## Objectives and Research Methods - Strength Prognosis JbyF

**Objectives**

- Enabling a prognosis of the joint strength for self-pierce riveting with semi-tubular rivets (SPR-ST) and clinching
- Extended experimental and numerical data describing the correlation of joint geometry and joint strength for both joining processes
- Weighting of the geometric characteristic values in relation to the joint strength
- Suitable data-based and analytical methods for the prediction of joint strength at mechanical joining

The following **research methods** will be used:

- Material characterisation:

The base material will be categorised based on the results of the standard tensile test. The flow curves for the simulation will be determined by a stack compression or other suitable test to identify a large strain flow curve of sheet metal.

- Metallographic analysis:

In the experiential investigations, the geometric parameters of the joints are determined using a cross section analysis. For this purpose, the joints are separated by cut-off grinding near the centre of the specimen and prepared for microscopic documentation of the joint by subsequent grinding and polishing processes. The individual geometric parameters are measured on the digital photographs using software.

- Testing of joint strength properties:

The joint strength is tested in accordance with DVS / EFB Merkblatt 3480 [DVS3480]. The joints are tested for shear and top tensile loads. In addition to the maximum forces in shear and head tensile direction, the energy absorption capacity of the joints is also included in the determination.

- Finite element (FE) modelling of the joining process:

In numerical FEM models the considered problem is divided into a finite number of subareas (e.g. partial bodies) of simple form, e.g. into many small cuboids or tetrahedra. These are the "finite elements". Their physical behaviour can be well calculated due to their simple geometry with known shape functions. The physical behaviour of the whole body is simulated by how these elements react to the forces, loads and boundary conditions and how loads and reactions propagate during the transition from one element to the neighbouring one through very specific problem-dependent continuity conditions which must be fulfilled by the combined functions. In this project the purpose of the numerical FE simulations is to acquire an extend amount of data about the SPR-ST and clinching process.

- Data-based algorithms:

The basic idea behind the approach of machine learning is to create models that can recognise patterns and derive laws with the help of larger amounts of data. The aim is to enable computers to learn connections without being programmed for a specific task. In supervised machine learning, the characteristics and the corresponding output values are specified in the training data. The learning algorithm uses the training data to create a model that allows the characteristics to be assigned to the output values as generally as possible. After the training, the model can be used to calculate the corresponding output values only by entering the characteristics.

- Analytical methods:

In essence there are two options to derive an analytical formula to predict the mechanical strength of a clinched joint or a SPR. The first one is to fit a mathematical function which embodies the governing parameters to a vast amount of experimental data. The drawback of this approach is the lack of physical back ground, and thus ignoring different failure mechanisms. A better approach lies trying to identify elementary deformation mechanisms via the so-called slab equilibrium technique. Both approaches are subjected to a thorough assessment in this project.