This sixth article in our series on the North American forging industry focuses on process modeling, simulation and analysis. Not that long ago, computers did not exist at a forge shop. Drafting boards defined the landscape of the engineering department. Production and quality records were kept on cards and various paper forms, requiring hours or days to extract basic data. Initial advancements were limited to a single operation, process or department. Eventually, paper gave way to computers, and today these digital tools are increasingly integrated. They have become mainstream with exceptional capabilities.
From our view, the North American forging industry has advanced at an incredible rate during the last two decades. New tools and capabilities have been developed and deployed throughout the industry. Equipment, process controls and heating methods have been discussed in prior articles. A few highlights of modeling and analysis technologies will be discussed, ranging from the incremental improvements in Computer Aided Design (CAD) systems to the tools to look for in the not-too-distant future.
CAD/CAM (computer-aided design/computer-aided manufacturing) has been used in the forging industry for 25 years. Most forge shops are on their second or third system. Systems have become more capable and robust over the years. CAD systems have matured from 2D wireframe tools to 3D solid models that are very easy to use. A useful capability is the ability to capture a point cloud from a coordinate measurement machine, generate a surface and use the results to compare with a simulation or nominal (target) geometry. In Figure 1, a surface representing a production forging is compared with a simulation result to calibrate model accuracy. These are superimposed and sliced for clarity. This is fairly routine today but was impossible even 15 years ago.
To further highlight how routine this has become, a hammer forging ran into a quality issue requiring a quick simulation. There were multiple “manual” operations to form a roller, which would have taken some time to set up. Since the problems appeared in the closed-die operations at the end, a roller was pulled out of production and scanned (30 minutes). It was subsequently converted to a usable model and set up as the workpiece in a simulation (10 minutes). A successful simulation was possible with this enhanced CAD functionality.
Simulation is well known in the forging industry and has evolved from simple 2D models to very detailed 3D simulations. A majority of forge shops use process models during the design process to predict the die fill, load and potential defects in a proposed design. Additionally, the tool has been invaluable and well documented in troubleshooting problem jobs. Over time, 2D simulations morphed into 3D. Single-operation models of the past are now capable of modeling multiple operations.
The development strategy going forward has been to study and understand the influence of changes to a forging design and process through the end of the manufacturing process and into the life cycle of the component in service. This is often referred to as simulating the “process chain.” The aerospace sector, with severe strength, toughness and fatigue-life requirements, has been influential in this development.
An illustration of simulating multiple operations was published in the “New Developments in Forging Technology” conference proceedings (2013, Fellbach, Germany). An IN-718 turbine disk was modeled from heating the billet (Figure 2) through the solution anneal (prior to quenching). Grain size was tracked throughout the process. Operations that had little or no influence on grain size, such as resting on dies a few seconds before forging, were neglected. After the initial setup, a change to any input can be made, with the entire simulation sequence restarted in a few minutes. This is a critical foundation for design of experiments (DOE), optimization and material modeling.
DOE and Optimization
DOE is a systematic approach to investigate a system or process. A series of structured tests are designed, with planned changes made to the input variables. The effects of these changes on a predefined output are then assessed. When DOE is applied to a multiple-operations simulation environment capable of modeling a process chain, a powerful tool emerges. Trade-offs can be observed between different stages of the process. Global optimums can be studied. More importantly, the response of each input can be evaluated. Finally, the optimization can be conducted in a more well-defined range.
Process simulation is an excellent platform for DOE. Since simulation is a mature and accurate tool in forging applications, it is understandable how a set of simulations can replace shop trials as a planned experiment. When dozens or hundreds of trials are contemplated, the use of simulation becomes compelling. A high degree of setup and data-mining automation makes the toolbox suitable for a production forge shop. DOE has been applied to existing processes, new design development and testing the robustness of a process.
Existing processes can be investigated using simulation. DOE can be applied to a process as an inverse calculation method to extract a value for friction, a heat-transfer coefficient or other process parameter. A more compelling application is applying it to a process that resulted in quality problems. In the inevitable problem-solving team meeting, one participant suggests that workpiece chill or die temperature is the problem. Another suggests it is transfer time, and a third is certain it is friction due to die wear. In traditional meetings, the senior manager has the best idea, which is immediately adopted.
The Tornado chart (Figure 3) shows the relative effect of three variables as the result of a simulation-based DOE study. In this case, the longest bar (bottom) represents the most important variable. Bars on the right of the center axis have a direct correlation, or the higher value produces a higher value of the defined metric. Bars on the left, such as the bottom bar, have an inverse relationship. In this case, the bottom bar shows that a higher friction value produced the least die fill in a forging.
In new designs, DOE provides the opportunity for an engineer to study alternatives at an early stage of the design. They can study the influence of key geometric elements of a buster or blocker on the final forging, including the likelihood of measurable defects.
Quantifying robustness before the first trial run has been an elusive target over the years. DOE has this one nailed. Forgers all know the expected variation in billet size, transfer time, furnace temperature, location and other process conditions. In the 2D surface-response plot (Figure 4, left), the red dot on the left axis represents a nominal run. The white zone shows a large region with total die fill for various friction and die temperature conditions. This process is robust for a wide range of variation. Conversely, the plot on the right shows a case where a design is tested against forging temperature and ejector position. In this case, there is a practical upper temperature limit, shown by the dotted line. The green ellipse is the only practical area to achieve a satisfactory part, which requires “threading the needle” with process control. This is not a robust process.
Optimization cases using simulation in bulk-metal forming have been published over the years in commercial finite-element modeling (FEM) software products. Optimization shares a data structure and operational strategy with DOE. In DOE, the entire sampling strategy is defined prior to the first simulation or experiment. With optimization, the “next” simulation or trial is based on the last solution. A control engine adjusts one or more parameters as it seeks an optimum solution based on an objective with defined constraints.
Process and design variations can span an entire process chain, including geometry, process conditions and material data. Constraints include the number of allowable simulations (samples) and predefined defects that can be avoided. Defects such as underfills or folds (laps) are penalized to help the user define a robust process. Optimization will not replace design engineers anytime soon, but optimization can be a valuable tool in the development of designs with a well-known geometry and process space.
Simulating the forging process includes the mechanical and thermal responses of and within the workpiece as the dies are closed by a press or hammer. The material behavior is dominated by the flow stress. Flow-stress models are defined by temperature, strain and strain rate in the form of an equation or table data. Modeling the actual microstructural development in thermo-mechanical processes is extremely complex.
Transformation models based on Time-Temperature-Transformation (TTT) or Continuous-Cooling-Transformation (CCT) data can predict the phase changes in alloy steels, which determine hardness and other mechanical properties. Since the phases have different volumes, this induces mechanical deformation in the form of heat-treat distortion.
A set of realistic but simplified models to predict grain size has been applied successfully to a number of materials over the years. These JMAK models account for grain growth at high temperatures and recrystallization during and after forging operations. This is important since fine-grain forgings generally have higher yield and ultimate tensile strengths in nickel alloys. The models can be used to study the trade-offs between forging temperature, billet size and various preform geometries. The objective is generally to achieve a fine-grain microstructure.
An illustration is shown in Figure 5, where a DOE was conducted on a forged turbine disk. The 3D surface response shows grain size (vertical axis) as a function of furnace temperature and a range of blocker geometries. The response shows that the preheat temperature is dominant, with temperatures above 1825°F (1000°C) resulting in increasingly large grains.
Meanwhile, development continues on very accurate models that depict precipitates, grain morphology, dislocations and texture. Texture leads to non-isotropic mechanical properties during forging and in service. Sophisticated material models are being developed and used by researchers in the “deep end of the pool.” One drawback is the data requirements, which would be daunting to production engineers and metallurgists.
Most forge shops have a few employees with decades of experience that can anticipate how a new part will go through the manufacturing process. Much of this comes from direct involvement and years of operational experience in producing challenging parts or processes. The “problem jobs” are generally remembered with better clarity than a routine part. This experience is essentially a form of analog regression analysis.
As companies moved data from cards and paper to integrated databases, its usefulness increased. Production, design and quality records can be merged with engineering data and test records. This makes it possible to study production results and trends that were previously limited to the intuition of very experienced staff. To illustrate the point, mechanical-test results are shown in the form of a scatter plot for parts produced from a specific material as a function of grain size (Figure 6, left image). Trends will be interpreted differently by each reviewer. Color coding the data by material supplier may or may not show a clear trend (Figure 6, middle image). By adding a regression analysis for each supplier, the trend is clarified (Figure 6, right image). Data analytics will continue to mature in the forging industry as it has in pharmaceuticals, aviation, automotive and finance. The forging applications are countless.
The forging industry is more technically advanced than most people outside of the industry realize. In fact, many within the industry fail to appreciate the magnitude of technical advancements. While a walk through a typical production facility might not look like a NASA control room, the two share many common features, some of which were discussed in this article. Future advancements are even more exciting.
Preparing this series of articles is clearly a team effort between SCRA Applied R&D, Scientific Forming Technologies Corporation (SFTC) and FORGE magazine. Jon Tirpak, the executive director of the Forging Industry Association – Department of Defense Manufacturing Consortium, and John Walters of SFTC appreciate the support received from the Defense Logistics Agency (DLA). “Forging is Advanced Manufacturing” was sponsored in part by the Defense Logistics Agency’s Procurement Readiness Optimization – Forging Advanced Systems and Technologies Program (PRO-FAST).
Co-author John Walters is vice president of Scientific Forming Technologies Corporation, Columbus, Ohio. He may be reached at 614-451-8330 or firstname.lastname@example.org. Co-author Jon D. Tirpak is the executive director of FDMC and FAST program manager. He is also president of ASM International. He may be reached at 843-760-4346, or email@example.com.