Date

2022-11-22

Author

Tüfekçioğlu, Murat Enis

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

291
views

112
downloads

Cite This

Rubberlike materials, due to their structure, can undergo very high strains during loading and exhibit highly non-linear behavior. After the load is removed, they usually return to their pre-deformation shape, in a sense, no energy is lost during their deformation. Due to these properties, Rubberlike materials can be modeled as hyperelastic materials. The modeling approach of many people dealing with rubber is to choose one of the appropriate free energy/strain energy functions developed for rubber in the literature. Material parameter values of the chosen model are determined from the parameter sets that give the best fit to the material tests. Sometimes to find the best-fitted model, several models are checked and compared. Currently, there are over 40 free energy functions defined in the literature and new ones are added to them every day. Finding the best-fitted model can become a tiresome process and data-driven material models can become very useful tools in this context. Thanks to the general approach they offer, they can enable easy and accurate modeling for rubberlike materials that exhibit very different behavior characteristics according to their chemical structure, curing process, additives, and the ambient conditions in which they are used. Data-driven hyperelasticity is a promising approach for the constitutive modeling of rubberlike materials because it enables the direct use of experimental data for the construction of the stress-strain response without using any specific analytical expression for the strain energy density function. Instead of defining a strain energy function with physically meaningful material parameters, the partial derivatives of the strain energy density functions, which are used in stress expressions, are replaced with appropriate B-spline interpolations. Those B-spline interpolations have a set of control points that are defined from various multiaxial loading scenarios such as uniaxial tension, pure shear, and (equi)biaxial tension deformations. These control points can also be defined as parameters of the data-driven material model. The convexity requirement is also enforced through those control points to ensure a convex and stable constitutive response. The thesis starts with the definitions of B-splines with different degrees and different control points and continues with the introduction of the developed B-spline algorithm. In the sequel, different kinematic approaches to be adopted in data-driven modeling are explained. The proposed data-driven models for each kinematic approach are tested and compared for different degrees and control point numbers using Kawabata and Treloar data sets. Finally, the success of the proposed models in capturing the mechanical behavior for different deformation modes and strain levels is demonstrated by the quality of fit criteria.

Subject Keywords

data-driven approach, hyperelastic material modeling, b-spline generation, invariant based material modeling, optimization

URI

https://hdl.handle.net/11511/101184

Collections

Graduate School of Natural and Applied Sciences, Thesis