Determination of the effect of laminated veneer lumber (LVL) moisture content on pressure resistance by artificial ıntelligence

• Main Authors: Eser Sözen, Kadir Kayahan, Timuçin Bardak
• Format: Article
• Language: English
• Published: Isparta University of Applied Sciences Faculty of Forestry 2021-06-01
• Series: Turkish Journal of Forestry
• Subjects: lvl; compression strength; moisture; artificial intelligence; decision trees; random forest; basınç direnci; rutubet; yapay zekâ; karar ağaçları; rastgele orman
• Online Access: https://dergipark.org.tr/tr/pub/tjf/issue/63166/888829

Wooden materials used in the building sector are exposed to different loading types and different strength depending on the place of use. The use of materials suitable for the type of loading affects important factors such as safety, performance and cost. Another important issue in wooden materials used in the building sector is wood-water relations. Moisture causes significant changes on the physical, mechanical and technological (hardness, wear) properties of wood. In this study, 5-layer LVL (Laminated Veneer Lumber) was produced from 2 mm beech (Fagus orientalis L.) veneer obtained by peeling process. Produced LVLs were subjected to four different moisture (0%, 12%, 18% and 25%) compressio strength in two different directions, perpendicular and parallel to the fibers. Using the data obtained from the specified moisture values, the pressure resistance values in other moisture amounts were estimated by artificial intelligence. Artificial Neural Networks (ANN), Decision Trees (DT) and Random Forest (RF) algorithms are used in the predictions. According to the mechanical test results, the highest compression strength value (51.96 N/mm²) was obtained in the loading parallel to the fibers of the samples with 0% moisture (oven dry). The lowest compression strength value (13.57 N/mm²) was determined in the loading vertical direction to the fibers of 25% moisture samples. The highest prediction success was obtained from the Random Forest algorithm with a value of R2 = 0.984. As a result, it has been determined that artificial intelligence techniques can be used successfully as a solution to predict the pressure resistance of LVLs at different humidity.