The empiric mathematical models to predict electrical properties of widely used natural woods in the industry by usi̇ng non-di̇stracti̇ve methods

In recent years, many applications in the industrial field have taken an interest in natural resources. Raw materials such as natural wood have been involved in many applications in microwave heating/drying systems, sustainable electronics, and lens applications. Therefore, material science engineering has a critical need for such research on the physical properties of natural wood. A total of nine commonly used natural woods have been prepared for this study. The selected natural wood can be classified into three softwoods (Cedar, Juniper, and Pine) and six hardwoods (Walnut, Sycamore, Aspen, Chestnut, Oak, and Eucalyptus). Each specimen has been carefully selected, without knots. The prepared wood materials were provided with a thickness variation between 0.4 mm and 0.6 mm, depending on the mechanical cut. Also, each wood material consisted of three samples with similar dimensions for the waveguide adapters that will be used in the measurement setup. However, the most important goal for industry is the response behavior of electrical properties under the variation of both moisture content and density conditions. In this study, the investigated electrical properties of the selected nine wood specimens were determined in the [2.17–6.0 GHz] frequency range. The measurements were illustrated in a suitable setup using VNA and three wave guide adapters (WR340, WR229, and WR159). A total of 49,500 softwood and 99,000 hardwood S-parameters have been collected, in which each specimen's measurement contains 500 raw data points. The collected s-parameters have been used in the wood specimen's dielectric properties calculations with MATLAB. The dielectric properties data results under different moisture and density conditions were used in generating an empirical mathematical model from the average data results of each measurement. The generated empirical mathematical models have a total of 36 empirical mathematical models constructed using softwood dielectric data results and 72 empirical mathematical models using hardwood dielectric data results. The presented natural wood specimens' empirical mathematical models were generated for both dielectric data results of permittivity and tangent loss. The empirical model from permittivity data results was generated with R2 over 0.9, which is considered to have a great prediction capability, while models from tangent loss results had R2 over 0.7, which is also considered a good prediction capability. Empirical mathematical evaluations were also performed to examine the performance of the generated models. For evaluation, a new set of wood specimens has been prepared with different thicknesses and different density conditions. This will allow the empirical model performance examination to be evaluated under such conditions. The evaluation results of empirical models performance in predicting the dielectric permittivity of natural wood materials were considered to be very good, with an accuracy of 85%. This prediction result of the generated empirical models can be relied on even under such different conditions from the previously prepared wood specimen sets that were used in generating these empirical models. The aim of this study is to have a detailed analysis of the selected natural woods and their electrical property variations with different moisture content and density levels, while the generated empirical models will provide the possibility of determining electrical properties without requiring any additional or destructive microwave measurement methods.