Control of Continuous Microwave Drying Process of Peanuts Using Remote Temperature Measurement

Process control methods are widely used in food industry for quality assurance purposes. The characteristics and requirements for various food processing operations are described by Mittal (1996) and McFarlane (1995). For peanut processing, a complex fuzzy control system for the roasting process, based in part on theoretical modeling (Landman et al., 1994) was developed by Davidson et al. (1999). The study reported here focuses on the process control for continuous microwave drying of peanuts, where the microwave energy delivered to the dryer can be changed as needed. The utilization of microwave energy in heating and drying applications has been addressed in many books (Decareau, 1985; Metaxa and Meredith, 1983). New designs, such as focusing structures and traveling wave (or planar) applicators, are currently used for heating of fluids (Coronel et al., 2003) and drying of agricultural commodities (Boldor et al. 2004). In the drying process, the most important processing parameter is product temperature. Traditional temperature sensors, such as thermocouples, cannot be used for measurements due to the inherent nature of microwaves. In addition, continuous monitoring of internal temperatures of seeds is impossible in continuous drying processes, even when using microwave-immune fiber optic temperature probes (Boldor et al., 2004). However, the correlation between the internal temperatures of peanuts and the surface temperature of the peanut bed (Table 1, Boldor et al., 2004) makes remote temperature measurement of peanut bed surfaces a feasible method for process control purposes. Remote surface temperature measurements are performed using either infrared thermocouples inserted at certain locations along the microwave waveguide, or a properly calibrated infrared imaging system (Goedeken et al., 1991). Materials and Methods Field dried peanuts (Runner and Virginia type) at moisture contents ranging from 22 to 52% (dry basis) were used in this study. Samples were shipped from USDA-ARS Peanut National Laboratory in Dawson, Georgia to the Department of Food Science at North Carolina State University during the months of September November of 2002. The microwave system included a curing chamber (a traveling wave planar applicator, IMS, Morrisville, NC) and a 5 kW microwave generator (IMS, Morrisville, NC). Air flow through the chamber was maintained at 25°C through an electrical fan and heater. The system was treated as a first order system with dead time, whose transient behavior can be represented in Laplace domain using the following transfer functions (Eqn.1) (Marlin, 2000): Infrared thermocouples Peanuts vz A B C F E D Microwaves Air flow ENTRANCE EXIT Figure 1. Schematic of the microwave drying system. Thermocouple groups In this study, process reaction curves as described in literature (Marlin, 2000) were used to determine the process parameters Kp, θ, and τp. The major advantage of this method is that the transfer functions of the sensor and the final control element are included in the model. The disadvantage is that the method is limited to first and second order systems with dead time. One of the most widely used process control method is feedback control using a PID (proportional, integral, derivative) algorithm, where response of the controller is proportional with the difference between the desired value (set point SP) and the measured value of the controlled variable (CV). The transfer function of the PID controller is: For determination of the initial control parameters Kc, τi, and τd, the authors used the Ciancone open-loop method (Ciancone and Marlin, 1990) for the two peanut varieties and the three initial moisture contents used in this study (Table 3). The lack of dead time for the sensors placed closer to entrance in the microwave applicator (Table 2) combined with the low signal-to-noise ratio permitted the use of a PI controller, with no derivative component. Table 1. Example of infrared thermocouples locations, grouping and relationship between internal and surface temperature for Runner type peanuts (Boldor et al., 2004). Gr. Sensor Location (m) Equation r 1 0.476 y = 1.77x 12.38 0.99 A 2 0.552 y = 1.81x 13.52 0.99 12 2.508 y = 1.77x 12.60 0.94 13 2.584 y = 1.87x 14.31 0.94 E 14 2.660 y = 1.90x 15.32 0.94 Table 2. Example of process parameters for Virginia and Runner type peanuts. Virginia Runners mc% Gr. A Gr. C mc% Gr. A Gr. C 44 6.292 4.287 52 4.771 3.222 Kp 22 6.277 4.288 33 4.824 4.096 44 0.288 1.4 52 0.45 1.6 τp 22 0.288 1.4 33 0.513 3.988 44 -0.038 -0.083 52 0.075 1.117 θ 22 -0.038 -0.083 33 0.038 1.354 Surface temperature sensors were based on infrared technology (Mullin and Bows, 1993, Goedeken et al., 1991). Infrared thermocouples (model OS36-T, OMEGA Engineering, Inc., Stamford, CN) were placed at various distances along the waveguide as shown in Figure 1 (Boldor et al., 2004). The surface temperatures were monitored and recorded through a data acquisition and control unit (HP34970A, Agilent, Palo Alto, CA) and a master software routine written in LabView (National Instruments Corp., Austin, TX), that also controlled the microwave generator, and monitored and recorded the power levels in the microwave curing chamber through power diodes (JWF 50D-030+, JFW Industries, Inc., Indianapolis, IN). The feedback control routine was developed in Labview, using a simulation of the first order process with dead-time. Once the optimum control parameters were determined, the control routine was added to the master Labview program. Results and Discussions Due to infrared thermocouple positioning in groups of 2 or 3 (Figure 1), the process reaction curves were averaged to reduce the number of controlled variables from 16 to 6 (A, B, C, D, E, and F respectively), and furthermore, the last two groups were dropped due to very long dead-times and low process gains (Figure 2). The values for the open loop process parameters are shown in Table 2. 1 s e K