Robust springback compensation success at Shanghai Volkswagen


Springback compensation is one of the key challenges in the field of die face engineering. We have mentioned the challenges with compensating springback several times within this blog (1 link, another link, and one more link) . In those earlier posts, we mentioned that the mode of the springback along with variation in production inputs often confounds attempts to compensate springback.

Yet, Shanghai Volkswagen was able to realize springback compensation for the decklid inner of their new “Rapid“ model, which resulted in a 65% reduction of tryout effort. Production start-up was also remakably smooth for the challenging inner panel; its first 8 months of production saw an effective reject rate of 0.0% from formability or springback issues.

Shanghai Volkswagen launched the new Rapid decklid with 0.0% forming issue reject rate

In the past, springback compensation was done manually, by following these steps:

  1. Build the tool prototype or production tools
  2. Produce prototype/tryout panels
  3. Take extensive measurements
  4. Modify tool geometry to compensate for springback
  5. Produce tryout/prototype panels
  6. Take extensive measurements
  7. Repeat as needed until acceptable parts are produced

This procedure was highly time-consuming and costly.

More recently, computer simulation has been used instead of prototype tools to recognize springback potential prior to the manufacture of the production “hard tool.” This can reduce some of the cost and unpredictability of springback compensation; but, during engineering, it is still not always possible to know how the process will behave in the production environment. Will the springback observed during engineering be the same once the realities of a physically imperfect press, varying ranges of material properties, and other production noise are introduced?

To address this challenge, Shanghai Volkswagen Automotive, a joint venture between Volkswagen Group and SAIC, decided to apply AutoFormplus software and methodology for “robust springback compensation.“ This new, simulation-based method helped to significantly reduce the number of physical compensation loops necessary as well as the tryout costs.

forming results through 5 stamping operations—OP20 through OP60—and the predicted springback after OP60

To obtain robust and reliable springback compensation, Shanghai Volkswagen Automotive implemented the following process:

  1. Process Engineering using AutoForm-ProcessExplorerplus
    1. process planning
    2. die face development using AutoForm-DieDesignerplus
    3. complete process simulation using AutoForm-Solverplus
    4. iterate until a safe nominal results
    5. springback measurement
  2. Robustness analysis using AutoForm-Sigmaplus
    1. input parameters for production noise (material, lube, pressure fluctuations, etc.)
    2. address any robustness issues through process improvement (beads, pad pressure, etc.)

AutoForm-Sigma predicts poor OP20 repeatability, splitting. Issue was identified and corrected prior to tool build

During the robustness analysis, noise variables were considered: variation in lubrication, material behavior, blank holder force, and blank position. It was found that there was a clear risk of cracking during the drawing operation, due to natural production fluctuation in these variables. The critical areas are marked in orange. These areas appeared to be safe in initial nominal simulations, but once production noise is considered, the instability of the safe result is visible. This result indicates that the process was not stable; significant thinning with risk of cracks would likely appear during tryout or production, requiring production downtime to make changes to the forming process or tool. Late recognition of these types of forming issues is extremely costly, as they delay production launch and cause unplanned downtime.

Robustness of splitting failure after Draw OP20 is unacceptable with original design, part will fail 44.4 per 1000 parts (cpk < 0.33)

Furthermore, robustness analysis showed significant variation in the amount of springback (areas A, B, C, and D below). Springback in the decklid inner panel was also unstable — meaning that if springback compensation had already been carried out, the part would often fail to match design intent, and the effort would have been wasted.

Areas with intermittent splitting likely when considering production noise

It was necessary to improve the process before carrying out springback compensation. The aim was to expand the process window and reduce the sensitivity of the process to noise variables, which are unavoidable and cannot be influenced during volume production.

Various changes were made:

  • die face addendum changes
  • modification of product geometry
  • panel stretch optimized
    • process parameter blank holder force
    • process parameter drawbeads

The formability result showed a safe result; the risks of cracks had been eliminated. Furthermore, the springback was now very stable and remained within acceptable limits regardless of the influence of noise variables.

After compensation is applied the decklid inner has reliable part shape, even with significant input noise variation

With the process stabilized, springback compensation can begin with the definition of the optimal compensation strategy.

To define the optimal compensation strategy, the engineer identifies where the springback originates and then minimizes the springback deviation. Springback, of the decklid inner, results directly from the deformation during the draw operation (OP20) and the flange and restrike operation (OP60). The other operations (OP30, OP40 and OP50) were trim and pierce only; therefore, the tool geometry was altered solely for OP20 and OP60.

Compensation was applied to both the Draw OP20 and forming tool OP60

After the preliminary work described above and only two iteration loops of compensation, the final geometric deviation between the sprung decklid inner part and the desired shape was well controlled within ±0.3mm in the simulation.

Number or needed tryout blanks and reject rate at start of production were significantly improved

Compared to the decklid inner of the previous model (New Lavida), it was found that validation of the Rapid used only one-third the number of tryout blanks .  This reduction in tryout blanks points to a significant reduction in material ordering and use; as well as press time, die maker effort, and quality inspection. During production, Shanghai Volkswagen has seen production reject rates fall from an already low 0.3% for the previous model to nearly 0.0 in the first 8 months of production.

This result confirmed that the compensation strategy was reliable.  Shanghai Volkswagen was so satisfied and impressed by the savings and quality improvements achieved that they plan to use robust springback compensation methods as a standard.



Please enter your comment!
Please enter your name here