Atmospheric freeze drying (AFD) is the lyophilization of a product at atmospheric pressure conditions and temperatures ranging generally between -15 and -5 ◦C (avoiding, thereby, ice melting). The quality of the obtained dried products is quite similar to the quality of products dried by vacuum freeze drying (VFD), but without the need of generating vacuum, maintaining temperatures around -50 ◦C in the condenser, or defrosting it. There are several ways to carry out AFD, such as the use of a fluidized bed or a tunnel conveyor. Nevertheless, AFD involves considerably longer drying times than VFD, and the process must be modified in some way in order to shorten them without loss of product quality at the same time. Moreover, since this process is usually carried out with air at very low temperatures, it can saturate rapidly. This situation leads to a reduction of the gradient of water concentration between air and product surface, and consequently, a diminution of the mass transfer rate. The use of an adsorbent material compatible with the food product (i.e., not toxic for human consumption) in a fluidized bed, could constitute an alternative for using other extra energy supplies (such as IR application, or heat pump). At the same time, the use of the adsorbent medium presents two additional advantages: the first, as the heat of adsorption of water vapour is of the same order of magnitude than sublimation heat of ice, no additional energy supply is necessary; second, it acts as adsorbent medium for generated water vapour, allowing air recirculation, which means an additional reduction of operative costs. In particular, non-food wheat bran is an interesting material to be applied as adsorbent in this process; this adsorbent is not only compatible with foodstuff, but also, since it is a by-product of the cereal processing industry, it is cheap and can be easily discarded (and reused, for example, in compost) without recovering it by means of a thermal treatment. Nonetheless, as it is the hard outer layer of cereals consisting of combined aleurone and pericarp, its particles exhibit a very irregular plane shape, with rests of grain brush and, in some cases, broken pericarp. These characteristics confer to the particle a rough surface and, as undesired consequence, the possibility of mechanical interaction during fluidization. However, when two different materials are fluidized in a fluidized bed, the mixture may undergo segregation, causing a poor contact between the adsorbent and the food particles. Thus, instead of using a traditional fluidized bed, a spout-fluid bed (an apparatus similar to the spouted bed, with lateral air injectors beside the main jet) may be utilized, and thereby enhancing mixing. On the other hand, CFD simulation would offer a great potential for simulating the AFD process, its optimization from the fluid dynamic point of view, and the design of new equipment. Various investigators have been working on the application of CFD models for simulating AFD in fluidized bed. However, in general, they simulated a single piece of foodstuff, but not the complete system with air, food material, and adsorbent (when it is applied). The general objectives of the PhD work are to determine the hydrodynamic conditions under which AFD in adsorbent fluidized bed is feasible, and to obtain a first approach to a CFD model of the process. Particularly, the study of the hydrodynamics of the process (non-food wheat bran fluidization behaviour, and mixing of binary mixtures) in a fluidized bed as well as in a spout-fluid bed is aimed from the experimental point of view, while the evaluation of the possibility of simulation by means of a CFD code of the AFD process by immersion in adsorbent medium in a fluidized bed is intended in the theoretical field. Unlike sand or other materials in which regular bubbles are formed, non-food wheat bran exhibits canalization or preferential air paths formation, and bed does not expand after overcoming minimum fluidization velocity. In addition, bran particle diameter is represented by a population distribution whose majority is Geldart B. Therefore, considering other bran particles physical characteristics such as rough surface and rest of grain brushes, it can be said that this "pseudo-cohesive" behaviour is caused principally by mechanical interactions rather than electrostatic forces as occurs in cohesive powders. In general terms, it can be observed a cyclical behaviour of channels generation and collapse. The number of channels and their shape depend on air superficial velocity as well as the bed position where they are formed. Anyway, in general they follow the Channel Generation and Collapse Cycle where two main stages are represented: I, generation, and II, collapse. Experiments emulating different stages of the AFD process with adsorbent application (using fresh food, partially lyophilized material, and completely lyophilized foodstuff) were done employing different food particles (peas, carrot discs, and potato slabs). Experiments were carried out in a 35 cm squared base fluidized bed and in a 20x10 rectangle base fluid-spout bed. The effects on segregation of air superficial velocity, product volumetric fraction, and particle shape were evaluated. For evaluating the segregation, segregation indexes form literature were evaluated, but some difficulties were found using them, besides it is not possible to obtain information about the segregation profiles with them. Thus, a novel way for characterizing segregation was proposed (the Three Thirds Segregation Indexes Set, TTSIS), consisting of three numbers that evaluate the distribution profile of a material of interest (food product, for the current case) and a fourth one that gives an idea of the segregation level. TTSIS was found the best tool for quantifying the segregation phenomenon, as it allows not only to measure the segregation level, but also classify the segregation pattern. As it was expected from the theory, it was evidenced that, even for a binary mixture composed by a pseudo-cohesive powder and a solid whose particles are considerably greater than the powder ones, the air superficial velocity plays a very important role in mixing. Particularly, at high air flows (2.6 umf-adsorbent for the analysed cases) uniform distribution of the material of interest are reached when dried foodstuff is used. Nonetheless, product density plays a fundamental role, since disuniform segregation profiles were obtained when fresh or partially lyophilized food material was used. Uniform mixing profiles were reached in the fluid-spout bed with a good circulation of the food particles along the bed during the fluidization. These results shown to be independent of the product density. Thus, this kind of bed should be used if an uniform mixing between adsorbent and food product is desired. Segregation phenomenon in channelling fluidized beds and the mixing process in fluid-spout beds might be explained by means of two food particle transport mechanisms (passive and active) and two movement blocking effects (floor and roof effects) observed during experimentation (video analysis), and the Channel Generation and Collapse Cycle. Regarding to CFD simulations, relatively reasonable results were obtained from the hydrodynamic point of view only at high air superficial velocity. However, specific models for cohesive or pseudo-cohesive powders are required if an accurate simulation of this kind of solids or binary mixtures integrated by them is intended. In sum, from the experimental results, emerged that the fresh product completely segregates toward the bed bottom in fluidized bed when mixtures containing food material without drying were used. Thus, a good contact between the material to be dried and the adsorbent (desirable for utilizing the adsorption heat for ice sublimation) would be not possible for AFD with use of adsorbent in fluidized bed applications. In contrast, using a fluid-spout bed maintains a very good mixing even if fresh food particles are used. Thus, beside the already known applications of this kind of beds for catalytic processes, its utilization for AFD with adsorbent medium seems to be an interesting and novel option for this process. Nevertheless, CFD simulation might be performed only for non-cohesive powders since the simulation of pseudo-cohesive materials fluidization is currently limited because of the lack of hydrodynamic models for this kind of solids.

Atmospheric Freeze Drying of Food in Fluidized Beds - Practical aspects and CFD simulation / Coletto, MAURICIO MIGUEL. - (2015). [10.6092/polito/porto/2588248]

Atmospheric Freeze Drying of Food in Fluidized Beds - Practical aspects and CFD simulation

COLETTO, MAURICIO MIGUEL
2015

Abstract

Atmospheric freeze drying (AFD) is the lyophilization of a product at atmospheric pressure conditions and temperatures ranging generally between -15 and -5 ◦C (avoiding, thereby, ice melting). The quality of the obtained dried products is quite similar to the quality of products dried by vacuum freeze drying (VFD), but without the need of generating vacuum, maintaining temperatures around -50 ◦C in the condenser, or defrosting it. There are several ways to carry out AFD, such as the use of a fluidized bed or a tunnel conveyor. Nevertheless, AFD involves considerably longer drying times than VFD, and the process must be modified in some way in order to shorten them without loss of product quality at the same time. Moreover, since this process is usually carried out with air at very low temperatures, it can saturate rapidly. This situation leads to a reduction of the gradient of water concentration between air and product surface, and consequently, a diminution of the mass transfer rate. The use of an adsorbent material compatible with the food product (i.e., not toxic for human consumption) in a fluidized bed, could constitute an alternative for using other extra energy supplies (such as IR application, or heat pump). At the same time, the use of the adsorbent medium presents two additional advantages: the first, as the heat of adsorption of water vapour is of the same order of magnitude than sublimation heat of ice, no additional energy supply is necessary; second, it acts as adsorbent medium for generated water vapour, allowing air recirculation, which means an additional reduction of operative costs. In particular, non-food wheat bran is an interesting material to be applied as adsorbent in this process; this adsorbent is not only compatible with foodstuff, but also, since it is a by-product of the cereal processing industry, it is cheap and can be easily discarded (and reused, for example, in compost) without recovering it by means of a thermal treatment. Nonetheless, as it is the hard outer layer of cereals consisting of combined aleurone and pericarp, its particles exhibit a very irregular plane shape, with rests of grain brush and, in some cases, broken pericarp. These characteristics confer to the particle a rough surface and, as undesired consequence, the possibility of mechanical interaction during fluidization. However, when two different materials are fluidized in a fluidized bed, the mixture may undergo segregation, causing a poor contact between the adsorbent and the food particles. Thus, instead of using a traditional fluidized bed, a spout-fluid bed (an apparatus similar to the spouted bed, with lateral air injectors beside the main jet) may be utilized, and thereby enhancing mixing. On the other hand, CFD simulation would offer a great potential for simulating the AFD process, its optimization from the fluid dynamic point of view, and the design of new equipment. Various investigators have been working on the application of CFD models for simulating AFD in fluidized bed. However, in general, they simulated a single piece of foodstuff, but not the complete system with air, food material, and adsorbent (when it is applied). The general objectives of the PhD work are to determine the hydrodynamic conditions under which AFD in adsorbent fluidized bed is feasible, and to obtain a first approach to a CFD model of the process. Particularly, the study of the hydrodynamics of the process (non-food wheat bran fluidization behaviour, and mixing of binary mixtures) in a fluidized bed as well as in a spout-fluid bed is aimed from the experimental point of view, while the evaluation of the possibility of simulation by means of a CFD code of the AFD process by immersion in adsorbent medium in a fluidized bed is intended in the theoretical field. Unlike sand or other materials in which regular bubbles are formed, non-food wheat bran exhibits canalization or preferential air paths formation, and bed does not expand after overcoming minimum fluidization velocity. In addition, bran particle diameter is represented by a population distribution whose majority is Geldart B. Therefore, considering other bran particles physical characteristics such as rough surface and rest of grain brushes, it can be said that this "pseudo-cohesive" behaviour is caused principally by mechanical interactions rather than electrostatic forces as occurs in cohesive powders. In general terms, it can be observed a cyclical behaviour of channels generation and collapse. The number of channels and their shape depend on air superficial velocity as well as the bed position where they are formed. Anyway, in general they follow the Channel Generation and Collapse Cycle where two main stages are represented: I, generation, and II, collapse. Experiments emulating different stages of the AFD process with adsorbent application (using fresh food, partially lyophilized material, and completely lyophilized foodstuff) were done employing different food particles (peas, carrot discs, and potato slabs). Experiments were carried out in a 35 cm squared base fluidized bed and in a 20x10 rectangle base fluid-spout bed. The effects on segregation of air superficial velocity, product volumetric fraction, and particle shape were evaluated. For evaluating the segregation, segregation indexes form literature were evaluated, but some difficulties were found using them, besides it is not possible to obtain information about the segregation profiles with them. Thus, a novel way for characterizing segregation was proposed (the Three Thirds Segregation Indexes Set, TTSIS), consisting of three numbers that evaluate the distribution profile of a material of interest (food product, for the current case) and a fourth one that gives an idea of the segregation level. TTSIS was found the best tool for quantifying the segregation phenomenon, as it allows not only to measure the segregation level, but also classify the segregation pattern. As it was expected from the theory, it was evidenced that, even for a binary mixture composed by a pseudo-cohesive powder and a solid whose particles are considerably greater than the powder ones, the air superficial velocity plays a very important role in mixing. Particularly, at high air flows (2.6 umf-adsorbent for the analysed cases) uniform distribution of the material of interest are reached when dried foodstuff is used. Nonetheless, product density plays a fundamental role, since disuniform segregation profiles were obtained when fresh or partially lyophilized food material was used. Uniform mixing profiles were reached in the fluid-spout bed with a good circulation of the food particles along the bed during the fluidization. These results shown to be independent of the product density. Thus, this kind of bed should be used if an uniform mixing between adsorbent and food product is desired. Segregation phenomenon in channelling fluidized beds and the mixing process in fluid-spout beds might be explained by means of two food particle transport mechanisms (passive and active) and two movement blocking effects (floor and roof effects) observed during experimentation (video analysis), and the Channel Generation and Collapse Cycle. Regarding to CFD simulations, relatively reasonable results were obtained from the hydrodynamic point of view only at high air superficial velocity. However, specific models for cohesive or pseudo-cohesive powders are required if an accurate simulation of this kind of solids or binary mixtures integrated by them is intended. In sum, from the experimental results, emerged that the fresh product completely segregates toward the bed bottom in fluidized bed when mixtures containing food material without drying were used. Thus, a good contact between the material to be dried and the adsorbent (desirable for utilizing the adsorption heat for ice sublimation) would be not possible for AFD with use of adsorbent in fluidized bed applications. In contrast, using a fluid-spout bed maintains a very good mixing even if fresh food particles are used. Thus, beside the already known applications of this kind of beds for catalytic processes, its utilization for AFD with adsorbent medium seems to be an interesting and novel option for this process. Nevertheless, CFD simulation might be performed only for non-cohesive powders since the simulation of pseudo-cohesive materials fluidization is currently limited because of the lack of hydrodynamic models for this kind of solids.
2015
File in questo prodotto:
File Dimensione Formato  
MCOLETTOthesis_PoliTo.pdf

accesso aperto

Tipologia: Tesi di dottorato
Licenza: Creative commons
Dimensione 9.86 MB
Formato Adobe PDF
9.86 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2588248
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo