1 123.8 125.5 6.01 60.8 111.2 16 105.4 106.3 114.9 120.3 128.0 5.94 61.4 98.2 17 114.4 105.8 114.1 123.7 125.6 6.06 61.3 99.5 18 123.9 105 114.9 121.6 127.7 5.95 59.1 104.8 19 118.9 107.4 112.4 122.8 127.6 6.07 61.2 103.2 20 97.2 104.1 113.9 122.2 125.3 6.00 61.3 95.9 21 104.9 105.8 114.1 121.6 126.5 6.05 61.5 97.9 22 103.4 105.9 112.9 122.0 126.8 6.05 62.0 97.7 23 103.6 106.1 113.3 121.4 126.7 6.02 61.6 98.1 24 105.4 106.0 113.2 122.0 126.2 6.02 61.4 98.3 25 99.9 105.0 110.7 123.1 128.1 5.97 63.5 96.3 26 127.7 104.4 114.7 121.4 129.5 5.98 57.8 107.8 27 103.5 104.4 110.4 120.5 125.9 6.07 60.3 102.7 28 129.0 106.4 110.3 123.8 128.2 6.00 61.1 109.2 29 98.1 105.3 112.6 122.3 127.5 6.02 63.3 95.2 30 98.8 105.5 113.1 122.1 127.3 5.99 62.9 95.3 31 99.3 105.6 112.4 122.3 127.8 6.06 63.4 95.7 32 105.5 104.5 111.3 121.2 129.5 5.98 62.4 100.3 33 111.5 107.5 112.7 121.5 129.6 6.03 61.9 100.9 84 Table 13. Optimization experiments for VRT processes to minimize Pt with F s < 55.3 min and 5.9 < F 0 < 6.1 min (z = 24 C°), the best result was the bold value (Pt = 106.9 min) Trial Ramp time RT^ R T 2 R T 3 R T 4 F 0 (min) Fs(min) Pt (min) 1 97.4 105.5 117.4 120.8 129.1 6.02 61.9 92.5 2 91.6 109.9 115.1 122.6 127.7 6.00 64.2 89.1 3 146.1 107.1 114.6 124.5 127 6.09 55.4 109.9 4 103.8 109.6 114.4 121.4 127.7 5.99 59.7 96.7 5 90.8 105.1 117.0 120.8 125.6 6.02 61.3 91.6 6 148.4 106.7 116.4 124.8 126.9 6.09 55.1 107.9 7 97.1 104.6 110.8 122.9 126.8 6.07 61.4 95.9 8 111.0 106.6 115.0 123.7 128.3 5.99 60.4 95.4 9 157.8 105 119.7 123.8 128.7 6.06 54.5 106.9 10 157.4 105.5 110.8 123.6 128.9 6.06 54.7 122.7 11 97.6 106.3 115.0 122.2 127.5 6.01 61.3 93.1 12 98.9 107.3 115.6 121.9 127.7 6.09 61.2 93.3 13 96.1 106.9 114.7 121.7 127.4 6.08 61.7 93.2 14 98:8 107.2 114.3 122.3 127.2 6.03 61.0 93.8 15 104.5 105.8 111.2 121.1 129.1 5.91 59.7 99.5 16 123.9 109.4 112.1 123.3 129.5 6.07 58.2 104.3 17 122.0 106.3 117.6 124.9 127.7 6.07 59.3 96.9 18 91.7 105.8 114.5 121.3 129.1 5.99 64.2 91.3 19 102.7 108.6 115.9 120.6 125.1 5.92 57.9 95.8 20 93.8 106.5 115.8 121.5 127.8 5.99 62.4 91.5 21 93.5 106.6 115.7 121.4 127.8 5.98 62.3 91.6 22 93.5 106.8 115.3 121.7 127.4 5.99 62.2 91.6 23 94.9 106.9 115.4 121.7 128.1 5.99 62.4 91.8 24 109.8 107.7 112.5 123.6 126.1 6.00 59.2 98.3 25 115.7 109.8 113.1 123.3 127.1 6.03 58.8 99.6 26 98.2 105.6 114.9 125.0 125.9 6.07 63.0 91.3 85 27 93.8 106.9 115.1 122.4 127.6 5.95 62.7 90.9 28 93.7 106.9 115.2 122.4 127.6 5.92 62.6 90.8 29 93.7 107.0 115.1 122.4 127.6 5.93 62.6 90.8 30 94.1 107.1 115.4 122.5 127.3 5.95 62.4 90.8 31 109.5 109.6 117.7 124.0 125.4 5.92 60.6 91.8 32 93.4 107.6 115.2 122.5 127.6 6.02 63.2 90.5 33 94.3 107.3 115.1 123.0 127.2 6.03 62.9 90.7 34 94.3 107.3 115.1 123.0 127.2 6.04 62.9 90.8 35 94.3 107.3 115.1 123.0 127.2 6.04 62.9 90.8 36 98.3 105.5 111.3 121.0 125.0 5.96 58.0 98.5 37 93.0 109.8 114.4 125.0 128.4 5.95 66.3 87.5 38 93.3 108.4 115.0 123.2 127.6 5.96 63.6 89.6 39 93.3 108.4 115.0 123.2 127.6 5.96 63.6 89.6 40 93.3 108.4 115.0 123.3 127.6 5.96 63.6 89.6 41 93.5 108.3 115.0 123.3 127.5 5.95 63.5 89.7 86 Table 14. Optimization experiments for VRT processes to minimize P t with F s < 59.7 min and 5.9 < F 0 < 6.1 min (z = 26 C°), the result was the bold value (P t = 97.5 min) Trial Ramp time RTi R T 2 R T 3 R T 4 F 0 (min) F s (min) Pt (min) 1 147.6 108.4 114.0 121.0 127.6 6.02 56.0 114.0 2 129.2 104.6 111.1 123.4 128.6 6.01 58.5 109.5 3 95.9 106.9 119.7 124.9 125.6 6.09 66.0 86.5 4 126.0 109.1 11.1 124.2 126.9 5.92 59.8 105.2 5 142.9 108.5 117.0 122.6 125.7 6.01 56.9 105.1 6 92.1 107.2 119.5 120.9 129.4 6.00 65.5 88.2 7 137.2 104.9 119.3 121.1 127.0 5.93 56.0 103.0 8 90.6 105.9 112.4 123.1 129.2 5.94 66.3 90.2 9 141.7 106.0 113.1 120.9 129.0 5.98 55.8 114.8 10 147.2 105.8 113.6 121.3 126.7 6.01 55.5 116 11 111.7 106.7 117.6 122.5 127.4 5.97 60.1 95.4 12 108.4 106.8 116.4 122.8 127.6 6.05 60.9 95.3 13 109.5 107.5 115.9 123.1 127.4 6.05 61.0 95.6 14 118.8 107.3 117.3 122.7 126.9 5.94 59.0 97.5 15 100.8 106.3 113.5 124.7 127.1 6.05 63.4 93.1 16 90.7 108.2 118.3 123.7 127.4 6.01 66.3 86.2 17 120.2 105.3 118.5 121.1 126.2 6.01 57.8 99.3 18 128.0 107.9 113.4 122.4 128.6 5.99 58.1 106.0 19 99.0 104.0 117.4 123.1 125.6 5.95 61.4 93.0 20 93.7 106.5 117.5 123.1 127.4 5.93 64.2 88.7 21 94.0 106.9 116.7 123.5 127.7 6.05 64.9 89.0 22 95.7 106.5 117.7 123.5 127.0 5.93 63.9 88.9 23 95.4 106.3 116.3 123.9 127.0 6.08 64.4 89.8 24 102.9 108.5 112.4 121.2 126.0 5.94 60.1 98.0 25 96.3 110.0 116.4 121.4 125.5 5.93 62.0 91.6 26 91.2 109.5 114.2 124.2 129.9 6.03 67.8 87.5 87 27 92.7 107.7 117.8 123.4 127.9 5.92 65.2 87.4 28 93.1 107.7 117.9 123.4 127.8 6.00 65.3 87.6 29 93.4 107.5 117.5 123.9 127.5 6.08 65.6 87.8 30 93.6 107.1 118.5 123.2 127.4 5.96 64.8 87.7 31 91.2 105.4 113.6 123.7 126.6 5.97 64.0 90.4 32 99.4 105.4 116.6 121.0 125.4 6.02 60.4 94.9 33 92.1 109.1 118.3 124.0 125.2 6.00 65.2 86.5 34 92.5 108.3 117.7 124.0 127.2 6.01 65.8 86.9 35 92.9 107.9 118.4 123.9 126.8 5.92 65.3 86.6 36 92.6 108.3 117.7 124.0 127.2 6.01 65.8 86.9 37 92.0 108.4 117.3 123.7 127.6 5.99 65.7 87.1 88 Table 15. Optimization experiments for VRT processes to minimize P t with F s < 66.3 min and 5.9 < F 0 < 6.1 min (z = 30 C°), the result was the bold value (P t = 88.2 min) Trial Ramp time RTi R T 2 R T 3 R T 4 F 0 (min) Fs(min) Pt (min) 1 146.5 105.4 118.9 120.1 127.4 6.05 59.5 107 2 130.5 107.5 116 121.1 129.1 5.96 60.9 104.8 3 127.1 109.8 114.7 123.1 127.7 6.06 62.9 102.4 4 118.8 106.1 116.0 124.4 128.1 6.06 63.3 98.1 5 143.5 105.4 111.9 123.2 125.6 6.02 61.5 115.2 6 139.5 104.2 114.4 121.9 128.2 5.95 58.7 112.7 7 100.1 107.8 114.2 124.5 129.2 6.00 66.2 91.5 8 129.6 108.9 117.3 121.6 128.6 5.95 61.3 101.1 9 92.7 109.0 118.0 121.5 125.4 6.00 65.3 89.6 10 100.6 108.8 118.6 124.2 125.1 6.02 65.8 88.6 11 108.4 108.1 116.8 123.2 127.3 6.03 64.0 93.9 12 107.9 108.3 116.3 123.5 127.1 6.01 64.1 93.8 13 110.0 108.9 116.6 123 127.2 4.92 63.6 94.3 14 113.7 108.5 116.9 123 127.0 5.93 63.1 95.4 15 98.3 107.4 112.6 124.7 127.1 6.00 65.6 92.3 16 98.1 107.1 114.8 122.2 127.0 6.03 64.5 93.3 17 90.7 109.3 118.1 123.7 127.4 6.03 67.7 86.0 18 112.8 106.4 118.4 121.2 126.2 5.91 61.9 96.4 19 111.4 109.3 113.3 124.0 126.3 6.03 64.1 97.3 20 96.9 105.1 117.1 123.1 125.6 5.96 64.3 91.1 21 96.2 108.0 117.2 123.4 126.5 6.01 65.6 89.4 22 95.8 107.9 116.9 123.5 126.1 6.01 65.5 89.5 23 96.5 108.4 116.3 123.7 126.9 5.98 65.7 89.5 24 97.3 107.7 116.1 124 126.9 6.03 65.7 90.0 25 98.9 110.0 115.1 123.8 127.2 6.01 65.9 90.7 26 95.6 106.2 118.6 121.5 126.7 6.05 65.0 90.7 89 27 95.5 107.6 117.0 123.2 129.6 6.03 66.6 94.2 28 90.4 107.4 115.8 121.6 127.9 6.09 66.6 89.9 29 96.0 108.5 117.4 123.7 126.4 5.99 65.9 88.6 30 95.3 108.7 117.7 123.3 126.3 5.91 65.7 88.3 31 95.2 108.6 117.8 123.3 126.1 5.96 65.7 88.5 32 95.3 108.6 117.6 123.3 126.2 5.98 65.8 88.6 33 90.7 106.0 119.9 124.3 129.1 6.03 68.7 88.9 34 98.1 107.1 113.4 121.2 129.2 6.06 65.4 94.8 35 94.5 108.7 117.7 123.5 126.5 6.06 66.4 88.1 36 95.4 108.8 117.9 123.6 126.2 6.02 66.2 88.2 37 95.6 108.7 117.9 123.6 126.3 6.00 66.1 88.2 38 95.6 108.8 117.9 123.7 126.3 6.04 66.3 88.1 90 Table 16. Optimization experiments for VRT processes to minimize P t with F s < 68.9 min and 5.9 < F 0 < 6.1 min (z = 32 C°), the best result was the bold value (P t = 87.5 min) Trial Ramp time RTi R T 2 R T 3 R T 4 F 0 (min) F s (min) Pt (min) 1 140.4 106.0 116.6 120.5 128.0 6.07 62.1 109.1 2 156.1 108.9 118.6 122.6 128.9 6.01 62.8 105.6 3 90.4 104.7 112.6 125.0 128.9 6.00 68.4 89.1 4 110.3 109.8 115.2 124.2 128.8 5.94 66.1 94.4 5 139.2 108.0 112.8 123.9 129.2 6.07 64.7 110.1 6 141.1 105.9 110.1 125.0 126.5 5.97 64.8 114.0 7 153.8 106.2 115.4 121.6 127.1 5.92 62.0 114.3 8 97.1 108.5 119.7 123.0 127.8 5.95 67.0 87.5 9 90.1 105.8 119.4 120.1 125.7 5.93 66.1 90.0 10 129.7 106.4 114.8 123.9 127.0 6.03 63.7 104.0 11 103.5 107.0 116.3 123.3 127.6 5.95 65.3 92.8 12 108.8 107.6 117.1 123.0 128.0 6.02 65.1 94.1 13 112.7 106.9 117.0 122.9 127.7 6.05 64.5 96.1 14 116.7 107.7 116.2 123.8 128.3 5.95 64.5 96.8 15 106.2 105.0 117.2 122.3 125.3 5.93 63.8 95.2 16 119.6 109.7 119.2 124.1 125.2 6.03 65.2 93.0 17 108.8 106.9 115.1 121.1 125.6 5.93 63.5 98.7 18 122.7 107.0 117.9 121.4 129.2 6.06 63.3 99.8 19 116.1 107.7 114.2 120.3 128.4 6.02 64.0 102.9 20 100.2 107.2 117.4 123.1 127.1 6.01 65.9 90.9 21 98.0 106.7 117.0 122.9 127.6 6.06 66.2 90.9 22 101.2 107.3 117.6 123.1 127.1 5.94 65.6 91.0 23 103.9 107.5 117.0 123.7 127.5 6.01 65.8 91.9 24 102.2 109.3 115.1 122.6 126.4 6.04 65.6 93.7 25 97.2 107.9 114.5 122.1 128.4 6.04 66.4 92.7 26 102.1 108.1 117.7 124.9 125.4 5.97 66.4 89.3 91 27 91.9 107.3 115.2 123.5 129.3 5.98 67.8 88.6 28 94.3 106.9 116.9 123.3 127.4 5.93 66.6 89.1 29 95.9 107.1 116.4 123.9 127.8 6.08 67.1 89.5 30 93.5 106.6 116.8 122.9 127.9 5.98 66.8 89.3 31 95.8 107.3 117.8 122.9 127.2 5.92 66.3 89.2 92 Table 17 showed the optimum VRT processes, considering the minimum process time for each surface z value. Table 17 showed that when the surface z values increased, the minimum process time of its optimum VRT process decreased. From table 17, the process times were changed with the different retort temperatures and different surface z values. Figure 18 showed the optimum VRT processes to yield the minimum process times at different surface z values. 93 Table 17. Comparison of the optimum VRT processes with minimum P t and F s in term of different z values z value Ramp time RTi R T 2 R T 3 R T 4 F 0 (min) F s (min) Pt (min) 24 157.8 105.0 119.7 123.8 128.7 6.06 54.5 106.9 26 118.8 107.3 117.3 122.7 126.9 5.94 59.0 97.5 28 98.8 105.5 113.1 122.1 127.3 5.99 62.9 95.3 30 95.6 108.7 117.9 123.6 126.3 6.00 66.1 88.2 32 97.1 108.5 119.7 123.0 127.8 5.95 67.0 87.5 94 Figure 18. The optimum VRT processes to yield the minimum P t of MC in terms of different z values. 95 Table 18 compared the process times of MC for the best C R T process and the best VRT process at different surface z values. From Table 18, the surface z value increased, their process times (P t) of the CRT and VRT processes all decreased. For the C R T processes, their optimum retort temperatures increased from 111 to 113 °C and their process times decreased from 148.1 to 124.8 min with an increase of z values from 24 to 32. For the VRT processes, their process times decreased from 106.9 to 87.5 min with an increase of z values from 24 to 32. From Table 18, it was found that the process times of the VRT processes were shorter than those of the CRT processes. When the z value was higher, the process times of the VRT processes decreased more from 29.5 to 50.6 min and their process times decreased from 23.6 to 34.2 % compared with those of the C R T processes. From Table 18, the surface cook values of the CRT processes were from 55.3 to 68.9 min but the surface cook values of the VRT processes were from 54.5 to 67.0 min. The surface cook values of the VRT processes also decreased from 0.2 to 1.9 min with the different z values. Then the surface cook values of the VRT processes all were lower than those of the CRT processes. 96 Table 1 8 . Comparison of P t for the optimum CRT and VRT processes in terms of different z values, (P t, min) z = z = z = z = z = / 2 4 C° 2 6 C° 2 8 C° 3 0 C° 3 2 C° 148 .1 148 .1 1 2 4 . 8 1 2 4 . 8 1 2 4 . 8 (CRT (CRT (CRT (CRT (CRT Ptfor the best C R T process = 1 1 1 ) = 1 1 1 ) = 1 1 3 ) = 1 1 3 ) = 1 1 3 ) Pt for the best VRT process 106.9 97.5 95.3 88.2 87.5 Decrease process time (min, Pt, CRT-Pt, VRT) 4 1 . 2 5 0 . 6 2 9 . 5 3 6 . 6 3 7 . 3 Decrease process time (%) 2 7 . 8 3 4 . 2 2 3 . 6 2 9 . 3 2 9 . 9 5 5 . 3 5 9 . 7 6 3 . 2 6 6 . 3 6 8 . 9 (CRT (CRT (CRT (CRT (CRT F S for the best C R T process = 1 1 1 ) = 1 1 1 ) = 1 1 3 ) = 1 1 3 ) = 1 1 3 ) F S for the best VRT process 5 4 . 5 5 9 . 0 6 2 . 9 66 .1 6 7 . 0 Decrease F S (min, F S , CRT-F S , VRT) 0 .8 0 .7 0 .3 0 . 2 1.9 Decrease F S (%) 1.5 1.2 0 .5 0 . 3 2 . 8 97 4. 8. Compare the results of CRT and VRT processes for MC When the aim was to find the minimum surface cook value for the C R T process and VRT process. The optimum CRT process of RT of 113 °C was chosen and its surface cook value was 63.2 minute. Its process time was 124.8 minutes. On the other hand, its minimum surface cook value of the optimum VRT process was 56.2 minutes and its process time was 109.8 minutes. Thus, the surface cook value of the optimum VRT process decreased about 11.1% than that of the optimum CRT process. In the mean time, the process time of the optimum VRT process was shorter than that of the optimum C R T process. The process time of the optimum CRT process was 124.8 min but the process time of the optimum VRT process was 109.8 min. The process time of the optimum VRT process decreased 15 min than that of the optimum CRT process. Figure 19 compared the changes of the retort temperatures and can center temperatures for the optimum CRT and VRT processes. 98 140.00 i 0.00 -I 1 1 1 0 45 90 135 180 Heating time (min) Figure 19. The optimum CRT and VRT processes of MC for the minimum surface cook values. RT and T c indicated retort temperature and can center temperature for the respective C R T and VRT computer simulations. 99 When the aim was to find the minimum process time, the optimum C R T process of RT of 113 °C was chosen for the z value of 28 C°, its process time was 124.8 min but the minimum process time of the optimum VRT process was 95.3 minutes. The process time of the optimum VRT process decreased about 23.6% than that of the CRT process with the same sterilization value of 6 minutes. In the mean time, its surface cook value of MC for the VRT process was smaller than that of the CRT process. The surface cook value of the optimum CRT process was 63.2 min but the surface cook value of the optimum VRT process was 62.9 min. The surface cook value of the optimum VRT process decreased 0.3 min than that of the optimum CRT process. Figure 20 compared the changes of the retort temperatures and can center temperatures of the optimum C R T and VRT processes. 100 140 i 0 -I 1 1 1 1 0 45 90 135 180 Heating time (min) Figure 20. The optimum CRT and VRT processes of MC for the minimum process time. RT and Tc indicated retort temperature and can center temperature for the respective CRT and VRT computer simulations. 101 Table 19 compared the results of surface cook values and process times with the optimum C R T and VRT processes for MC. In terms of the surface cook value of MC, the optimum VRT process decreased surface cook value and improved surface quality compared with the optimum CRT process. In term of the process time of MC for the thermal processing, the optimum VRT process decreased process time than that of the optimum CRT process. From this table, it was found that different VRT processes had different effects on the surface quality and process time. 102 Table 19. The optimum C R T and VRT processes of MC (z=28 C°) in terms of the minimum surface cook value and the minimum process time Thermal processes CRT=113°C Optimum VRT process 1 Optimum VRT Process 2 Process time P t (min) 124.8 109.8 95.3 Save the process time (Pt, min) / 15.0 29.5 Save the process time (%) / 12.0 23.6 Surface cook value (F s, min) 63.2 56.2 62.9 Decrease the surface cook value (F s, min) / 7.0 0.3 Decrease the surface cook value (%) / 11.1 0.5 Optimum VRT process 1: VRT process for the minimum surface cook value, the process time was smaller than that of CRT process. Optimum VRT process 2: VRT process for the minimum process time, the surface cook value was smaller than that of CRT process. 103 4. 9. Confirmation of Optimum CRT and VRT Processes in an Actual Steam Retort The optimum CRT process of RT of 113 °C was chosen because the minimum surface cook value for MC was obtained. The optimum VRT processes with the minimum surface cook value or minimum process time were chosen through the Retort Program and R C O program. Retort experiments in an actual steam retort were done to confirm the results of the computer simulation. A. Comparison of Sterilization Value (F0) Calculation of the sterilization value (F 0) was done from actual retort experiments based on the results of computer simulations. Table 20 showed that the sterilization values from the retort experiment were slightly higher than those of the predicted results from the computer simulation. In the actual retort experiments, some factors such as initial product temperature, cooling water temperature, and retort temperature were less precisely controlled than in the computer simulation. The sterilization values from the computer simulation and retort experiments were different. But the results were generally within one standard deviation of empirical values. 104 Table 20. Sterilization values (F 0) for MC with three process runs for each treatment and calculations done by using improved general method Process Number of individual .cans tested (retort experiments) F 0 , min, mean (standard deviation) F 0 , min, Computer Simulation CRT=113°C 9 6.3 ±0.6 6.0 ±0.1 CRT=121 °C 8 6.5 ±0 .9 6.0 ±0.1 Optimum VRT 8 6.2 ±0 .3 6.0 ±0.1 process 1 Optimum VRT 9 6.3 ±0 .5 6.0 ±0.1 process 2 Optimum VRT process 1: VRT process for the minimum surface cook value, the process time was smaller than that of CRT process. Optimum VRT process 2: VRT process for the minimum process time, the surface cook value was smaller than that of CRT process. 105 B. Comparison of the Surface Color Parameters In this project, a surface z value of 28 C° was found for surface color change of MC. The optimum CRT for a z value of 28 C° was 113 °C and the optimum VRT processes were determined by the Retort program and the R C O program. Next it was necessary to confirm these results in an actual retort. Following the retort experiments, the surface color of M C was measured to confirm the results of the computer simulation. Table 21 shows the surface color parameters L, a and b values from different actual retort processes. For the C R T process, 113 °C was chosen as the optimum RT. The surface color L, a and b values of the MC decreased and the surface of MC appeared dark. For the VRT process, the surface color L, a and b values decreased less than those of the C R T process and the surface of MC appeared less dark. From Table 21, it can be seen that the surface color of MC for the VRT processes was significantly better than that of the CRT processes. That is to say the VRT processes improved the surface quality compared to the CRT processes. In order to quantitatively evaluate whether there was a consistent difference in the surface color L, a and b values by the different thermal processes, a paired t-test was performed on the means comparing between L (CRT=113) and L (VRT 1), between L (CRT=113) and L (VRT 2), between L (CRT=13) and L (CRT=121), and between L (VRT1) and L (VRT 2). Also a paired t-test was performed to compare between the a and b values with different thermal processes (table 21). 106 Table 21. The surface color parameters of MC in terms of the different CRT and VRT processes (confirmation experiment results and t-test) L , a and b value (mean 1 ± SD) L , a and b value (mean 2 ± SD) p values Significant or not L C RT=113 = 59.48 ±0.18 LCRT=121= 58.47 ±0.10 p<0.05 Significant L C R T = I I 3 = 59.48 ±0.18 L VRT 1= 60.78 ±0.24 p<0.05 Significant L C R T = I I 3 = 59.48 ±0.18 L V R T 2 =60 .27 ±0.22 p<0.05 Significant L VRT 1=60.78 ±0.24 LVRT2=60.27 ±0.22 p<0.05 Significant a C R T = n 3 = 8.45±0.21 a CRT=I2I= 8.29 ± 0.28 p>0.05 Not significant a C R T = n 3 = 8.45±0.21 a V R T i = 8.61 ±0.24 p>0.05 Not significant a CRT=II3 = 8.45 ± 0.21 a VRT 2=8.53 ± 0.19 p>0.05 Not significant a V R T I = 8.61 ±0.24 a VRT 2=8.53 ±0.19 p>0.05 Not significant b C RT=i i3 = 23.31 ±0.13 b CRT=I2I= 23.00 ± 0.14 p<0.05 Significant b C R T = i i 3 = 23.31 ±0.13 b VRT 1=23.94 ±0.21 p<0.05 Significant b C R T = i i 3 = 23.31 ±0.13 b VRT2=23.59 ±0.20 p<0.05 Significant b VRT 1=23.94 ±0.21 b VRT 2=23.59 ±0.20 p<0.05 Significant V R T 1: the optimum V R T process 1 for the minimum surface cook value, the process time is smaller than that of C R T process. V R T 2: the Optimum V R T process 2 for the minimum process time, the surface cook value is smaller than that of C R T process. 107 C. Confirmation of the Surface Cook Values of MC Table 22 compared the surface cook values of the C R T and VRT processes with the retort experiments and the computer simulations. From this table, it was found that the surface cook values of the CRT and VRT processes with the retort experiments were slightly higher than those of the computer simulations. In the actual retort experiments, some factors such as initial product temperature, cooling water temperature, and retort temperature were less precisely controlled than in computer simulation. Here the effect of the thermal treatment must be integrated for every point in the container. Thus, actual tests or simulation work must be carried out to determine the effect of a processing temperature change and the effect may vary depending on the container tested. 108 Table 22. Comparison of surface cook values (F s) of MC in terms of computer simulation and retort experiments (three process runs for 8-10 cans, based on the sterilization value F 0 of 6.0 min Retort experiments (average values of 8-10 cans) Computer simulations F s (min, CRT=113°C) 66.4 ±3.4 63.2 F s (min, CRT=121 °C) 80.9 ±2.7 77.1 F s (min, Optimum VRT process 1) 57.7 ± 1.5 56.2 F s (min, Optimum VRT process 2) 64.0 ±2.1 62.9 Optimum VRT process 1: VRT process for the minimum surface cook value, the process time is smaller than that of CRT process. Optimum VRT process 2: VRT process for the minimum process time, the surface cook value is smaller than that of CRT process. 109 C H A P T E R V C O N C L U S I O N S Based on the experiment results, the surface color changes of MC were related to the heating temperature and heating time. If the heating temperature or heating time was increasing, the surface color of MC became more dark and the surface color parameter L, a, b values all changed. Lightness (L value) of MC was considered the most important factor to affect human color judgment for the MC in this study. The surface color parameter L value change of MC tested was described by first order reaction kinetics. The surface z value of MC was 28 C°. R C O program, when combined with Retort Program, provided a convenient, efficient method for choosing the optimum VRT thermal process. Optimum VRT process was proved to reduce surface cook value of MC and reduce the surface color change, while maintaining the sterilization value (F 0) of the total process. The optimum VRT processes reduced surface cook value by 4.9-7.7 minutes, depending on different z values. This corresponded to a reduction of 8.9-11.2% compared with the optimum CRT processes and also process time of the optimum VRT processes was shortened compared to the C R T processes for MC. In terms of the surface color change of MC, the optimum VRT process improved surface quality (MC with less darkening of color (higher L, a, b values) compared with the optimum C R T process. From the experimental results, the optimum VRT processes reduced process time by 29.5 to 50.6 minutes depending on the different surface z values. This corresponded to a reduction of 23.6 to 34.2 % compared with the best CRT process, depending on the different z values. The results of this study demonstrated that process times of the VRT processes for MC were shortened and also the surface cook value was slightly decreased compared to the C R T processes for MC. 110 Actual steam retort experiments confirmed that the optimum VRT process was indeed superior to the optimum C R T process for MC. From this study, the conclusion was that the optimum VRT processes were better than the optimum CRT processes for the conduction-heated canned foods. In the near future, this study will do more research. Other different conditions, such as different temperature, different can size or different foods or products will be considered for the surface color change characteristics. Other factors to affect on the surface color changes of canned foods will be considered. This study only considered the L value changes of surface color of MC, a value and b value changes or combination of L, a and b value changes will be considered for researching the surface z value. Other quality characteristics, such as thiamine retention, flavor retention and so on, will be compared by using the CRT and VRT processes. I l l APPENDIX A: Terminology and Abbreviations in Thermal Processing a : thermal diffusivity. a = thermal Conductivity/(specific heat * density), is inversely proportional to fh where the proportionality constant is related to the container geometry (m2/s). a w : water activity. ANOVA: Analysis of variance. b: the half-height of the can (mm). cold spot: the slowest heating point in a can of food or in a retort. come-up time: the time between the start of heating and the time when the retort reaches processing temperature. C: the measured color scale value of the product. C 0 : the measured color scale value of the product at the beginning. Commercial sterility: free of all viable microorganisms of public health significance and of all other organisms capable of growth at normal storage temperatures. Some thermophilic bacteria may still survive in commercially sterile products but these grow at temperatures only above 100 °F and are not of public health significance. CRT: constant retort temperature. D value: the decimal reduction time, usually in minutes. This is the time at a lethal temperature, which will destroy 90% of the population of the target organisms. Color D value is defined as the time at a specified temperature required for a 90% change in a numerical color value. fc: Cooling rate index. The index of the cooling curve, such as (log (T c-T w) versus time) and is numerically equal to the negative reciprocal of the slope. fh: Heating rate index. The minutes required for the heat penetration curve (log (T r-T c) versus time) to traverse one-log cycle in temperature difference. It is numerically equal to the negative reciprocal of the slope of the semi-log plot. F 0 : Process lethality or sterilization value. The equivalent, in terms of minutes at 250 °F (121.1 °C), of all lethal heat received by the cold spot in a container. This lethality is calculated with a z value of 18 F° (10 C°). F 0 = ZLAt. F s : The accumulated surface cook value of MC, F s = ZQAt. 112 g value: The unaccomplished temperature, the temperature difference at given time between the cold spot temperature of a container and retort temperature, g=RT-T c. Hunter a: The HunterLab colorimetric scale, the a scale measures the redness (+a value) or greenness (- a value) of the color of the product. Hunter b: The HunterlLab colorimetric scale, the b scale measures the yellowness (+ b value) or blueness (-b value) of the color of the product. Hunter L: The HunterLab colorimetric scale, the L scale measures the lightness (L=100) or darkness (L=0) of the color of the product. k: the reaction rate constant for base e (natural logarithms; death rate constant in the Arrhenius model, k=2.303/D). K: thermal conductivity (W/m °K) L: Lethal rate expressed as minutes at the reference temperature per minute at the center temperature of the container, L = 10 T c " T r e f / z . low acid food: food with a natural pH higher than 4.5. MC: Macaroni and cheese. MR: Maillard reaction. MRPs : Maillard reaction products. P t : Operator's process time. The time from when the retort reaches processing temperature until the steam is turned off and cooling started. Q: surface cook rate of the product surface, Q = 10 T s - T r e f / Z . r: the radius of the can (mm). R C O : random centroid optimization program (Dou et al., 1993; Nakai et al., 1998) Retort Program: A finite difference computer model for conduction with a cylindrical container (Durance et al. 1997). Rho or p value: Fraction of sterilization value, which occurs up to the time the steam is turn off. RT or T r: retort temperature. Saturated steam: 100% water vapor at a temperature equal to the boiling point of water at the prevailing pressure. t: the heating time (min). tv: the ramp time of VRT process (min). T c : the slowest heating point product temperature of the container, 113 Tf: the center-point temperature of the can at time of steam-off. T\\ Initial retort temperature, Tj = 22.5 °C, T p : Initial product temperature, the temperature of the can contents when it enters the retort, T p= 20 °C, T ref: Reference temperature for microorganisms or reference temperature for surface color of product, T r e f = 121.1 °C, T s : the surface temperature of canned foods, T w : cooling water temperature, T w = 10 °C, Vent time: The time at the beginning of a retort run necessary for complete removal of air from a retort, meeting both a time and a retort temperature requirement. VRT: variable retort temperature. X i : a factor for the R C O program. z value: The temperature difference required for a thermal death time or D value to change by one order of magnitude. It is a measure of the sensitivity of the relevant reaction or event to change in temperature. 114 A P P E N D I X B : Processing Conditions for Computer Simulation Model C R T process: Product: MC (macaroni and cheese) Thermal diffusivity, a = 3.625 x 10"7 m 2/s Thermal conductivity, =19.00 (W/m °K) Can dimension: Diameter = 82 mm, Height = 111.6 mm Lethality: Sterilization value, F 0 = 6.0 minutes, z value of microbial thermal death, z m = 10 C°, z value of thermal darkening of the surface of MC, z s = 28 C° Heat penetration parameters: fh = 58 minutes, fc = 77.3 minutes Operation Conditions, Initial retort temperature, Tj = 22.5 °C Initial product temperature, T p= 20 °C, Form of cans, cylindrical cans (307 x 409), normally 3 7/16 inch diameter and 4 9/16 inch height. Cooling water temperature, T w = 10 °C, Reference temperature for microorganisms, T r ef = 121.1 °C, Reference temperature for surface color of product, T r e T = 121.1 °C, VRT process: Product: MC (macaroni and cheese) Thermal diffusivity, a = 3.625 x 10"7 m 2/s Thermal conductivity, =19.00 (W/m °K) Can dimension: Diameter = 82 mm, Height = 111.6 mm Lethality: Sterilization value, F 0 = 6.0 ± 0.1 minutes, Z m = 1 0 C ° , z s = 28 C°, Heat penetration parameters: fh = 58 minutes, fc = 77.3 minutes Operation Conditions, Initial retort temperature, Ti = 22.5 °C, 115 Initial product temperature, T p= 20 °C, Cooling water temperature, T w = 10 °C, Form of cans, cylindrical cans (307 x 409), Time-temperature profile limits, for VRT, 104 -130 °C, Ramp time for V R T processes, tv = 90-160 min, Reference temperature for microorganisms, T r ef =121.1 °C, Reference temperature for surface color of product, T r ef = 121.1 °C. 116 R E F E R E N C E S Afaghi, M., Ramaswamy, H. S., Prasher, S. O. (2001) Thermal process calculations using artificial neural network models. Food Research International. 34(1), 55-65. Afaghi, M., Ramaswamy, H. S. (2001) Artificial neural network simulation of ball and Stumbo formula methods of thermal process calculations. Journal of Food Science and Technology, India. 38(5), 439-446. Ahmed, J . , Shivhare, U. S., Raghavan, G. S. W. (2000) Rheological characteristics and kinetics of color degradation of green chilli puree. Journal of Food Engineering. 44, 239-244. Ahmed, J . , Shivhare, U. S. (2001) Thermal kinetics of color change, rheology, and storage characteristics of garlic puree/paste. Journal of Food Science. 66 (5), 754-757. Aishima, T., Nakai, S. (1986) Centroid mapping optimization: a new efficient optimization for food research and processing. Journal of Food Science. 51, 1297-1300,1310. Almonacid, M. S. F., Simpson, R., Torres, J . A. (1993) Time-variable retort temperature profiles for cylindrical cans: batch process time, energy consumption, and quality retention model. Journal of Food Process Engineering. 16 (4), 271-287. Ames, J . M. (1990) Control of the Maillard reaction in food systems. Trends in Food Science and Technology. 1(6), 150-154. Ariahu, C. C , Ogunsua, A. O. (1999) Effect of thermal processing on thiamine retention, in-vitro protein digestibility, essential amino acid composition, and sensory attributes of periwinkle-based formulated low acid foods. Journal of Food Processing and Preservation. 23, 121-134. Avila, I. M. L. B., Silva, C. L. M. (1999) Modelling kinetics of thermal degradation of color in peach puree. Journal of Food Engineering. 39, 161-166. 117 Baisier, W. M., Labuza, T. P. (1992) Maillard browning kinetics in a liquid model system. Journal of Agricultural and Food Chemistry. 40 (5), 707-713. Banga, J . R., Perez-Martin, R. I., Gallardo, J . M., Casares, J . J . (1991) Optimization of the thermal processing of conduction-heated canned foods: study of several objective functions. Journal of Food Engineering. 14(1), 25-51. Bates, L , Ames, J . M., MacDougall, D. B., Taylor, P. C. (1998) Laboratory reaction cell to model Maillard color development in a starch-glucose-lysine system. Journal of Food Science. 63 (6), 991-996. Bee, G. R., Park, D. K. (1978) Heat penetration measurement for thermal process design. Food Technology. 32(4), 56-58. Berry, B. W. (1998) Cooked color in high pH beef patties as related to fat content and cooking from the frozen or thawed state. Journal of Food Science. 63(5), 797-800. Box, M. J . (1965) A new method of constrained optimization and a comparison with other methods. Computer Journal. 8, 42-45. Buckholz, L. L., Daun, H., Stier, E. (1980) Influence of roasting time on sensory attributes of fresh roasted peanuts. Journal of Food Science, 45, 547-554. Carabasa, G. M., Ibarz, R. A. (2000) Kinetics of colour development in aqueous fructose systems at high temperatures. Journal of the Science of Food and Agriculture. 80, 2105-2113. Chen, Q., Koh, H. K., Park, J . B. (1999) Color evaluation of red pepper powder. Transactions of the A S A E . 42(3), 749-752. Chen, C. R., Ramaswamy, H. S. (2002a) Color and texture change kinetics in ripening bananas. Lebensmittel Wissenschaft und Technologie. 35 (5), 415-419. Chen, C. R., Ramaswamy, H. S. (2002b) Modeling and optimization of variable retort temperature (VRT) thermal processing using coupled neural networks and genetic algorithms. Journal of Food Engineering. 53 (3), 209-220. Cornwell, C. J . , Wrolstad, R. E. (1981) Causes of browning in pear juice concentrate during storage. Journal of Food Science. 46, 515-518. Dou, Jinglie, Toma, S., Nakai, S. (1993) Random-centroid optimization for food formulation. Food Research International. 26, 27-37. 118 Downing, D. L. (1996) A complete course in canning and related processes. Book II. Microbiology, packaging, H A C C P and ingredients. Baltimore, MD 21218-4547, USA; CTI Publications Inc. Driscoll, R. H., Madamba, P. S. (1994) Modelling the browning kinetics of garlic. Food Australia. 46 (2), 66-71. Durance, T. D. (1995) Principles of Food Process Science I. Food 308 Course Manual. 73-175. UBC Access Guided Independent Study, Continuing Studies, UBC, Vancouver, BC , Canada Durance, T. D., Collins, L.S. (1991) Quality Enhancement of Sexually Mature Chum Salmon Oncorhynchus Keta in Retort Pouches. Journal of Food Science. 56 (5), 1282-1286. Durance, T. D. (1997) Improving canned food quality with variable retort temperature processes. Trends in Food Science & Technology. 8 (4), 113-118. Durance, T. D., Dou, J . , Mazza, J . (1997) Selection of variable retort temperature processes for canned salmon. Journal of Food Process Engineering. 20 (1), 65-76. Ecklund, O. F. (1956) Correction factors for heat penetration thermocouples. Food Technology. 10(1), 43-47. Fellows, P. J . (1998) Food Processing Technology. Principles and Practice. First edition, C R C Press. Boca Raton, FL, 33431 USA Fogliano, V., Monti, S. M., Musella, T., Randazzo, G. (1999) Formation of colored Maillard reaction products in a gluten-glucose model system. Food Chemistry. 64, 1-7. Francis, F. J . (1991) Color measurement and interpretation. In Instrumental methods for quality assurance in foods. Edited by D. Y. C. Fung and R. F. Matthews. Marcel Dekker Inc. New York, NY, USA. Friedman, M. (1996) Food browning and its preservation: An overview. Journal of Agricultural and Food Chemistry. 44, 631-653. Giese, J . (2000) Color measurement in foods as a quality parameter. Food Technology. 54(2), 62-63. 119 Hendrickx, M. E., Van Genethten, K., Tobback, P. (1990) Optimization quality attributes of conduction heated foods, a simulation approach. In Engineering and Food: Preservation Processes and Related Techniques, 2, edited by W. E. L. Spess and H. Shubert. Elservier Applied Science, New York, USA. Hendrickx, M., Silva, C , Oliveira, F., Tobback, P. (1993) Generalized (semi) -empirical formulae for optimal sterilization temperatures of conduction-heated foods with infinite surface heat transfer coefficients. Journal of Food Engineering. 19 (2), 141-158. Hildenbrand, P. (1980) An approach to solving the optimal temperature control problem for sterilization of conduction-heating foods. Journal of Food Process Engineering. 3,123-142. Hofmann, T. (1998) Studies on the influence of the solvent on the contribution of single Maillard reaction products to the total color of browned pentose/alanine solutions - a quantitative correlation using the colour activity concept. Journal of Agricultural and Food Chemistry. 46 (10), 3912-3917. Holdsworth, S. D. (1985) Optimization of thermal processing - a review. Journal of Food Engineering. 4,89-116. Hutchings, J . B. (1994) Food Color and Appearance. Blackie Academic and Professional Publication. Glasgow, U. K. 375-376. Ibarz, R. A., Pagan, J . , Garaz, S. (1999) Kinetic models for color changes in pear puree during heating at relatively high temperatures. Journal of Food Engineering. 39,415-422. Little, A. C. (1976) Physical measurements as predictors of visual appearance. Food Technology. 30 (10), 74, 76-77, 80, 82. Lopez, A. (1981) A complete course in canning. I. Basic information on canning. Baltimore, Maryland, USA; Canning Trade, Inc. Lopez, A., Pique, M. T., Boatella, J . , Romero, A., Ferran, A. (1997) Influence of drying conditions on the hazelnut quality: III. Browing. Drying Technology. 15,989-1002. Lund, D. B. (1977) Design of thermal processes for maximizing nutrient retention. Food Technology. 31 (2), 71-75. 120 Lund, D. B. (1982) Applications of optimization in heating processing. Food Technology. 36 (7), 97-100. Lund, D. B. (1991) Heat processing. In Principles of Food Science. Part II. Physical Principles of Food Preservation. Edited by M. Karel, O. w. Rennema and D. D. Lund. 31-92. Marcel Dekker, Inc., New York, USA. Mabon, T. J . (1993) Color measurement of food. Cereal Foods World, 38 (1) 21-25. Maesmans, G. , Hendrickx, M., Weng, Z., Keteleer, A., Tobback, P. (1990) Endpoint definition, determination and evaluation of thermal processes in food preservation. Belgium Journal of Food Chemistry and Biotechnology. 45 (5), 97-100. Martins, S. I. F. S., Jongen, W. M. F., Boekel, M. A. J . S. van (2001) A review of Maillard reaction in food and implications to kinetic modeling. Trends in food science and technology. 11, 364-373. Maskan, A., Kaya, S., Maskan, M. (2002) Effect of concentration and drying processes on color change of grape juice and leather. Journal of Food Engineering. 54, 75-80. Mauron, J . (1981) The Maillard reaction in food: a critical review from the nutritional standpoint. Progress in Food and Nutrition Science. 5(1-6), 5-35. Morgan, S. L., Deming, S. N. (1974) Simplex optimization of analytical chemical method. Analytical Chemistry. 46,1170-1181. Nakai, S. (1981) Comparison of optimization techniques for application to food product and process development. Journal of Food Science. 47, 144-152, 157. Nakai, S., Koide, K., Eugster, K. (1984) A new mapping super-simplex optimization for food product and process development. Journal of Food Science. 49, 1143-1148, 1170. Nakai, S. (1990) Computer-aided optimization with potential application in biorheology. Journal of Japanese Society of Biorheology. 4, 143-152. Nakai, S., Dou, J . , Lo, K.V., Seaman, C. H. (1998) Optimization of site-directed mutagenesis. 1. New random-centroid optimization program for Windows 121 useful in research and development. Journal of Agricultural and Food Chemistry. 46, 1642-1654. Nakai, S., Saeki, H., Nakamura, K. (1999) A graphical solution of multimodal optimization to improve food properties. International Journal of Food Properties, 2(3), 277-294. Namiki, M. (1988) Chemistry of Maillard reactions: recent studies on the browning reaction mechanism and the development of antioxidants and mutagens. Advances in Food Research. 32. Academic Press, Inc. San Diego, California, 92101, USA. Noronha, J . , Hendrickx, M., Suys, J . , Tobback, P. (1993) Optimization of surface quality retention during the thermal processing of conduction heated foods using variable temperature retort profiles. Journal of Food Processing and Preservation. 17 (2), 75-91. Noronha, J . , Loey, A. van, Hendrickx, M., Tobback, P. (1996a) Simultaneous optimization of surface quality during the sterilization of packed foods using constant and variable retort temperature profiles. Journal of Food Engineering. 30 (3/4), 283-297. Noronha, J . , Loey, A. van, Hendrickx, M., Tobback, P. (1996b) An empirical equation for the description of optimum variable retort temperature profiles that maximize surface quality retention in thermally processed foods. Journal of Food Processing and Preservation. 20 (3), 251-264. Nursten, H. E. (1986) Aroma compounds from the Maillard reaction. In Developments in Food Flavor. Edited by G. G. Birch and M. G. Lindley. 173-190. Elsevier, Amsterdam. Ohlsson, T. (1980) Optimal sterilization temperatures for sensory quality in cylindrical containers. Journal of Food Science. 45 (6), 1517-1521. Ozdemir, M., Devres, O. (2000) Kinetics of color changes of hazelnuts during roasting. Journal of Food Engineering. 44,31-38. Palombo, R., Gertler, A., Saguy, L. (1984) A simplified method for determination of browning in dairy powders. Journal of Food Science. 49(6), 1609-1613. Ramesh, M. N. (1995) Optimum sterilization of foods by thermal processing-a review. Food Science & Technology Today. 9 (4), 217-227. 122 Rizvi, S. S. H , Acton, J . C. (1982) Nutrient enhancement of thermostabilized foods in retort pouches. Food Technology. 36 (4), 105-108. Romero, C , Garcia, P., Brenes, M., Garrido, A. (1995) Color and texture changes during sterilization of packed ripe olives. International Journal of Food Science and Technology. 30,31-36. Sablani, S .S. , Ramaswamy, H.S., Prasher, S. O. (1995) A neural network approach for thermal processing applications. Journal of Food Processing and Preservation. 19, 283-301. Samaniego-Esguerra, C. M., Boag, I. F., Robertson, G. L. (1991) Kinetics of quality deterioration in dried onions and green beans as a function of temperature and water activity. Lebensmittel Wissenschaft und Technologie. 24, 53-58. Sandberg, Gary (1991) Thermal processing of low-acid foods with sub-zero initial temperatures. M. Sc. Thesis, The University of British Columbia, BC, Canada. Sandberg, G. M. M., Durance, T. D., Richard, P., Leung, P. H. (1994) Thermal processing canned meat with sub-freezing initial temperatures: thermal and microbial validation. Journal of Food Science, 59 (4), 693-696, 719. Shin, S., Bhowmik, S. R. (1995) Thermal kinetics of color changes in pea puree. Journal of Food Engineering. 24, 77-86. Silva, C. L. M., Hendrickx, M., Oliveira, F., Tobback, P. (1992) Optimum sterilization temperatures for conduction heating foods considering finite surface heat transfer coefficients. Journal of Food Science. 57 (3), 743-748. Silva, C. L. M,, Oliveira, F. A. R., Hendrickx, M. (1993) Modelling optimum processing conditions for the sterilization of prepackaged foods. Food Control. 4(2), 67-73. Silva, C. L. M., Oliveira, F.A. R., Lamb, J . , Torres, A. P., Hendrickx, M. (1994) Experimental validation of models for predicting optimal surface quality sterilization temperatures. International Journal of Food Science and Technology. 29 (3), 227-241. Spinak, S. H., Wiley, R. C. (1982) Comparisons of the general and Ball formula methods for retort pouches process calculations. Journal of Food Science. 47, 880-. 123 Stumbo, C. R. (1973) Thermobacteriology in food processing. Academic Press, New York, London. Teixeira, A. A., Dixon, J . R., Zahardnik, J . W., Zinsmeister, G. E. (1969) Computer optimization of nutrient retention in the thermal processing of conduction-heated foods. Food Technology. 23, 845-853. Teixeira, A., Stumbo, Charles R., Zahrad, John W. (1975a) Experimental evaluation of mathematical and computer models for thermal process evaluation. Journal of Food Science. 40, 653-655. Teixeira, A., Zinsmeister, George E., Zahrad, John W. (1975b) Computer simulation of variable retort control and container geometry as a possible means of improving thiamine retention in thermally processed foods. Journal of Food Science. 40, 656-659. Teixeira, A. A., Tucker, G. S. (1997) On-line retort control in thermal sterilization of canned foods. Food Control. 8(1), 13-20. Terajima, Y., Nonaka, Y. (1996) Retort temperature profile for optimum quality during conduction-heated of foods in retortable pouches. Journal of Food Science. 61 (4), 673-678. Vazquez-Arteaga, M. C. (1990) Computer-aided formula optimization. MSc thesis, The University of British Columbia, Vancouver, BC , Canada. Weemaes, C. A., Ooms, V., Loey, A. M., van, Hendrickx, M. E. (1999) Kinetics of chlorophyll degradation and color loss in heated broccoli juice. Journal of Agricultural and Food Chemistry. 47(6), 2404-2409. Wijewikreme, A. N., Kitts, D. D., Durance, T. D. (1997) Reaction conditions influence the elementary composition and metal chelating affinity of nondialyzable model Maillard reaction. Journal of Agricultural and Food Chemistry. 45(2), 4577-4583 Xing, B. K. (2002) Maillard browning reaction of glucose-lysine model system in a microwave field. U B C thesis. Young, K. E. (1984) Thiamine degradation in a luncheon-type ham product thermally processed in retort pouches and cans. M.Sc. Thesis, The University of British Columbia, Vancouver, BC , Canada.