This conference paper was submitted for presentation at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

The main objective in design of a polymer extrusion die is to develop a die channel geometry which gives a uniform velocity distribution at the die exit. The uniform velocity at the die exit is required to minimize the extrudate distortion after the polymer exits the die. In the past, most extrusion dies were designed by a trial-and-error approach using the experience of the die designer. With the development of computationally efficient three-dimensional finite element flow simulation software for extrusion die design, designers can now virtually fine-tune their extrusion die before the die is machined. This can reduce the development time for extrusion dies by 40- 50%. However, virtual fine tuning of extrusion dies still requires the die designers to modify the die geometry themselves after each flow simulation using their experience. To eliminate this need for the die designer to modify the geometry after each flow simulation, in the present work, a die optimization software, optiXtrue, is used in the present work to automatically improve the geometry of a profile die after each flow simulation. The software then simulates the flow again, in the improved die geometry. This cycle of geometry improvement followed by a flow simulation is repeated till a geometry with a uniform exit velocity distribution is obtained in this automatic die optimization process. Besides eliminating the need for designer intervention for die improvement after each flow simulation, this extrusion die optimization software further reduces the development time for extrusion dies, and also provides a better die geometry than the geometries obtain by trial-and-error or by virtual finetuning. For a complex extruded profile shape, the initial die geometry developed by an experienced die designer, as well as the die geometry optimized by the optimization software were machined to validate the predictions from the software. The initial die geometry had a large variation in velocity at the die exit, whereas the die geometry optimized by the optimization software resulted in a uniform exit velocity distribution. The predicted exit velocity distribution in the initial as well as optimized die geometry matched accurately with the velocity distribution at the die exit in the experiments.