Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Experimental validation of a data-driven fracture mechanics approach
The aim of this project is to validate a recently proposed fracture propagation data-driven approach for both rate-independent and rate-dependent crack phenomena using experimental data.
The constitutive relationships describing the material behavior in mechanics usually come from the attempt to distillate analytical rules from discrete experimental observations. This procedure involves uncertainties that interact with those inherent to the experimental testing leading to unpredictable effects.
The recent development of new techniques such as machine learning and/or neural networks allows overcoming this problem by directly substituting the constitutive relationships with raw experimental data avoiding. In this context, a theoretical framework dealing with the propagation of a fracture into an otherwise linear elastic material has been recently proposed but still lacks experimental validation.
The main aim of this project is to validate the fracture propagation data-driven approach for both rate-independent and rate-dependent crack phenomena using experimental data either from the literature or from lab experiments. The experimental results in terms of load, displacement and crack length will be compared to the numerical results coming from the application of the proposed framework. Further, the collected database can be used to study the effect of data variance and scattering on the main mechanical quantities.The experimental results in terms of load, displacement and crack length will be compared to the numerical results coming from the application of the proposed framework. Further, the collected database can be used to study the effect of data variance and scattering on the main mechanical quantities.
The constitutive relationships describing the material behavior in mechanics usually come from the attempt to distillate analytical rules from discrete experimental observations. This procedure involves uncertainties that interact with those inherent to the experimental testing leading to unpredictable effects.
The recent development of new techniques such as machine learning and/or neural networks allows overcoming this problem by directly substituting the constitutive relationships with raw experimental data avoiding. In this context, a theoretical framework dealing with the propagation of a fracture into an otherwise linear elastic material has been recently proposed but still lacks experimental validation.
The main aim of this project is to validate the fracture propagation data-driven approach for both rate-independent and rate-dependent crack phenomena using experimental data either from the literature or from lab experiments. The experimental results in terms of load, displacement and crack length will be compared to the numerical results coming from the application of the proposed framework. Further, the collected database can be used to study the effect of data variance and scattering on the main mechanical quantities.The experimental results in terms of load, displacement and crack length will be compared to the numerical results coming from the application of the proposed framework. Further, the collected database can be used to study the effect of data variance and scattering on the main mechanical quantities.
1. Literature study on fracture mechanics and data-driven approaches\\
2. Collection of experimental results databases and/or design of the experimental tests\\
3. Validation of the model-free data-driven fracture mechanics approach\\
4. Critical discussion and analysis of advantages and limits\\
After implementing the main tasks, the project may move on in multiple possible directions (e.g., extension to a multidimensional framework), giving room for inputs from the student's side.
1. Literature study on fracture mechanics and data-driven approaches\\ 2. Collection of experimental results databases and/or design of the experimental tests\\ 3. Validation of the model-free data-driven fracture mechanics approach\\ 4. Critical discussion and analysis of advantages and limits\\ After implementing the main tasks, the project may move on in multiple possible directions (e.g., extension to a multidimensional framework), giving room for inputs from the student's side.
Dr. Pietro Carrara
Computational Mechanics Group
Tannenstrasse 3, CLA J 17.1
E-mail: pcarrara@ethz.ch
Dr. Pietro Carrara Computational Mechanics Group Tannenstrasse 3, CLA J 17.1 E-mail: pcarrara@ethz.ch