PhD Defence Ngoc Anh Vu | Accurate overbraiding simulations for complex mandrels - On the role of yarn interactions

Accurate overbraiding simulations for complex mandrels - On the role of yarn interactions

Ngoc Anh Vu is a PhD student in the department Production Technology. (Co)Promotors are prof.dr.ir. R. Akkerman and dr.ir. W.J.B. Grouve from the faculty of Engineering Technology.

Overbraiding is a process to manufacture tubular structural components at a high deposition rate. Composite product development using overbraiding and subsequent infusion can be both costly and time-intensive. This procedure often requires multiple iterations to meet design specifications. To reduce the associated expenses and time investments, simulation tools for overbraiding processes prove to be valuable. Various approaches exist, ranging from straightforward analytical models to computationally intensive finite element methods. Earlier developed kinematic models provide fast computations for complex geometries, making them suitable for design purposes. These models neglect yarn interactions, which reduces the computational cost, but also compromises the accuracy of the model predictions.

The aim of this thesis is to develop an efficient yarn interaction model to enhance the accuracy of overbraiding simulations for practical design of components with complex mandrel shapes, improving the accuracy but without severely impairing the computational performance of the kinematic model. To this end, the following three problems are addressed:

First, a yarn interaction model was developed for biaxial and triaxial braid patterns. The model considers the stick-slip process at the cross-over point between yarns and at the yarn-ring contact. Subsequently, the solution was implemented in a multiple contact points model employing a fast iterative frontal approach to solve force equilibrium for the braid as a whole. Validations of the overbraiding models against experimental data, encompassing axisymmetric and non-axisymmetric cases, demonstrate a substantial improvement in predictions compared to earlier published simulation results.

Second, an experimental setup was developed to measure the yarn-to-yarn coefficient of friction as a function of the inter-yarn angle and normal contact force. Friction between carbon and glass contact pairs was investigated under both dry and water-lubricated conditions. A mesoscopic friction model was introduced to capture the measured impact of the inter-yarn angle and normal force on the dynamic friction coefficient. Notably, observations indicate the formation of a water bridge at the contact interface during wet testing. Furthermore, it is argued that the additional contribution of capillary forces results in consistently higher friction coefficients for yarns lubricated with water compared to dry yarns.

Finally, overbraiding simulations for both biaxial and triaxial processes were conducted to investigate the impact of process configurations and friction data on the simulation outcomes. The findings revealed effects of the yarn interactions on braid angles during both the steady and unsteady states of the process. This improved accuracy of the models is achieved with increased computational cost, yet remaining within acceptable range for design and optimization purposes.