Computer-aided Process Engineering in the Snack Food Industry
Michael Nikolaou
Chemical Engineering Department
Texas A&M University
College Station, TX 77843-3122
Presented at
CPC-V, Tahoe City, January 1996
Abstract
Food science and engineering traditionally have been the basis for the development of the process technology related to the manufacture of snack foods. While these disciplines remain important, new challenges for the snack food industry have created an incentive to explore the potential of recent advances in computer-aided process engineering (CAPE). Such challenges include consistent quality, productivity, safety, and environmental and consumer friendliness. In this paper, we briefly describe basic snack food production processes, and compare them to processes with which most chemical engineers are more familiar. A number of snack food process engineering tasks for which computers and chemical engineering principles can provide powerful aids are outlined, and the fundamental and practical problems associated with these tasks are identified. Examples of our experience with theoretical and applied developments for specific snack food processes are provided.
Keywords
Computer aided process engineering, Snack food.
Introduction
Snack foods figure eminently in America's dietary and business profiles. In 1993, Americans consumed 3.48 billion lb of salty snacks – potato chips, corn and tortilla chips, extruded and fabricated snacks, and multigrain chips (Snack Food, June 1994, p. 42) – an amount corresponding to 13.5 lb per capita. The 1993 ex-factory snack food sales were $42 billion (Snack Foods, 1994), the retail value of which was close to $60 billion. While the US. snack food market is dominated by giant corporations, many small companies are flourishing by concentrating on specialty snacks.
New developments are taking place, given the benefits of new knowledge in nutrition science and new process technologies (Shukla, 1994; Tettweiler, 1991). Snack foods and processes are being thoroughly revamped or redesigned, either under regulatory mandate or under the food industry's self-imposed guidelines for healthful public nutrition. This is a formidable activity, necessitating the use of talent from several disciplines, including chemistry, biology, food science, agricultural & food engineering, and chemical & process engineering. Computer-aided process engineering (CAPE), in particular, is an important tool that can be used effectively in the development of flexible, safe, cost-effective, high quality snack food production processes. CAPE can contribute solutions to a broad range of snack food process technologies. For salty snacks such technologies include ultrafiltration and reverse osmosis for concentration of fresh fruits and vegetables; extrusion for blending, mixing, cooking, reacting, in-process flavor production, forming, and texturizing; and microwave processing for drying, dehydration, proofing, and puffing (Shukla, 1994).
This paper is an attempt to give an overview of the snack food industry from a CAPE perspective. The author's experience with the snack food industry during the last four years has been both fruitful and enjoyable. Hoping to broaden the discussion between the CAPE and snack food communities, the author discusses open problems and possible future directions throughout the text.
Snack Food Products
There is a multitude of products that the snack food industry produces. The list is growing steadily, as competition forces companies to introduce snacks with refined features, such as new raw material basis, improved texture, shape, color, flavor, and nutritional content. Basic product categories are
potato chips; meat-based snacks; popcorn-based snacks; puffed snacks; corn/tortilla chips and simulated potato chips; nut-based snacks.
Snack Food Processes and Equipment
There is a relatively small number of processes that appear most frequently in snack food production plants. A brief description of the most important processes follows. The discussion emphasizes elements that are important from a CAPE viewpoint.
Extrusion
Food extrusion simultaneously achieves three functions: (a) mixing of raw materials – e.g. various ground grains and water, (b) shearing/kneading, and (c) heating and/or cooking, using the heat released by friction inside the extruder (Harper, 1981; Frame, 1984). For some extrusion processes, the last function may be followed by puffing, i.e. abrupt expansion of high-temperature plasticized and gelatinized cornmeal from a pressurized chamber into the atmosphere (Matz, 1975). The control of extrusion cooking processes is a formidable challenge, because of variability of raw materials, process modeling difficulties, frequent wear of equipment, and various external and internal disturbances. A case study is discussed in a subsequent section of this paper.
Frying
This is a process in which a snack is cooked by floating or being immersed in hot oil that provides heat at high transfer rates. Continuous fryers are used for large scale operations (~5,000 lb/hr throughput), while batch (kettle-style) fryers are experiencing a comeback in small scale enterprises (<200 lb/hr). The frying fat or oil becomes a significant component of the end product, varying from as little as 10% by weight in breaded fish sticks to 40% in potato chips (Matz, 1975). Responding to consumer preferences, the snack food industry is now emphasizing good-tasting low-fat snacks. The design and operation of manufacture processes for such snacks poses a plethora of exciting CAPE problems.
Baking
In baking processes snacks are cooked by heat transferred through the air by convection, conduction, or radiation. The effectiveness of each of these mechanisms varies with oven and product design, and is generally lower than that of heat transfer based on liquids such as oil. Baking does not contribute any added fat to the finished snack. This may result in more healthful products, a fact that explains the recent surge in popularity of baked snacks.
Drying
Drying is required for several snacks to develop the right crispness. Puffed snacks, for example, must be dried after extrusion in order to obtain moisture content sufficiently low (to ~4%) to insure good texture and storage stability. Because a puffed snack with moisture level above or below normal feels stale or spongy, there is a need for accurate control of the drying process..
Packaging
Challenges include accurate control of the snack weight contained in each package, and safe extension of shelf storage life.
Miscellaneous
(a) Nut Processing equipment: Sorters, blanchers, roasters, coolers. (b) Oil, Powder, and Granule Applicators: Oil and cheese sprayers, powder dispensers, electrostatic salters, coating tumblers. (c) Transfer and Storage Equipment. (d) Measuring and Weighing Equipment
Comparison with Other Industries
Raw materials variability
Because most raw materials for the snack food industry are natural products, they can exhibit substantial variability. This may be due to the variety of a particular crop, location of cultivation, weather conditions during crop growing and harvesting, and storage conditions. The chemistry of natural products can be particularly complex. Similar concerns exist for other process industries, such as the petroleum, metal, and pulp and paper industries, that rely on the primary production sector for raw materials (e.g., oil, metal ores, and timber).
Unit operations
While the majority of the traditional chemical processes deal mostly with fluids, snack food processes deal mostly with solids or highly viscous fluids. Heat transfer is common in snack food production. Mechanical or mass transfer based separations are not uncommon. Cleaning, peeling, slicing, washing, crushing, and seasoning are common solids handling unit operations.
Snack Food Challenges and Relevant CAPE Tools
Product and process development
Computer modeling of new snack products and processes can substantially speed up their development. Examples are (a) the design of healthful snacks; (b) the development of flexible processes that can reliably operate on a multitude of raw materials and are capable of producing a variety of products; (c) the design of environmentally benign snack food processes. Examples of environmental issues are (a) the treatment of wastewater containing considerable amounts of starch and peel fragments resulting from potato peeling, slicing, and washing operations; (b) the containment of oil or particulate solid containing fumes resulting from frying. In addition to design tasks, process models can be used for personnel training, and process control.
Scheduling
Unlike oil and petrochemical products, snacks can be stored for a relatively limited time only. In addition, demand may vary unexpectedly and quite rapidly. This creates the need for just-in-time production methods. Given that production equipment is limited, the selection of what snack to produce, at what quantity, and in which production line, become particularly important scheduling issues.
Process control
Variables to control include temperature, pressure, moisture, oil content, color, shape, and texture. Image-based control is not uncommon. It should be stressed that, unlike oil and petrochemicals, snack food blending is hardly ever possible, a fact that renders it imperative for the product to be within specifications during as large a percentage of the production time as possible. Controlled variables may be difficult or impossible to measure on-line, thus creating a need to infer values from secondary measurements, and, ultimately, a need for new sensors. Quality control laboratory results, produced at ordinary but infrequent intervals, need to be integrated with on-line process control methods. During frying, for example, one needs to adjust the amount of fresh oil exchanged with used oil, to counter the effects of free fatty acids (released from the oil triglycerides and imparting a foamy or even soapy feel to the oil) and rancid-tasting compounds produced from oil oxidation (Robards et al., 1988). Control systems that compensate for equipment wear and can indicate when equipment maintenance is necessary need to be perfected. In extrusion cooking, for example, there is a need to know when screw wear can no longer be compensated for by feedback control and equipment maintenance is needed.
An adverse result of effective control systems is that they may offer few opportunities for novice operators to understand the dynamics of a process in depth. This situation might jeopardize the smooth operation of a process when circumstances are encountered that cannot be handled autonomously by the process control system, thus requiring human intervention (Brooks, 1995).
Statistical methods
Statistics can be used to extract workable knowledge from an abundance of process data available from real time measurements or in databases. One such case, as previously mentioned, is integration of statistical quality control results from the quality control laboratory with on-line process control. Statistics are also used extensively in translating subjective sensory judgments about the quality of a snack to numerical values of appropriate variables (Vickers, 1987; Defreitas and Molins, 1988; Prinyawiwatkul, 1993).
Software engineering and artificial intelligence
An idea is to use as many forms of information available in the plant as possible, such as past operator experience, heuristics, complex food chemistry, and qualitative objectives and constraints. Computer system integration, from the factory floor to the corporate headquarters, is a task that is showing signs of becoming feasible.
Case Studies
Extrusion process
Twin screw extrusion has found widespread application in extrusion cooking, because of several advantages, most important of which is the possibility of good process control (Straka, 1985). However, varied usage of the term control in the past seems to have contributed to misunderstandings, and to have stifled progress in research on extrusion cooking control, both in industry and academia. What most studies refer to as extrusion control, means, in reality, design of an extrusion process with the ability to operate at certain steady state operating conditions (Matz, 1975, p. 135; Falcone and Phillips, 1988). For example, numerous studies have concentrated on controlling pressure inside an extruder. While an extruder must be able to reach high enough pressures, regulation of pressure does not guarantee product quality, because the latter is affected by numerous other factors, not all of which can be taken into account. Direct feedback (or feedback-feedforward) control based on direct measurement of variables associated with product quality clearly provides better opportunities for regulatory control.
The measurement of variables associated with product quality raises two fundamental issues:
1. What variables can be used to infer product quality?
2. How can these variables be measured on-line?
The first question can be addressed using statistical approaches, that relate human judgment of the product quality to number-valued variables. The second question is related to the development of sensors and analytical methods for product evaluation. Experience has shown that common sense does not always provide reliable solutions to the above problems (Vickers, 1987).
In our experience we dealt with an extrusion cooking system for puffed snacks. We first collaborated with our industrial sponsor to address the two questions raised above. Subsequently, we identified the following fundamental questions to address:
How can one configure a control system?
How much accuracy is needed in a process model for control purposes? When do I stop identification experiments?
Are constraints important for the controller, thus requiring an on-line optimization based control system?
Are nonlinear modeling and control required?
How can a model be refined under closed-loop control, with minimum penalty on product quality?
Addressing the above questions requires good understanding of both the underlying physical process and the current state of the art in control theory. It is evident, however, that the above questions are, by no means, confined to snack food process control, but emerge naturally in a wide range of process control problems.
For the particular problem, we decided to work in a model predictive control (MPC) framework. Excellent reviews on MPC and related issues are provided by Lee (1995) and Mayne (1995) elsewhere in these proceedings. Therefore, we do not attempt to present the state of the art or review related work by other investigators in this particular area. Instead, we summarize below our contributions to research in this field, and explain the significance and usability of our results.
Robust stability of constrained MPC – On the basis of the seminal work of Rawlings and Muske (1993), theoretical results that can be used for the design of constrained MPC systems with robust stability and performance were obtained in Genceli and Nikolaou (1993) for MPC with l1 objective, and Vuthandam et al. (1995) for MPC with quadratic objective. These results were extended to nonlinear MPC systems with second-order Volterra models (Genceli and Nikolaou, 1995a) and to non-square MPC systems (Sarimveis et al., 1995a, b). A comparison between simulation and experiment for a puffed snack extrusion cooking system is shown in Fig. 1.
Performance of constrained MPC – A common theme of the above results is that they create a connection among (a) a process model and its accuracy, expressed as upper and lower bounds in the model's parameters; (b) the optimization and control horizon lengths; (c) the minimum values of the input move suppression coefficients in the on-line objective function. For overdamped or "slightly" underdamped systems, the smallest values of the input move suppression coefficients allow the most aggressive control action, thereby producing the tightest control. This realization has the following implications:
Control system configurations that result in the smallest values of the input move suppression coefficients, for the same level of modeling uncertainty, are preferable.
Identification experiments should produce modeling accuracy (e.g., bounds for model parameters) such that the resulting closed-loop performance bound can be satisfactory.
To decide whether to use a linear or nonlinear model in constrained MPC, one can compare the closed-loop performance bounds corresponding to the different uncertainty levels of the two classes of models.
Simultaneous MPC and Identification (MPCI) – A first attempt for simultaneous constrained MPC and process identification is presented in Genceli and Nikolaou (1995b) and Shouche et al. (1995). The idea is to conduct the on-line optimization over a set of process inputs that satisfy a persistent excitation condition, in addition to all standard MPC constraints. The resulting on-line optimization problem is non-convex. An approach to its solution, based on the solution of a series of semi-definite programming problems is shown in Genceli and Nikolaou (1995b). The importance of MPCI is that it performs process identification under closed-loop MPC, without resorting to external excitation of the closed loop. Therefore, the perturbation of the process output is kept at a minimum level, while, at the same time, process identification remains feasible.
Multiscale MPC – This is a recent effort to unify phenomena, control objectives, and constraints that occur at different time scales. The goal is to handle situations where frequency- or time-domain analyses alone cannot capture the essential features of signals that have time-varying frequency spectra. A first attempt is described in Koulouris et al. (1995), where a wavelet-domain framework is employed for external disturbance prediction and subsequent incorporation in the objective function of an MPC system.
Potato chip frying process
Frying is one step of an integrated potato chip manufacture line that includes a sequence of operations: cleaning, peeling and spotting, slicing, washing, frying, seasoning, and packaging. The exposure of a potato chip to hot oil during frying has several effects (Mottur, 1989):
Starch, the predominant food component of potato (up to 99.5% of dry weight), is gelatinized, thus rendering the potato more easily digestible.
The water content of the slice is reduced from about 80% to about 1-2%, resulting in a desirable crisp texture and stability from microbiological spoilage.
Cooking oil is absorbed into the chip, enhancing flavor and texture.
The level and variety of flavor compounds are greatly increased over those present in the raw potato.
The design and operation of fryers poses a variety of CAPE problems such as the following:
How can the process be optimized through better design and/or operation (e.g., for minimum oil uptake by the potato chip)?
At what level of detail should a first-principles process model be developed for design and optimization purposes?
At what level of detail should a process model be developed for control purposes?
How accurate does a process model have to be?
How can on-line and quality control laboratory measurements, obtained at different time scales, be used effectively in process control?
What control strategy provides adequate control, without excessive design and maintenance requirements?
Tight process control of snack chip frying offers significant economic incentives. It is estimated that reduction of potato chip losses by 1% of throughput would reduce production costs by over $3 million per year for the single largest US snack producer alone (Brooks, 1995). There are interesting process control problems associated with potato chip frying, as outlined below:
Moisture control – It is important to end the frying process so that final moisture will be in the critical 1-2% range. A 1.4 mm thick slice at 180 C will reach this range after about 2 minutes of fry time. If moisture is above 2%, crispness will suffer. If it is below 1% it will result in excessive oiliness, dark color, and scorched flavor (Mottur, 1989).
Oil content control – Typical finished potato chips contain 30-40% oil. Since potato tissue contains only 0.1% fat, nearly all of the oil in the finished chip is due to absorption from the frying medium. If the oil content rises above 40% the chip becomes unpleasantly greasy with an oil-soaked appearance. For chips designated as low-fat, maintaining oil content below the level advertised on the product's package poses a constraint that must be satisfied throughout production.
Color control – There are two components in the color of the finished chip: (a) background color, and (b) dark spots. Background color, mostly due to starch caramelization, is associated with the length of frying period, oil temperature, and slice thickness (Talburt and Smith, 1975). Background color is relatively straightforward to control at a golden yellow level, using feedforward/feedback control. Dark spots, a feature undesirable to consumers, are due to the Maillard reaction of reducing sugars (mostly sucrose, fructose, and glucose (Yada et al., 1985)) with aminoacids, that occurs during frying. While the extent of the Maillard reaction can be partially controlled, it is mostly the sugar content of the potato that determines whether dark spots will appear or not. Sugar concentration may increase during potato storage at low temperature (2-4 C) due to slowing of the breakdown of sugars to carbon dioxide and water (Snackfood Association, 1987; Blankson et al., 1988). Sugar content can be lowered back to an acceptable level (<0.2-0.4%) by reconditioning (storage at >13 C for several weeks prior to use). It is clear that knowledge of the sugar concentration before frying can be valuable for use in feedforward-feedback control schemes.
The above process control tasks are challenging because of factors such as the following:
large dead-time (a few minutes, required for passage of chips through the fryer);
time-varying dynamic behavior (e.g., due to oil degradation, or variation of raw materials due to different sources and/or storage conditions (Evensen et al., 1988; Louwes and Neele, 1987; Sieczka and Maatta, 1986));
nonlinearity (e.g., due to chemical reaction kinetics);
strong coupling of variables (e.g. moisture and oil content of the cooked chip);
constraints on process variables (e.g., specifications for "low-fat" product designation).
In addressing the above issues, we started with the development of a first-principles model for a continuous fryer (Feng et al., 1995). The objective was to develop a model usable in the design of control systems for several variations of continuous fryers, without requiring excessive experimentation for each particular fryer. The challenge for such a model is to decide what physical and chemical phenomena to incorporate, and at what level of detail. There have been several studies on individual phenomena that occur during frying, such as
chip texture formation (Pravisani and Calvelo, 1986) ;
chip color formation (Buera et al., 1987; Coffin et al., 1987; Dahlenburg, 1982; Habib and Brown, 1957; Lee et al., 1984; Leszjowiat et al., 1990; Lyman and Mackay, 1961; Marquez and Anon, 1986; Picha, 1986; Pravisani et al., 1986; Roe and Faulks, 1991; Roe et al., 1990; Talburt and Smith, 1975);
oil penetration into the chip (Farkas, 1994; Keller et al., 1986; Lamberg et al., 1990);
oil uptake by the chip (Gamble et al., 1993; Miller et al., 1975; Nonaka et al., 1977; Pinthus et al., 1993; Sayer et al., 1975; Shouche and Feng, 1995; Talburt and Smith, 1975);
heat and mass transfer in the chip (Keller et al., 1986; Mittelman et al., 1984; Rice and Gamble, 1989).
A list of models was constructed and a heuristic sensitivity analysis was conducted, to determine the effects of various model parameters on the prediction of controlled variables. A model was selected that showed very good agreement between predicted and measured values for controlled variables. The computer implementation of the model was done in a Matlab-Simulikn environment. Simulations produced by the model are shown in Fig. 2. Testing of the model in an MPC system is underway.
Conclusions
There is hardly an economic commodity more important for our lives than food. Snack foods, in particular, are a substantial part of Americans' diet. Salty snack consumption could triple in a decade, as eating on the run and grazing are becoming more common (Shukla, 1994). This makes the improvement of snack foods and processes a high-priority economic opportunity and social responsibility. The US snack food industry is hard at work addressing the marketing and technological challenges associated with the production of better snacks. This is a multifaceted endeavor, requiring talent and tools from several disciplines. The academic CAPE community, strongly represented in chemical engineering, can be a substantial contributor to that effort. Cross fertilization of ideas common in the chemical and food industries can offer to each industry a fresh view of related challenges and possible solutions. Collaboration between academia and the snack food industry can identify exciting research problems to work on. This can steer research to fruitful directions for the production of better snacks.
Acknowledgment – This paper is based on work partially supported by NSF, Shell Development, Frito-Lay, and the Texas Advanced Technology Program. The author gratefully acknowledges that support.
References
Blankson, J. E., Coffin R. H., Yada R., Lougheed E. C. (1988). The Effects of Moderate Carbon-Dioxide Levels in Storage upon Potato-Chip Color, Canadian Journal of Plant Science, Vol 68, Iss 2, pp 567-568.
Brooks, A. (1995). Personal Communication with the Author.
Buera, M.P., J. Chirife, S.L. Resnik and R.D. Lozano, (1987). J. Food Sci., 52, 4, 1059.
Coffin, R.H., R.Y. Yada, K. L. Parkin, B. Grodzinski and D.W. Stanley, (1987). J. Food Sci., 52, 3, 639.
Dahlenburg, A.P., (1982). Food Tech. in Australia, 34, 11, 544.
Defreitas, Z., Molins R. A. (1988). Development of Meat Snack DIPS - Chemical, Physical, Microbiological and Sensory Characteristics, Journal of Food Science, Vol 53, Iss 6, pp 1645-1649.
Evensen, K. B., Russo J. M., Braun H. (1988). Predicting Potato-Chip Quality and Yield, HortScience, Vol 23, Iss 3, pp 728-728.
Falcone, R. G., Phillips R. D. (1988). Effects of Feed Composition, Feed Moisture, and Barrel Temperature on the Physical and Rheological Properties of Snack-Like Products Prepared from Cowpea and Sorghum Flours by Extrusion, Journal of Food Science, Vol 53, Iss 5, pp 1464-1469.
Shouche, M. S., H. Genceli, and Michael Nikolaou (1995). Model Predictive Control and Identification of Time-Varying Processes, submitted to Automatica.
Farkas, B. E., (1994). Ph.D. Dissertation, UC, Davis.
Feng, W Y., M. Shouche, and M. Nikolaou (1995). Modeling and Predictive Control of a Continuous Fryer, paper 173j, AIChE Annual Meeting, Miami Beach, FL.
Frame, N. D. (1994). The Technology of extrusion cooking, 1st ed., Blackie Academic & Professional.
Gamble M.H , P. Rice, and J.D. Selman, (1987). Int. J. of Food Sci. and Tech., 22, 233-241.
Genceli, H., and M. Nikolaou (1993). Robust Stability Analysis of Constrained l1-Norm Model Predictive Control, AIChE J., 39, 12, 1954-1965.
Genceli, H., and M. Nikolaou (1995a). Design of Constrained Model-Predictive Controllers with Volterra Series, AIChE J., 41, 9, 2098-2107.
Genceli, H., and M. Nikolaou (1995b). A New Approach to Simultaneous Model Predictive Control and Identification, AIChE J., accepted.
Habib, A. T , and Brown, H. D, (1957). Food Technol., 11, 85.
Harper, J. M. (1981). Extrusion of foods, CRC Press, Boca Raton, Fla..
Keller C. F., Escher., and J. A. Solms, (1986). Lebensmittel-Wissenschaft & Technologie, 19, 4, 346-348.
Koulouris, A., M. Nikolaou, and Geo. Stephanopoulos (1995). Multiscale MPC, journal paper in preparation.
Lamberg, I., B. Hallstrom, and H. Olsson, (1990). Lebensmittel-Wissenschaft & Technologie, 23, 4, 245-251.
Lee, C.M., B. Sherr and Y.N. Koh, (1984). J. Agric. Food Chem., 32, 379-382.
Lee. J. H. (1995). Recent Advances in Model Predictive Control and Other Related Fields, CPC V Proceedings.
Leszjowiat, M.J., V. Barichello, R.Y. Yada, R.H. Coffin, E.C. Lougheed, and D.W. Stanley, (1990). J. Food Sci., 55, 1.
Louwes, K. M., Neele A. E. F. (1987). Selection for Chip Quality and Specific-Gravity of Potato Clones - Possibilities for Early Generation Selection, Potato Research, Vol 30, Iss 2, pp 241-251.
Lyman, S. and A. Mackay, (1961). American Potato J, 38:51.
Marquez, G and M. C. Anon, (1986). J. Food. Sci., 51, 1, 157.
Matz, S. A. (1975). Snack food technology, 1rd ed., Van Nostrand Reinhold, New York; 3rd ed. 1993.
Mayne, D. (1995). Optimization in Model Based Control, CPC V Proceedings.
Miller, R.A., J.D. Harrington and G.D. Kuhn, (1975). Am. Potato J., 52, 379.
Mittelman, N., S. Mizrahi, and Z. Berk, (1984). Engineering and Food, Ch. 12, B. M. McKenna (Ed.), Elsevier 109-116.
Mottur, G. P. (1989). A Scientific Look At Potato-Chips - The Original Savory Snack, Cereal Foods World, Vol 34, Iss 8, pp 620-626.
Nonaka, M., R. N. Sayre., and M. L. Weaver, (1977). Am. Potato J., 54, 151-159.
Orr, P.H., Janardan K. G. (1990). A Procedure to Correlate Color Measuring Systems Using Potato-Chip Samples, American Potato Journal, Vol 67, Iss 9, pp 647-654
Picha, D. H. (1986). Influence of Storage Duration and Temperature on Sweet-Potato Sugar Content and Chip Color, Journal of Food Science, Vol 51, Iss 1, pp 239-240.
Picha, D.H., (1986). J. Food Sci., 51, 1, 239.
Pinthus, E. J., P. Weinberg, and I. S. Saguy, (1993). J. Food. Sci., 58, 1, 204.
Pravisani, C. I., and A. Calvelo, (1986). J. Food Science, 51, 3, 614 .
Prinyawiwatkul, W., Beuchat L. R., Resurreccion A. V. A. (1993): Optimization of Sensory Qualities of an Extruded Snack Based on Cornstarch and Peanut Flour, Food Science and Technology-Lebensmittel-Wissenschaft & Technologie, Vol 26, Iss 5, pp 393-399.
Rawlings, J. B., and K. R. Muske (1993). The Stability of Constrained Receding Horizon Control, IEEE Trans. AC, 38, 10, 1512.
Rice, P., M. H. Gamble, (1989). Int. J. of Food Sci. and Tech., 24, 183-187.
Robards, K., Kerr A. F., Patsalides E. (1988). Rancidity and Its Measurement in Edible Oils and Snack Foods - A Review, Analyst, Vol 113, Iss 2, pp 213-224.
Roe, M.A., and R.M. Faulks, (1991). J. Food Sci., 56, 6, 1711.
Roe, M.A., R.M. Faulks and J.L. Belson, (1990). J. Sci. Food Agric., 52, 207-214.
Sayre, R. N., M. Nonaka, M. L. Weaver, (1975). Amer. Potato J., 52, 73-82.
Shukla, T. P. (1994). Future Snacks and Snack Food-Technology, Cereal Foods World, Vol 39, Iss 9, pp 704-705.
Sieczka, J. B., Maatta C. (1986). The Effects of Handling on Chip Color and Sugar Content of Potato-Tubers, American Potato Journal, Vol 63, Iss 7, pp 363-372.
Snackfood Association, (1987). Fifty Years: A Foundation for the Future, Alexandria, VA.
Sowokinos, J. R., Knoper J. A ., Orr P. H., Varns J. L. (1987). Influence of Potato Storage and Handling Stress on Sugars, Chip Quality and Integrity of the Starch (Amyloplast) Membrane, American Potato Journal, Vol 64, Iss 5, pp 213-226
Straka, R. (1985). Twin-Screw and Single-Screw Extruders for the Cereal and Snack Industry, Cereal Foods World, Vol 30, Iss 5, pp 329-332.
Strock, H., C. O. Ball, S. S. Chang, and E. F. Stier, (1966). Food Technology, 20, 4, 193-196.
Talburt, W. F., and O. Smith (1975). Potato Processing, 3rd Ed., AVI Publishing, Westport, Connecticut.
Tettweiler, P. (1991). Snack Foods Worldwide, Food Technology, 45, 2, pp 58-62.
Vickers, Z. M. (1987). Sensory, Acoustical, and Force-Deformation Measurements of Potato-Chip Crispness, Journal of Food Science, Vol 52, Iss 1, pp 138-140.
Vuthandam, P., H. Genceli, and M. Nikolaou (1995). Performance Bounds for Robust Dynamic Matrix Control with End-Condition, AIChE J., 41, 9, 2083-2097.
Warner, K., Orr P., Parrott L., Glynn M. (1994). Effects of Frying Oil Composition on Potato-Chip Stability, Jounal of the Americal Oil Chemists Society, Vol 71, Iss 10, pp 1117-1121.
Yada, R. Y., Stanley D. W., Fitts M., Coffin R. H., Leszlowiat M. J. (1985). Effect of Sucrose, Glucose and Fructose Content of Ontario Grown Potato-Tubers on Chip Color, American Potato Journal, Vol 62, Iss 8, pp 450-451.
Figure 1a. Twin-screw extruder.
Figure 1b. Response of product quality variables to setpoint changes, for a closed loop with a 3x2 controller.
Figure 1c. Response of product quality variables to change in the raw material, for a closed loop with a 3x2 controller.
Figure 2a. Overview of the fryer model.
Figure 2b. Steady state profiles along the fryer predicted by the model.
Figure 2c. Fryer dynamics predicted by the model.
Figure 2d. Simulation of closed-loop response to step change in the thickness of potato slices.
Figure 2e. Simulation of response of manipulated inputs to step change in the thickness of potato slices.