Integration of a complex control system into a numerical process model for the simulation of ring rolling processes

在传统的环件轧制过程有限元模拟中，必须在模拟前确定整个过程中所有工具的运动。由于在实际过程中，所有的运动都是由复杂的闭环控制系统根据实际过程变量和预先选择的控制策略自适应控制的，因此不可能预先规定所有的运动。因此，在研究的第一部分中，将工业控制器的复杂控制算法集成到环件轧制的数值过程模型中。为此，将环锭轧机制造商提供的预编译版工业控制器与环锭轧制有限元仿真相耦合。在模型中，工具的所有平移和旋转速度都根据当前传感器值（位置、力和扭矩）调整为执行器，这些值在有限元模拟中计算。此外，该控制在考虑初始用户要求（轧制策略、期望的环生长速率和几何形状）的同时考虑了机器的力和扭矩限制。新方法首次实现了具有真实机器特定运动历史的完整环轧制过程的模拟，而无需使用来自实验的运动变量。由于工业控制系统被设计为在机器的功率极限下执行过程，因此正确估计力和扭矩在反馈控制仿真中是至关重要的。由于这些值并不总是准确预测的，该提案第二阶段的主要目标是研究和改进有限元模型中的力和扭矩计算。一方面，这涉及到热场计算的评价和必要时的改进。另一方面，必须评估在何种条件下，辊缝之间可能发生的静态软化机制会显著影响流动行为。两种材料的软化行为有很大的不同，将通过实验分析大和小运行时间之间的辊隙。利用这些实验的模拟，可以通过提取不同元素的局部热历史和力学历史来计算再结晶。由于在某些情况下，虽然力的计算是正确的，但模拟扭矩与实验结果不一致，因此需要验证轧制扭矩测量的准确性。由于所提供的工业控制器只包含一个相关的工业滚动策略作为黑盒，因此并不总是能够确定仿真和实验之间微小差异的原因。此外，这种方法的另一个缺点是控制算法不能增强和修改。因此，将实现一个具有类似于工业控制器的技术上合理的控制算法的模型。

In conventional Finite Element Simulation of the ring rolling process the motions of all tools have to be defined for the entire process prior to simulation. Since in the real process all motions are adaptively controlled by sophisticated closed-loop control systems according to actual process variables and preselected control strategies, it is not possible to prescribe all motions beforehand. Therefore, in the first part of the investigation the complex control algorithms of an industrial controller have been integrated into a numerical process model of ring rolling. For this purpose, a precompiled version of an industrial controller, which was provided by the manufacturer of ring rolling mills, has been coupled to the FE simulation of ring rolling. Within the model all translational and rotational velocities of the tools are adapted as actuators according to current sensor values (positions, forces and torques), which are calculated within the FE simulation. Furthermore, the control considers the force and torque restrictions of the machine while taking into account initial user requirements (rolling strategy, desired ring growth rate and geometry). The new approach enables for the first time the simulation of a complete ring rolling process with realistic machine specific kinematic history without using kinematic variables from the experiment. Since industrial control systems are designed to perform the process at the power limit of the machine, a correct estimation of forces and torques is of utmost importance in feedback controlled simulation. As these values were not always accurately predicted, the main goal of the second phase of the proposal is to investigate and improve the force and torque calculation in FE model. On the one hand this involves evaluation and if necessary improvement of the thermal field calculation. On the other hand it must be evaluated under which conditions static softening mechanisms, which might occur between the roll gaps, influence the flow behavior significantly. Two materials with highly different softening behaviors will be analyzed by experiments with large and small runtimes between the roll gaps. Using the simulations of these experiments the recrystallization can be calculated by extracting the local thermal and mechanical history of distinct elements. Since in some cases, the simulated torque differs from the experimental results though forces are computed correctly, the accuracy of the rolling torques measurement should be verified. As the provided industrial controller contains only one relevant industrial rolling strategy as a black-box, it is not always possible to identify the reasons for the minor differences between simulations and experiments. Furthermore, another drawback of this approach is that the control algorithms cannot be enhanced and modified. Therefore, a model with technologically reasonable control algorithms similar to the industrial controller will be implemented.