Elsevier

Applied Thermal Engineering

Volume 40, July 2012, Pages 326-336
Applied Thermal Engineering

Simulation of initial cooling rate effect on the extrudate distortion in the aluminum extrusion process

https://doi.org/10.1016/j.applthermaleng.2012.02.012Get rights and content

Abstract

The cooling process after aluminum extrusion is crucial for the strengthening of AlMgSi-type alloys but can induce distortions caused by fast cooling, especially for complex or critical sections. The optimum cooling conditions fall within the so-called quench window between two limitations (high strength and lower distortion). Appropriate modeling, using both physical and numerical methods, can guide us to achieve a better cooling strategy in extrusion plants. In this paper, we focus on the study of the optimal strategy of the initial cooling for selected cases.

A simple extruded section in the actual scale was simulated by a reasonably realistic numerical model. The section was cooled by a water spray from one side (top side) and still air on the other side. The effect of the initial cooling was evaluated by defining a measure called the “front width”, and the distortion mechanism for different front widths was studied. The effect of some selected parameters (e.g., the thickness, the stop time and the extrusion speed) were also taken into consideration.

The simulation results showed that a combination of distortion stages (concave or convex) played a main role in determining the final shape and distortion magnitude. If the process parameters (e.g., the front width, extrusion speed and others) are set in such a way that could restrict the accumulation of distortions at subsequent zones, then the section suffers fewer distortions. For example, if the cooling begins sharper the extrudate suffers smaller distortion. A longer stop time is beneficial for less distortion when it is long enough to ensure that the section in the cooling zone is cooled to room temperature. The extrusion speed has varying effects and could be optimized for any profile thickness. However, even when optimizing the other parameters, the thicker section might be distorted more than the thinner section.

Highlights

► Real scale simulation of cooling process in aluminum extrusion has been proposed. ► Highly realistic condition was defined. ► Both the first and the second billet were considered. ► The distortion mechanism was illustrated and discussed.

Introduction

Aluminum extrusion is capable of producing long solid or hollow sections with very complex cross-section geometries [1]. Cooling immediately after extrusion is crucial for the nucleation of high densities of strengthening particles in alloys, such as AlMgSi, allowing them to obtain their full strength potential. This is feasible because a hot extrusion temperature is high enough for Mg and Si to remain in solution before cooling. Therefore, using on-line cooling to room temperature followed by aging makes separate solution treatment unnecessary [2]. Consequently, most of the 6XXX alloys can be simply naturally or artificially aged if cooling is performed at a proper rate [3].

The actual cooling occurs in three stages: first from a die to a quench box, then quenching period (generally in water or forced air) and finally by an air cooling to the ambient temperature [3]. When a rapid cooling is desired, water is the best medium because it is inexpensive and easy to use. For example, 6061 and especially 6082 alloys require water quenching [3] to achieve the best mechanical properties. The thermo-mechanical history can be a combination of solution treatment and strengthening, which offers potential economic savings but requires the control of a large set of parameters. From an economic standpoint, the process eliminates one solution treatment stage, which could save 1.6 J/kg [3]. Handling is also reduced, resulting in both labor and energy savings. In terms of extrudate quality, the process also minimizes the possibility of coarse surface grains and, in general, produces a non-recrystallized oriented structure that possesses superior properties to those obtained by a separate solute soak treatment [3].

The intensity of the cooling is characterized by heat transfer coefficient (HTC). In case of water spay, HTC is highly temperature-dependent [1]. Its variations with temperature are generally classified into four different regimes as shown in Fig. 1: film boiling, transition boiling, nucleate boiling and natural convection [4], [5].

Perfect uniform and tailored cooling cannot be achieved from ∼550 °C to 50 °C due to limitations in nozzle technology and geometrical positioning as well as inherent limitations due to variations in thickness and the inability to cool inside the hollow section. Therefore, there is non-uniform cooling both across the section and along the length of the section. This non-uniform cooling may lead to large temperature gradients causing thermally induced distortions and high residual stresses [6]. The challenge is to reduce the distortions, such as banana shaping or twisting. Therefore, temperature management is important for the aluminum extrusion process, and, if correctly planned, large monetary savings can be achieved, which can represent 25–30% of the total processing costs [3].

Appropriate modeling, using both physical and numerical methods, can help us to achieve a better cooling strategy in extrusion plants. It is necessary to define boundary conditions similar to the actual conditions to obtain relevant results for the real process [7].

The optimum cooling conditions can be obtained by cooling through the so-called quench window between the two limitations, which is illustrated schematically in Fig. 2 [6]. In reality, it is a very difficult to control complex section temperature that goes through the quench window. As the figure shows, at a higher temperature (e.g., above ∼450 °C for aluminum alloys), shape distortion is an important issue, and metallurgical aspects are not critical. This may lead manufacturers to begin the cooling process (when the section temperature is high) in a “gentle” way and to not apply the full capacity of the quench box during its entrance. When the section has cooled to the critical temperature (e.g., ∼250 °C) from a metallurgical point of view, the section should be quenched at the full capacity of the quench box. However, this solution must be substantiated.

Previously, Deiters and Mudawar [8] and Rozzi et al. [9] studied spray quenching of an extruded L shape section, in order to approach uniform cooling of the extrudate. Järvstråt and Tjøtta [6] made a FE model for cooling of aluminum extrusions and investigated on the extrudate thickness varying effect on the distortions due to cooling. Becker et al. [10] tried to predict the distortion and residual stress numerically for a rectangular bar quenched from one surface by water.

Pietzsch et al. [11] presented a thermo-mechanical model for simulating the transient fields of the temperature, microstructure, stress, strain, and displacement during quenching of steel profiles. Kaymak [12] introduced a thermo-mechanical FE model and used it for simulating the transient fields of the temperature, microstructure, stress, strain, and displacement during a heat treatment process of long profiles with different cross sections. Nallathambi et al. [13], [14] worked on mathematically formulating quenching process, too. They simulated the process by means of a nonlinear finite element technique which included the coupling of thermal, metallurgical, and mechanical fields.

Prior to this study, Kristoffersen [4] studied the material model and heat transfer parameters of water cooling and air cooling using a simulated 2D model. Walle [5] also studied the effect of the moving cooling source from one end of the sample to the other using a 3D modeling of laboratory experiments. Bikass et al. [15], [16] followed that work by studying the effect of non-uniform cooling across the width. These former studies had focused on small-sized samples with constant length. However, it was shown that the effect of the sample size and length variation during the cooling process was not negligible for longitudinal distortions [17].

In the present work, a 2D finite element method (FEM) by using ABAQUS was used to simulate the actual scale cooling process inside the quench box. The extrudate length variation during the extrusion process was considered. The process conditions and parameters were selected to be similar to the real process in plants. The main purpose of this paper was to illustrate the effect of the initial cooling rate on the distortion mechanism due to non-uniform cooling throughout the thickness. This effect was also studied for the related parameters (e.g., thickness and extrusion speed).

Section snippets

Case design

The extrudate length is usually 20–40 m in extrusion plants and can be produced from one billet (especially at high extrusion ratios). After extruding one billet, the process is paused for a time (approximately 10–20 s) to load the next one. This time is called the dead cycle time (DC).

A puller is used to guide the extrudate from the press along the run-out table. According to our case design, this puller could not enter the quench box. For the first extrudate, the quench box must be raised up

Material and process parameters

The material used was 6082 alloy. The physical properties of the alloy were considered constant in the models and were given in Table 2. The mechanical properties (including strength, elastic modulus and Poisson ratio) were considered to be strain-, strain rate- and temperature-dependent (as shown in Fig. 6). The extrusion process parameters were selected to be similar to the typical real process in plants (Table 3). These experimental data collected by Kristoffersen [4].

Cooling conditions

The extrudate was cooled by water that was sprayed on the top face and by still air on the bottom face. The HTC for still air (Fig. 1b) varied slightly with temperature but in a linear way. However, the HTC for the water spray varied strongly from high temperatures (above 400 °C) to room temperature throughout the four regimes as shown in Fig. 1a. These regimes were simplified in the present work. The HTC of the nucleate boiling regime were merged with the natural convection regime and was

The finite element model

ABAQUS/Standard was used to simulate the process by using coupled thermal-displacement analysis. Because cooling across the extrudate width (200 mm in this case) was uniform, a 2D model was selected to decrease computational time. Four quadratic brick elements (CPE8T) available in the ABAQUS library were used throughout the thickness. To define the mesh size along the length, preliminary simulations were performed and the proper size was selected.

The material model used in this study was

Distortion mechanism

We first considered a section that was simply cooled from the top surface. In this simple case, two stages of displacements occurred due to non-uniform cooling throughout the thickness. The resultant out of straightness of the extrudate induced by every point displacement was considered as “distortion”. In the first stage, the longitudinal contraction on the top surface was faster and larger than the bottom surface and this caused the bowing of the section below (Stage 1 in Fig. 8). As the top

Conclusion

To obtain results relevant for an actual aluminum extrusion cooling process, both the first and second extrudates were simulated with respect to the distortions. This study focused on the effect of the initial cooling rate by introducing a parameter called “the front width”. The simulation results showed that the initial cooling rate had a significant effect on the distortion magnitude. We showed that, generally, if the cooling process began sharper, the longitudinal distortions caused by

Acknowledgements

This work was a part of the project run by Hydro in cooperation with SINTEF Materials and Chemistry. It was partly funded by the Research Council of Norway (RCN), and the authors would like to acknowledge their support.

References (18)

There are more references available in the full text version of this article.

Cited by (7)

  • Effect of welding current on morphology and microstructure of Al alloy T-joint in double-pulsed MIG welding

    2015, Transactions of Nonferrous Metals Society of China (English Edition)
View all citing articles on Scopus
View full text