Computer modeling of forging processes can help predict and determine parameters like starting billet size or die wear patterns.

Two actual case studies are used to illustrate the utility and effectiveness of the finite element method as a manufacturing tool.

FIGURE 1. The hot-forged yoke platter prior to trimming the flash.


Forging companies are always looking to gain a competitive edge. Raw material costs are up, schedules are tight and competition is fierce. New ways to control costs are always being sought. The optimization of forging processes is an excellent place to start, and process simulation software is available to help. Process modeling simulates the forging process on the computer. Based on the finite element method, this type of analysis is very good at modeling the complex metal flow that occurs during forging. Before dies are even cut, users of process simulation routinely answer the following questions:
  • Do the parts fill out correctly?
  • Are flow defects going to occur?
  • Are the dies likely to crack?
  • Is die wear going to be an issue?
When forging processes are developed on the computer instead of on the shop floor several things happen – unnecessary design trials are avoided, tooling costs are reduced and production equipment can be used to make production parts instead of being tied up in testing. In addition, simulation allows engineers and designers to “look inside” the process to gain information unavailable in a shop-floor trial. The velocity of the flowing material, the amount of work hardening and the development of contact with the tools are important factors that are difficult to obtain in a physical trial. All of these, however, can help a designer or engineer optimize their forging processes. This article details several examples of how forging companies have used DEFORM-3D – a process simulation package – to improve their forging practices.

FIGURE 2. The large-diameter billet produced excessive flash (highlighted in red).

DETERMINING OPTIMAL STARTING STOCK SIZE

American Axle & Manufacturing (AAM) in Detroit, Mich., is a Tier One manufacturer of driveline components. One part they produce is a weld yoke (Figure 1), which is formed in a three-piece platter forging. The platter is produced in one hit on a mechanical press followed by a flash trimming operation.

The typical time from receipt of an order to the first shipment was eight weeks. The lead times for both the dies and the steel were relatively long. Shop-floor trials to determine the optimal billet size were not an option since there was not enough time to order the steel after the dies arrived. The steel for the parts needed to be on hand prior to the first production trials. The annual production of this part was 200,000 pieces, requiring 1.25 million pounds of steel. The sourcing of material for this part was, therefore, not a trivial matter.

The goal for American Axle was to determine the optimal starting stock size for this part. Several criteria were used to determine which billet geometry was the best option:
  • The die cavity should be completely filled. Parts with underfills would not meet the finished part specification.
  • Flash should be produced around the entire flash land. Inadequate fl ash makes the parts hard to trim and also indicates a non-robust process.
  • Excessive flash should be avoided. Temperature, forming pressure and sliding velocity all go up with increased flash, thereby increasing the amount of tool wear.
  • Defects should be avoided. Laps, shear bands or other flaws would result in scrap.


Oversized Billet

A large-diameter billet was initially simulated (Figure 2). This billet filled out the die cavity nicely, but an excessive amount of flash was observed. In addition to the high tool wear, the cost of steel made this volume of wasted material unacceptable.

FIGURE 3. The small-diameter billet does not completely fill the die cavity.

Undersized Billet

To reduce the amount of flash, a smaller-diameter billet was simulated (Figure 3). Unfilled regions on the center yoke were observed. When die contact is viewed as a shaded contact patch, the unfill is clearly seen. In addition, there was not enough fl ash around the parts for a clean trim. Using information gained from these simulations, an optimal diameter was determined that had an acceptable amount of flash around the parts.

FIGURE 4. The short billet does not complete fill the last yoke.

Short Billet

Next, the proper cut length of the billet had to be decided. For these yokes, extra material had to be allowed at the front of the die so that the operator could hold the platter with his tongs. In one simulation, the billet was too short and not enough material was available at the back of the platter (Figure 4). This left an unfill in the third yoke.



FIGURE 5. Billet positioning was found to contribute to this fold defect.

Billet Positioning

Even after the billet geometry is optimized, there is still a chance that defective parts can be produced. One process variable is the position of the billet in the die. Changes in the billet location have an impact on material fl ow and therefore on final part quality. A simulation was run to see what would happen if the billet was placed closer to the operator than usual. In this scenario, a fold defect was observed to develop in the last yoke. The close-up image shows how DEFORM’s automatic fold-detection system draws attention to the folding area (Figure 5).



FIGURE 6. The nominal billet filled out the yokes nicely and didn't have excessive flash.

Nominal Billet

When the diameter and the length of the billet had been optimized and the positioning in the die was correct, a simulation of the forming process showed that the parts came out looking great. The die cavity filled without defects. The flash was not excessive but was ade-quate for trimming and process robustness (Figure 6). This case study is an excellent example of how designers and engineers use simulation to help with their day-to-day responsibilities.

MINIMIZING TOOL WEAR

FormTech Industries, formerly part of Metaldyne, is also a Tier One supplier of automotive components. FormTech makes a precision spindle forging, and they wanted to minimize the amount of tool wear they were experiencing. In general, there are four main modes of die failure. Dies can fail due to catastrophic failure, plastic deformation, low cycle fatigue or wear. The simulation of tool wear is an advanced application of process simulation, mainly because tool wear is not well understood. There is currently research underway to study the physics and deformation mechanisms involved in the wear process. One of the most commonly used wear models applicable to forging dies is the Archard model:

In this model, tool wear (W) is a function of the interfacial pressure (p), the sliding velocity (v), the hardness of the tool material (H) and some experimentally calibrated coefficients (a,b,c,K). Once these coefficients are calibrated, the amount of tool wear can be accu-rately calculated. This is often reported as the amount of wear per part or as the total number of parts that can be produced prior to a certain wear depth occurring.

Another option is to use generic coefficients in the wear model. In DEFORM, the default values are typical ones for tool steel dies.

Using these values, users can compare different preforms or tool designs to see what impact the changes have on wear.

FIGURE 7. Spindle formed using the production process

Standard Preform

The production forging process for the spindle (Figure 7) used three forming stations. The first station flattened the preform, the second did the majority of the forming and the final station finished forming the features. Since most of the deformation was occurring in the second station, the punch from that station was exhibiting high wear.

The wear shown on the real punch and on the simulation punch was compared. The wear pattern seen in production was accurately predicted in the simulation. The wear on the protrusion was intuitive, but the half-moon-shaped wear might not have been anticipated (Figure 8).

The production runs were made using what FormTech called a “standard” preform. From a press productivity standpoint, the proc-ess using this preform was not optimal. Due to the large amount of load needed to form the part in station 2, parts could not be formed in stations 1 and 3 at the same time.

FIGURE 8. Wear comparison between real and simulated tools

Cone Preform

In an attempt to decrease the amount of tool wear, a “cone” preform was investigated. This design required more deformation in station 1 and therefore less deformation and load in station 2. Only two stations were required for this design, so parts could be formed in both at the same time. If this design was successful, the productivity of the press would be doubled.

Unfortunately, the simulation showed that the wear in the fi nal station was even worse than that seen in the production process using the standard preform. Even though the productivity benefits were enticing, this cone preform could not be utilized due to the in-creased tool wear that would occur.

FIGURE 9. A fold developed when using a round preform.

Round Preform

Next, a “round” preform was simulated. A process using this preform would have the same press productivity as the original process, but it was hoped that the wear to the tooling would be reduced. Tool wear became a non-issue when it was found that a fold developed in the fi nal station (Figure 9).

FIGURE 10. Tool wear comparison using standard, cone, and round preforms.

Modified Round Preform

Until the fold developed, the tool wear in the round preform process looked promising. Therefore, a modified round preform was developed to try to get rid of the folding. The simulation showed that the modified preform did indeed produce a good spindle.

When the tools from the three different preform simulations are compared, it is easy to see that using the round preform significantly reduced the amount of wear in the finished tools (Figure 10). FormTech noted that the optimal process would actually involve a preform somewhere between the cone and round shapes. With such a geometry, the productivity benefi ts of the cone preform and the wear benefits of the round preform could be realized in one process. Even though the mechanisms of wear are not well understood, this ex-ample shows that process simulation can still be used to solve challenging tool-wear issues.