Artificial neural-network optimisation of nail size and spacings of plywood shear wall

The racking performance of shear walls, which is one of the most important elements of light-frame wooden structures, is affected by many factors such as the type of sheathing material, thickness, fibre direction and the size and spacing of the fasteners. Determining the most suitable production par...

Full description

Saved in:
Bibliographic Details
Published inWood material science and engineering Vol. 18; no. 1; pp. 97 - 106
Main Authors Demir, Aydin, Demirkir, Cenk, Ozsahin, Sukru, Aydin, Ismail
Format Journal Article
LanguageEnglish
Published Abingdon Taylor & Francis 02.01.2023
Taylor & Francis Ltd
Subjects
Artificial neural network
Artificial neural networks
Construction industry
Experimentation
Fasteners
Mathematical models
nail size
nail spacing
Neural networks
Optimization
Parameters
Plywood
Prediction models
racking performance
shear wall
Shear walls
Sheathing
Statistical analysis
Thickness
Wooden structures
Online AccessGet full text

Cover

More Information
Summary:The racking performance of shear walls, which is one of the most important elements of light-frame wooden structures, is affected by many factors such as the type of sheathing material, thickness, fibre direction and the size and spacing of the fasteners. Determining the most suitable production parameters is extremely necessary in terms of both time and cost. Therefore, it is aimed to predict the optimum nail size and spacing that gives the best racking performance of plywood shear walls produced with different production parameters using artificial neural networks in this study. The racking performances of shear walls produced with plywood with different wood species, thickness and fibre directions were determined according to ASTM E 72 - 13a standard and the maximum load and displacement values were obtained for each wall model as a result of the test. The prediction models having the best prediction performance were determined by means of statistical and graphical comparisons. It was observed that the prediction models yielded very satisfactory results with acceptable deviations. As a result, the findings of this study could be employed effectively in the building industry to reduce time, energy and cost for experimental studies within the range of experimentation conducted.
ISSN:1748-0272
1748-0280
DOI:10.1080/17480272.2021.1992648