Issue 
Manufacturing Rev.
Volume 6, 2019



Article Number  20  
Number of page(s)  17  
DOI  https://doi.org/10.1051/mfreview/2019019  
Published online  12 July 2019 
Research Article
Rolling paths design assisted by targettemperature driven intelligent FE simulation of radialaxial ring rolling
^{1}
Research & Development Institute of Northwestern Polytechnical University in Shenzhen, Shenzhen 518057, PR China
^{2}
School of Materials Science and Engineering, Northwestern Polytechnical University, Xi'an 710072, PR China
^{*} email: glgglg66@nwpu.edu.cn
Received:
16
April
2019
Accepted:
24
June
2019
The microstructures of hardtodeform materials such as titanium alloy are very sensitive to temperature change in hot working process. During ring rolling process, unreasonable rolling paths will lead to drastic temperature change in local region of ring, thus damaging the microstructure and performance of rolled ring. This work proposes a method for designing the rolling paths which could accurately control the ring temperature by targettemperature driven intelligent FE simulation. The main idea of targettemperature driven intelligent simulation is introduced. An intelligent 3DFE model for TA15 titanium alloy ring rolling is established under ABAQUS/Explicit environment. The rolling paths under different initial conditions are obtained by intelligent FE simulations. The influence rule of initial conditions on rolling paths is revealed. The temperature control effects and change under different initial conditions are discussed. Considering the temperature control effects, a feasible range of initial ring temperature is suggested. Using the proposed method, the quick and accurate design for the rolling paths in radialaxial ring rolling process is realized. It is of great significance for the design and optimization of rolling paths and the accurate regulation of ring temperature in actual production.
Key words: Radialaxial ring rolling / rolling paths / targettemperature driven intelligent simulation / temperature control / TA15 titanium alloy
© X. Li et al., Published by EDP Sciences 2019
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
1 Introduction
With the rapid development of aerospace industry, the demand for aerospace equipment with light weight but high performance has becoming increasingly urgent [1]. TA15 titanium alloy rings, as the core component of aero engine casting, have been extensively employed in aircraft enginebuilding, rocket and space engineering, due to its unique performances of high specific strength, excellent thermal stability, welding performance and strong corrosion resistance [2–4]. The radialaxial ring rolling has become an irreplaceable advanced manufacturing technology for hard wrought alloy like TA15 titanium alloy, because of its some advantages such as high quality, high efficiency and considerable saving in energy and material cost [5–7].
TA15 titanium alloy is very sensitive to deformation temperature, leading to complicated microstructure evolution with the thermalmechanical coupling effects [8]. The radialaxial ring rolling is a complicated dynamic forming process with multifactor coupling effects of geometrical, physical and boundary nonlinearity. During ring rolling process, the ring temperature depends on the coupling effects of many factors, such as initial forming temperature, deformation heat, friction heat and heat loss [9,10]. It is difficult to accurately control the ring temperature during the whole process. The rolling paths define the motions of various rolls in ring rolling process, which have crucial influences on the ring temperature, thus greatly affecting the microstructure and performance of titanium alloy components [11–13]. Therefore, it is crucial to obtain the reasonable rolling paths by which the ring temperature can be controlled within a desired range during ring rolling process, thus improving the microstructure and performance of rolled ring.
As we all know, the finite element simulation technology has become a powerful means to study the plastic forming process [14–16], by which many research has been performed for ring rolling process. Using dynamic explicit FE simulation, the reasonable ranges of the rotational speed of main roll and feed rate of mandrel were determined [17], respectively. This lays a foundation for the study on hot radialaxial rolling of large titanium alloy rings. Anjami et al. reported that changing the sizes of forming rolls could considerably affect the strain and temperature distribution, thus affecting the forming quality of the rolled ring [18], which could serve as a guide for the design of rolls sizes and optimization of the hot ring rolling of TC4 titanium alloy. In hot rolling of large rings of titanium alloy, the ratio of roll radii has a significant effect on the geometric shape of deformation zone [19]. Moreover, there exists an optimized range of the ratio of roll radii, which plays a decisive role in achieving large titanium alloy rings with high quality. However, in these studies, the motions of all rolls which decide the rolling paths are mainly defined by production experience. This does not accord with the actual production. Moreover, the ring temperature could not be controlled accurately. In addition, the trial and error method is always employed in current numerical modeling and simulation. This not only leads to lower efficiency, but also could not accurately control the temperature of ring during rolling process.
To overcome the shortcomings of traditional FE simulations, an intelligent simulation method driven by the control objective has been proposed [20]. In this method, the motions of various rolls could be driven automatically during the FE analysis by the control objective. It has been successfully employed to control the rolling force during radialaxial ring rolling of supersized aluminum alloy ring [21,22]. Similarly, taking ring temperature as the control objective, the intelligent control of ring temperature could be realized in radialaxial ring rolling process. The essence of accurate temperature control is to acquire the rolling paths which satisfy the temperature control objective.
To realize accurate control for ring temperature, this work proposes a method to design the rolling paths by the targettemperature driven intelligent FE simulation. The main idea for targettemperature driven intelligent FE simulation is introduced. Through VUAMP subroutine development under ABAQUS/Explicit environment, the intelligent temperature control method is integrated into the FE solver. An intelligent 3DFE model for radialaxial ring rolling of TA15 titanium alloy is established. By intelligent FE simulations, the rolling paths under different initial conditions, namely the initial ring temperatures and initial ring growth velocities, are obtained. The temperature control effects under designed rolling paths are discussed. The influence rules of initial conditions on rolling paths and temperature control effects are revealed. This work provides an innovative method for the design and optimization of rolling paths and the accurate regulation of ring temperature in radialaxial ring rolling process.
2 Targettemperature driven intelligent FE simulation of radialaxial ring rolling
2.1 Idea of targettemperature driven intelligent FE simulation
During ring rolling process, the ring temperature depends on the comprehensive effects of initial ring temperature, deformation heat, friction heat and heat loss by interaction, convection and emissivity. According to the energy conversion in physics, the deformation heat mainly derives from the work done by external forces. When other forming conditions remain unchanged, the faster the deformation velocity of material, the greater the rolling force. The amount of work done per unit time increases, thus the deformation heat per unit time increases. Meanwhile, the rolling time will be reduced due to fast deformation, thus the heat loss decreases. This eventually leads to the rise of ring temperature. The slower the deformation velocity of material, the smaller the rolling force. The amount of work done per unit time decreases, thus the deformation heat per unit time decreases. The rolling time becomes long due to the slow deformation, thus the heat loss increases. This eventually leads to the drop of ring temperature.
The ring blank is deformed under the coordinate motions of all the rolls in ring rolling process. Therefore, the ring temperature can be controlled by adjusting the motions of all the rolls (rolling paths) to regulate the heat generation and heat loss. Based on this, the targettemperature driven intelligent FE simulation of radialaxial ring rolling is proposed in this work. Without predefining rolling paths, the intelligent simulation can realize the adaptive control of the rolls by detecting ring temperature. Driven by the temperature control objective, the rolling paths can be adjusted realtimely to maintain the ring temperature within the desired range. The main idea of targettemperature driven intelligent FE simulation of radialaxial ring rolling is shown in Figure 1.
From Figure 1, a desired range of ring temperature (T _{objmin}∼T_{objmax}) is determined to be the temperature control objective according to the microstructure and properties of the studied material before FE calculation. As the initial conditions of intelligent simulation for ring rolling, the initial ring temperature T _{0} and initial ring growth velocity v_{D} _{0} are preset. The preset initial ring temperature T _{0} is determined according to the material behaviors, and the preset initial ring growth velocity v_{D} _{0} is predesigned to ensure a successful establishment of ring rolling process. The main roll is used to drive the motion of ring, whose rotational speed is n _{1}. The rolling curve, which describes the relation between the instantaneous height h and the instantaneous thickness b of ring during ring rolling process, is predefined as h = I(b), as shown in Figure 2. The instantaneous outer diameter D_{t} and thickness b could be monitored realtimely in each increment by the virtual sensors. Then the instantaneous height of ring h can be calculated by the relation of h = I(b).
The feed rate of mandrel v_{f} , feed rate of upper axial roll v_{a} and withdrawal speed of axial rolls v_{w} can be determined by equation (1):(1)where v_{D} is the growth velocity of ring diameter; D _{0}, d _{0}, b _{0} and h _{0} are the outer diameter, inner diameter, thickness and height of ring blank, respectively; b and h are the instantaneous thickness and height of ring, respectively; I′(b) is the slope of rolling curve h = f (b), which describes the instantaneous assignment of deformation amount between axial and radial directions.
The rotational speed of axial rolls and the instantaneous position of guide rolls could be determined by equation (2):(2)where n _{1} and R _{1} is the rotational speed and radius of the main roll, respectively; D_{t} is the instantaneous outer diameter of ring during ring rolling process; θ is the half cone angle of the axial rolls; P = g _{3}(D_{t} ) denotes the relation between the instantaneous positions of guide rolls P and the instantaneous outer diameter of ring D_{t} . P stands for the instantaneous position vector of guide rolls in ring rolling process.
From equations (1) and (2), it can be seen the motions of all the rolls could be driven by the ring growth velocity v_{D} and the instantaneous outer diameter of ring D_{t} during ring rolling process [7]. Therefore, the adaptive regulation of ring temperature could be realized by developing a ring growth velocity model relating to the ring temperature.
To develop the ring growth velocity model relating to the ring temperature, the mathematical relations between ring temperature (T _{ringmin} and T _{ringmax}) and ring growth velocity v_{D} , initial temperature of ring T _{0} and the instantaneous outer diameter of ring D_{t} are established firstly, which is described as follows.
Through intelligent simulations under different v_{D} , T _{0} and D_{t} , the information of ring temperature is extracted from the FE calculation results through the VUAMP subroutine. By sorting the temperatures of all elements in each increment, the instantaneous minimum temperature and maximum temperature of ring (T _{ringmin} and T _{ringmax}) are obtained, respectively. Then by polynomial regression analysis, the established mathematical model can be expressed as(3)
By substituting temperature control objective (T _{objmin} and T _{objmax}) for T _{ringmin} and T _{ringmax} in equation (3), the equations about v_{D} can be obtained. Then solving the obtained equations, the ring growth velocity model relating to the temperature control objective (T _{objmin} and T _{objmax}), instantaneous outer diameter of ring D_{t} and initial temperature of ring T _{0} can be expressed as(4)
Considering the equations about v_{D} above may have no solutions, the established ring growth velocity model has five cases: (5)
where v_{D} _{min} and v_{D} _{max} are the minimum and maximum values of ring growth velocity determined by the penetrating and gripping conditions during ring rolling process [14].
Based on this model, by comparing the obtained ring temperature (maximum and minimum temperature) with the temperature control objective, the ring growth velocity can be adjusted realtimely to maintain the ring temperature within the desired range, as shown in Figure 1. If both the maximum temperature and minimum temperature are within the desired temperature range (T _{ringmin }> T _{objmin} and T _{ringmax }< T _{objmax}), it is thought that current ring temperature meets the temperature control objective. Therefore, the ring growth velocity v_{D} remains the current value unchanged. Inversely, if the maximum temperature or minimum temperature is beyond the temperature control objective, the developed ring growth velocity model will be adopted to adjust the current rolling paths to control ring temperature back to the desired range. At this time, for T _{ringmin }< T _{objmin}, if f _{3 }(T _{objmin}, D_{t} , T _{0}) ≥ v_{D} _{max} or in the case of no solution, the maximum ring growth velocity v_{D} _{max} allowed by the gripping conditions will be employed to increase the generation of deformation heat; for T _{ringmax }> T _{objmax}, if f _{4} (T _{objmax}, D_{t} , T _{0}) ≤ v_{D} _{min} or in the case of no solution, the minimum ring growth velocity v_{D} _{min} allowed by the penetrating conditions will be employed to decrease the generation of deformation heat.
From the developed v_{D} model, the temperature control objective (T _{objmin} and T _{objmax}), instantaneous outer diameter of ring D_{t} and initial temperature of ring T _{0} are the parameters affecting the ring growth velocity. The former is predefined according to the intrinsic properties of material. The instantaneous outer diameter D_{t} could be monitored in real time by virtual sensors. The initial ring temperature T _{0} is predefined, which is a key factor significantly affecting the ring temperature. Additionally, as an initial designed variable to ensure the success of rolling process, the different values (v_{D} _{0} _{1}, v_{D} _{0} _{2}) of v_{D} _{0} will lead to obvious change of subsequent ring growth velocity, as shown in Figure 3. This eventually affects the rolling paths and the ring temperature. Therefore, the influence rules of the two factors (T _{0} and v_{D} _{0}) on the rolling paths and temperature control effects will be discussed in this work.
Fig. 1
Main idea of targettemperature driven intelligent FE simulation of radialaxial ring rolling. 
Fig. 2
Rolling curve for intelligent FE simulation of radialaxial ring rolling. 
Fig. 3
Change of growth velocity of ring diameter with the rolling time. 
2.2 Targettemperature driven intelligent FE model of radialaxial ring rolling
To validate the feasibility of the intelligent temperature control idea proposed above, FE simulation experiments for radialaxial ring rolling process are performed. The established 3D coupled thermomechanical FE model in this work is shown in Figure 4.
The blank material used in FE model is TA15 titanium alloy, whose density and Poisson ratio are 4450 kg/m^{3} and 0.39, respectively. The temperaturedependent physical properties of TA15 titanium alloy, including Yong's modulus, specific heat, thermal conductivity and linear expansion coefficient, are derived from the reference [23]. The constitutive relationship reported by Shen [24] is employed to describe the plastic deformation behavior of TA15 titanium alloy.
In the model, the coupled thermomechanical elasticplastic dynamic explicit approach and mass scaling are adopted to accelerate the computation. The ring is meshed by the coupled thermomechanical hexahedron element with eight nodes (C3D8RT). An adaptive meshing technology is employed for the whole ring blank to maintain a highquality mesh throughout the analysis. The reduction integration and hourglass control are utilized to shorten the computation time and avoid the zeroenergy model induced by the bending model of deformation.
The ring blank and all the rolls are treated as deformable body and rigid bodies, respectively. Eight contact pairs exist between the ring and the rolls. The friction conditions and contact heat conduction at the interface of every contact pair are defined. The modified Coulomb friction model is adopted and the friction coefficients are assumed to be constant during the whole analysis. The guide rolls and supporting rolls are assumed to be smooth surfaces, thus the friction coefficients on them are zero. The needed simulation parameters are listed in Table 1.
Through VUAMP subroutine development under ABAQUS/Explicit environment, the intelligent temperature control method above is implemented into the 3DFE model of radialaxial ring rolling for TA15 titanium alloy. The ring growth velocity can be adjusted in real time by tracing the temperature control objective, thus maintaining ring temperature within the desired range during the whole process. As the initial conditions of intelligent simulation, different T _{0} and v_{D} _{0} will affect the rolling paths, thus resulting in difference in temperature control effects of ring.
Fig. 4
3DFE model for radialaxial ring rolling. 
FE simulation parameters.
3 Results and discussion
To obtain the rolling paths under different initial conditions and study the influence rules of different initial conditions on the rolling paths and temperature control effects, the intelligent FE simulations are performed under different T _{0} and v_{D} _{0}. Based on the intelligent temperature control idea shown in Figure 1, the temperature control objective (T _{objmin}~∼T_{objmax}), initial temperature of ring T _{0} and initial ring growth velocity v_{D} _{0} need to be predefined before FE calculations. The phase transformation temperature of TA15 titanium alloy is 985 °C [25], and its forgingending temperature should be larger than 850 °C [26]. To keep better comprehensive mechanical performance and low deformation resistance, the lower limit and upper limit of temperature control objective are set to be 900 °C and 980 °C, respectively. Considering the temperature drop during the transfer and rolling process, the range of initial ring temperature T _{0} is set to be 900–970 °C. Based on the gripping and penetrating conditions of steady rolling [14], the range of initial ring growth velocity v_{D0} is determined to be 6–35 mm/s. Here, initial ring temperature T _{0} is set as 915 °C, 920 °C, 925 °C, 930 °C, 935 °C, 940 °C, 945 °C and 950 °C, respectively. Initial ring growth velocity v_{D} _{0} is set as 6 mm/s, 20 mm/s and 35 mm/s, respectively.
3.1 Rolling paths by targettemperature driven intelligent simulation
3.1.1 Rolling paths under different initial conditions
During the targettemperature driven intelligent simulation, the ring growth velocity v_{D} with respect to the rolling time can be output by the VUAMP subroutine development. The feed rate of mandrel v_{f} and feed rate of upper axial roll v_{a} with respect to the rolling time can be extracted from the output database after FE calculation. Then the variation curves of ring growth velocity, feed rates of mandrel and upper axial roll with respect to the rolling time, namely the rolling paths, can be obtained.
Considering the rolling time is different while the outer diameter of rolled ring is the same under different initial conditions, the variation curves of ring growth velocity, feed rates of mandrel and upper axial roll with respect to the outer diameter of ring D_{t} are also given. Here, the outer diameter of ring D_{t} can be detected by virtual sensors and output by the VUAMP subroutine during intelligent simulation of ring rolling.
Figures 5–7 show the designed rolling paths under different initial ring temperatures when the initial ring growth velocity v_{D} _{0} is 6 mm/s, 20 mm/s and 35 mm/s, respectively. As analyzed above, the initial ring temperature T _{0} and initial ring growth velocity v_{D} _{0} are key parameters influencing the rolling paths. Therefore, the influence rule of T _{0} and v_{D} _{0} on the designed rolling paths will be discussed below.
Fig. 5
Rolling paths designed by targettemperature driven intelligent simulation under different initial ring temperatures when v_{D} _{0} = 6 mm/s: (a), (c) and (e) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with time, respectively. (b), (d) and (f) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with ring outer diameter, respectively. 
Fig. 6
Rolling paths designed by targettemperature driven intelligent simulation under different initial ring temperatures when v_{D} _{0} = 20 mm/s: (a), (c) and (e) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with time, respectively. (b), (d) and (f) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with ring outer diameter, respectively. 
Fig. 7
Rolling paths designed by targettemperature driven intelligent simulation under different initial ring temperatures when v_{D} _{0} = 35 mm/s: (a), (c) and (e) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with time, respectively. (b), (d) and (f) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with ring outer diameter, respectively. 
3.1.2 Influence rule of initial conditions on rolling paths
Under the same initial ring growth velocity v_{D} _{0}, for the initial temperature of 915 °C, 920 °C, 925 °C and 930 °C, the curves of ring growth velocity, feed rate of mandrel and feed rate of upper axial roll are almost coincident, as shown in Figures 5–7. This is because the minimum temperature of ring T _{ringmin} will be less than the lower limit of temperature control objective within a very short time when the initial ring temperature is too low. The ring growth velocity model v_{D} = f _{3}(T _{objmin}, D_{t} , T _{0}) will be employed to regulate the ring temperature. However, v_{D} belongs to the case of no solution at this time. According to the ring growth velocity model in equation (5), to avoid further drop of the minimum temperature of ring, the maximum value of ring growth velocity (v_{D} _{max}) determined by the gripping conditions is directly employed to adjust the motions of all rolls, thus generating more deformation heat for increasing the minimum temperature of ring as soon as possible. That is to say, the same ring growth velocity (v_{D} _{max}) is employed for the initial temperature of 915 °C, 920 °C, 925 °C and 930 °C, thus leading to the same rolling paths. Therefore, the curves of ring growth velocity, feed rate of mandrel and feed rate of upper axial roll are almost coincident.
When the initial temperature of ring T _{0} > 930 °C, the ring growth velocity first increases then decreases and finally increases again. Driven by the ring growth velocity, the feed rates of mandrel and axial roll also change accordingly. This can be explained as follows: at the initial stage of rolling process, the minimum temperature of ring T _{ringmin} drops below the lower limit of control objective T _{objmin}, so the ring growth velocity increases to generate more deformation heat. Due to the increase of initial temperature of ring, the maximum temperature of ring T _{ringmax} becomes easy to exceed the upper limit of control objective T _{objmax}. To decrease the maximum temperature of ring, the ring growth velocity decreases to reduce the generation of plastic deformation heat. At last, the ring growth velocity increases again to maintain the minimum temperature of ring within the desired range.
In addition, when initial ring temperature T _{0} > 930 °C, the variation trends of ring growth velocity and the feed rates of mandrel and axial roll under different initial ring temperatures are similar. However, with the increase of initial temperature of ring, the ring growth velocity and the feed rates of mandrel and axial roll decrease. This will lead to the rolling time becoming longer, thus reducing the heat generation and increasing the heat loss to maintain the ring temperature within the control objective.
Under different initial ring growth velocities v_{D} _{0}, the ring growth velocity shows a significant difference during the whole process. This agrees well with the analysis in Figure 3. The greater the initial ring growth velocity, the more deformation heat at the initial period of rolling process, resulting in the ring temperature being easier to exceed the upper limit of temperature control objective. Therefore, the ring growth velocity under v_{D} _{0} = 35 mm/s first decreases to reduce the deformation heat. Moreover, the time under lower velocity becomes longer until the ring temperature meets the control objective. The feed rates of mandrel and upper axial roll also decrease accordingly. This eventually makes the rolling paths become different under different v_{D} _{0}.
In a word, according to different initial conditions (T _{0} and v_{D} _{0}), the ring growth velocity can be adjusted adaptively by tracing the temperature control objective during the targettemperature driven intelligent FE simulation. Driven by the ring growth velocity, the motions of mandrel and upper axial roll will also be adjusted accordingly. In the actual production, the ring growth velocity is the main parameter the ring rolling mill needs to input. If the ring growth velocity is determined, the rolling paths can be obtained. Therefore, the targettemperature driven intelligent simulation can realize the quick design for rolling paths in radialaxial ring rolling. Different rolling paths will lead to different deformation behaviors, thus influencing the temperature of ring. The temperature control effects under the designed rolling paths are discussed below.
3.2 Temperature control effects under designed rolling paths
3.2.1 Temperature control effects
In FE model, the ring is meshed by a total of 30240 elements. The temperatures of all the elements are sorted from high to low in each increment. Here, the temperature of the 29938th element (30240 × 99% = 29938) is defined as T _{ring99%}. If T _{ring99%} is maintained above the lower limit of the control objective T _{objmin} during the whole process, it is indicated that the temperatures of ninetynine percent elements have been controlled accurately. The minimum temperature mainly exists on the surface of ring, while the low temperature part of ring will be cut off during the subsequent machining process in actual production. This means the temperature control for this part almost can be neglected. Therefore, the ring temperature can be considered to meet the temperature control requirement during the whole process.
Different initial conditions will lead to great difference of rolling paths, so the temperature control effects under different initial ring temperatures T _{0} and initial ring growth velocities v_{D} _{0} are different. Figures 8–10 display the temperature control effects under different initial ring temperatures when initial ring growth velocity v_{D} _{0} is 6 mm/s, 20 mm/s and 35 mm/s, respectively. The influence rule of initial conditions of T _{0} and v_{D} _{0} on temperature control effects is discussed below.
Fig. 8
Control effects of ring temperature under designed rolling paths when v_{D} _{0} = 6 mm/s: (a) T _{0} = 915 °C; (b) T _{0} = 920 °C; (c) T _{0} = 925 °C; (d) T _{0} = 930 °C; (e) T _{0} = 935 °C; (f) T _{0} = 940 °C; (g) T _{0} = 945 °C; (h) T _{0} = 950 °C. 
Fig. 9
Control effects of ring temperature under designed rolling paths when v_{D} _{0} = 20 mm/s: (a) T _{0} = 915 °C; (b) T _{0} = 920 °C; (c) T _{0} = 925 °C; (d) T _{0} = 930 °C; (e) T _{0} = 935 °C; (f) T _{0} = 940 °C; (g) T _{0} = 945 °C; (h) T _{0} = 950 °C. 
Fig. 10
Control effects of ring temperature under designed rolling paths when v_{D} _{0} = 35 mm/s: (a) T _{0} = 915 °C; (b) T _{0} = 920 °C; (c) T _{0} = 925 °C; (d) T _{0} = 930 °C; (e) T _{0} = 935 °C; (f) T _{0} = 940 °C; (g) T _{0} = 945 °C; (h) T _{0} = 950 °C. 
3.2.2 Influence rule of initial conditions on temperature control effects
Under the same initial ring growth velocity like v_{D} _{0} = 6 mm/s in Figure 8, the ring temperature cannot be accurately controlled under the rolling paths of 915 °C and 950 °C. For T _{0} = 915 °C in Figure 8a, the minimum temperature of ring T _{ringmin} drops rapidly below the lower limit of the control objective T _{objmin}. Driven by the ring growth velocity model, the ring growth velocity increases to generate more deformation heat. However, the generated plastic deformation heat is not enough to compensate for the temperature drop because the initial temperature is too low, leading to T _{ring99%} being lower than T _{objmin} during almost the whole process. Therefore, the ring temperature cannot meet the control objective. For T _{0} = 950 °C in Figure 8h, the maximum temperature of ring T _{ringmax} cannot be reduced in time because the initial temperature is too high, leading to longer time (larger than 10 s) above the upper limit of control objective T _{objmax}. Due to the lower ring growth velocity during the whole process, the rolling time increases, resulting in a great increase of heat loss. This makes the deviation value of T _{ring99%} from T _{objmin} is relatively large (greater than 10 °C) at the end of rolling process. This means that the control effects of ring temperature are not satisfactory under the rolling paths when T _{0} equals to 950 °C.
For the initial ring temperature of 920 °C, 925 °C and 930 °C in Figure 8b–d, the ring growth velocity rises rapidly to increase the deformation heat when the minimum temperature of ring T _{ringmin} is lower than the lower limit of temperature control objective T _{objmin}, which eventually maintains the minimum temperature of ring near the lower limit of control objective. However, with the increase of ring outer diameter, the rotational speed of ring becomes slower and the time interval passing through the deformation zones increases. Owing to the increase of heat dissipation area, the heat loss to ambient increases. The temperature drop can't be compensated by the deformation heat due to the lower initial temperature and the more heat loss. The minimum temperature of ring T _{ringmin} drops continuously at the end of rolling process. However, T _{ring99%} has been maintained above the lower limit of control objective T _{objmin}. Therefore, the ring temperature could meet the temperature control objective under the rolling paths when T _{0} equals to 920 °C, 925 °C and 930 °C.
For the initial ring temperature of 935 °C, 940 °C and 945 °C in Figure 8e–g, the ring growth velocity increases to avoid the further decrease of the minimum temperature of ring at the early stage of ring rolling. However, the maximum temperature of ring T _{ringmax} exceeds the upper limit of control objective T _{objmax} due to the higher initial temperature and the accumulation of deformation heat in the ring. Drive by the ring growth velocity model, the ring growth velocity decreases to reduce the generation of deformation heat. When the generated deformation heat is less than heat loss to ambient, the maximum temperature of ring T _{ringmax} starts to decrease until T _{ringmax} meets the temperature control objective. After this, to raise the minimum temperature of ring T _{ringmin}, the ring growth velocity increases again. This means the ring growth velocity realizes the intelligent control for both the minimum and maximum temperatures. Besides, T _{ring99%} is maintained above the lower limit of temperature control objective almost in the whole process, especially for 935 °C. Though T _{ring99%} when T _{0} equals to 940 °C and 945 °C is lower than T _{objmin} due to more heat loss to ambient, the temperature deviation value is quite small (less than 10 °C). Therefore, the temperature control effects under the rolling paths when initial temperature T _{0} equals to 935 °C, 940 °C and 945 °C are considered acceptable.
From Figures 8–10, under different initial ring growth velocities v_{D0} , the similar conclusions can be obtained. For the initial temperature of 915 °C and 950 °C, the ring temperature cannot be accurately maintained within the desired range under the corresponding rolling paths. For the initial temperature of 920 °C–945 °C, it is considered that the control effects of ring temperature are all satisfactory under the corresponding rolling paths. It is revealed that initial temperature T _{0} exists a feasible range of 920 °C−945 °C for realizing the intelligent control of ring temperature. Within this feasible range, as long as the initial ring growth velocity v_{D} _{0} meets the gripping and penetrating conditions, the ring temperature can be controlled within the desired range. This provides an important reference for the determination of initial temperature of ring in actual production of ring rolling of TA15 titanium alloy.
However, as discussed above, different initial ring growth velocities affect the rolling paths, thus leading to the difference of ring temperature control effects. From Figures 8–10, under different initial ring growth velocities and the same initial temperature of ring, the higher the initial value of ring growth velocity v_{D} _{0}, the earlier the maximum temperature of ring T _{ringmax} exceeds the upper limit of the control objective T _{objmax} due to the fast accumulation of plastic deformation heat and the less heat loss. Driven by the ring growth velocity model, the subsequent ring growth velocity decreases early to reduce the deformation heat. The lower the initial value of ring growth velocity v_{D} _{0}, the earlier the minimum temperature of ring T _{ringmin} is less than the lower limit of the control objective T _{objmin} due to the less plastic deformation heat and the more heat loss. The subsequent ring growth velocity increases early to avoid further temperature drop with the driving of ring growth velocity model.
3.2.3 Temperature evolution under designed rolling paths
To further study the temperature distribution and temperature control effects under the designed rolling paths, Figures 11–13 display the temperature evolution with rolling time under the rolling paths which could realize accurate temperature control when v_{D} _{0} = 6 mm/s, 20 mm/s and 35 mm/s, respectively.
Obviously, the low temperature region is mainly distributed on the surface of ring due to its direct heat transfer behavior with the rolls and external environment. The temperature inside the ring is relatively higher than other regions owing to the less heat loss.
From Figures 11–13, under the same initial ring growth velocity, for the initial ring temperature of 920 °C, 925 °C and 930 °C, the maximum temperature of ring T _{ringmax} increases gradually with the rolling time, which can be attributed to the continuous accumulation of plastic deformation heat caused by the increase of ring growth velocity. The minimum temperature of ring T _{ringmin} first increases and then decreases at the late stage of rolling process, which is in good agreement with the results obtained in Section 3.2.2.
For the initial ring temperature of 935 °C, 940 °C and 945 °C, the maximum temperature of ring T _{ringmax} first increases and then decreases with the rolling time. As discussed above, the maximum temperature of ring T _{ringmax} is easy to exceed the upper limit of temperature control objective T _{objmax} when the initial ring temperature T _{0} is higher. The ring growth velocity will decrease to reduce the deformation heat at this time. With the effect of intelligent control of rolling paths, the maximum temperature of ring T _{ringmax} is maintained below the upper limit of temperature control objective T _{objmax}. For the minimum temperature of ring, only the temperature in the ring corner becomes less than the lower limit of temperature control objective T _{objmin}. This indicates that the ring temperature can be maintained within the desired range of 900–980 °C throughout the ring rolling process.
Besides, from Figures 11–13, it can be seen that all the rings keep a high geometric quality during the targettemperature driven intelligent simulation of radialaxial ring rolling process. The rolled rings with higher circularity are obtained, which reveals that the whole rolling process can maintain a higher stability under the effect of rolling paths designed by targettemperature driven intelligent FE simulation.
The analysis above indicates not only the ring temperature can be controlled within the reasonable range, but the whole rolling process can keep a higher stability, using the targettemperature driven intelligent FE simulation. This means the reasonable rolling paths which could realize intelligent temperature control and ensure process stability simultaneously can be designed directly only by one intelligent simulation. The obtained results could provide a meaningful reference for the design and optimization of rolling paths and accurate regulation of ring temperature in the actual industrial production of ring rolling, thus greatly improving the production efficiency and the quality of rolled ring.
Fig. 11
Temperature evolution under designed rolling paths when v_{D} _{0} = 6 mm/s: (a) T _{0} = 920 °C; (b) T _{0} = 925 °C; (c) T _{0} = 930 °C; (d) T _{0} = 935 °C; (e) T _{0} = 940 °C; (f) T _{0} = 945 °C. 
Fig. 12
Temperature evolution under designed rolling paths when v_{D} _{0} = 20 mm/s: (a) T _{0} = 920 °C; (b) T _{0} = 925 °C; (c) T _{0} = 930 °C; (d) T _{0} = 935 °C; (e) T _{0} = 940 °C; (f) T _{0} = 945 °C. 
Fig. 13
Temperature evolution under designed rolling paths when v_{D} _{0} = 35 mm/s: (a) T _{0} = 920 °C; (b) T _{0} = 925 °C; (c) T _{0} = 930 °C; (d) T _{0} = 935 °C; (e) T _{0} = 940 °C; (f) T _{0} = 945 °C. 
4 Conclusions
Through targettemperature driven intelligent FE simulation, this work proposes a method for designing the rolling paths which could realize accurate control of the ring temperature. The main conclusions are drawn as follows:

The main idea of targettemperature driven intelligent simulation is introduced. In this idea, the initial ring temperature T _{0} and initial ring growth velocity v_{D} _{0} are two key parameters influencing the rolling paths. By VUAMP subroutine development, an intelligent 3DFE model for TA15 titanium alloy ring rolling is established under ABAQUS/Explicit environment.

By targettemperature driven intelligent simulations, the rolling paths under different initial conditions (T _{0} and v_{D} _{0}) are obtained, and the influence rule of T _{0} and v_{D} _{0} on rolling paths is revealed. The higher the initial ring temperature T _{0}, the smaller the ring growth velocity v_{D} . The greater the initial ring growth velocity v_{D} _{0}, the earlier the ring growth velocity v_{D} decreases. This method could realize the quick and accurate design for the rolling paths in radialaxial ring rolling process.

The temperature control effects under the designed rolling paths are discussed and the influence rule of T _{0} and v_{D} _{0} on temperature control effects is revealed. The simulation results indicate the ring temperature could be maintained within the desired range when the initial ring temperature T _{0} is within the feasible range of 920–945 °C. This provides a meaningful reference for accurate regulation of ring temperature in the actual production.
Acknowledgments
The authors would like to thank the support of the National Natural Science Foundation of China (No. 51575448, 51875468) and the Research & Development Institute of Northwestern Polytechnical University in Shenzhen (JCYJ20170815162709770).
References
 R. M'Saoubi, C. Nobel, W.M. Sim, High performance cutting of advanced aerospace alloys and composite materials, CIRP Ann. Manuf. Technol. 64 (2015) 557–580 [CrossRef] [Google Scholar]
 S. Zhu, H. Yang, L.G. Guo et al., Effect of cooling rate on microstructure evolution during α/β heat treatment of TA15 titanium alloy, Mater. Char. 70 (2012) 101–110 [CrossRef] [Google Scholar]
 P.F. Gao, H. Yang, X.G. Fan et al., Microstructure evolution in the local loading forming of TA15 titanium alloy under nonisothermal condition [J], J. Mater. Process. Technol. 212 (2012) 2520–2528 [CrossRef] [Google Scholar]
 D. He, J. Zhu, S. Zaefferer et al., Effect of retained beta layer on slip transmission in Ti6Al2Zr1Mo1V near alpha titanium alloy during tensile deformation at room temperature, Mater. Des. 56 (2014) 937–942 [CrossRef] [Google Scholar]
 J.M. Allwood, A.E. Tekkaya, T.F. Stanistreet, The development of ring rolling technology, Steel Res. Int. 76 (2005) 111–120 [CrossRef] [Google Scholar]
 H. Yang, M. Wang, L.G. Guo et al., 3D coupled thermomechanical FE modeling of blank size effects on the uniformity of strain and temperature distributions during hot rolling of titanium alloy large rings, Comput. Mat. Sci. 44 (2009) 611–621 [CrossRef] [Google Scholar]
 S. Zhu, H. Yang, L. Guo et al., Research on the effects of coordinate deformation on radialaxial ring rolling process by FE simulation based on inprocess control, Int. J. Adv. Manuf. Technol. 72 (2014) 57–68 [CrossRef] [Google Scholar]
 Z. Sun, H. Yang, X. Ou, Effects of process parameters on microstructural evolution during hot ring rolling of AISI 5140 steel, Comput. Mater. Sci. 49 (2010) 134–142 [CrossRef] [Google Scholar]
 G. Zhou, L. Hua, J. Lan et al., FE analysis of coupled thermomechanical behaviors in radialaxial rolling of alloy steel large ring, Comput. Mater. Sci. 50 (2010) 65–76 [CrossRef] [Google Scholar]
 X. Tang, B. Wang, H. Zhang et al., Study on the microstructure evolution during radialaxial ring rolling of IN718 using a unified internal state variable material model, Int. J. Mech. Sci. 128 (2017) 235–252 [CrossRef] [Google Scholar]
 S.L. Semiatin, S.L. Knisley, P.N. Fagin et al., Microstructure evolution during alphabeta heat treatment of Ti6Al4V, Metal. Mater. Trans. A 34 (2003) 2377–2386 [CrossRef] [Google Scholar]
 Y.G. Zhou, W.D. Zeng, H.Q. Yu, An investigation of a new nearbeta forging process for titanium alloys and its application in aviation components, Mater. Sci. Eng. A 393 (2005) 204–212 [CrossRef] [Google Scholar]
 X.G. Fan, H. Yang, P.F. Gao, Throughprocess macromicro finite element modeling of local loading forming of largescale complex titanium alloy component for microstructure prediction, J. Mater. Process. Technol. 214 (2014) 253–266 [CrossRef] [Google Scholar]
 L. Guo, H. Yang, Towards a steady forming condition for radialaxial ring rolling, Int. J. Mech. Sci. 53 (2011) 286–299 [CrossRef] [Google Scholar]
 X. Han, L. Hua, G. Zhou et al., FE simulation and experimental research on cylindrical ring rolling, J. Mater. Process. Technol. 214 (2014) 1245–1258 [CrossRef] [Google Scholar]
 C. Wang, H.J.M. Geijselaers, E. Omerspahic et al., Influence of ring growth rate on damage development in hot ring rolling, J. Mater. Process. Technol. 227 (2016) 268–280 [CrossRef] [Google Scholar]
 M. Wang, H. Yang, Z.C. Sun et al., Analysis of coupled mechanical and thermal behaviors in hot rolling of large rings of titanium alloy using 3D dynamic explicit FEM, J. Mater. Process. Technol. 209 (2009) 3384–3395 [CrossRef] [Google Scholar]
 N. Anjami, A. Basti, Investigation of rolls size effects on hot ring rolling process by coupled thermomechanical 3DFEA, J. Mater. Process. Technol. 210 (2010) 1364–1377 [CrossRef] [Google Scholar]
 M. Wang, H. Yang, L. Guo et al., Effects and optimization of roll sizes in hot rolling of large rings of titanium alloy, Rare Metal Mater. Eng. 38 (2009) 393–397 [CrossRef] [Google Scholar]
 L. Guo, F. Wang, L. Liang et al., New idea and advances in intelligent simulation and optimization of highperformance ring rolling process, J. Netshape Form. Eng. 9 (2017) 1–11 [in Chinese] [Google Scholar]
 D. Yang, Research on intelligent modeling and simulation of radialaxial rolling process for large 2219 aluminum alloy ring based on force feedback control, Northwestern Polytechnical University, 2017 [in Chinese] [Google Scholar]
 L. Liang, L.G. Guo, X.C. Li et al., Intelligent simulation for realtimely forcecontrolled radialaxial rolling process of supersized aluminium alloy rings, Proc. Manuf. 15 (2018) 105–112 [Google Scholar]
 B. Huang, C. Li, L. Shi et al., China materials engineering canon [M]. Chemical Industry Press, Beijing, China, 2005, pp. 566–577 [in Chinese] [Google Scholar]
 C. Shen, Research on material constitution models of TA15 and TC11 titanium alloys in hot deformation processes, Northwestern Polytechnical University, 2007 [in Chinese] [Google Scholar]
 S. Zhu, Deformation and microstructure evolution in whole process of radialaxial ring rolling of TA15 titanium alloy [D]. Northwestern Polytechnical University (2015) [in Chinese] [Google Scholar]
 X. Song, Study on the forming features and process of large scale TA15 titanium alloy aviation structural parts, Chongqing University, 2011 [in Chinese] [Google Scholar]
Cite this article as: Xuechao Li, Lianggang Guo, Yifan Wang, Lei Liang, Rolling paths design assisted by targettemperature driven intelligent FE simulation of radialaxial ring rolling, Manufacturing Rev. 6, 20 (2019)
All Tables
All Figures
Fig. 1
Main idea of targettemperature driven intelligent FE simulation of radialaxial ring rolling. 

In the text 
Fig. 2
Rolling curve for intelligent FE simulation of radialaxial ring rolling. 

In the text 
Fig. 3
Change of growth velocity of ring diameter with the rolling time. 

In the text 
Fig. 4
3DFE model for radialaxial ring rolling. 

In the text 
Fig. 5
Rolling paths designed by targettemperature driven intelligent simulation under different initial ring temperatures when v_{D} _{0} = 6 mm/s: (a), (c) and (e) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with time, respectively. (b), (d) and (f) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with ring outer diameter, respectively. 

In the text 
Fig. 6
Rolling paths designed by targettemperature driven intelligent simulation under different initial ring temperatures when v_{D} _{0} = 20 mm/s: (a), (c) and (e) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with time, respectively. (b), (d) and (f) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with ring outer diameter, respectively. 

In the text 
Fig. 7
Rolling paths designed by targettemperature driven intelligent simulation under different initial ring temperatures when v_{D} _{0} = 35 mm/s: (a), (c) and (e) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with time, respectively. (b), (d) and (f) are ring growth velocity, feed rate of mandrel and feed rate of upper axial roll with ring outer diameter, respectively. 

In the text 
Fig. 8
Control effects of ring temperature under designed rolling paths when v_{D} _{0} = 6 mm/s: (a) T _{0} = 915 °C; (b) T _{0} = 920 °C; (c) T _{0} = 925 °C; (d) T _{0} = 930 °C; (e) T _{0} = 935 °C; (f) T _{0} = 940 °C; (g) T _{0} = 945 °C; (h) T _{0} = 950 °C. 

In the text 
Fig. 9
Control effects of ring temperature under designed rolling paths when v_{D} _{0} = 20 mm/s: (a) T _{0} = 915 °C; (b) T _{0} = 920 °C; (c) T _{0} = 925 °C; (d) T _{0} = 930 °C; (e) T _{0} = 935 °C; (f) T _{0} = 940 °C; (g) T _{0} = 945 °C; (h) T _{0} = 950 °C. 

In the text 
Fig. 10
Control effects of ring temperature under designed rolling paths when v_{D} _{0} = 35 mm/s: (a) T _{0} = 915 °C; (b) T _{0} = 920 °C; (c) T _{0} = 925 °C; (d) T _{0} = 930 °C; (e) T _{0} = 935 °C; (f) T _{0} = 940 °C; (g) T _{0} = 945 °C; (h) T _{0} = 950 °C. 

In the text 
Fig. 11
Temperature evolution under designed rolling paths when v_{D} _{0} = 6 mm/s: (a) T _{0} = 920 °C; (b) T _{0} = 925 °C; (c) T _{0} = 930 °C; (d) T _{0} = 935 °C; (e) T _{0} = 940 °C; (f) T _{0} = 945 °C. 

In the text 
Fig. 12
Temperature evolution under designed rolling paths when v_{D} _{0} = 20 mm/s: (a) T _{0} = 920 °C; (b) T _{0} = 925 °C; (c) T _{0} = 930 °C; (d) T _{0} = 935 °C; (e) T _{0} = 940 °C; (f) T _{0} = 945 °C. 

In the text 
Fig. 13
Temperature evolution under designed rolling paths when v_{D} _{0} = 35 mm/s: (a) T _{0} = 920 °C; (b) T _{0} = 925 °C; (c) T _{0} = 930 °C; (d) T _{0} = 935 °C; (e) T _{0} = 940 °C; (f) T _{0} = 945 °C. 

In the text 
Current usage metrics show cumulative count of Article Views (fulltext article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 4896 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.