Computer-aided design of a dielectric insert supporting uniformity of fast microwave heating

Abstract

Purpose This paper aims to introduce and illustrate a computational technique capable of determining the geometry and complex permittivity of a supplementary dielectric insert making distributions of microwave-induced dissipated power within the processed material as uniform as possible. Design/methodology/approach The proposed technique is based on a 3D electromagnetic model of the cavity containing both the processed material and the insert. Optimization problem is formulated for design variables (geometrical and material parameters of the insert) identified from computational tests and an objective function (the relative standard deviation [RSD]) introduced as a metric of the field uniformity. Numerical inversion is performed with the method of sequential quadratic programming. Findings Functionality of the procedure is illustrated by synthesis of a dielectric insert in an applicator for microwave fixation. Optimization is completed for four design variables (two geometrical parameters, dielectric constant and the loss factor of the insert) with 1,000 points in the database. The best three optimal solutions provide RSD approximately 20 per cent, whereas for the patterns corresponding to all 1,000 non-optimized (randomly chosen) sets of design variables this metric is in the interval from 27 to 136 per cent with the average of 78 per cent. Research limitations/implications As microwave thermal processing is intrinsically inhomogeneous and the heating time is not a part of the underlying model, the procedure is able to lead only to a certain degree of closeness to uniformity and is intended for applications with high heating rates. The initial phase of computational identification of design variables and their bounds is therefore very important and may pre-condition the “quality” of the optimal solution. The technique may work more efficiently in combination with advanced optimization techniques dealing with “smart” (rather than random) generation of the data; for the use with more general microwave heating processes characterized by lower heating rates, the technique has to use the metric of non-uniformity involving temperature and heating time. Practical implications While the procedure can be used for computer-aided design (CAD) of microwave applicators, a related practical limitation may emerge from the fact that the material with particular complex permittivity (determined in the course of optimization) may not exist. In such cases, the procedure can be rerun for the constant values of material parameters of the available medium mostly close to the optimal ones to tune geometrical parameters of the insert. Special manufacturing techniques capable of producing a material with required complex permittivity also may be a practical option here. Originality/value Non-uniformity of microwave heating remains a key challenge in the design of many practical applicators. This paper suggests a concept of a practical CAD and outlines corresponding computational procedure that could be used for designing a range of applied systems with high heating rates.