Experimental and numerical modelling of mechanical properties of 3D printed honeycomb structures

• Published in: Measurement : journal of the International Measurement Confederation 2018-02, Vol.116, p.495-506
• Main Authors: Panda, Biranchi, Leite, Marco, Biswal, Bibhuti Bhusan, Niu, Xiaodong, Garg, Akhil
• Format: Article
• Language: English
• Subjects: 3-D technology; 3D printing (3DP); Acoustic insulation; Acoustic properties; Aerospace industry; Automobile industry; Automotive engineering; Cells; Cellular structure; Cellular structures; Computational intelligence (CI); Design parameters; Fuel consumption; Fused deposition modeling; Genetic algorithms; Honeycomb construction; Honeycomb structures; Insulation; Mathematical models; Mechanical Properties; Modulus of elasticity; Neural networks; Numerical methods; Parameter estimation; Regression analysis; Response surface methodology; Statistical analysis; Stiffness; Strength to weight ratio; Surface analysis (chemical); Three dimensional models; Two dimensional models
• ISSN: 0263-2241; 1873-412X
• DOI: 10.1016/j.measurement.2017.11.037

•Cellular Honeycomb parts is fabricated using fused deposition modelling.•Effect of design parameters on strength of honeycomb parts is undertaken.•Experiments are performed to study strength by varying cell size and thickness.•Numerical modeling using GP and ANS is applied to formulate models.•Surface analysis and optimization is performed to optimize the strength. In recent years, 3-D printing experts have laid emphasis on designing and printing the cellular structures, since the key advantages (high strength to weight ratio, thermal and acoustical insulation properties) offered by these structures makes them highly versatile to be used in aerospace and automotive industries. In the present work, an experimental study is firstly conducted to study the effects of the design parameters (wall thickness and cell size) on the mechanical properties i.e yield strength and modulus of elasticity (stiffness) of honeycomb cellular structures printed by fused deposition modelling (FDM) process. Further, three promising numerical modelling methods based on computational intelligence (CI) such as genetic programming (GP), automated neural network search (ANS) and response surface regression (RSR) were applied and their performances were compared while formulating models for the two mechanical properties. Statistical analysis concluded that the ANS model performed the best followed by GP and RSR models. The experimental findings were validated by performing the 2-D, 3-D surface analysis on formulated models based on ANS.