Analyze of leaf springs with parametric finite element analysis and artificial neural network


Creative Commons License

Yavuz S., ÖZKAN M. T.

JOURNAL OF POLYTECHNIC-POLITEKNIK DERGISI, cilt.25, ss.827-842, 2022 (ESCI) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 25
  • Basım Tarihi: 2022
  • Doi Numarası: 10.2339/politeknik.1074516
  • Dergi Adı: JOURNAL OF POLYTECHNIC-POLITEKNIK DERGISI
  • Derginin Tarandığı İndeksler: Emerging Sources Citation Index (ESCI), TR DİZİN (ULAKBİM)
  • Sayfa Sayıları: ss.827-842
  • Anahtar Kelimeler: Heavy vehicles, suspension, conventional leaf spring, artificial neural networks
  • Gazi Üniversitesi Adresli: Evet

Özet

Leaf spring especially used in heavy vehicles. Leaf spring provide to increase the strength of heavy vehicles chassis and some components, absorb the shock loading due to some road condition and absorb the vibration. Because of these features, leaf springs are most used suspension element for the heavy vehicles. For this purpose, parametric rectangular cross sectional leaf spring design was obtained. For leaf spring design, FEA models of different variations were created and analyzed primarily between 1-10 layers using ANSYS software. The leaf spring layers and their dimensions were taken in varying ways according to the manufacturer's catalog. The effect of the number of leaf spring layers on element resistance and deformation was also simulated by changes in stresses. The number of layers between 1-10 and spring models of different section sizes were modeled parametrically in the ANSYS program and different variations were created by applying different load sizes. Mesh optimization of the model was performed in ANSYS software and all variations were solved. An Artificial Neural Networks Model was developed using stress-strain values depending on the design type and loading conditions obtained. Thus, depending on the number of layers, section measurement sizes and loading sizes, the spring constant (K), stress-strain values were estimated with high precision. Using the ANN model developed, the designer has put in place an approach that can be achieved quickly, easily and at a minimized design costs.