Anyone that has designed processes to manufacture seamless rolled rings knows how challenging it is to produce them defect-free and with the required properties. Even rings with a simple rectangular cross section can suffer from any number of defects such as conicity, waviness, dishing, non-circularity (ovality) and cavities.
Modern customers are demanding more complicated profiled rings that reduce downstream machining requirements, however, and the greater the profile complexity, the more rings may suffer from additional defects such as profile nonfill/overfill as well as laps and folds. Further complicating the production is the fact that all rolled rings can suffer from problematic dynamic behavior such as climbing or beating.
Until recently, the simulation of rolled rings would not reliably show these defects. The problem was that these defects did occur, and if the simulation did not show them, then we were just wasting time running the simulation. We needed software that could show rolled-ring defects during the design and development phase so that the rolling process would produce quality rings without wasting time and money on shop-floor trials. The many challenges to accurately and reliably identify defects in simulations of rolled rings have now been overcome, and simulation has become a necessary step in the design and production of defect-free rings.
It is very important to simulate the entire process, from heating the raw billet through the end of the rolling process (Figure 1). This is because slight irregularities in the upset/forge/piercing operations can lead to defects in the final product.
The simulation of the process chain for these preliminary operations has been mature for many years, and these operations are very easily simulated with good accuracy by almost any simulation software.
Once the pierced blank has been simulated, however, setting up the equipment for an accurate rolling simulation requires that the software account for all the control parameters that might be included on a real ring-rolling mill. QForm software can now accurately simulate any kind of rolling operation with tool and workpiece rotation on equipment such as single-stand mills, double-stand mills, mills for rolling inside of the main roll or wheel rolling mills (Figure 2). To get accurate results, it is critical for simulation software to consider even the most minute particularities of the control of each of these machines.
A rolling simulation can be set up so that the control parameters used in the simulation are exactly like those used by a real rolling mill (Figures 3 and 4). In fact, the simulation interface allows direct input of the same parameters used in many commercially available rolling mills. Once you get the simulation to produce a ring that meets your requirements, you can simply transfer the values from the simulation directly into your real rolling mill so that the simulated and the real rolling mills run on exactly the same control parameters.
In a ring-rolling simulation on a double-stand mill, for example, the motion of the mandrel and axial rolls can be set up in any number of ways. Their control can be dependent on a traditional lamination curve; by the velocity of the mandrel and axial rolls; by the circular motion of the mandrel; or defined as a universal drive that can have any motion desired. You can use curves such as height dependent on thickness or ring-growth speed dependent on the diameter (Figure 5).
The velocity of the mandrel and axial rolls can be controlled based on the outer diameter of the ring (e.g., a closed-loop electromechanical drive system) or pre-set not to vary during the process (e.g., a hydraulic drive, Figure 6).
The rotational speed of the main roll can be entered as rpm (angular velocity) or as linear velocity based on the diameter of the roll. The horizontal displacement of the axial rolls can be set so the velocity is always half that of the ring-growth speed. It can also be dependent on the current ring diameter, or it can start at some specified diameter and then move with the ring-growth speed.
The load on the guide rolls can be set as a constant or as a curve, and even the deviation of the guide rolls can be controlled accurately in a simulation (Figure 7).
Even subtle differences in the height of the table from one side to the other can be adjusted and are considered in the simulation. The table itself can be a solid model from CAD, or it can simply be defined as a boundary condition. The initial positioning of the tools, the tool dynamics and the stop conditions for the simulation are flexible and easy to define.
Even though the volume of a rolled ring and the required tooling can be quite large, the contact zones between the tools and the ring are relatively small. To ensure accuracy, a very fine mesh is needed on both the tool and the ring at these deformation zones.
To have this fine a mesh throughout the entire ring and tools would make the simulation very slow. Further complicating the mesh is the fact that this deformation zone is constantly moving as the ring makes many revolutions through the tooling. So, the displacement of each node in the ring must be accurately calculated and translated as the ring deforms, grows and rotates.
These problems are overcome in QForm by several methods. The program uses a dual-mesh method (Figure 8) where a “mechanical mesh” is used to accurately calculate the deformation. This non-uniform mesh is very fine in the deformation zones and quite large in most of the ring (where no deformation is taking place), which gives good accuracy and high speed.
The second concurrent mesh is a so-called “geometry mesh” that is very fine throughout the entire ring. This mesh keeps the precise shape of the ring, and it is used for all the thermal calculations and for transferring strain and mechanical data back and forth between the “mechanical mesh” as it enters and leaves the deformation zones.
Another trick to improving the accuracy and speed in the software is that the Euler approach is used in the rolls and mandrel and the Lagrange approach is used in the ring itself. The Euler approach tracks specific locations in space through which the material flows in time. It is metaphorically characterized by sitting on the bank of a river and watching the water flow by. The Lagrange approach focuses on tracking a specific material point as it moves through space and time. It is metaphorically characterized by sitting in a boat and drifting down the river.
Even though the mesh is stationary in the tools, velocity influences the heat transfer in the rolls, which means that the heat is carried by conductivity and mass motion. This allows for fast simulations with accurate and realistic tool behavior (Figure 9).
Since rolled rings are generally used for critical components, it can be very useful to understand how the forming process affects the ring’s microstructure. Modern rolling simulations can include microstructure evolution through the entire forming process, from billet heating through the final product shape. The resulting data can include dynamic and metadynamic recrystallization, and the graphic output of the program should clearly show grain sizes on the surface as well as in a cross section of the final rolled product.
Verification of Accuracy
The accuracy of these methods of simulation have been verified by numerous industrial producers (Figures 11, 12, 13 and 14). Simulations of rectangular and profiled rings have shown very close correspondence to the actual rolled rings.
Tests have also been performed where the ring was rolled without the conical axial rolls (Figure 13) so that the horizontal faces of the ring were free to form to whatever shape occurred. This is a useful test of simulation accuracy because, since there are no axial rolls to control the material flow, the shape of the horizontal surfaces is formed only by the dynamics of the rolling mill acting on the flow stress of the material being rolled. In these tests the simulation of the same process yielded a shape virtually indistinguishable from the actual rolled ring.
Clearly, ring-rolling simulations have matured to the point where they are a necessary tool for the efficient production of defect-free rings. Most defects encountered in production can now reliably be detected in a simulation, and solutions for the defects can be quickly implemented and verified in the simulation.
The optimized ring-rolling technology chain – from billet heating to final rolled shape – developed via simulation virtually ensures a defect-free product, and the information from the successful simulation can be used to set the operating parameters of your rolling mill. The simulation calculates microstructure and material flow in the ring as well as the load in every tool component of the mill, which prevents possible overloading and inevitably improves the life of the tools.
Co-author Tom Ellinghausen is President of Forge Technology Inc., Woodstock, Ill.He can be reached at 815-337-7555 or firstname.lastname@example.org.
Co-author Nikolay Biba, Ph.D., is Director of Micas Simulations Limited, Oxford, UK. He can be reached at Nick@QForm3D.com. For additional information visit www.ForgeTechnology.com and www.qform3d.com.