Die structure optimization for a large, multi-cavity aluminum profile using numerical simulation and experiments

Abstract

The steady state extrusion process of a large, multi-cavity aluminum profile for high-speed train was simulated using the Arbitrary Lagrangian Eulerian (ALE) algorithm. The simulation result of the initial design shows that the material flow velocity in the bearing exit is non-uniform. Comparing the obtained extrudate front end of the simulation with the experimental results of the initial design, it is clear that the metal in thick-walled regions of the profiles in both results flows faster than that in the other regions of the profiles. Moreover, the distorted metal in the six inclined ribs shown in both results are very similar to each other. Aiming at a uniform flow velocity, four kinds of structure optimizations were proposed and simulated. The methods of resizing portholes, adding bosses, chamfering mandrels, and adjusting the length of the bearings can balance the local metal flow velocity effectively. Finally, a perfect simulation result with exit velocity difference of 18.7 mm/s and little distortion was obtained. Through the effective prediction of flow velocity in the simulation extrusion, the qualified profiles with few trial and error extrusions were manufactured successfully. The case study demonstrates that the ALE method is a viable predictive tool for die design, and the approach is applicable to the extrusion of other alloys for any other extrudate shapes.

Introduction

Nowadays, the demand for aluminum profiles used in high-speed trains is growing rapidly worldwide. However, the production of large, thin-walled aluminum hollow profiles entails the completion of a very complicated and technical process. If the die is not well designed, the extruded product is likely to be twisted, warped, or curved along its length. On the other hand, a well-designed die yields a product with no distortion and within the required tolerance; this is usually achieved by having an extruding profile with uniform exit velocity. In practice, the exit velocity control is achieved through the empirical method or the traditional trial and error approach. However, this consumes time and money; in addition, dies cannot be repaired in some instances and have to be discarded. This design and manufacture method does not satisfy the requirement of the modern industry for short cycle production and low cost. Using numerical simulation of the aluminum profile extrusion process, the deformation, states of stress and strain, along with the distributions of flow velocity and temperature, can be obtained when materials pass through the extrusion die. Therefore, the defects of actual extrusion process can be predicted, thus allowing immediate correction on the initial design before manufacturing. This has become a trend in modern aluminum extrusion industry.

In recent decades, many papers have been published on the subject of extrusion process using Finite Element (FE) analysis. Most of them are based on the Lagrangian method and the Eulerian method. Jo has studied the effects of the initial billet temperature, bearing length, tube thickness and extrusion ratio on extrusion load and weld quality by means of the Lagrangian method [1]. Fang has investigated the effects of die bearing and ram speed on extrudate temperature, dimensional precision, surface quality and extrusion pressure during a typical industrial profile extrusion process using the Lagrangian method and the results had been validated by extrusion experiments [2]. Lee has studied the effects of chamber shapers of porthole die on elastic deformation and extrusion process in condenser tube extrusion based on the Lagrangian method [3]. Liu has optimized a large diameter thin-walled aluminum profile extrusion die design using the Lagrangian method [4]. Wu has simulated and optimized the extrusion process of an aluminum rectangular hollow pipe using the Eulerian method [5]. Kim has studied the lubrication effect on the backward extrusion of thin-walled rectangular aluminum case with large aspect ratio using the Eulerian method [6]. Chen has optimized extrusion process parameters and die structure based on the Eulerian method, artificial neural network, and genetic algorithm [7]. However, to this day, these two methods have only been moderately successful in the aluminum profile extrusion processes. They both have their own limitations and are restricted to simple and symmetric profiles. For the Lagrangian method, its mesh is rather easy to distort due to its large deformation so that the mesh regeneration is inevitable. As a result, inaccuracies arise in the solution and termination of the calculation. On the other hand, the Eulerian method, in which the mesh is fixed, is not suitable for the description of the movement of the free surface. In addition, because of a huge number of elements, the calculation time is unbearable, especially for simulating a complex, thin-walled hollow profile extrusion process. Meanwhile, the Arbitrary Lagrangian Eulerian (ALE) method is a combination of the Lagrangian and the Eulerian methods. It can move arbitrarily since its mesh is not attached to any material unlike the Lagrangian method. It is also not fixed in space unlike the Euler method. Therefore, the ALE method is effective in simulating the extrusion process with large deformation, in which the defect of mesh distortion and the difficulty of free surface tracking can be avoided [8], [9], [10]. In recent years, studies have been made on the ALE method extrusion simulation. For instance, Yang has optimized the flow guide and die land during the practical extrusion processes [11]. Zhang has investigated the effects of stem speed on metal flow behavior at die exit, temperature distribution, extrusion force, welding pressure [12]. Bastani has discussed the effect of different process parameters on the flow balance and temperature evolution of the extruded sections [13]. Kumar has investigated the average extrusion pressure in cold case for lead alloy (70Pb30Sn) billets and in hot case for commercial grade aluminum (Al 2024) billets respectively [14]. Lof has described the resistance in an equivalent bearing without using a large number of elements [15], while Zhuang has simulated the sheet metal extrusion process [16]. In a previous study, we simulated various pocket designs during a thin-walled aluminum profile extrusion using the ALE algorithm [17]. However, the analysis of a large, non-symmetric and complex thin-walled multi-cavity aluminum profile has not been presented so far.

In this work, a large multi-cavity aluminum profile die for high-speed train is designed and simulated using an ALE-based commercial package called HyperXtrude. The defects existing in the initial design are identified. The following extrusion experiment further confirmed its accuracy. After dozens of optimizations, a perfect simulation result with the minimum exit velocity difference and distortion is obtained. Referring to the prediction of flow velocity in the simulation extrusion, the qualified profiles with few trial and error extrusions were manufactured successfully. This work offers a reference for the die design of large and complex cross-section aluminum profiles extrusion.