J-integral analysis of cord-rubber serpentine belt using neural-network-based material modelling

• Published in: Fatigue & fracture of engineering materials & structures Vol. 28; no. 10; pp. 847 - 860
• Main Authors: SONG, G., CHANDRASHEKHARA, K., BREIG, W. F., KLEIN, D. L., OLIVER, L. R.
• Format: Journal Article
• Language: English
• Published: PO Box 1354, 9600 Garsington Road, Oxford OX4 2XG, UK Blackwell Science Ltd 01.10.2005; Blackwell Science; Wiley Subscription Services, Inc
• Subjects: Application fields; Applied sciences; Automotive parts; belt; Exact sciences and technology; Finite element analysis; finite-element; fracture; Fractures; J-integral; Neural networks; neural-network; Polymer industry, paints, wood; rubber; Technology of polymers; J integral; finite-element; belt; Rupture; Mechanical properties; Neural network; Modeling; fracture; neural-network; Finite element method; J-integral; Rubber; Application; Artificial intelligence
• ISSN: 8756-758X; 1460-2695
• DOI: 10.1111/j.1460-2695.2005.00917.x

ABSTRACT A known factor that limits the performance of automotive front‐end accessory serpentine belt drive is cracking of the elastomer located in the rib tip. In this paper, fracture experiments were conducted using single‐edge notched tension (SENT) specimens to study the fracture behaviour of a belt rib compound. A finite‐element modelling method utilizing singular elements for crack in rubber solid was proposed and implemented in both plane‐stress and 3D solid models using ABAQUS. A newly developed neural‐network‐based model was used to represent a nonlinear elastic belt rib rubber compound. The crack finite‐element model, along with the neural‐network‐based material model, was verified with analytical and experimental results. A global–local finite‐element procedure was developed to evaluate the J‐integral for mode‐I through‐the‐thickness crack in V‐ribbed belt rib. Effects of pre‐crack length, pulley pre‐load and backside pulley displacement were investigated.