Viskoelastik sıkıştırma modelleri ile malzeme karakterizasyonu ve vakum infüzyon (VI) işlemi modellemesi

In the Vacuum Infusion (VI) process, a dry fabric preform is compacted between a single-sided mold half and a vacuum bag, and then impregnated with thermoset resin to manufacture polymeric fiber-reinforced composite parts. Use of vacuum bag as the upper mold half reduces the cost, but more importantly it allows to manufacture large parts such as wind turbine blades. Compared to two-sided mold manufacturing processes such as Resin Transfer Molding (RTM), VI suffers from significant thickness variation because of the non-stiff upper mold half. Manufacturing of parts with small dimensional tolerances require compaction characterization of the fabric preforms which can be used in the process modeling by coupling fabric preform and the resin flow. Viscoelastic models have advantage over elastic models since they have the ability to model the time dependent compaction of the fabric. Since the fabric undergoes significant change in thickness with time even at constant pressure especially during relaxation stage, using elastic models greatly decreases the accuracy of the VI simulation. With viscoelastic models, it is possible to model both filling and post filling stages of the VI process with a greater accuracy than with elastic models. With this increased accuracy, filling time and final thickness distribution can be estimated correctly. This study contributes to the composite manufacturing literature with the following achievements: (1) a compaction characterization procedure is designed to mimic the compaction behavior of fiber reinforcements used in VI; (2) an automated characterization experimental setup is constructed to reduce human error and increase the repeatability; (3) experimental characterization data allows to have a database for two fabric types (random and woven); (4) five viscoelastic compaction models (Maxwell, Kelvin-Voigt, Zener, General Maxwell and Burgers) with dampers and nonlinear spring elements, and an elastic compaction model are used to investigate different stages of VI; (5) comparison of the models allows to study the contribution of different elements of the compaction models and thus understand their physical responses in different stages. 8 different sets of experiments are conducted on each fabric type to investigate the effect of different compaction parameters (such as loading/unloading rates, relaxation pressure and whether the specimen is dry or wet) on the results. Differential equation (DE) solvers are used on the derived system equations stage by stage to overcome discontinuity problems at the stage transitions. It is seen that Burgers model outperforms other 4 models in terms of the parameter count and 𝑅2. However, Burgers model undergoes infinite deformation under a constant load, which is not physically correct. Hence, Generalized Maxwell model with 𝑛 = 2, which mimics the plastic deformation with slow viscous response, should be used if prolonged time of settling is expected. By coupling a resin flow model and a viscoelastic compaction model (preferably Generalized Maxwell with n=2 or Burgers depending on which stage(s) of VI is desired to be modeled) in a complete VI process model, it is possible to simulate more accurate gradual pressure and thickness changes with time during the application of control actions in VI post-filling stage. Elastic models fail in these situations because they cannot model time dependent response, and thus viscoelastic models of this study should be preferred.