A Novel Heat-Assisted Three-Roll Incremental Rolling System for Flexible Rolling

Traditional methods for manufacturing rods and tubes include drawing, and two-high caliber rolling [1,2]. In the conventional drawing (Fig. 1(a)), the outer diameter of the workpiece is reduced by pulling it through a die. A benefit of conventional drawing is its simple machine design, but a major drawback of this method is that the geometry and profile of the product are highly dependent on die geometry. The workpiece diameter is constrained by die size and dies with different diameters need to be replaced in multipass processes. In two-high rolling (Fig. 1(b)), a workpiece is deformed by several two-high rolling mills. Drawback of this method is the large space requirements [3]. To overcome these drawbacks, three-roll incremental rolling processes have been developed [4,5]. Three-roll incremental rolling is a dieless process in which the workpiece passes through the roll gap to achieve reductions in diameter. Unlike traditional rod manufacturing methods, in three-roll incremental rolling, the position of the rolls is continuously adjustable. This allows for (1) different and flexible initial and final diameters, (2) cost- and time-efficient prototyping, and (3) lower space requirements [3,4].

Efforts have been made to explore the three-roll incremental rolling process, for example, on the effects of the process parameters. Shih et al. [6] established a finite element (FE) model of the rolling process to discuss the effect of rotational speed, offset angle, and inclination angle of the rolls on the rolling force and the exit velocity of the workpiece. A study of the thread/helix angle of the spiral marks and the end cavity of the rolled product was performed by Shih and Hung [3] by combining experimental and numerical analysis. Luo et al. [7] performed FE simulations and demonstrated that friction and surface conditions can significantly affect the productivity of the process. Others focused their investigations on the shape of the rolls. Hwang et al. [2] analyzed the effects of rolling conditions on spiral marks both numerically and experimentally and found that convex-shaped rolls can lead to a smoother surface than conical or spherical-headed rolls. The effect of roll structure on rolling stability was discussed by Zeng et al. [8].

It is worthwhile to note that most research on three-roll rolling has been carried out numerically. In the studies involving experiments, plasticine is most commonly selected as the workpiece material due to the difficulties with testing metallic materials on an experimental machine. However, plasticine is very soft, resulting in differences in deformation behavior as compared to metals. To reduce the diameter of metallic materials using lab-size equipment, forming can be conducted at high temperatures to induce thermal softening and improve plastic flow [9]. Several heat-assisted forming methods have been developed, for example, furnace heating, laser heating [10,11], induction heating [12], and electrical heating [13]. Additionally, the achievable sectional diameter of the product is rarely discussed in the literature. Dimension information is important since in some applications small diameter tubes are needed, for example, cladding tubes in nuclear reactors.

The objective of this work is to build an experimental heat-assisted three-roll incremental rolling system capable of reducing the outer diameter of the workpiece down to 6 mm aimed at two purposes: (1) evaluating the feasibility of this technology for this size range and (2) creating an experimental platform for studying the underlying physical mechanisms. To support the rolling of difficult-to-roll materials both/either induction heating and/or electric current-based heating are applied to the workpiece to assist the rolling process during forming. The following sections first discuss the machine design and specifications. Then a series of experimental results are presented to demonstrate the feasibility of the newly designed machine. Finally, an FE model of three-roll rolling is described to analyze the process.

The developed rolling system enables the production of rods/tubes with high process flexibility and the ability to simultaneously heat the workpiece during the forming process. Figure 2 shows the mechanical structure design of the heat-assisted three-roll incremental rolling system. The rolling system includes four subsystems for rotary rolling, radial feeding, axial input and output feeding, and heating. The basic specifications of the rolling machine are listed in Table 1. Note that in the tube rolling process, the inner hole is maintained by a mandrel, while the outer diameter of the tube is formed by the rotation and radial feeding motions of the rolls and the axial feeding of the rod/tube. A detailed description of the key subsystems follows below.

The rod/tube is formed by three rolls which are situated 120 deg apart with respect to each other. As shown in Fig. 3, each roll is rotatable in its bearing housing. The head of each roll is replaceable without having to remove from the bearing housing because of their cross-grove engagement design. The rotations of the three rolls are driven by three subchains. Each subchain consists of two-phased universal joints and a spline shaft to transmit the high rolling torque. In the gearbox, the three subchains are driven by three pinions respectively. The pinions are engaged with a central slewing ring with an external gear. The fourth pinion at the bottom is connected to an AC helical gear motor to provide the source of power. The overall rotary power transmission of the rolls was designed in a compact form.

In Fig. 4, the radial feeding mechanism, based on a standard lathe three-jaw chuck, is shown. Each bearing housing is attached to one of the jaws of a lathe chuck. The radial feeding motion is provided by the motions of the jaws, which are driven by the inner spiral gear mechanism. An external servo motor with a helical gear head is attached to accurately drive and control the rotation of the spiral gear. The jaws are simultaneously movable inward/outward along the radial direction to provide a radial feeding motion of the rolls. The radial feeding mechanism is fixed on the base.

FE simulations were initially conducted to guide the mechanical design of the axial feeding system. Figure 5 shows the comparison of the spiral marks under different axial feeding rates. As it can be seen, obvious spiral marks would appear if the axial motion of the workpiece is driven by pure rolling fiction (Fig. 5(a)). To obtain smooth formed surfaces, the axial feeding motion of the rod needs to be properly controlled. As seen in Fig. 5(b), shallower spiral marks are obtained if the axial motion is reduced to a constant value of, e.g., 1 mm/s in the figure. Details regarding the simulation model will be introduced in a later section.

According to the above analyses, an axial drag mechanism was developed for controlling the axial feeding speed. As shown in Fig. 6(a), the input axial linear motion is provided by a linear stage, which is actuated by a servo motor with a spur gear head. The output motion is constrained by a passive linear roll guideway (Fig. 6(b)). Both the input/output rod holders can passively rotate in the bearing housings. The assembly of the input and output linear motions was calibrated for parallelism and collinearity to ensure a continuous rolling process.

The machine includes two heating systems that can be used independently or in unison, to be described next.

The schematic of the induction heating system is shown in Fig. 7(a), where a 2500 W induction heater is located on the input side of the machine. During the rolling process, the rod is preheated by the high frequency induction heater. Heating is achieved through a multiturn water-cooled copper coil surrounding the rod (Fig. 7(b)). The surface temperature of the rod was measured by an infrared (IR) camera (TIM160 Imager, Micro-Epsilon). The infrared temperature readings were calibrated using a thermocouple. A representative thermal image captured during a rolling process is shown in Fig. 7(c). The advantages of this heating method are its rapid heating, local heating, and low hardware cost, but its application is restricted to ferromagnetic metals.

The schematic of the electric heating system is shown in Fig. 8. To realize the electrically assisted rolling process, electric current needs to pass through the rod. Two basic current paths were designed. The current can be applied by either passing it from one end of the rod to the three rolls or through the two ends of rod only. A pulsed power supply (THDM-III, Hongqiao metal products Co. Ltd) capable of supplying square pulses of up to 3000 A and frequencies from 100 Hz to 800 Hz is utilized to provide electric current. All electrical terminals were designed with graphite blocks in a slip ring manner to accommodate the existing rotation of the rod and rolls. Electrical insulation was added at the connections of the bases and electrical terminals. Since heat is generated near the bearing housings in the electrically assisted rolling process, a cooling system was provided in the bearing housings. Serially connected cooling channels are used in the body of the bearing housings of the rolls as shown in Fig. 9 to facilitate the circulation of liquid coolant.

For the overall design of the new EA rolling machine, one of the most important considerations is related to the rolls and their relation to the rod/tube. The profile of the rolls was designed as a convex shape. This shape has been demonstrated as useful for producing smoother surfaces compared with those produced using conical-shaped rolls or spherical-headed rolls [2]. The roll–rod relation is characterized by the skew and roll angles. The skew angle is the angle between the axes of the roll and rod. Due to the skew angle, α (Fig. 10(a)), the rod moves forward as the rolls rotate. The roll angle is the angle between the axis of the rod and the plane of the rod–roll contact area. The roll angle, β (Fig. 10(b)), causes plastic deformation of the rod.

The contact area between the roll and rod is illustrated in Fig. 11(a). To determine the key geometric relationships between the roll and the rod that define the area of contact, the surface of the rod along with the contact trace generated by one of the rolls is unfolded and represented in a plane (Fig. 11(b)). To ensure that the roll–rod contact areas cover the whole surface of the rod during forming, the projection (line L) of the width of the contact trace (line l) onto the unfolded planar section of the rod needs to be larger than one-third of the circumference of the rod.

The projection length is affected by two factors, i.e., the width of the contact trace and the skew angle. The larger the width of the contact trace and the smaller the skew angle, the larger the projection length. However, in the actual design, the choice is not arbitrary and is limited by pragmatic constraints as they pertain to the interference between realistically dimensioned bearing housings needed to transmit the required forming loads/power, which, in turn, depend on the desired forming objectives. It is evident from the above that a larger roll diameter is needed to increase the width of the contact trace, and thus, a larger bearing housing is required. On the other hand, the rolls need to be moved close enough to reach all the way down to the 6 mm outside diameter of the rod without interference which also limits the size of the bearing housing to avoid interference. Similarly, decreasing the skew angle increases the risk of interference between the bearing housings. Through an iterative design process, by comparing the projection length with the circumference of the rod and checking the interference of the rolling mechanism (Fig. 12) in the CAD model, the skew angle (50 deg) and the roll profile (Fig. 13) were finally determined.

Initial performance testing and evaluation of the newly fabricated heat-assisted three-roll incremental rolling system (Fig. 14(a)) was performed by using low carbon steel rods with an initial diameter of 25.4 mm. Bearing steel 52,100 was used for the rolls rotated at a rolling speed of 1.256 rad/s. The experimental setup on the input side is shown in Fig. 14(b). Before the rolling process, the workpiece was preheated to about 500° C by the induction heater as shown in Fig. 15(a). During the rolling process, the outside diameter of the rod was reduced incrementally by 1 mm in each rolling pass until the desired final diameter (i.e., 6 mm) was reached except for the first pass. In the first pass, the three rolls were located to achieve a roll gap of 24.4 mm (the initial diameter of the rod is 25.4 mm). However, the true reduction in the diameter of the rod after the first pass was only 0.4 mm due to backlash and machine compliance. They were compensated by setting the actual roll gap smaller than the ideal roll gap to achieve a 1 mm reduction of the rod diameter in the following passes. Therefore, the number of effective rolling passes was 20 ((25.4 − 6 − 0.4) mm/1 mm + 1 = 20). Each rolling pass was divided into three substeps. The rod moved forward by pure rolling friction in the first two substeps and was then fed at a constant axial speed of 1 mm/s in the third substep. The first substep of each incremental rolling pass left spiral marks on the outer diameter as demonstrated in the first substep of the sixth rolling pass in Fig. 15(b) as an example. After the first substep, the rod was manually rotated by 60 deg (i.e., half the angle between the rolls) compared to its initial rolling phase and rolled for the second time. Through this second rolling step, the amplitude of the ridges of the spiral marks was reduced. Subsequently, by controlling the axial feeding speed in the third substep, the spiral marks can be mostly removed after each rolling pass. Shown in Fig. 16(a) are images of the rolled final product. The diameter was reduced from 25.4 mm to 6 mm after rolling, demonstrating that this newly fabricated rolling system achieved the expected goals. The final surface profile was measured using a Bruker Alicona InfiniteFocus microscope. Detailed surface finish and the corresponding surface profile are shown in Figs. 16(b) and 16(c), respectively. As it can be seen from the figures, the final part of the surface has no obvious spiral marks.

One representative spiral mark after the very first pass is shown in Fig. 17(a). Three spiral marks from the first pass were measured for repeatability. The average depth is 0.198 mm (Fig. 17(b)). The control system utilizes a power monitor device for sensing the total power required to drive the AC helical gear motor for rotating the three rollers. Voltage and current consumptions were 89 V and 2.5 A, respectively, without load, and 94 V and 4.1 A, respectively, with load in the first pass. Therefore, the power consumed to deform the workpiece during the first pass is calculated as 94 V × 4.1 A − 89 V × 2.5 A = 162.9 W.

To further investigate the rolling process, FEM simulations of the very first rolling pass were conducted using the commercial finite element software abaqus/explicit. The simulation model is shown in Fig. 18. The diameter of the rod was set to be 25.4 mm, the same as used in the experiments. Both rolls and rods were modeled as deformable parts and were meshed with C3D8R elements. Properties of the roll material, 52,100 bearing steel, at room temperature, and properties of the rod, low carbon steel, at 500° C were used in the simulation. Young’s modulus of 52,100 bearing steel and low carbon steel were set to be 200 GPa and 146 GPa, respectively. Poisson’s ratio of these two materials was both set to be 0.3. The flow curve of the low carbon steel is given in Fig. 19. The friction coefficient of the roll and rod contact pair was set to 0.3. A mass scaling factor of 3000 was applied to accelerate the simulations. The boundary conditions included: (1) the rod rotates and displaces about and along its axis, while other degrees of freedom are restricted; and (2) the rolls rotate around their own axis at a constant speed of 1.256 rad/s.

Figure 20 shows the deformed specimen after the first roll pass. The depth of the spiral marks calculated for the first roll pass from the simulations was 0.207 mm, which matches very well with the experimental depth of 0.198 mm (Fig. 17(b)). Note that the depth of the spiral marks was defined as d = r_i − r_m, where r_i = 12.7 mm is the initial radius of the workpiece, and r_m is the minimum distance between the surface of the rod and its axis.

Figure 21 shows the simulated reaction moment per roll along its axis and can be used to calculate the total power consumption: P = 3 × M × ω = 3 × 40 Nm × 1.256 rad/s = 150.7 W, where M is the reaction moment per roll and ω is the angular velocity of the rolls. When the depth of the spiral marks obtained from the simulations and the experiments are very close in value, the power from the simulations (150.7 W) is slightly lower than that in the experiments (162.9 W). Power in the experiment is the sum of the power consumed to keep the machine running and the power consumed to deform the workpiece. However, power in the simulations is only the power used to deform the workpiece. This is believed to be the main reason for the discrepancy between the simulations and experiments. But overall, the simulations did well in predicting the reaction moment and deformation; the general trends were captured by the model with less than a 10% error. Therefore, the developed model can be used as a tool to guide the selection of process parameters. For example, the maximum achievable diameter reduction in each rolling pass can be determined through the relationship between the diameter reduction and the torque of the rolls. Note that these simulations did not consider the thermal exchange between the workpiece and the rolls, and the roll temperature was assumed to be room temperature. Future improvement will require the use of fully coupled thermal-stress analyses.

A novel heat-assisted three-roll incremental rolling system was designed to produce rods and tubes with various outer diameters using the same set of rolls. An induction heater and power supply for electrically assisted heating were incorporated into the system to heat the workpiece during the rolling process. This design allows for a high degree of flexibility in changing the roll to rod distance and the forming temperature during the process. Therefore, this system provides a means to achieve both local geometric and material property control. Both experimental and numerical studies were performed to study the process. The results have demonstrated that a smoother surface can be obtained by controlling the axial motion of the rod. For future work, a comprehensive array of testing will be conducted using this rolling system to study the effects of process parameters on the quality of the produced rods. The numerical models and results can provide useful guidance for the selection of process parameters and understanding the resulting microstructure of the rolled product.

This work was funded by the DOE NEUP program (DE-NE0008409).

There are no conflicts of interest.

The authors attest that all data for this study are included in the paper.