An Improved Optimization Model to Predict the MOR of Glulam Prepared by UF-Oxidized Starch Adhesive: A Hybrid...

Area of Science:

• Materials Science
• Wood Engineering
• Adhesives Technology

Background:

• Glued laminated timber (glulam) is a critical engineered wood product.
• Optimizing glulam properties requires understanding adhesive performance and wood characteristics.
• Urea-formaldehyde (UF) resins are common wood adhesives, but their properties can be modified.

Purpose of the Study:

• To evaluate the bending strength (modulus of rupture - MOR) of glulam.
• To investigate the influence of UF resin molar ratios and modified starch adhesive concentrations on glulam strength.
• To utilize artificial neural networks (ANN) for predicting glulam bending strength.

Main Methods:

• Glulam specimens were prepared using plane tree (Platanus Orientalis-L) wood layers bonded with UF resin at varying formaldehyde to urea molar ratios.
• Modified starch adhesive with different sodium hypochlorite (NaOCl) concentrations was incorporated.
• Multilayer perceptron (MLP) models within an ANN framework were employed to predict MOR.
• Genetic algorithms were combined with ANN for optimization.
• Fourier-transform infrared spectroscopy (FTIR) and X-ray diffraction (XRD) were used for characterization.

Main Results:

• The feed-forward neural network, utilizing Levenberg-Marquardt algorithms, demonstrated excellent prediction accuracy for MOR.
• Statistical validation metrics (R², RMSE, MAPE) confirmed the reliability of the ANN predictions.
• Optimization using genetic algorithms and ANN showed minimal differences between experimental and estimated optimal values.
• FTIR and XRD analyses provided insights into the input effects on glulam properties.

Conclusions:

• ANN models, particularly MLP with Levenberg-Marquardt, are highly effective for predicting the bending strength of modified glulam.
• The study successfully optimized glulam properties by considering UF resin ratios and modified starch adhesive concentrations.
• The findings contribute to the development of enhanced glulam with predictable performance characteristics.