Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Extrusion of starches: Expansion and modeling with neural networks
Abstract
Extrusion of starches has evolved over the past 50 years with an enormous amount of research. Yet there are many unsolved problems related to the extrusion of starches. Extrusion of starches involves subjecting starch granules, with plasticizers/water, to various levels of shear, heat and pressure along a single screw or set of screws in an extruder to melt them. And finally forcing the melt through a die constriction to the atmosphere releasing the pressure and leading to an expanded product. Starch is a complex biopolymer and due to its complex behavior under different extrusion conditions it is difficult to model the process of starch extrusion. Attempts have been made to model different aspects of extrusion individually, by keeping various associated aspects constant. Starch, in itself, varies a lot depending on its source and also can be chemically modified to attribute specific characteristics to it. Expansion during extrusion of starches is a unique phenomenon that is utilized to enormous extent in manufacture of food and biofoams for packaging. Numerous studies have reported on improving expansion and thus physical and mechanical properties of extrudates. Those studies have concentrated on the properties of the final stable extrudates obtained during extrusion. The actual expansion phenomenon has not been explained satisfactorily. The studies reported have considered the final (stable) expansion to arrive at various conclusions, overseeing the phenomena occurring during expansion and its effects. It is assumed that expansion is driven by bubble growth. There have been some reports attributing expansion to the die swell or the viscoelastic nature of the dough melt. In this dissertation, emphasis has been given to the study of expansion in detail. Expansion has been studied to a great extent and a hypothesis for the same has been proposed. The complete extrusion process can be viewed as a multiple input and multiple output system (MIMO) when it needs to be modeled taking into consideration all the extrusion parameters. This MIMO system can be more effectively modeled using neural networks compared with the statistical models. In this dissertation an attempt was made to model the extrusion process using neural networks in two approaches. Extrusion process was viewed as a one-step process (Input/Process parameters and Output/Extrudate characteristics) and as a two-step process (Input/Process parameters, System Parameters and Output/Extrudate characteristics). These models were more robust than statistical models and were used to predict the extrudate characteristics given a set of input/process parameters. If the neural network models are trained with a wide range of data sets from various types of feed materials and compositions, different extruders and different sets of extrusion conditions, then generalized models can be developed. Neural network models can be used effectively for extrusion process control.
Subject Area
Agricultural engineering|Food science|Mechanical engineering
Recommended Citation
Ganjyal, Girish M, "Extrusion of starches: Expansion and modeling with neural networks" (2003). ETD collection for University of Nebraska-Lincoln. AAI3117799.
https://digitalcommons.unl.edu/dissertations/AAI3117799