Prediction of Optimum Expanded Polystyrene Densities for Best Thermal Insulation Performances of Polystyrene Composite Particleboards by Using Artificial Neural Network

Abstract

The objective of this study is to predict the optimum expanded polystyrene (EPS) densities for the best insulation properties of the particleboards manufactured with waste EPS instead of formaldehyde-based adhesives used in particleboard production with artificial neural network (ANN). For this purpose, the waste EPS particles of five different densities were used in the production of composite particleboards. The experimental dana used in the study were obtained from the previous study. Half of the beech, poplar, alder, pine and spruce chips were dried in a drying oven and the other half were naturally conditioned at room temperature, and then 18 mm thick three-layer composite particleboards were produced. The thermal conductivity of panels was determined according to ASTM C 518. The prediction model with the best performance and acceptable deviations was determined by using statistical and graphical comparisons between the experimental data and the prediction values obtained as a result of ANN analysis. Then, using this prediction model, the thermal conductivity coefficient values were estimated for the intermediate EPS densities that were not experimentally tested. According to the analysis findings, the thermal insulation performance for both beech and spruce polystyrene composite particleboards (PCP) panels increased with using of waste EPS foams with a density of 30 kg/m3. The lowest thermal conductivity values were obtained from the EPS waste foams with the density of 18, 13 and 22 kg/m3 for the PCP panels produced with poplar, alder and pine in the natural drying, respectively. In the technical drying, these values were found to be 15, 14 and 11-13 kg/m3, respectively. Technical drying showed much better thermal performance than natural drying while poplar indicated the best performance among the wood species.